92230 v1 KAZAKHSTAN Agricultural Insurance Feasibility Study Agriculture and Rural Development Unit Sustainable Development Department Europe and Central Asia Region June 2012 2012 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 www.worldbank.org Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The findings, interpretations, and conclusions expressed in this paper do not neces- sarily reflect the views of the Executive Directors of The World Bank or the governments they repre- sent. The World Bank does not guarantee the accuracy of the crop production and yield data and crop insurance financial data included in this work and on which basis some analyses have been made and from which some conclusions have been drawn and recommendations made. The World Bank cannot be held responsible for any insurance result or other financial outcome which might arise from any decisions or actions taken by insurers or any other party as a consequence of the contents of this report. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Cover photos: The World Bank Photo Library. Cover and Layout design: Duina Reyes Kazakhstan Agricultural Insurance Feasibility Study Agriculture and Rural Development Unit Sustainable Development Department Europe and Central Asia Region June 2012 Table of Contents ........................................................................................................... 5 Acknowledgments. Abbreviations.................................................................................................................... 6 ......................................................................................................... 8 Executive Summary. ............................................................................................ 8 Context and Scope of the Study. Crop Risk Assessment.......................................................................................................... 10 ................................................................. 11 Review of Compulsory Crop Insurance Program. Strategy and Options for Strengthening the Current Crop Insurance Scheme.................... 15 ................................................................. 28 Opportunities for New Crop Insurance Products. Tailoring Crop Insurance to the Needs of Small Farmers in South Kazakhstan................... 33 Agricultural Insurance Feasibility Study: Summary of Recommendations for Improvement of the Obligatory Crop Insurance Scheme in Kazakhstan.................. 36 ................................................. 41 Chapter 1: Introduction and Objectives of the Study. Importance of Agriculture in Kazakhstan.............................................................................. 42 Agricultural Crop Production in Kazakhstan......................................................................... 42 Government Policy for Agriculture........................................................................................ 44 Exposure of Agriculture to Natural and Climatic Disasters................................................... 44 Government Objectives for Crop Insurance......................................................................... 45 Objectives and Scope of the Study...................................................................................... 46 Outline of the Report . .......................................................................................................... 48 Chapter 2: Crop and Weather Risk Assessment......................................................... 49 Objectives and Scope of Agricultural Crop and Weather Risk Assessment......................... 49 .................................................... 49 Data Availability for Crop and Weather Risk Assessment. Climate and Agro-Ecological Regions ................................................................................. 52 Overview of Spring Wheat Crop Production in Kazakhstan................................................. 54 .......................................... 61 Key Climatic Perils and Impact on Crop Production and Yields. Assessment of Crop Production Risk Exposures................................................................. 65 Chapter 3: Review of Kazakhstan Crop Insurance Program..................................... 70 Policy and Regulatory Framework for Crop Insurance......................................................... 70 Compulsory Crop Insurance Policy Terms and Conditions.................................................. 71 ....................................... 78 Government Financial Support to Crop Insurance in Kazakhstan. Performance Assessment: Technical Results, Liabilities, Reinsurance ............................... 80 Assessment of the Technical, Operational, and Institutional Features of the Compulsory Crop Insurance Program.................................................................................. 92 Evaluation of Crop Insurance Effectiveness for Key Stakeholders....................................... 98 Agricultural Insurance Feasibility Study  3 Chapter 4: Strategy and Options for Strengthening the Current Crop Insurance Program. ...................................................................................................... 101 Phase 1: Returning the Obligatory Crop Insurance Scheme to Profitability and Financial Stability................................................................................................................ 102 Phase 2: Transition toward a Market-Based Crop Insurance System................................ 120 Phase 3: Transformation into a Fully Commercial Crop Insurance Scheme . .................... 130 Chapter 5: Opportunities for New Crop Insurance Products................................... 145 .............................................................................................. 145 Named-Peril Crop Insurance. Area-Yield Index Crop Insurance........................................................................................ 151 Crop Weather Index Insurance .......................................................................................... 167 Chapter 6: Tailoring Crop Insurance to the Needs of Lower-Income Smaller Farmers........................................................................................................... 184 Identification of Appropriate Crop Insurance Products...................................................... 184 Farmer Segmentation and Crop Insurance......................................................................... 186 Tailoring Crop Insurance for Different Client Levels............................................................ 187 Organizational and Operational Systems for Small Farmer Crop Insurance...................... 192 Identification of Operational Linkages to Bundle Programs............................................... 199 Chapter 7: Fiscal Implication of Various Insurance Products on the GKR Budget.................................................................................................................. 201 Bibliography.................................................................................................................. 209 4  Kazakhstan Acknowledgments T he report was prepared by the World Hydro Meteorological Service of Kazakhstan Bank in partnership with the Second (Kazhydromet, KHM); the National Agency of Agricultural Post Privatization Assis- Statistics; the Agency for Financial Market and tance Project (APPAP II) and was authored by a Financial Institutions Regulation and Control; team led by Sandra Broka (Senior Rural Finance the National Space Agency; JSC; the Union of Specialist, ECSS1, World Bank) and Meiram Farmers of Kazakhstan; the Scientific Research Akchukakov (Program Coordinator, APPAP II). Institute of Economy of Agro-Industrial Com- The team was composed of Ramiro Iturrioz (Se- plex and Development of Rural Territories; the nior Agricultural Insurance Specialist, Insurance A. Barayev Kazakh Scientific and Research for the Poor Program, GCMNB, World Bank— Institute of Grain Farming; Akimat of Enbek- Technical Leader); Talimjan Urazov (Operations shilder rayon; Akimat of Bulandinskky rayon; Officer, ECSS1, World Bank); Charles Stutley Akimat of Altynsarin rayon; Kazakh Instrakh (Consultant, ARMT, ARD, World Bank); An- Halyk Group Insurance Company; Pana Insur- drea Stoppa (Consultant, ARMT, ARD, World ance Company; Grain Insurance Company; Bank); Bakhyt Sattybaeva (Consultant, APPAP “Agro-Insurance” Mutual Insurance Society; II); Lunara Umralinova (Consultant, APPAP “SFK-Insurance” Mutual Insurance Society; II); Marina Gabdulinova (Consultant, APPAP and “Dostyk 05” Company Ltd. II); Aigerim Malik (Consultant, APPAP II); and Arka Consulting, the local consultant firm se- The authors are grateful to the peer reviewers lected by APPAP II for this project. John Nash (Lead Economist, LCSSD, World Bank), Olivier Mahul (Program Coordinator, The team acknowledges the contributions of FCMNB, World Bank), and Gary Reusche (Op- all stakeholders, including the Ministry of Ag- erations Officer, IFC). riculture (MoA) and, in particular, the Depart- ment of Investment Policy and External Rela- The team gratefully acknowledges funding sup- tions and the Department of Strategic Planning port from the Commodity Risk Management in Agribusiness and Innovation Policy; the JSC Multi-donor Trust Fund. Fund for Financial Support for Agriculture, Agricultural Insurance Feasibility Study  5 Abbreviations A&O Administrative and Operating Agroseguro Agrupación Española de Entidades Aseguradoras de los Seguros Agrarios Combinados (Spanish Group of Insurance Entities of the Combined Agrarian Insurance) AIC Agricultural Insurance Company of India APPAP II Second Agricultural Post Privatization Assistance Project ARD Agriculture and Rural Development Department, World Bank ARKS National Agency of Statistics ARMT Agriculture Risk Management Team AYII Area-Yield Index Insurance Centner Grain production unit equivalent to 100 kilograms CJSC Kazakh Actuarial Center CoV Coefficient of Variation CRAM Crop Risk Assessment Model EEL Each and Every Loss EKO East Kazakhstan Oblast FAPRAC Fund for the Care of Rural Population Affected by Weather Contingencies (Mexico) FFSA Fund for Financial Support for Agriculture GDP gross domestic product GMFP Grameen Fisheries and Livestock Foundation (Bangladesh) GNPI Gross Net Premium Income GRK Government of the Republic of Kazakhstan GRP Group Risk Plan (United States) HTR Hydrothermal Ratio 6  Kazakhstan IU Insured Unit K Humidity factor KHM Hydro Meteorological Service of Kazakhstan (alternatively known as Kazhydromet) KZT Kazakhstan tenge (monetary currency of Kazakhstan) LIC Loss of Investment Costs (crop insurance policy) LIF Livestock Insurance Fund (Bangladesh) MFI Microfinance Institution MoA Ministry of Agriculture MPCI Multiple-Peril Crop Insurance MT Metric tonne, equivalent to 1,000 kilograms MU Managing Underwriter MUC Managing Underwriting Company NAIS National Agricultural Insurance Scheme (India) NCSRT National Center of Space Research and Technologies NDVI Normalized Difference Vegetation Index NGO Nongovernmental Organization NKO North Kazakhstan Oblast NSA National Space Agency Oblast Administrative Region of Kazakhstan PML Probable Maximum Loss PPP Private-Public Partnership Rayon Administrative area within an oblast SIF Self-Insured Fund (Mexico) SKO South Kazakhstan Oblast TSI Total Sum Insured VaR Value at Risk WII Weather Index Insurance WKO West Kazakhstan Oblast WTO World Trade Organization Agricultural Insurance Feasibility Study  7 Executive Summary is lost due to drought and other perils every year, Context and Scope of the Study valued at KZT 66.5 billion (US$443 million).1 In 1998, an extreme drought year, physical loss- 1. Agriculture is a very important socioeco- es were on the order of 7 million metric tons, nomic sector in Kazakhstan. Approximately 7.3 equivalent to as high as 42 percent of the total million people (47.2 percent of the total popula- expected value of wheat production. Very severe tion) currently live in rural areas, and agricul- drought losses were experienced in spring wheat ture employs more than 22 percent of the labor most recently in 2008 and again in 2010. force in the country. Agriculture contributes 5.92 percent of Kazakhstan’s gross domestic 3. The Government of the Republic of Ka- product (GDP). The country is one of the major zakhstan (GRK) introduced a national compul- global producers and exporters of grains (main- sory crop insurance scheme in 2005 in order ly wheat). Other principal agricultural products to provide grain producers and other farmers include meat, wool, cotton, and milk. Farming with a minimum level of protection against areas occupy more than 220 million hectares catastrophic climatic events. The scheme was (about 74 percent of the country’s total area), enacted through the Law on Compulsory Crop of which cereal-growing areas occupy about 13 Insurance, which is dated March 10, 2004 and million to 14 million hectares. The major grain became operational in 2005. The law established crop is spring wheat, which is grown predomi- the terms and conditions for implementation of a nantly under extensive low-cost production sys- compulsory loss of investment costs (LIC) insur- tems in northern and central Kazakhstan. ance policy providing comprehensive protection against the loss of production costs invested in 2. In northern and central Kazakhstan, growing a range of strategic grain, oilseed, and spring wheat production and yields are highly other field crops. The Kazakhstan crop insurance influenced by climatic and biological factors. scheme is based on a public-private partnership On account of the very uncertain climatic condi- (PPP) implemented by the private commercial tions, Kazakhstan has the highest year-on-year and mutual insurance companies and supported variation in national average wheat production by government financial subsidies on claims. The and yields of any major wheat-producing and implementing agencies are the Ministry of Ag- -exporting country. Drought is the most perva- riculture (MoA), through the Direction of Stra- sive peril, affecting rain-fed crop production in tegic Planning, the Fund for Financial Support northern Kazakhstan, and severe droughts are ex- for Agriculture (FFSA), the private commercial perienced every two to five years. Spring wheat crops can also be damaged by hailstorms, early autumn frost, pests, and diseases. This study es- 1  Kazakhstan’s currency is the tenge. This report timates that, on average, about 14.71 percent of uses a current 2011 exchange rate of KZT 150 = the total value of the national spring wheat crop US$1.00, unless otherwise stated. 8  Kazakhstan insurance companies, the farmer mutual crop in- many commercial insurers have ceased to sup- surance associations, and the local authorities in port the scheme. From an operational viewpoint, each oblast and rayon. Under the PPP, the GRK the scheme is very costly to administer, and it is provides financial contributions to the crop in- underrated in several regions. As the terms and surance scheme through an indemnity fund—the conditions of cover are determined by govern- FFSA—and indemnifies 50 percent of the costs ment and fixed by law, the insurance companies of all crop insurance claims. have little say in risk acceptance and underwrit- ing decisions, and today only three companies 4. The GRK’s decision to introduce a na- continue to support the scheme. Finally, private tional compulsory crop insurance scheme for commercial insurers and farmer mutual insurers the most important crops grown in the country are very exposed to catastrophic losses, as the should be viewed in the context of government’s scheme is not currently reinsured against excess policy toward agriculture. The GRK recognizes losses. the importance of the agriculture sector in diver- sifying economic growth, reducing rural pov- 6. During 2011, the World Bank, under erty, and improving food security, with a strong the risk management component of the Agri- emphasis on maintaining rural welfare for the cultural Post Privatization Assistance Project country’s predominantly small farmers, as set (APPAP II), performed a comprehensive study out in the strategic Three-Year Plan for Agri- to review, refine, and improve the compulsory culture, 2009–11. Government’s introduction crop insurance scheme in Kazakhstan. The of compulsory insurance was designed mainly study reviewed the Kazakhstan compulsory crop to guarantee that small farmers and agricultural insurance scheme and made recommendations wage laborers would have a minimum level of for its strengthening and transition over time financial protection in the event of catastrophic to a more market-based agricultural insurance crop production losses. Since the late 1990s, the system. The study also assessed the potential to government has significantly increased its finan- introduce new crop insurance products and pro- cial support to the crop sector, providing direct grams to complement the existing LIC policy. input subsidies, subsidized credit, and output The study included the following specific com- price support. Under its goal of meeting World ponents: Trade Organization (WTO) accession terms, Ka- zakhstan will need to reduce its direct subsidies a. Review of the compulsory crop insur- to agriculture; however, WTO legislation does ance scheme in Kazakhstan. A detailed not prohibit government from increasing its fi- diagnostic review was carried out of the nancial subsidies to agricultural crop and live- technical basis of the LIC policy as well stock insurance. as the institutional and organizational features, of the public private partner- 5. Overall, the crop insurance scheme has ship scheme, and its operating systems, not performed well. Over the past six years, the procedures, and financial performance of scheme achieved very high levels of uptake, but the PPP scheme. This review identified also encountered major operational and financial a series of key issues and challenges for problems. Crop insurance penetration is very scheme management to address. high in Kazakhstan, averaging 74 percent of the eligible cropped area for the period 2005–10. b. Identification of a phased strategy to The high uptake is a function of the compulsory transform this scheme into a finan- nature of the scheme. Notwithstanding the high cially sustainable market-oriented sys- level of penetration, insurance results were very tem. Drawing on local expertise and in- poor over the past three years, with the result that ternational experience and best practice, Agricultural Insurance Feasibility Study  9 a phased strategy was identified for the study was to identify ways to tailor the next three to five years to strengthen provision of crop insurance to the needs the scheme, return it to profitability, and of resource-poor farmers located in south- transform it into a sustainable market- ern Kazakhstan. Tole-bi rayon in South based system that is supported both by Kazakhstan (SKO) was selected for more the public sector and by international re- in-depth study. insurers. f. Brief Assessment of the Fiscal Implica- c. Assessment of agricultural risk. A for- tion of the various crop insurance prod- mal crop risk assessment was performed ucts on the GKR budget. for spring wheat, the country’s most im- portant export crop. This risk assessment Crop Risk Assessment was intended to assist policy makers, planners, and crop insurers in the plan- 7. A detailed risk assessment was conduct- ning, design, and rating of new crop in- ed of weather risks and their impact on spring surance products. Owing to the size of the wheat crop production and yields in northern country, it was agreed that the scope of Kazakhstan. The risk assessment comprised the risk assessment would be limited to the following components: (a) a review of the spring wheat grown in the eight most im- availability and quality of time-series crop pro- portant oblasts of Kazakhstan: Kostanay, duction data and the availability of weather data Akmola, North Kazakhstan (NKO), East in Kazakhstan for spring wheat risk assessment Kazakhstan (EKO), West Kazakhstan and insurance design and rating purposes; (b) a (WKO), Pavlodar, Karaganda, and Ak- review of climatic and agro-ecological regions tobe. and spring wheat crop production systems in the selected oblasts of Kazakhstan; (c) a detailed d. Analysis of crop insurance products. statistical analysis of spring wheat production Four new types of crop insurance prod- and yields and the climatic constraints to pro- ucts were analyzed that, in the future, duction, including an analysis of rainfall data could either complement or replace the and the relationship to national and rayon-level current LIC policy: (1) individual grower spring wheat crop production and yields; and fi- multiple-peril crop insurance (MPCI), nally (d) application of a crop risk assessment (2) crop hail cover, (3) area-yield index model (CRAM) that uses time-series rayon-lev- insurance (AYII), and (4) weather index el production and yield data to estimate values at insurance (WII). For the purpose of the risk, expected losses, and expected claims costs feasibility studies, the research on area- for spring wheat in the eight selected oblasts in yield index insurance was performed for Kazakhstan. This latter analysis is very relevant spring wheat grown in the eight selected to crop insurers’ understanding of risk accumu- oblasts of Kazakhstan. For crop hail and lation and maximum expected losses in spring weather index insurance products, the re- wheat. Full results of the risk assessment are search focused on spring wheat grown in presented in chapter 2. Altynsarinski and Auliyekolski rayons in Kostanay; Aktogayskiy and Zhelezinski 8. The crop risk assessment of rayon-level rayons in Pavlodar; and Bulandinski and crop production and yields for spring wheat in Enbekshilderski rayons in Akmola. the northern and central regions of Kazakh- stan found that this crop is heavily exposed to e. Identification of challenges for develop- losses caused by droughts. This is evidenced by ing crop insurance for small farmers in the average loss cost estimated by the CRAM Kazakhstan. The final objective of this for the 17-year period from 1994 up to 2010 of 10  Kazakhstan KZT 66.5 billion (14.71 percent of the total val- MPCI and area-yield index crop insurance prod- ue at risk of spring wheat production) and a cal- ucts for Kazakhstan (as detailed in chapters 4 culated 1-in-100-year probable maximum loss and 5). (PML) of KZT 246.8 billion (54 percent of the gross value of production for spring wheat). The Review of Compulsory Crop highest average annual expected losses in spring wheat apply to Aktobe (22 percent of the spring Insurance Program wheat crop value) and to WKO (40 percent of the spring wheat crop value) located in western 10. Chapter 3 of this report provides a de- Kazakhstan. Except in a few rayons situated tailed review of the technical, operational, in- in the north of Aktobe and in the southwest of stitutional, and financial features of and chal- WKO, spring wheat average yields are both lenges faced by the compulsory crop insurance low and highly variable in most of the rayons scheme. located in these two western oblasts, which re- ceive much lower and erratic rainfall. Converse- Technical Challenges ly, spring wheat production is much less risky in the northern oblasts of NKO, Kostanay, and 11. The Kazakhstan compulsory crop insur- Akmola, which receive higher and more stable ance policy is a loss of yield policy that indemni- precipitation. fies the insured when the value of the harvested production falls short of the costs invested in growing the crop due to the action of insured 9. The need for accurate and independent perils. The LIC policy has several advantages: measurement and recording of crop produc- (a) it provides comprehensive MPCI protection tion, crop yield, and weather data at local up to the farmer against the loss of production costs to national levels is critical to the design, rat- invested in growing the crop, and (b) it can be ing, and implementation of any crop insurance used in situations where there is inadequate or scheme. On the basis of this study, it is apparent no information on the historical crop yields of that the GRK has very efficient meteorological, individual farmers. However, it also has several agricultural, and statistical services. The avail- potential drawbacks, including the need for in- ability of production and weather data is, in gen- field yield-based loss assessment where partial eral, very good. Kazakhstan has a modern and losses are involved and the difficulty of estab- efficient national meteorological service, Hydro lishing objectively the salvageable amount of Meteorological Service (alternatively known as the crop, its sale value, and whether this salvage Kazhydromet or KHM), which provided time- value exceeds the insured investment costs lead- series rainfall data for a sample of weather sta- ing to a claim. tions in the selected rayons. These data enabled a detailed assessment of rainfall and yield rela- 12. The coverage levels provided by the LIC tionships and provided the basis for the design crop insurance scheme in Kazakhstan are ex- and rating of prototype WII products. There is, tremely low and do not provide adequate lev- however, a key constraint at present: the density els of financial protection to the farmer in the of weather stations in such a vast territory as Ka- event of loss. Although the compulsory scheme zakhstan is inadequate to implement a national offers a series of optional sum insured levels for commercial micro-level crop WII scheme in the farmers to choose from, based on different crop near future. Kazakhstan also has good statistical technology levels and costs of production vary- records of spring wheat crop area, production, ing from low to high production costs, in prac- and yield at the rayon, oblast, and national lev- tice practically all farmers in Kazakhstan elect els, which are collected by the National Agency the cheapest or lowest option for sum insured of Statistics (ARKS). These data enabled the de- coverage—about KZT 3,500 (slightly less than sign and rating of prototype individual grower US$25) per hectare nationally for spring wheat Agricultural Insurance Feasibility Study  11 in 2010—because this option has the lowest pre- located in such high-risk areas that they would mium. On average, farmers only insure between normally be considered uninsurable by commer- 20 and 30 percent of their total production costs. cial crop insurers. The very low levels of crop insurance cover are, therefore, often inadequate to put farmers back 16. Policy sales are currently permitted to into production in the event of a major crop continue up to the time of sowing, when farm- loss. ers are in a good position to predict whether the growing season will be poor, and this exposes 13. The methodology for rating LIC insur- the program to moral hazard. Where preexist- ance premiums in Kazakhstan should be re- ing drought conditions are developing, farmers vised. Currently, crop insurance premiums are may modify their behavior by (a) electing to buy calculated for each crop at the oblast level, when the maximum level of “normative costs” sum they should to be calculated at the rayon level insured in the expectation of receiving a claims or even higher levels of disaggregation in order payment or (b) incurring less than their normal to take into account differences in risk either at level of expenditure on crop husbandry and in- the rayon level or at the individual farmer level. puts because they know they are likely to lose Crop insurance premium rates are fixed by law. their crops, in which case, they can expect to They were last adjusted in 2008 and now need claim on their insurance policies (this is termed to be adjusted on an actuarial basis. Crop insur- moral hazard). ance premium rates are set for each type of crop and group of oblasts according to minimum and 17. Private insurance companies do not maximum rates, and insurers are not permitted have their own network of locally based quali- to charge higher rates even when they are actu- arially required. fied agronomists to inspect the insured farms at the time of sowing. Such inspections are needed to confirm whether the farmer has complied with 14. The basis of indemnity on the LIC policy the correct sowing practices, seed rates, and so requires strengthening. Crop output valuation forth. As such, cover is open to moral hazard. prices are determined at the time of harvest, as opposed to being preagreed and specified The costs of establishing such a network and in the policy wording. This means that neither inspecting each and every farm would be pro- the insurers nor the government can calculate hibitively high for the insurers, which, under the their liability in the event of claims: when out- current rating system, are not able to increase put prices are low, the policy is much more ex- rates to cover their administrative and operating posed to losses than when prices are high. This (A&O) expenses. issue should be addressed by introducing fixed, preagreed harvest valuation prices at the time of 18. Loss adjustment requires the participa- signing the policy agreement. tion of several parties, is expensive, and some- times lacks transparency. In Kazakhstan, up Operational Challenges to five persons are involved in adjusting crop losses at the local level, and this is a very time- 15. The obligatory nature of the scheme pre- consuming and costly exercise. There is a need vents underwriters from exercising proper risk to rationalize the loss assessment procedure and selection and control over the underwriting of to reduce its costs. their crop insurance portfolios. Insurance com- panies are obliged to accept all crop insurance Institutional Challenges risks (that is, individual farmers) even when the farmers are poor risks (that is, they do not use the 19. The Kazakhstan compulsory crop insur- correct crop production or husbandry practices ance scheme is a public-private partnership. It for growing the insured crop) or the farms are is highly regulated by government and is under- 12  Kazakhstan written by the private commercial (and mutual) Government Support insurance sector, with financial claims subsidy support provided by government through the 22. The GRK provides major financial sup- FFSA. As such, the insurance companies are port to the compulsory crop insurance scheme. unable to accept or reject individual crop risks This support is provided in two ways: (a) by (farmers), which is a fundamental principle of compensating the insurers for 50 percent of all crop underwriting. In most countries where ag- the claims incurred and (b) by funding the A&O ricultural insurance PPPs exist, the main role of expenses of the FFSA. The 50 percent claims government is to provide legal and regulatory compensation fund is administered by the FFSA, support and financial subsidies. which is responsible for monitoring and manag- ing the financial transactions of this insurance 20. Crop insurance in Kazakhstan is writ- scheme on behalf of the government and for ap- ten by commercial insurers and farmer mutual proving the claims reimbursements to individual insurance associations. The level of participa- insurance companies. Over the past six years, tion of the private insurance companies is very the GRK provided KZT 4.7 billion to the FFSA, low (only three commercial insurers participate of which 93 percent was allocated to settling the in the scheme), and there is a danger that, un- 50 percent of claims and 9 percent to paying the less the scheme can be returned to profitability, A&O expenses of the FFSA. During this period, the private insurance companies may cease to the FFSA reimbursed the insurance companies a provide their support altogether. Conversely, the total of KZT 3.84 billion, equivalent to 46.7 per- participation of farmer mutual crop insurance cent of total claims paid or an average of KZT associations is gaining importance. As of 2011, 641 million (US$4.3 million) per year. more than 38 farmer mutual associations were offering crop insurance in Kazakhstan. Financial Challenges 21. Private insurance companies and farmer 23. The compulsory crop insurance pro- mutual associations providing crop insurance gram in Kazakhstan experienced poor overall are not equally regulated. While private insur- underwriting results over the period 2005 to ance companies are regulated by the Agency for 2010. The long-term average loss ratio for the Financial Market and Financial Regulation and six-year period was 140 percent, and in four of Control, the farmer mutual crop insurance as- the six years, the scheme operated at a financial sociations are regulated separately by the Law loss, with a loss ratio (gross claims to premium) on Mutual Insurance. Nonlife private insurance exceeding 100 percent. The average net loss companies are required to have minimum capi- ratio for the insurance companies after receiv- tal of KZT 1.2 billion (US$8.0 million) in order ing the 50 percent government claims subsidies to operate and are frequently monitored on their was 75 percent. Over the past three years (2008 solvency, net retentions, and implementation of to 2010), the results deteriorated badly, with an risk management procedures. Conversely, the average loss ratio of 182 percent (99 percent af- farmer mutual associations offering crop insur- ter government claims subsidies); the worst loss ance are not subject to minimum capital require- was experienced in 2010, with a loss ratio of 261 ments, controls over net retentions, or solvency percent. requirements. In 2010 several mutuals report- edly did not collect enough crop insurance pre- 24. Scheme performance varies widely across miums to pay the full amount of claims incurred different regions, and the very poor underwrit- during the 2010 crop season. There is a need to ing results in Aktobe and WKO are making ensure that the private companies and mutuals the scheme financially unviable. The pattern are regulated equally under the crop insurance of claims varies widely by geographic region. scheme. The best-performing oblast is NKO, which over Agricultural Insurance Feasibility Study  13 the past six years contributed 22 percent of total on their 50 percent retained claims, and they scheme liability but only 3 percent of claims and are extremely exposed to catastrophic drought had a long-term loss ratio of only 24 percent. At losses. This is a major issue that will need to be the other extreme, Aktobe and WKO in western addressed under any future reform of the Ka- Kazakhstan, which collectively accounted for zakhstan crop insurance scheme. only 4.8 percent of total scheme liability, in- curred 41 percent of all claims and had six-year Evaluation of Crop long-term loss ratios of 381 and 507 percent, re- Insurance Effectiveness spectively. These two oblasts are severely preju- for Key Stakeholders dicing the financial viability of the national crop insurance program, and measures for controlling 26. The Kazakhstan compulsory crop in- the claims costs in these two oblasts must be in- surance scheme was launched with very well- troduced. intentioned social and economic objectives, but it is failing to meet the requirements of 25. The private insurance companies and its key stakeholders, including farmers, crop mutuals do not have access to reinsurance insurers, and government. The program was Figure 1 Key Issues Facing Main Stakeholders Involved in the Compulsory Crop Insurance Scheme Farmers • Dissatisfaction with compulsory nature of crop insurance. • Reluctance to purchase insurance in spite of being compulsory. • Extremely low levels of coverage. • Difficulties in understanding the insurance agreement especially the basis of indemnity and loss assessment procedures. Government Insurers: • Very good social intentions, but compulsory nature of the • Limitations to perform risks selection/underwriting. program is very unpopular with most farmers. • No incentives to invest in underwriting/claims • Intention of obligatory insurance in early years to develop crop management infraestructure. insurance market has not been achieved. • Uncertainties about financial exposure. • Uncertainties about financial exposure. • Difficulties to gain access to reinsurance. • Uncertainties about budgeting claims participation in • Concern by private commercial insurers that they do not face a level catastrophic events. playing field with the Farmers’ Mutual Crop Insurance Associations. • Crop Insurance has been unprofitable to date. Source: Authors. 14  Kazakhstan originally conceived as a mechanism to ensure that the Law on Compulsory Crop Insurance that all farm workers and small peasant farm- be amended to permit insurance companies ers would receive a minimum indemnity in the greater flexibility in determining the premium event of crop failure due to drought or other nat- rates and other terms and conditions of cover ural and climatic perils. The obligatory nature they provide under the LIC policy. While it is of the scheme was intended to be a short-term assumed that crop insurance will continue to be measure that would enable the development of obligatory for farmers in the short term, the law a sound and stable crop insurance market based should be amended to permit insurers to set their on a partnership between the private and public own terms, conditions of cover, and, especially, sectors, while at the same time providing time to premium rates (see chapter 4). educate farmers in the benefits of crop insurance so that they would continue to purchase cover Measures to Strengthen the Design and once the scheme was made voluntary. After six Rating of the LIC Policy years, the scheme has failed to develop a strong crop insurance market and to educate farmers. 29. Practical measures were identified for The fact that the program is very unpopular with strengthening the design of the LIC policy. many farmers also suggests that it is failing to These measures are listed in box 1; full details meet its social objectives. Finally, government are provided in chapter 4, and key recommenda- faces major uncertainties over its financial expo- tions for strengthening are highlighted in box 1. sure to claims (figure 1). 30. There is a need to consider increasing Strategy and Options for the sums insured. Currently, the LIC policy does not provide adequate levels of protection Strengthening the Current Crop for the majority of farmers. If the levels of sum Insurance Scheme insured coverage were to increase under this program, farmers would receive higher levels of 27. A phased approach is recommended for protection, but this would have to be accompa- strengthening and improving the obligatory nied by increases in the premium rates that in- crop insurance scheme in Kazakhstan and for surers would have to charge and in the amount gradually converting it into a fully market- of premium farmers would have to pay. With in- based commercial crop insurance system. creased sums insured, the insurance companies The proposed approach involves three distinct would face a much higher financial liability in phases: (a) strengthening the current obligatory the event of severe drought losses, and it would scheme and achieving financial stability (short be essential for them to have comprehensive re- term, one to three years); (b) moving toward insurance protection. For government, the higher a market-based crop insurance system (short coverage levels would mean a correspondingly term, one to three years); and (c) becoming a higher budgetary allocation to cover the 50 per- fully market-based pool crop insurance system cent claims reimbursements. The sums insured backed up by international reinsurance under in each rayon should be set according to (a) a PPP suitable for Kazakhstan (medium term, the actual production costs of different types of three to five years). farmers in each rayon and (b) the risk exposure in each rayon. Phase 1: Short-Term Measures to Strengthen the Existing Compulsory 31. There is a need to revise the crop pre- Scheme mium rating methodology in order to (a) intro- duce a system of rayon-level premium rates and Measures to Strengthen the Legal and (b) update the premium rates on an actuarial Regulatory Framework basis. The six-year long-term gross loss ratio at end-2010 was 140 percent (equal to a 70 percent 28. In the short term, it is recommended loss ratio net of the government’s 50 percent Agricultural Insurance Feasibility Study  15 Box 1 Measures to Strengthen the Design of the Loss of Investment Costs Policy Criteria for acceptance of risk and compulsion of cover. Even if the government decides in the short term to maintain compulsory crop insurance for all producers, special consideration will need to be given to farmers located in Aktobe and WKO. Insured perils. Current coverage should be maintained to include loss or damage to crop production due to “adverse weather events,” as defined. Sales cutoff date. A policy sales cutoff date of April 1 should be introduced. Cover period. The cover period should be from the time of crop emergence and full stand establishment (for example, 10 centimeters for wheat) through to completion of the harvest. Insured unit (IU). The IU is currently defined as the “individual field.” For farms of less than 250 hectares, the IU should perhaps be redefined as “the total area of all fields of the same crop grown in the same location or farm.” Sum insured. Government should amend the law to permit insurance companies to have the option of establishing an agreed sum insured with each farmer according to its own circumstances and insurance requirements. Premium rates. Government should amend the obligatory law to permit each insurance company to set its own premium rates for each crop and zone. Actuarial rating should be introduced to reflect differences in risk exposures between rayons in each oblast and possibly differences in levels of technology use and risk exposures between farmers. Bonus-malus system. It is recommended that underwriters introduce a bonus-malus system on the compulsory crop insurance scheme. The objectives of the bonus-malus are (a) to introduce rating of individual farmers and (b) to reduce the submission of speculative claims notices by farmers. Basis of indemnity and claims settlement. The crop sales price, which is used to value actual production in the event of a partial loss, should be preagreed and based on an average historical farm-gate sales price for each crop in each rayon and stated in the policy wording. Source: Authors. claims reimbursement), suggesting that, on av- ly, it is recommended that a bonus-malus system erage, the scheme has operated on a breakeven be introduced into the crop rating methodology basis. The premium rates were last revised in (see chapter 4).2 2008; however, the performance of the scheme deteriorated very badly over the past three years, Measures to Improve Scheme Profitability with an average loss ratio of 182 percent (loss ratio of 91 percent net of the government’s 50 32. The financial viability of the crop insur- percent claims subsidy). In some oblasts, includ- ance scheme is being adversely affected by the ing NKO and Kostanay, the scheme performed inclusion of Aktobe and WKO. Over the past six very well, but in others, including Aktobe, WKO, and Zhambyl, it performed very poorly. There are also major differences in the pattern of 2  Under a bonus-malus system, a farmer who does losses among rayons in individual oblasts and a not submit a claim is rewarded at the renewal date of need to revise the oblast rating methodology by his policy by receiving a reduction in his premium rate (termed the “bonus”), while a farmer who submits a introducing a system of rayon-level actuarially claim sees an increase in his rate at the time of re- determined premium rates for each crop. Final- newal (termed the “malus”). 16  Kazakhstan years Aktobe and WKO collectively accounted insurance policy is structured, production will for only 4.8 percent of total scheme liability, but always be very heavily exposed to loss and the contributed 41 percent of total claims and had technical and commercial premium rates that actual loss ratios of 381 and 507 percent, respec- would have to be charged will be so high (an tively. Over this period, if Aktobe and WKO average commercial rate prior to application of had not been included in the scheme, this would government claims subsidies of between 37 per- have led to a reduction in total premium earn- cent in Aktobe and 66 percent in WKO) as to ings of only 13 percent, but would have led to a make the scheme commercially unviable. There huge savings in claims costs for insurers of 40 are, however, a few areas where spring wheat percent of actual claims (or a reduction in claims production is more stable and where crop insur- of KZT 3.3 billion—US$22.0 million) and a re- ance might be considered in the future, albeit at duction in the six-year long-term gross loss ratio very high rates, including the rayons of Taskal- from 140 to 96 percent. Following application isky and Kaztalisky located in the southwest of of the government 50 percent reimbursement WKO and the rayons of Karagaly and Martuk of claims, the “as if” loss ratio for the scheme located in northern Aktobe. would have been reduced to about 48 percent (table 1). These two oblasts (in common with much of the northern and central regions of Ka- Options for WKO and Aktobe zakhstan) were traditionally livestock grazing regions that were converted to cereal (mainly 33. The first option for insurance compa- spring wheat) production during Soviet times. nies to consider would be to continue operat- The underlying problem in most of Aktobe and ing the LIC scheme in Aktobe and WKO on the WKO is that soils are mostly poor and average understanding that government will reimburse annual rainfall is very low: in a normal year, them for a higher percentage of claims in these rainfall is barely adequate for growing spring two oblasts. Currently, government reimburses wheat in most rayons in these two oblasts. As insurers for 50 percent of the value of all paid such, these oblasts are very marginal for spring claims. Under this option, the GRK would agree wheat crop production. No matter how a crop to indemnify a higher percentage of claims in- Table 1 Actual Results of the Crop Insurance Scheme with and without Aktobe and WKO Actual portfolio Modified portfolio|(without Item % reduction (with Aktobe and WKO) Aktobe and WKO) Number of policies 140,961 134,193 −5 Total insured area (number of 73,770,915 69,453,803 −6 hectares) Sum insured (thousands of KZT) 242,631,438 231,048,598 −5 Premiums (thousands of) 5,861,958 5,102,670 −13 Average premium rate (%) 2 2 −9 Claim payments (thousands of 8,222,776 4,910,314 −40 KZT) Loss ratio (%) 140 96 −31 Loss cost (%) 3.39 2.13 −37 Source: Authors based on FFSA crop insurance data for 2005–10. Agricultural Insurance Feasibility Study  17 curred in Aktobe and WKO only. The insurance losses every other year, with loss ratios (claims industry has asked government to consider rais- to premium ratios) as high as 1,000 percent in ing its share to 70 or 75 percent of total claims major drought years. It is therefore necessary in these two oblasts in 2012. Over the past six for government to consider whether to continue years, government’s 50 percent contribution to promoting annual crops in these oblasts, but to claims costs amounted to about KZT 4.1 billion use a separate disaster relief fund to compensate or an annual average of KZT 0.685 billion per farmers in severe drought years. The cost of this year. If government agreed to increase its share disaster relief compensation program for farm- of claims in Aktobe and WKO, the “as if” an- ers in Aktobe and WKO, assuming the same nual total cost to government would increase as 100 percent original claims costs and compensa- follows under the two options: (a) under option tion levels as under the current LIC insurance A, 75 percent claims share in Aktobe and WKO, the cost would increase to KZT 0.823 billion scheme, would be exactly the same as the 100 per year, an increase of 20 percent, and (b) un- percent claims compensation option outlined der option B, 100 percent claims share in Ak- previously: the average cost to government for tobe and WKO, the cost would increase to KZT these two oblasts would be about KZT 550 mil- 0.961 billion per year, an increase of 40 percent. lion (US$3.7 million) per year, with a peak of Option B shows the maximum expected costs about KZT 1.6 billion (US$10.7 million) in to government if it wishes to continue includ- a very severe drought year, such as 2010. Op- ing Aktobe and WKO in the LIC scheme and tions for some form of macro-level catastrophe assumes that insurance companies act purely drought weather index insurance (WII) cover as administrators in these two oblasts, but do are also examined in this report. not accept any claims liability: the “as if” 100 percent claims cost to government in these two Measures to Strengthen the Scheme’s oblasts would average about KZT 550 million or Operating Systems and Procedures US$3.7 million per year. While this option may provide government a short-term solution to re- 35. In conjunction with the changes in de- taining the support and underwriting capacity of sign identified to strengthen the LIC policy, the insurance industry and increasing the prof- administrators of the scheme should consider itability of the program, it does not tackle the a series of potential measures to improve un- fundamental issue in the medium to long term, derwriting and claims operating systems and which is that commercial spring wheat crop in- procedures and to reduce the costs of these surance is not financially viable in most of the operations. Box 2 summarizes some of the key rayons located in Aktobe and WKO. operational areas that require strengthening. In the specific case of loss assessment, the current 34. The second option for government to procedures, whereby a committee of up to five consider would be to take Aktobe and WKO out of the crop insurance scheme altogether and to individuals from five different organizations is establish a separate disaster relief scheme for involved in the in-field loss assessment, must be producers of annual crops in these two oblasts. rationalized and made more cost-effective, and The argument for removing Aktobe and WKO the insurance companies should be given a more from the scheme centers on the following: (a) central role in loss assessment. At the same time, these areas receive very marginal rainfall and impartiality and fairness must be maintained for therefore are not suitable for growing annual both the insured and the insurer. In the short grain crops and (b) from a crop insurance view- term, Kazakhstan does not have independent point, these oblasts (with the possible exception firms of certified and approved loss adjusters of a few rayons listed above) are effectively un- who specialize in crop loss assessment, but in insurable, with six-year long-term loss costs of the medium term, options should be explored for 22 and 40 percent, respectively, and catastrophic creating such specialists. 18  Kazakhstan Box 2. Options for Streamlining and Reducing the Costs of Policy Marketing, Preinspections, and Loss Assessment Introduce preinspections. A system of sample preinspections should be considered for large farms in northern Kazakhstan. Preinspections are needed in order to minimize moral hazard. Streamline loss notification and loss assessment procedures and make them more cost-effective. The law should be modified to simplify the loss notification obligations of the insured, to enable loss assessment functions to be managed by the insurance companies, and to streamline procedures and reduce the costs of loss assessment. A bonus-malus system should be introduced to dissuade farmers from submitting claims except where a major loss has occurred and is likely to lead to an indemnity. Use remote sensing to support in-field loss assessment. The National Space Agency (NSA) already applies remote sensing to estimate crop sown area, production, and yield for the MoA and to monitor crop status during the growing season. The crop insurance scheme managers should review potential supporting roles by the NSA in loss assessment. Review policy marketing and distribution channels with a view to reducing costs. Currently most policies are sold through local agents located in each oblast and subregion. The agents are currently paid 10 percent brokerage by law. Alternative crop insurance marketing and distribution channels should be promoted in order to reduce costs, including sales though cooperatives and farmer associations, input dealers, rural banks, and grain merchants. Offer farmer awareness programs on crop insurance. In conjunction with the proposed improvements in operating systems and procedures, greater emphasis needs to be placed on farmer awareness programs about the basis of insurance and indemnity in the current LIC program. Source: Authors. Measures to Strengthen the Institutional insurance associations should in the future be Framework regulated by the Agency for Financial Market and Financial Institutions Regulation and Con- 36. In the short term, a major challenge is to trol and be required to follow the same guide- find ways to encourage more private commer- lines with regard to capital requirements and cial insurance companies to support the crop insurance reserves as commercial insurance insurance scheme. Several measures identified companies. Failing to do so creates biased com- under this study should be attractive to local in- petition and increases the chances that mutual surers, including (a) the introduction of actuari- associations will fail to meet their financial ob- ally determined premium rates, (b) the measures ligations. to reduce insurers’ liability in Aktobe and WKO, and (c) recommendations for strengthening loss Measures to Cap Catastrophic Losses assessment procedures and for giving insurers (Excess of Loss Protection) more direct control over this important function. However, in the medium term, it is also probable that insurers will insist on being given greater 38. Both the crop insurance companies and control over risk selection and underwriting if the farmer mutual crop insurance associations they are to join the scheme. in Kazakhstan are very exposed to catastrophic losses that exceed their reserves, and options 37. There is also a need to create a level for enhanced reinsurance protection need to playing field for commercial insurers and the be considered. The GRK currently provides free farmer mutual insurance associations. Mutual proportional reinsurance protection equal to 50 Agricultural Insurance Feasibility Study  19 percent of the claims to the private insurance and 100 percent of gross net premium income companies and mutuals in Kazakhstan. However, (GNPI). This analysis was conducted separately, neither the private insurers nor the mutuals have first assuming that Aktobe and WKO are includ- any reinsurance protection on their 50 percent ed in the scheme and then assuming that they are retentions, and they are therefore very exposed excluded. Although this is a preliminary analy- to major systemic drought losses. In the case sis, it is considered robust and suggests that the of the private insurance companies, their abil- indicative costs of providing full value (that is, ity to absorb catastrophic losses is much higher up to 100 percent of the total sum insured) ag- than that of the mutuals because of their much gregate stop-loss protection in excess of 100 larger size, their capital and claims reserves, and percent and 70 percent priorities would be on their diversified nonlife insurance portfolios, the order of KZT 0.27 billion (US$1.8 million) under which crop insurance only represents a to KZT 0.31 billion (US$ 2.1) per year (see fraction of their overall premium earnings and chapter 4). overall liability. In 2010 there is evidence that some mutuals could not meet their claims obli- Premium Subsidies gations in full because the claims exceeded their premium earnings: they therefore had to prorate 40. If government were to switch its support down claims. There is an urgent need to design from a share of 50 percent of claims to some catastrophe excess of loss reinsurance protection form of nonproportional excess of loss protec- for both the private insurance companies and the tion, the insurers would need to increase their mutual associations on their 50 percent reten- premium rates and the GRK would need to tions through some form of nonproportional or consider whether such rates are affordable or stop-loss reinsurance protection. whether premium subsidies are needed. Over the past six years, government’s 50 percent of 39. In the short term, it is unlikely that the claims liability was an average of KZT 685 mil- Kazakhstan obligatory crop insurance scheme lion per year (US$4.6 million), but in severe will be able to meet the standards required to drought years such as 2010, it was much higher, attract international reinsurers; therefore, any at KZT 1.4 billion (US$9.3 million), and the nonproportional reinsurance solutions will total cost was KZT 4.1 billion (US$27.3 mil- probably have to be provided by the GRK. Some lion) in claims compensation. If instead, over indicative rating analyses were carried out for the past six years, government had provided 50 a GRK aggregate (that is, over the whole crop percent premium subsidies, the total cost would insurance scheme) stop-loss reinsurance protec- have been somewhat higher, at KZT 5.86 billion tion3 for spring wheat under the assumptions of (US$39.1 million) total, or an average of KZT the existing LIC scheme sum insured and actu- 0.98 billion (US$6.5 million) per year. arially determined rates for priority levels of 70 41. The GRK should study very carefully the issues surrounding premium subsidies before 3  Stop-loss reinsurance is a type of nonproportional deciding whether to switch from subsidizing reinsurance protection that is designed to cap an in- claims to subsidizing premiums. In the current surer’s claims liability at a preagreed amount (value), system, government compensates 50 percent of which is referred to as the priority. Any claims above the priority are then transferred (ceded) to the rein- the claims costs and then caps premium rates surer to settle, up to the limit of the reinsurer’s liability, at approximately 50 percent of the technically which is defined in the reinsurance agreement. The required rates. In some regions of the country, priority is often expressed as a percentage of the pre- current premium rates are above the actuarially mium income that is underwritten by the insurance scheme, net of any policy cancellations and returns required rates and in others actual rates are far of premium; hence the term gross net premium in- too low. On the one hand, this results in distorted come. crop insurance price signals in the market, and, 20  Kazakhstan on the other hand, the 50 percent claims compen- 44. MPCI and LIC insurance products are sation does not provide local insurers with the ca- slightly different. The main differences be- tastrophe protection they need on their retained tween an individual grower MPCI product and claims. Finally, international reinsurers are not the LIC policy include (a) the establishment of willing to support an underpriced scheme. While a preagreed insured yield at the time of policy the authors are very cautious about recommend- subscription (the insured yield is usually calcu- ing premium subsidies, it would be preferable lated as a percentage of the individual farmer’s in Kazakhstan to have an actuarially rated and historical average or normal crop yield or the commercially priced program and then for gov- local area average yield), (b) a preagreed unit ernment to decide whether to provide financial valuation price, which is applied to the insured support in the form of premium subsidies, to yield to calculate the sum insured, and (c) loss provide nonproportional stop-loss reinsurance assessment, which involves measuring the ac- protection in the short term, and to promote the tual yield, comparing it to the insured yield, participation of international reinsurers in the and indemnifying the amount of shortfall at the medium term. preagreed valuation price. Insurance and indem- nity procedures that are based on loss of yield Phase 2: Transition toward a Market- are potentially much more transparent and un- Based Crop Insurance System derstandable for farmers than the existing LIC policy, and loss assessment is also much more 42. The proposed transition over the next objective, as yield loss is measured rather than few years to a market-based crop insurance expected shortfall in production costs and then system in Kazakhstan is centered on (1) the in- compared to the estimated value of the remain- troduction of individual grower MPCI, either ing crop (salvage revenue). as a complement to or as a substitute for the current LIC policy and (2) the introduction of 45. The international experience is that formal excess of loss reinsurance protection for MPCI is very popular with farmers, but be- the crop insurance industry. It is assumed that cause the premiums associated with this prod- crop insurance would continue to be obligatory uct are high, most schemes depend on govern- for farmers during this interim phase and that ment support in the form of premium subsidies. Aktobe and WKO would no longer be included in the scheme because spring wheat cannot be The international literature on MPCI often high- commercially insured in these two oblasts. It is lights the drawbacks encountered under volun- intended that farmers in these two oblasts would tary schemes of antiselection and moral hazard, be protected by a separate disaster relief mecha- the difficulties of establishing average farmer nism. yields and corresponding premium rates, the high costs of premiums requiring government Individual Grower MPCI Cover for Spring support in the form of premium subsidies, and Wheat the often very high costs of individual grower in- field loss inspection and loss assessment. While 43. The transition from the existing LIC these arguments are indeed valid, they apply as cover to a more standardized individual grower equally to the existing LIC policy in Kazakh- MPCI policy would be relatively simple. The stan. Currently, issues of antiselection are less crop insurers of the LIC policy have gained con- of a problem because the scheme is obligatory siderable experience in underwriting loss of yield for all farmers, but because it is a loss of yield multiple-peril crop insurance and in conducting multiple-peril scheme, it shares the drawbacks in-field loss assessment to establish actual yields of other MPCI schemes. and the amount of loss. This experience would enable them relatively easily to design, rate, and 46. In Kazakhstan, the main challenge for implement individual grower MPCI. introducing individual grower MPCI centers Agricultural Insurance Feasibility Study  21 on the procedures for establishing an individ- cations for insurers’ and government’s liabil- ual grower “normal average yield” and then ity in the event of severe drought losses. For establishing premium rates for different MPCI these reasons, a detailed analysis is conducted in coverage levels in each rayon. On the basis of chapter 4 of the PML associated with increased international experience, it is believed that the coverage levels for the spring wheat program. quality of the rayon-level crop production and The analysis suggests that, for a maximum 50 yield data in Kazakhstan is good enough to en- percent coverage level, the expected losses that able the 17 years of data on spring wheat yield to might occur every 10 years could be on the order be used to design and rate an individual grower of KZT 36.3 billion (US$242 million, or 17.13 MPCI program. As the historical rayon-level percent of the value of the total sum insured, data on spring wheat crop production and yield TSI), and for a 1-in-100-year PML, the expected are available for both production enterprises and loss could be on the order of KZT 99.7 billion commercial farmers, separate coverage levels (US$665 million, or 47 percent of TSI). and premium rates can be offered for each type of farmer, if required. 50. Given the exposure of spring wheat pro- duction to catastrophic drought risks in Ka- 47. A detailed rating analysis was conducted zakhstan, it is extremely unlikely that the in- for MPCI cover for spring wheat grown in the six main oblasts (excluding WKO and Aktobe) surance sector would be willing to assume the located in northern Kazakhstan. The statistical increased liabilities implied under the proposed rating methodology used in this study to estab- MPCI program unless the government is will- lish individual grower MPCI rates conforms ing to provide reinsurance support for this ini- to the MPCI rating procedures used by the in- tiative. Currently, government is providing 50:50 ternational crop insurance industry. The spring quota-share reinsurance protection to the private wheat MPCI rates presented in this report are commercial crop insurers and mutual crop insur- indicative of commercial premium rates for a ers, but, as previously noted, this protection does 60 percent target loss ratio, but final decisions not cap their exposure to catastrophic losses. over rates will be taken by insurers and their Therefore, for the purposes of this study, some reinsurers. For insured yield coverage of 20 to preliminary analyses were conducted for non- 30 percent of yield, the MPCI rates would be proportional stop-loss reinsurance protection for roughly comparable to the full (unsubsidized) the spring wheat MPCI program. LIC premium rates, but for higher levels of cov- erage, rates would be correspondingly higher. 51. Some preliminary modeling was con- See chapter 4. ducted to establish the indicative pricing for aggregate stop-loss reinsurance protection for 48. It is likely that, if individual grower the spring wheat MPCI program. The modeling MPCI is introduced into Kazakhstan, the GRK was conducted assuming full-value protection will need to consider subsidies for the higher and priority levels of 70, 100, and 150 percent premiums. Chapter 4 makes some preliminary of Gross Net Premium Income, GNPI, for the estimates of the costs to government under dif- four levels of MPCI insured yield coverage. The ferent uptake scenarios. analysis found that, for a full-value aggregate stop-loss reinsurance protection for losses in ex- Risk Financing and Reinsurance cess of 100 percent of GNPI, the stop-loss rein- surance pricing would be on the order of KZT 49. If individual grower MPCI is introduced 991 million (US$6.6 million) for 20 percent for spring wheat and other crops, and higher coverage, rising to KZT 5.8 billion (US$38.7 coverage levels and sums insured are offered million) for 50 percent coverage. If the lower 70 to farmers, this would have important impli- percent of GNPI priority were adopted, the ag- 22  Kazakhstan gregate stop-loss pricing would be correspond- surance should be a voluntary class of insurance. ingly higher. In many countries (including India, the Philip- pines, and Mexico), the links between public 52. In phase 2, it is possible that some of the sector provision of crop credit and crop insur- larger private commercial crop insurers would ance are very close, and banks make their lend- be able to arrange their own reinsurance pro- ing conditional on the farmer having crop insur- grams with international reinsurers. If these ance in place at the time of receiving the loan. insurance companies can demonstrate that, even In other words, crop insurance is mandatory for under an obligatory crop insurance scheme, they borrowers, but voluntary for nonborrowers. are underwriting a balanced portfolio by risk and region, are adopting technically based MPCI 55. If crop insurance is made voluntary in rating, and have strengthened their MPCI loss Kazakhstan, it is likely that there will be a ma- assessment systems and procedures, then they jor reduction in the demand for crop insurance should be able to arrange both proportional and in the short term, while the farming sectors ad- nonproportional reinsurance through leading in- just to the realities of a demand-driven volun- ternational reinsurers of this class of business. tary crop insurance system. At this stage, it is not possible to predict how great the contraction Phase 3: Transformation into a in demand for voluntary crop insurance may be, Commercial Crop Insurance Pool but it is likely to be significant. Under a volun- Scheme Supported by International tary system, crop insurers would be free to select Reinsurance which types of farmers, which crops, and which regions they are willing to underwrite. 53. In the final phase, it is assumed that the scheme will be transformed into a fully mar- Rationale for and Features of an ket-based PPP agricultural insurance scheme. Agricultural Insurance Pool in Kazakhstan Central features of such a scheme would include (a) voluntary crop insurance, (b) formation of an 56. Coinsurance pools for agricultural in- agricultural pool coinsurance system to crowd in surance have proved to be very popular with private commercial insurers (and possibly mutu- private and mutual insurers in many countries. al crop insurers if they can conform to insurance Most notable are the Agroseguro pool in Spain,4 market regulations), (c) a formal program of in- the TARSIM pool in Turkey, the livestock insur- ternational reinsurance protection for the pool, ance pool in the Philippines, the hail insurance pool in Austria, and various other pool arrange- and (d) suitable backing by the government, ments in Argentina, China, Malawi, Mongolia, which might include financial support for pre- and Ukraine. mium subsidies or catastrophe risk financing. 57. The rationale for recommending the for- Transition from Obligatory to Voluntary mation of a coinsurance pool in Kazakhstan Crop Insurance centers on several key factors. These include the following: 54. As part of the transition to a market- based crop insurance system, policy makers in a. The very small number of private com- Kazakhstan will need to consider making crop mercial companies that are currently insurance voluntary. Kazakhstan is one of a small minority of countries to adopt obligatory crop insurance and is almost unique in trying to 4  Agroseguro stands for Agrupación Española de implement obligatory crop insurance through Entidades Aseguradoras de los Seguros Agrarios the private commercial insurance sector. Inter- Combinados (Spanish Group of Insurance Entities of national experience suggests that agricultural in- the Combined Agrarian Insurance). Agricultural Insurance Feasibility Study  23 supporting this scheme and the need to e. The difficulties of arranging commercial crowd in commercial insurers if crop in- international reinsurance protection for surance is to remain a viable proposition individual Kazakhstan insurance com- in Kazakhstan. LIC/MPCI crop insurance panies with very different underwriting is a catastrophe class of business, and standards and portfolios. Under a pool many insurance companies are reluctant agreement, the MUC would purchase a to risk their capital on it; however, under single reinsurance program, which would a pool agreement, individual companies lower the transaction costs and thus the can participate with very small shares of cost of reinsurance. the overall risk if they wish. 58. There are also potential drawbacks to b. The prohibitively high start-up invest- introducing pools to underwrite agricultural ment costs for individual insurance insurance. Classical economic theory would ar- companies in creating their own internal gue that forming a pool (with monopolistic or crop underwriting and claims depart- oligopolistic tendencies) reduces competition, ments and then in developing regional particularly over the pricing of insurance prod- networks of marketing and sales agents ucts. These arguments do not apply directly to to sell the product and trained crop in- Kazakhstan, where there is currently no com- spectors and loss assessors to implement petitive market for agricultural insurance, a the scheme. Pools offer the opportunity single product is available, and premium rates to create a single centralized insurance are fixed by law. While initially each member underwriting and claims management and of the pool would offer standard crop insurance loss assessment capability (often termed a policies and uniform technical premium rates, managing underwriter company, MUC); the aim over several years would be to expand individual members of the pool contrib- the range of crop insurance products available ute to the costs of running the MUC, in the Kazakhstan market, to achieve economies while benefiting from the advantages of of scale in key areas such as loss assessment, to economies of scale. lower commercial premium rates overall, and to improve the insurance services provided to c. The lack of common standards at the re- farmers. gional level in the underwriting of crop risks, especially with regard to the in- 59. The proposed Kazakhstan agricultural field loss assessment capabilities of indi- insurance pool would involve the active par- vidual insurance companies and farmer ticipation of the public and private sectors. mutual insurance associations. Under a The central feature of the new system would pool agreement, the MUC would be re- be the creation of an agricultural coinsurance sponsible for coordinating all underwrit- pool, which would be designed to underwrite all ing and loss adjustment activities and for classes of agricultural insurance business. It is ensuring that common standards are ad- recommended that the pool coinsurers also cre- opted throughout the country. ate a separate managing underwriting company (MUC), which would be responsible for un- d. A lack of consistency in crop rating and derwriting the scheme, handling premiums and competition, which is driving down the loss assessment, settling claims, and negotiating crop insurance premium rates to unsus- reinsurance on behalf of members of the pool. tainable levels. Under a pool agreement, Members would be responsible for marketing all insurers would issue standard crop crop insurance through their sales distribution insurance policies and would adopt the networks. The public sector, including the insur- same premium rates for each crop in each ance regulator, the MoA, and the FFSA, would zone and region. play very important roles in the proposed sys- 24  Kazakhstan Figure 2 Organizational Framework of the Proposed Agricultural Insurance Pool Scheme Agency for Financial Market Find for Financial Support Ministry of Agriculture and Financial Institutions to Agriculture (FFSA) Regulation and Control Financial Support: Policy, planning, premium subsidies, research and Insurance legal catastrophe, reinsurance development and regulatory Pool Management Board Kazakhstan Non-Life Insurance Companies Agricultural International Reinsurers Insurance Pool Farmer Mutual Crop Insurance Associations Managing Underwriting Company Farmers Associations, Cooperatives, Large Farmer production Enterprises Rural Banks and other Aggregators Small and Medium Farmers Source: Authors. tem. The FFSA would continue to act as the and (d) provide information and advice to farm- main public sector implementing agency for the ers. An outline institutional framework for the scheme, but would no longer participate in field- agricultural insurance pool is shown in figure 2, level loss assessment; instead, it would (a) coor- which draws on the organizational structures of dinate with the crop insurance pool’s MUC in the the Spanish and Turkish agricultural insurance development of the technical studies required to pools. design new crop, livestock, forestry, and aqua- culture insurance policies and programs, (b) 60. Details of the proposed operating sys- manage the FFSA and disburse funds (including, tems and procedures for the Kazakhstan agri- as appropriate, premium subsidies and catastro- cultural insurance pool are set out in chapter phe reinsured claims payments) to the MUC on 4. The MUC would be responsible for the func- behalf of the pool coinsurers, (c) maintain crop tions of product design and rating, underwriting insurance underwriting and claims databases, and risk acceptance, claims administration, and Agricultural Insurance Feasibility Study  25 loss assessment and for negotiating common ac- average 40 percent coverage level and projected count reinsurance protection on behalf of and re- uptake rates of 10 percent the first year, rising to porting to the pool coinsurers. The potential for 50 percent after five years. On this basis, total the MUC to organize and train an independent scheme liability in year 1 might be on the order crop loss assessment capability in standardized, of KZT 17 billion (US$113 million), rising to timely, and accurate field-based loss assessment KZT 85 billion (US$567 million) by year 5, with procedures is one of the major potential advan- corresponding commercial premium in year 1 of tages of a pool system. The MUC, as a single nearly KZT 2.1 billion (US$14.0 million), rising entity acting on behalf of all pool coinsurers, to KZT 10.7 billion (US$71.3 million) by year should be able to achieve major economies of 5. The provisional estimates of the costs to gov- scale and A&O cost savings. ernment of 50 percent premium subsidies would be nearly KZT 1.07 billion (US$7.1 million) in Government Support to the Kazakhstan year 1, rising to KZT 5.35 billion (US$35.7 mil- Agricultural Insurance Pool lion) by year 5. Estimates are also provided for premium subsidies of 25 and 65 percent of pre- 61. As part of the transition to a market- mium (table 2). based agricultural insurance pool system in Kazakhstan, the role of public sector support Risk Financing and Reinsurance for this scheme should be reviewed. To date, the GRK’s main support for the obligatory crop 64. It is recommended that in the future the insurance scheme has been in the form of free agricultural insurance pool would purchase quota-share reinsurance of 50 percent of the in- common account reinsurance protection from curred claims, but this is likely to change under international reinsurers in order to protect the the transition to a market-based agricultural in- program against catastrophic losses. The sup- surance scheme. port from international reinsurers could include both proportional and nonproportional reinsur- 62. The GRK has several potential roles ance. It is likely that, in the initial stages, inter- to play in supporting the pool. These roles in- national reinsurers will only be willing to pro- clude (a) providing legal and regulatory sup- vide layered stop-loss reinsurance protection in port, including a review and reform of the Law order to limit their liability to catastrophe claims on Compulsory Crop Insurance, (b) enhancing and that the GRK may therefore need to provide crop insurance data and information systems catastrophe reinsurance for low-frequency but and infrastructure support, for example, invest- high-severity losses. An example of layered in- ing in weather stations, (c) supporting the insur- surance and reinsurance is presented in figure 3 ance companies through implementation of crop for the proposed spring wheat MPCI program. insurance awareness programs for farmers, (d) Full results of this analysis are presented in providing catastrophe risk financing, and (e) chapter 4. subsidizing premiums. Full details are reviewed in chapter 4. 65. The access to the international agricul- tural reinsurance markets will benefit the local Financial Estimates of the Voluntary Crop industry by making available the expertise and Insurance Scheme services of specialized reinsurers. The interna- tional agricultural reinsurers can provide servic- 63. Some provisional financial estimates of es and expertise critical for the development of total sum insured, premium, and costs of pre- agricultural insurance schemes, particularly dur- mium subsidies to government were calculated ing the first years of operation. International ag- for the spring wheat MPCI program over the ricultural reinsurers can provide their expertise next five years. The estimates were based on a and services to the local industry in the fields of voluntary MPCI scheme for spring wheat with an agricultural insurance product research and de- 26  Kazakhstan Table 2. Five-Year Estimates of Voluntary MPCI Uptake, Total Sum Insured, Premium Income, and Costs of Premium Subsidies (40% Coverage Level) MPCI 40 % MPCI uptake Item coverage and Year 1 Year 2 Year 3 Year 4 Year 5 100% uptake 10% 20% 30% 40% 50% Total sum insured 169,697 16,970 33,939 50,909 67,879 84,849 Total commercial 21,410 2,141 4,282 6,423 8,564 10,705 premium Cost of GRK premium subsidies 25% of premium 5,353 535 1,071 1,606 2,141 2,676 50% of premium 10,705 1,071 2,141 3,212 4,282 5,353 65% of premium 13,917 1,392 2,783 4,175 5,567 6,958 Probable maximum loss 1 in 100 years 66,470 6,647 13,294 19,941 26,588 33,235 1 in 250 years 79,688 7,969 15,938 23,906 31,875 39,844 Source: Authors. Figure 3. Example of Agricultural Risk Layering 50 Goverments 45 40 Reinsurers Size of the Loss 35 Risk Transfer 30 25 Insurance Companies 20 15 Cooperatives/Mutuals Risk Pooling 10 5 0 Agricultural Producers Risk Retention m ic l e al or rg iu ph Sm in ed La tro M M as at C Type of Event Source: Mahul and Stutley 2010. Agricultural Insurance Feasibility Study  27 velopment, pricing and underwriting, and claims than 100 years in Argentina, Australia, Europe, management. New Zealand, and North America. Today there is a large body of accumulated experience with Opportunities for New Crop crop hail damage-based insurance and indem- nity products. Wordings are readily accessible Insurance Products through international associations of hail insur- ers, and premium rates can initially be set based 66. As part of this World Bank study, an as- on transferred international experience. As long sessment was conducted of the potential to de- as suitably high each and every loss (EEL) de- sign and implement new crop insurance prod- ductibles (or franchises) are maintained, the ucts. These products include named-peril frost premium rates are generally not high. Finally, and hail cover, area-yield index insurance, and standardized damage-based loss assessment pro- weather index insurance for specific types of cedures can be obtained from the international Kazakhstan farmers and for different regions, hail associations, and training can be provided to according to the key risk exposures. Chapter 5 local staff. This experience could be transferred of the report presents the findings of this assess- very easily to Kazakhstan and then be tailored ment. by the crop insurers to meet local requirements. Features for the design of a spring wheat hail Named-Peril Crop Hail Insurance for insurance policy for Kazakhstan are outlined in Spring Wheat Farmers chapter 5. 67. Crop hail was the second most important 69. Crop hail insurance should be relatively cause of insured claims on the obligatory crop easy to design and implement in Kazakhstan insurance scheme over the past six years, and as a commercial crop insurance product. A this study found that farmers in specific regions preliminary hail rating exercise conducted for of Kazakhstan may be interested in a hail-only spring wheat in selected rayons in Kazakhstan crop insurance policy. Hail is a moderate to se- suggests that it should be possible to design hail vere problem in many parts of Kazakhstan, with cover at affordable rates. In low hail-risk regions peak months of hail exposure occurring between for a standard 6 percent damage excess (fran- May and July (chapter 2). Over the past six years chise), average hail rates of between 2.5 to 3.5 of operation of the obligatory LIC scheme, hail percent should be feasible, rising to between 5 was the second most important cause of loss af- and 6 percent in medium hail-risk regions. Since ter drought, accounting for about 2.5 percent of this is a noncatastrophe crop insurance product, the total area lost due to insured perils. On the it should be relatively easier for the crop insur- basis of the field visits conducted as part of this ance companies and possibly the farmer mutual study, farmers in some regions have expressed insurance associations to underwrite this product interest in a voluntary scheme of hail-only in- with limited access to reinsurance protection. It surance for spring wheat, including pockets of is likely that the demand for single-peril crop hail medium to high hail risk in parts of NKO, Ak- insurance for spring wheat will be quite low in mola, Kostanay (for example, in the rayons of the initial stages of implementation, as hail risk Altynsarin and Kostany), and Pavlodar. exposure is not as widespread as drought risk exposure. There would be an important start- 68. Crop hail insurance is a simple and well- up cost—namely, to design the suitable loss as- understood class of crop insurance, and it is sessment procedures and then to identify a core widely applied throughout the world to wheat of loss assessors who would receive specialist and a wide range of cereals, horticultural, and training in hail loss assessment procedures in tree fruit crops. Single-peril hail insurance is wheat. Finally, it is likely that there will be de- the simplest and best-known type of indemnity- mand for hail protection for other crops—for based crop insurance; it has operated for more example, cotton and horticultural crops—grown 28  Kazakhstan in southern Kazakhstan; over time there should cost savings in AYII loss assessment, as this is be potential to develop and expand a crop hail not conducted on an individual farmer and field- portfolio in Kazakhstan. by-field basis, but rather according to a preagreed random sampling of crop yields on plots within Area-Yield Index Insurance for Spring the IU (the rayon). However, the AYII product Wheat has one important drawback: “basis risk” or the potential difference between the insured area- 70. On the basis of this feasibility study, yield outcome and the actual yields achieved by there may be considerable potential in Kazakh- individual insured farmers within the IU. Basis stan to design and implement crop AYII as an risk arises when an individual grower incurs se- alternative or complement to the existing indi- vere crop yield losses due to a localized peril, vidual grower LIC and the proposed individual such as hail, but these localized losses do not af- grower MPCI programs. Outline proposals are fect the average yield of the district or rayon, so presented in chapter 5 of this study for a pro- the farmer who has incurred crop damage does totype area-yield index insurance, AYII,product not receive an indemnity. In addition, basis risk and program for spring wheat grown in Kazakh- arises when the crop production and yields of stan. The AYII proposals relate only to the six individual farmers in the same rayon are highly main wheat-growing oblasts of Akmola, EKO, heterogeneous and some farmers whose aver- Karaganda, Kostanay, NKO, and Pavlodar; Ak- age yields are above the area average receive tobe and WKO and their rayons are excluded on indemnities, even though they have not incurred account of their commercially uninsurable risk any significant reduction in yield or loss on their exposures. own farms. Features, Advantages, and Disadvantages 73. AYII is potentially a flexible crop insur- ance product that can be implemented at the mi- 71. AYII aims to overcome many of the draw- cro level for individual farmers or at the meso backs of traditional individual grower MPCI. level to protect the credit portfolio of a regional The key feature of this product is that it does financial institution. In Kazakhstan, there may not indemnify crop yield losses at the individual be scope to design AYII both as a micro-level field or grower level; instead, it makes indem- individual grower product for medium to large nity payments to growers according to yield loss wheat (or other cereal) producers and as a meso- or shortfall against an average area yield (the in- level product to protect the loan portfolios that dex) in a defined geographic area (for example, cooperatives or microfinance institutions (MFIs) the total sown area of spring wheat grown in a offer to large numbers of small rural households single rayon) and which is termed the Insured in individual rayons in southern Kazakhstan Unit, UI. In Kazakhstan, it is proposed that the (discussed further in chapter 6). AYII product would operate at the rayon level. 74. AYII has been adopted in several coun- 72. In the context of Kazakhstan, a key po- tries, including India, the United States, and tential advantage of AYII over individual grow- more recently Brazil and Ukraine, and is be- er MPCI is the ability to offer higher levels of ing tested in Peru, the Philippines, and Viet- insured-yield coverage at lower rates because nam. India has more than 30 years of experience losses are adjusted against an area-yield index with implementing AYII for food crops and and not at the individual farmer level. Other oilseeds under the National Agricultural Insur- advantages of the AYII approach are that it mini- ance Scheme (NAIS), which is a public sector mizes moral hazard and antiselection bias and program for small and marginal farmers that is reduces administrative costs. These advantages linked on an automatic or compulsory basis to offer the potential to market this product at lower the provision of crop credit and is heavily sub- premium costs to farmers. There are also major sidized by government in the form of capped Agricultural Insurance Feasibility Study  29 premium rates and government compensation of calculated technical and indicative commercial excess claims. As such, there are many similari- premium rates are presented in chapter 5. ties between the NAIS and the current scheme in Kazakhstan. The NAIS is the world’s largest 76. Under a spring wheat AYII program for crop insurance program and is currently insur- Kazakhstan, the coverage level in each rayon ing about 25 million farmers each year. Since should be set in accordance with (a) the under- 2010, it has been undergoing major transforma- lying exposure to and frequency of risk and (b) tion, both to strengthen the AYII policy and to the commercial premium rate that the target move the scheme toward a market-based sys- farmers can afford. In order for a crop insurance tem, and many of the issues and lessons from the scheme to be affordable to farmers, the levels of NAIS are of potential interest to crop insurance insured yield coverage in each rayon should be planners in Kazakhstan. In the United States, set to enable commercial premiums of no more AYII has been offered for a wide range of ce- than about 10 percent; on this basis, coverage real and oilseed crops for more than a decade, levels of up to 50 percent could be offered in but it is a relatively small program in contrast most oblasts or rayons, and levels as high as 70 to the popular individual farmer MPCI covers percent could be offered in NKO (chapter 5). that are available. In Ukraine, which has spring wheat production systems very similar to those Operational Considerations of Kazakhstan, attempts have been made since 2002 to introduce both AYII for cereals and 77. The procedures followed by the ARKS weather index insurance, but with limited suc- for estimating actual average yield in each ray- cess. In the case of AYII, there were difficulties on are technically sound for the implementa- in accessing quality oblast and rayon time-series tion of AYII. However, it is recommended that, crop production and yield data because of the if AYII is implemented, the insurance companies disruptions surrounding independence and be- enter into a formal agreement with the ARKS to cause the product was not widely accepted by provide the results of their crop-cutting yield es- farmers, who preferred individual grower MPCI timates for each rayon. Under an AYII scheme, cover. The experience in Ukraine can provide it is also likely that insurers and their reinsurers useful lessons for any future planning and de- will wish to implement some form of indepen- sign of AYII for Kazakhstan. See chapter 5 for dent monitoring of the area-yield estimation pro- further details on the AYII programs in India and cedures at the rayon level to verify that standards Ukraine. of accuracy in the measurement of yields are maintained. Insurance companies may also wish Design, Coverage Levels, Sums Insured, and to establish formal agreements with the National Premium Rating Space Agency in order to use its remote-sensing services to estimate crop sown area, production, 75. In Kazakhstan, the proposed AYII policy and yields and to monitor crop status during the for spring wheat would operate at the rayon growing season. level. This is the lowest level of disaggregated time-series crop area, production, and yield Financial Estimates and Reinsurance data available through the ARKS. A preliminary premium rating exercise was conducted for the 78. Some provisional financial estimates spring wheat AYII program for all eligible ray- were calculated for AYII cover for spring ons in the six oblasts. The rating exercise was wheat, assuming a voluntary program and 5 based on 17 years of historical time-series data percent incremental uptake rates per year over on crop production and yield for spring wheat. the next five years. Under the assumptions of a This exercise adopted internationally recog- 50 percent insured yield coverage and 5 percent nized rating procedures, and full details of the uptake of AYII insurance per year over the next 30  Kazakhstan five years, the total sum insured might rise from hedging the drought exposure of spring wheat KZT 10.6 billion in year 1, with corresponding in the north of Kazakhstan is technically fea- premium income of KZT 692 million, to KZT 53 sible. However, challenges related to the scale billion in year 5, with premium income of KZT of implementation of, commercial viability of, 3.5 billion. The costs to government of different and farmers’ interest in WII may limit the ap- levels of premium subsidy were also estimated. plication of this class of insurance products. The These uptake estimates are extremely ambitious details of the analysis are presented in chapter for a voluntary insurance scheme and would 5, where the prototypes developed for this study need refinement following a more detailed study are illustrated and the operational and commer- of AYII demand. cial challenges of WII are discussed. Conclusions on AYII Features, Advantages, and Disadvantages 79. AYII for spring wheat is technically and 83. The essential feature of WII is that the operationally feasible in Kazakhstan. However, insurance contract responds to an objective until further research has been conducted into parameter (for example, measurement of rain- the potential demand for this cover, it is very fall or temperature) at a defined weather sta- difficult to predict likely uptake rates under a tion during an agreed-upon time period. The voluntary crop insurance program. parameters of the contract are set so as to cor- relate, as accurately as possible, with the loss of 80. Farmers’ demand for and willingness to a specific type of crop suffered by the policy- holder. All policyholders within a defined area pay for AYII crop insurance will also have to be receive payouts based on the same contract and studied further before any decisions are made measurement at the same station, eliminating the to proceed with the design of an AYII program. need for field loss assessment. This feasibility study identified a very low level of interest in the obligatory LIC crop insurance 84. The suitability of WII for transferring scheme by farmers, and it is probable that this weather risks depends on the strength of the would apply equally to voluntary crop insurance correlation between the weather parameter and in the future. Similarly, Ukraine has experienced the crop yield and the spatial correlation of the low demand for this voluntary product. It is risks. WII is best suited to transferring weather therefore recommended that the key stakehold- risks, where these risks are well correlated over a ers in Kazakhstan conduct a formal study to as- widespread area and there is a strong correlation sess demand for AYII. between weather and crop yield. The strongest relationships typically involve a single crop, a 81. AYII for spring wheat could possibly be marked rainy season, and no irrigation. WII is underwritten. This could be either a meso-level less suited to the transfer of weather risks where product designed to protect the seasonal loan conditions are more complex. Localized risks, portfolio of agencies that are lending to cereal such as hail, or areas with microclimates (for producers (banks, input suppliers, or MFIs) or example, in mountainous areas) are not suitable macro-level AYII cover operated by govern- for WII. Similarly, the scope for WII is limited ment to protect small family farms in southern where crop production is affected by many or Kazakhstan. complex causes of loss or where pests and dis- eases are major influences on yields. Weather Index Insurance Opportunities for Kazakhstan 85. The features of spring wheat crop pro- duction in the north of Kazakhstan indicate 82. The analysis carried out in this feasibility that WII is a potentially suitable risk transfer study found that developing WII contracts for product for this crop. Spring wheat is almost a Agricultural Insurance Feasibility Study  31 monoculture in northern Kazakhstan, the region ducted for drought peril in spring wheat pro- enjoys a marked rainy season, and crop produc- duction in six rayons in northern Kazakhstan. tion is fully rain fed. Furthermore, the analysis The prefeasibility analysis was performed based presented in this report of cumulative rainfall on rainfall information provided by the KHM deficits and spring wheat yields for a sample of for nine weather stations situated in the selected stations and rayons, in general, found very high rayons or in neighboring rayons that fully com- correlations, which suggests that there is consid- plied with the best practice for the design and erable scope for the GRK to continue to research operation of WII. The prefeasibility analysis the development of rainfall deficit WII cover for followed a widely used methodology developed spring wheat in the northern region. by the Agriculture Risk Management Team (ARMT) of the World Bank and specific addi- 86. As for AYII, basis risk is the key con- tional indexing procedures that were developed straint of WII. “Basis” can be defined as the for the spring wheat environment of northern difference between the loss experienced by the Kazakhstan. farmer and the payout triggered by the weather index. It could result in a farmer experiencing a 89. WII may be technically feasible for yield loss without receiving a payout or receiv- transferring drought risk in spring wheat crop ing a payout without experiencing a loss. WII production in northern Kazakhstan. In six out works best where losses are homogeneous in the of the nine combinations of rayon-level yield defined area and are highly correlated with the and weather station data analyzed, it was pos- weather peril. sible to develop meaningful rainfall deficit WII structures. However, while this applies to the 87. WII can be retailed at different business areas surrounding the specific weather stations levels. At the micro level, the policyholders (the analyzed, full-scale implementation of a micro- insurer’s customers) are individual farmers, households, or small business owners who pur- level (farm-level) WII program may be hindered chase insurance to protect themselves from po- by the relatively low density of the weather tential losses caused by adverse weather events. station network. Taking as a reference the nine At the meso level, the insurance policy is issued cases examined, in Kazakhstan, distances be- to an organization with economic interests that tween contiguous weather stations start from a are contingent on the results of agricultural ac- minimum of 70 kilometers (against conventional tivities, including, for example, an input suppli- wisdom that, for rainfall WII, stations should be er or a financial institution or cooperative that no farther than 20 to 25 kilometers apart), which lends to the rural sector and wants to protect it- is probably too far for granting full micro-level self against eventual default of the loans given WII coverage of the entire territory. to farmers due to unfavorable weather condi- tions during the crop season. At the macro level, 90. The potential implementation of macro- the insurance policy is settled to a government level or meso-level WII contracts, which are or a national organization. The insured interest less reliant on the weather station network in the case of macro-level coverage is usually than micro-level WII contracts, could have a related to government disaster relief for small good chance of being implemented relatively farmers (macro-level index programs in Mexico rapidly. Although it may be possible to over- are presented in box 6.1 in chapter 6) or food come the structural constraints related to the security issues. insufficient density of weather stations in the medium term, widespread full-scale implemen- Weather Data and Contract Design tation of WII at the farm level does not seem to be realistic in the short term. In this regard, the 88. Under this study, a prefeasibility analy- implementation of WII contracts at the meso or sis of a prototype WII product design was con- macro levels, which may be less influenced by 32  Kazakhstan the density of the weather network, could have Conclusions on WII a greater chance of being rapidly implemented than the micro-level approach. 94. The analysis carried out in this feasibil- ity study found that offering WII for spring Basis Risk wheat in the north of Kazakhstan is technically feasible. Despite the positive technical findings, 91. Preliminary analyses conducted under the actual density of the weather measurement this study suggest the existence of a relevant network does not make full-scale implementa- basis risk for individual farmer WII cover. tion of farm-level WII a realistic option in the Basis risk is defined as the potential mismatch short term. While potential actions to address between the actual financial losses due to an in- this constraint may be undertaken, including the sured event suffered by the farmer on his farm GRK’s current investments in new weather sta- and the payouts received from the insurance. tions, for the time being individual farmer WII During the study, a basis risk assessment of WII cannot be considered a readily implementable was performed for a limited sample of farmers alternative to the current LIC insurance scheme. situated within 25 kilometers of the selected In addition, the high potential cost of WII prod- weather stations for the 2010 spring wheat crop ucts developed in the analysis, together with a year. The results of this analysis suggest the ex- preliminary analysis of the patterns of basis risk, istence of basis risk and highlight the importance suggests the need to carry out further research to assess the potential interest of farmers in such an of carefully evaluating the basis risk embedded approach. in farm-level WII products. Costs Tailoring Crop Insurance to the Needs of Small Farmers in 92. If WII were implemented to cover South Kazakhstan drought for spring wheat crop production in northern Kazakhstan, it would be a relatively 95. The final section of this report presents expensive product. The analysis carried out in some of the international lessons and experi- the feasibility study found that indicative com- ence on strategies and programs to address the mercial premium rates of the technically feasi- agricultural risk transfer and insurance needs ble contract structures range from 13.7 percent of small farmers. These lessons may be applica- in relatively good crop areas to 24.2 percent in ble to the small household mixed crop and live- marginal areas. This is a clear indication of the stock farming sectors, which are mainly located high potential cost of using WII contracts to in- in southern Kazakhstan. sure against drought. Identifying Appropriate Crop Insurance Meso-Level WII and Reinsurance Products 93. A specific form of meso-level application 96. One size does not fit all. To date, crop in- of WII could be the use of index contracts as surers in Kazakhstan have offered a single LIC reinsurance coverage for insurance companies crop insurance product mainly to medium and involved in agricultural insurance programs. large cereal producers located in northern Ka- The analysis carried out in the study found that it zakhstan. There has been very little debate about is possible to structure drought insurance prod- the appropriateness of this product for small and ucts for spring wheat production by calibrating marginal farmers in southern Kazakhstan. weather indexes on rayon-level yield records. These structures could technically form the ba- 97. A wide range of crop insurance prod- sis of a reinsurance transaction. ucts is available internationally, and this report Agricultural Insurance Feasibility Study  33 recommends that Kazakhstan’s crop insurers strument that can be designed to transfer risk should aim to develop and introduce several at different levels of aggregation. Index prod- of these alternative products in the future. The ucts can be offered at three levels: (a) at the in- products that are recommended for individual dividual farmer level (termed micro-level insur- farmers include named-peril (hail) crop insur- ance), (b) at an intermediate level of aggregation ance, MPCI loss of yield insurance, area-yield as financial business interruption protection for index insurance, and crop weather index insur- banks and other lending organizations such as ance. The suitability of each type of crop insur- cooperatives and MFIs (meso-level insurance), ance product should be studied carefully in the and (c) at the regional or national level as in- context of the types of farmers (commercial, surance for regional or national governments semicommercial, subsistence) and farming against major systemic perils such as drought systems (irrigated, nonirrigated) prevailing in (macro-level insurance). southern Kazakhstan. 101. In Kazakhstan, there may be opportuni- 98. Traditional individual farmer MPCI is a ties for government to use macro-level index risk management tool that is often appropriate insurance as a catastrophe drought insurance for commercial and semicommercial farmers; mechanism. For nearly a decade Mexico has however, it cannot provide solutions for sub- operated a system of AYII and WII catastrophe sistence farmers. There is much evidence today climatic insurance programs that provide in- that traditional MPCI does not work for small surance protection to state-level governments and marginal farmers and usually ends up being against crop failure among small-scale subsis- heavily subsidized by governments. Few small tence farmers. This macro-level index approach subsistence farmers producing food crops for may offer solutions to severe drought losses in on-farm family consumption can afford crop in- WKO and Aktobe that are not be insurable un- surance, which explains why governments offer der a commercial individual farmer insurance premium subsidies. In Kazakhstan, crop insur- program. The application of such a macro-level ance is unlikely to be a useful intervention for cover for small subsistence farmers in SKO ap- the very small rural households of subsistence pears more limited because farming systems producers. are highly heterogeneous and do not lend them- selves to macro-index solutions. 99. For subsistence farmers, it may be much more cost-effective for governments to exam- Insurance Delivery Systems ine alternative social safety nets or, where they elect to use insurance, to consider some form 102. Insurance companies throughout the of macro-level weather index program, to per- world face major challenges in trying to iden- mit early payments to be made in the event of tify cost-effective ways of delivering and ad- a major natural disaster. To date, several coun- ministering agricultural crop and livestock tries, including Ethiopia, Malawi, and Mexico, insurance programs for small farmers. In Ka- have designed macro-level rainfall deficit index zakhstan, this problem is accentuated by the covers to provide national or regional govern- very low sums insured under the obligatory ments with immediate cash liquidity following a crop insurance scheme, with an average sum in- natural disaster and to enable the government to sured over the past five years of only KZT 3,287 provide an early response. (US$22) per hectare and an average premium rate of only 2.42 percent, generating an average Tailoring Crop Insurance to Different premium of about KZT 80 (US$0.53) per in- Clients sured hectare. In other words, a farmer with only 100 hectares would, on average, generate a total 100. Crop index insurance (including both premium of about US$53, and this amount is far AYII and WII) is potentially a very flexible in- too low to enable Kazakhstan’s commercial crop 34  Kazakhstan insurers to cover their A&O costs on such small the obligatory LIC crop insurance scheme in farm units. Kazakhstan. These mutuals have very limited fi- nancial reserves, and none is formally protected 103. In southern Kazakhstan, there may be by any form of insurance or reinsurance. The in- considerable potential for commercial insur- dividual mutuals are therefore very exposed to ers to enter into a “partner-agent” relationship catastrophic losses that exceed their reserves. In with rural organizations (such as the coop- the event that claims exceed their reserves, as eratives or MFIs) that have an existing rural happened in 2010, the mutuals had to pro rata distribution network and a large membership down each claim settlement made to members of farmers. Under a partner-agent model, the who incurred losses (as reported in chapter 4). insurance company enters into a formal con- International experience shows that, when catas- tractual agreement in which the agent assumes trophe claims occur, the mutual often cannot pay responsibility for marketing and promoting the and eventually collapses. insurer’s policies to its membership, for collect- ing premiums from the insured and paying these 105. If the mutuals are to remain solvent and over to the insurer, for notifying claims to the in- to underwrite crop insurance, livestock insur- surer, and, in some cases, for distributing claims ance, or both for small and marginal farmers settlement payments to the insured. Usually the in Kazakhstan, ways of providing some form of insurer agrees to pay the agent a commission for catastrophe reinsurance protection must be de- its services. This model would potentially enable veloped. In the short term, it is unlikely that the the private commercial insurers in Kazakhstan private commercial insurance sector in Kazakh- to deliver crop insurance more cost-effectively stan or international reinsurers will be willing to large numbers of small and medium farm- to provide excess of loss reinsurance protection ers. Such a model could also be used to deliver to the mutuals. It is likely that such a program livestock insurance to the small mixed cropping will have to be offered through the public sector. and livestock farmers located predominantly in Given the FFSA’s experience with administer- southern Kazakhstan. ing financial claims subsidies on the obligatory LIC program, the FFSA would be best placed to Mutual Insurance administer some form of excess of loss program for the mutual crop insurance associations. In 104. The farmer mutual insurance associa- this context, policy makers in Kazakhstan may tions have been heavily promoted by the GRK wish to study the Mexican “fondos de asegura- since 2008, and if their financial status could miento” (self-insurance funds, SIFs) program, be strengthened they might be the ideal in- which is a very successful small farmer mutual stitutional vehicle for underwriting Kazakh- crop and livestock insurance scheme that is for- stan’s small and marginal crop and livestock mally reinsured by Agroasemex, the national producers. Currently, about 38 farmer mutual agricultural reinsurance company. See chapter 6 crop insurance associations are underwriting for further details of the SIF program. Agricultural Insurance Feasibility Study  35 Agricultural Insurance Feasibility Study: Summary of Recommendations for Improvement of the Obligatory Crop Insurance Scheme in Kazakhstan The current situation in Kazakhstan is as follows: • An obligatory crop insurance scheme was initiated in 2005. Insurers (commercial companies and farmer mutual associations) offer a single loss of investment costs (LIC) policy for major crops, including cereals. The financial results of the scheme have deteriorated in recent years. • The Government of Kazakhstan (GRK) contributes 50 percent of claims costs through the Fund for Financial Support of Agriculture (FFSA). Over the past six years (2005–10), the GRK’s 50 percent claims subsidy amounted to a total of KZT 4.1 billion. In the worst year—2010—it amounted to KZT 1.4 billion. • The existing fiscal cost of the subsidy for the GRK is KZT 685 million per year on average. Ex- cluding Aktobe and WKO, it is about KZT 409 million per year. • In addition, the GRK pays for the operating costs of the FFSA, which average KZT 46 million per year. The following are the main options for consideration under the proposed phased approach. • Strengthen the legal and regulatory framework • Strengthen of the design and rating of the LIC policy • Introduce measures to improve profitability (reconsider insurability of WKO and Aktobe) Phase 1 (short • Strengthen operating systems and procedures term): Strengthen • Strengthen institutions the existing • Cap catastrophe losses compulsory scheme • Consider implementing GRK premium subsidies instead of contributing 50% of claims cost • Introduce individual grower multiple-peril crop insurance (MPCI) • Introduce higher levels of insured yield coverage and sums insured Phase 2: Transition • Introduce actuarially calculated rates (implications for affordability and government premium toward a market- subsidies) based crop insurance • Continue to strengthen the scheme’s operating systems and procedures system • Introduce excess of loss reinsurance protection for the crop insurance industry • Transition to voluntary crop insurance • Improve the current public-private partnership for crop insurance Phase 3 (medium • Create a crop coinsurance pool term): Transform • Conduct research and development of new insurance products into a commercial • Strengthen operating systems and procedures pool crop insurance • Introduce a formal program of international reinsurance protection for the pool scheme 36  Kazakhstan Phase 1 (Short Term): Strengthen the Existing Compulsory Scheme Options for Main Considerations Fiscal implications consideration recommendations Strengthen Amend legislation Assuming that crop insurance will continue to the legal and to allow insurance be obligatory in the short term, the law should regulatory companies to set terms be amended to permit insurers to set their framework and conditions own terms and conditions (premium rates and sums insured). Strengthen of the Increase sum insured The LIC policy does not provide adequate design and rating values levels of protection to farmers. of the LIC policy Revise the rating Premium rating should be revised to reduce methodology in order to the loss ratio to 60% and allow some profit update the rates on an in the scheme (which currently experiences actuarial basis a 70% loss ratio net of 50% claims reimbursement). Revise the rating Rayon-level rates would help to avoid cross- methodology to subsidization between different rayons and introduce rayon-level oblasts. Farmers in each rayon should pay for premium rates the risk that applies in that rayon. Strengthen the terms Terms and conditions should be aligned and conditions of the with international best practice in order to current insurance policy avoid moral hazard and attract international reinsurers. Introduce Remove WKO and Over the past six years, Aktobe and WKO The additional cost of measures Aktobe from the accounted for only 4.8% of total scheme moving from government to improve insurance scheme liability, but contributed 41% of claims valued 50% financing of claims profitability at KZT 3.312 billion, of which government paid to 100% financing of KZT 1.656 billion in 50% claims compensation claims in Aktobe and or an average of KZT 276 million a year. The WKO is KZT 276 million study considered two options for Aktobe a year (based on past and WKO: establish a separate disaster relief claims pattern) or a total scheme or provide 100% government claims cost for these oblasts of reinsurance support. KZT 552 million a year. Strengthen operat- Streamline and reduce Streamlined procedures would improve ing systems and the costs of marketing, underwriting and claims operating systems procedures preinspections, and loss and procedures and reduce the costs of these assessment procedures operations. The goal is for farmers to pay for risks, not transaction costs. This is important for phase 2. Strengthen Create incentives Incentives are needed to reverse the exodus institutions for private insurance of private insurance companies. Insurance companies to support companies will likely remain in the scheme the crop insurance if certain conditions are in place (actuarially scheme determined premium rates and freedom to select their own risks). Agricultural Insurance Feasibility Study  37 Create a level playing Mutual (farmer) insurance associations should field for commercial be regulated under the Agency for Financial insurers and the mutual Market and Financial Institutions Regulation insurance associations and Control and be required to follow the same guidelines with regard to capital requirements and constitution of insurance reserves as insurance companies. Cap catastrophic Design catastrophe Private insurers and mutual farmer The additional cost losses (excess of excess of loss associations do not have reinsurance to cap the losses in loss protection) reinsurance protection protection and are very exposed to major excess of the collected for private insurance systemic drought losses. In the short term, premiums is about KZT companies and farmer it is unlikely that the scheme will be able to 270 million per year.a mutual insurance meet the standards required by international associations, based on reinsurers; therefore, the GRK will have to their risk retentions provide nonproportional reinsurance. Consider premium Assess whether farmers If government switches to some form of A 50% premium subsidy subsidies can afford the rate nonproportional excess of loss protection, scheme would cost KZT increases associated premium rates will rise, and premium 980 million a year or with higher levels of subsidies may be necessary. KZT 295 million a year coverage and whether more than the current premium subsidies are cost of assuming 50% needed liabilities (KZT 685 million a year). a. Values assuming the current spring wheat portfolio in Akmola, Kostanay, NKO, Pavlodar, EKO, Aktobe, and WKO. Phase 2: Transition toward a Market-Based Crop Insurance System Options for Main Considerations Fiscal implications consideration recommendations Introduce Engage in MPCI The main challenges for introducing MPCI individual grower product research and center on the procedures for establishing multiple-peril crop development an individual grower “normal average yield” insurance (MPCI) and then premium rates for different MPCI coverage levels for each type of farmer and each rayon. For insured yield with coverage levels of 20 to 30% of yield, the MPCI rates would be roughly comparable to the full (unsubsidized) loss of investment costs (LIC) premium rates, but for higher levels of coverage, MPCI rates would be correspondingly higher. A major objective of introducing MPCI would be to increase the levels of yield protection purchased by farmers from about 13% of average yield (approximately 1 centner per hectare) to about 30% of average yield (3 centners per hectare); in the best wheat-growing areas (NKO, Kostanay), even higher levels of yield coverage should be affordable. 38  Kazakhstan Consider mechanisms MPCI is very popular with farmers, but in Assuming obligatory crop of government support most schemes is dependent on government insurance, the estimated for the scheme support in the form of premium subsidies. additional cost of providing It is likely that, if individual grower MPCI 50% premium subsidies for is introduced into Kazakhstan, the GRK the same level of coverage will need to consider financing premium as the current LIC policy is subsidies. The fiscal implications for KZT 686 million on average. government are if the levels of insured yield This amount would be KZT protection and sums insured are increased, 1.9 billion for 20% coverage the corresponding premium subsidies will and KZT 5.3 billion for 30% also increase significantly. coverage.a Introduce excess Introduce nonpropor- If MPCI is introduced and higher coverage Assuming obligatory crop of loss reinsurance tional stop-loss reinsur- levels and sums insured are offered to insurance, the estimated protection for the ance protection for the farmers, the liabilities assumed by insurers additional cost to cap losses crop insurance MPCI program and the GRK will increase significantly. in excess of the collected industry Insurers and government will need premiums for the same to protect their risk retention against level of coverage as the catastrophic losses, and nonproportional current scheme is KZT 409 reinsurance may have an important role to million. This amount would play. be KZT 991 million for 20% coverage and KZT 2.24 billion for 30% coverage.a a. Values assuming current spring wheat portfolio in Akmola, Kostanay, NKO, Pavlodar, and EKO (excluding risks from Aktobe and WKO). Phase 3 (Medium Term): Transform into a Commercial Pool Crop Insurance Scheme Options for Main Considerations Fiscal implications consideration recommendations Move to Give insurers the A voluntary scheme will allow insurers to perform voluntary crop opportunity to select risk selection, which is the basis of all insurance. insurance risks (that is, to select Insurers’ technical results from underwriting which types of farmers, agricultural insurance will depend on how efficient which crops, and which they are in selecting and underwriting risk. It is ex- regions they are willing pected that this will incentivize insurance compa- to underwrite) nies to invest in human resources and information infrastructure related to agricultural insurance. Link crop insurance with Linking crop insurance with other services other services received received by farmers will reduce transaction costs by farmers (for example, due to cost sharing in the delivery channel. In rural credit, government some countries (Brazil, India) crop insurance is programs targeting mandatory for farmers seeking to obtain rural farmers, and so forth) credit. Public banks use crop insurance as collateral for the loans given to farmers and, in some cases, as a risk selection tool. Farmers improve their access to rural credit. Insurers are willing to participate because there is a minimum volume of guaranteed business. Agricultural Insurance Feasibility Study  39 Improve Have the Kazakhstan The primary role of the GRK in the PPP will the current PPP for agricultural be to create an enabling environment for the public-private insurance define clear development of crop insurance, including partnership roles for the private addressing market imperfections and providing (PPP) for crop sector and public sector financial stability. The insurance sector will provide insurance participants expertise on risk management, underwriting, and loss adjustment and capacity to operate the scheme. Pool risks Create a crop The potential benefits of an insurance pool include coinsurance pool among the ability to underwrite a much broader and larger nonlife insurers in book of business and the potential to achieve a Kazakhstan much better geographic spread of risk than if each company were operating independently. Other benefits include economies of scale in developing new products and programs, underwriting risks, adjusting claims, and, purchasing reinsurance. Strengthen Develop crop Stronger systems and procedures would Costs will be operating underwriting expertise assist the pool in developing underwriting quantified in due systems and and rationalize MPCI expertise, strengthening in-field loss assessment course procedures in-field loss assessment procedures, and seeking cost savings through systems and procedures appropriate training courses. Introduce a Have the GRK promote The GRK should encourage companies The GRK’s cost of formal program insurers to purchase participating in the pool to purchase proportional, purchasing stop- of international quota-share, stop- nonproportional stop-loss, or both types loss nonproportional reinsurance loss reinsurance, or of reinsurance treaties in the international reinsurance in excess protection for both to protect the reinsurance market. of the premiums the pool liabilities arising from collected for 30% the risks written by the coverage in northern coinsurance pool Kazakhstan is KZT 224 million and KZT 95 million for MPCI and AYII products, respectively, assuming 10% uptake and is KZT 448 million and KZT 190 million, respectively, assuming 20% uptake. a. Values assuming current spring wheat portfolio in Akmola, Kostanay, NKO, Pavlodar, and EKO (excluding risks from Aktobe and WKO). 40  Kazakhstan Chapter 1: Introduction and Objectives of the Study 1.1. The Republic of Kazakhstan is located in West Siberian lowland in the north to the desert Central Asia. It is the ninth largest country in of Kyzylkum and the mountain range of Tien the world, with an area of 2,724,900 square ki- Shan in the south. lometers (1,049,150 square miles). Landlocked, Kazakhstan shares common borders with China, 1.2. With gross domestic product (GDP) of the Kyrgyz Republic, Turkmenistan, Uzbeki- US$6.3 billon, Kazakhstan is the largest econ- stan, and the Russian Federation; the total length omy in Central Asia. Formerly part of the So- of its borders amounts to 12,187 kilometers. The viet Union, Kazakhstan gained independence in territory of Kazakhstan stretches some 3,000 ki- 1991 and is composed of 14 regions, known as lometers from the low reaches of the Volga River oblasts. Map 1.1 shows the administrative divi- in the west to the foothills of the Altai Mountains sions of Kazakhstan. in the east, and about 2,000 kilometers from the Map 1.1 Administrative Divisions of Kazakhstan Source: Wikipedia. Agricultural Insurance Feasibility Study  41 Importance of Agriculture in than 220 million hectares (about 74 percent of the country’s total area), of which cereal-grow- Kazakhstan ing areas occupy about 13 million to 14 million hectares. The area under pastures totals 185.7 1.3. Agriculture is a very important econom- million hectares or 68 percent of the total farm- ic sector in Kazakhstan. Agriculture contributes ing area. Individual farmers, collective farms, 5.92 percent of Kazakhstan’s GDP (Economist and organizations use 81 percent of all farming Intelligence Unit 2011a). 5 Although agricultural areas and 98 percent of all pastures. output contracted sharply during the transition following independence in 1991, output has re- 1.6. Livestock is also a very important agricul- covered steadily since 1998. During 1998–2010, tural activity. For a country with a long nomadic total agricultural output increased at an average history, it is not surprising that stockbreeding is annual rate of 5.21 percent. Most of this growth the traditional and dominant agricultural sector. was a consequence of increases in crop produc- No less than three-quarters of all agricultural tion (+84 percent), while livestock grew in real land is used for grazing. Sheep breeding is pre- terms a modest 7 percent. Despite the recovery dominant, while cattle breeding and the rearing of agricultural production, agriculture’s share in of pigs, horses, and camels are also well de- the economy shrank because of strong growth in veloped. Animal husbandry typically accounts other sectors, particularly the extractive oil in- for about 45 percent of the production value in dustry. Kazakhstan agriculture. Primary meat products include beef, veal, chicken, horse, lamb, pork, 1.4. The agriculture sector is also very im- and rabbit. portant from a social perspective in Kazakh- stan. Approximately 7.3 million people (47.2 percent of the population) live in rural areas.6 Agricultural Crop Production in Agriculture employs more than 22 percent of the Kazakhstan labor force. Although Kazakhstan is classified as a middle-income country, rural poverty is wide- 1.7. Kazakhstan is an important producer spread. The transition from a centrally planned and exporter of crops, especially high-quality to a market economy brought about a decline in wheat. In 2010 Kazakhstan was the sixth largest many state-provided agricultural services, and wheat exporter, by volume, in the world. Average enduring problems of soil degradation and lack annual production of wheat is about 13 million of infrastructure have resulted in abandonment tons, but output is highly dependent on weather and underuse of agricultural land. and, in recent years, has fluctuated between 10 million and 17 million tons. Between 2 million 1.5. The natural and climatic conditions in and 8 million tons of wheat are exported annu- Kazakhstan are favorable for agriculture. The ally, mainly to destinations in Europe (includ- country is one of the major global producers and ing Russia and Ukraine), northern Africa, and exporters of grains (mainly wheat). Other prin- Central Asia. Kazakhstan also produces around cipal agricultural products include meat, wool, 2 million tons of barley and a small amount of cotton, and milk. Farming areas occupy more oats, corn, and rice, but wheat is by far the coun- try’s most important commodity. The production of oilseeds (sunflower seed and rapeseed) is in- creasing, but total oilseed output remains well 5  GDP values are calculated as average for the pe- below 1 million tons. The country also grows a riod 2006–10. small amount of cotton in southern Kazakhstan, 6  Embassy of the Republic of Kazakhstan in the with annual lint output of about 100,000 tons. United States, http://www.kazakhembus.com/index. php?page=national-goals-and-initiatives. 42  Kazakhstan 1.8. Crop production consists mostly of grain neous with regard to farm structure and pro- crops in northern Kazakhstan. About 75 per- ductivity. The northern region is dominated by cent of the country’s wheat is produced in three large agricultural enterprises (also termed pro- oblasts in north-central Kazakhstan: Kostanay, duction enterprises) specialized in crop produc- Akmola, and North Kazakhstan (NKO); see map tion, whereas smaller mixed peasant farms (also 1.2 for the main crop-producing areas. Kostanay termed commercial farms) and household plots alone plants about 4 million hectares of wheat predominate in the south. The total agricultural per year. Spring wheat occupies 95 percent of land area increased by about 6 percent between the total wheat area in Kazakhstan and virtu- 2003 and 2007, including 4 percent growth in ally all of the wheat in the three north-central the amount of arable land and 6 percent growth oblasts. Minor grains include spring barley and in the amount of hayfields and pastures. The oats (which are grown in the same region as share of agricultural land farmed by production spring wheat), winter wheat (southern Kazakh- enterprises declined from 59 to 50 percent dur- stan), and rice (southern Kazakhstan, mostly in ing 2003–07, while land farmed by individual Kzyl-Orda). Oilseed area has nearly doubled commercial farms increased from 41 to 49 per- in the past five years but still accounts for only cent. Commercial farmers slightly reduced their about 6 percent of the country’s total crop area. arable land (by 0.14 million hectares), while Sunflower, which is Kazakhstan’s main oilseed increasing their hayfields and pastures (by 8.47 crop, is grown mostly in eastern Kazakhstan. million hectares) between 2003 and 2007. In Rapeseed is grown in north-central Kazakhstan. contrast, production enterprises increased their Cotton is grown only in southern Kazakhstan. arable area (by 0.92 million hectares) and de- creased their hayfields and pastures (by 5.14 The crop farming sector is highly heteroge- million hectares). Map 1.2 Main Crop Production Areas for Wheat in Kazakhstan and Russia Source: Authors based on ARKS. Agricultural Insurance Feasibility Study  43 1.9. The agro-industrial sector of Kazakh- conditions for rural welfare based on optimizing stan still suffers from several problems that rural settlements by improving the capacity of result in low productivity and low profitability. rural territories through integrated rural devel- Labor efficiency in agriculture is five times low- opment. er in Kazakhstan than in Eastern Europe, lower even than in Russia and Ukraine. Although in 1.11. The GRK has significantly increased its the process of being updated, the fleet of agri- support to the agriculture sector since the end cultural machinery is aged. According to the of the 1990s (World Bank 2010a). This is evi- National Agency of Statistics (ARKS), a high denced by the introduction of support measures portion of Kazakhstan’s current fleet (including including a significant increase in budget alloca- 77 percent of its tractors and 59 percent of its tion and institutional improvements, such as new harvesters), was more than 15 years old at the laws and reorganized institutions. The budget time of the 2006 agricultural census. Improve- allocated to agriculture reached KZT 390.7 bil- ments in crop management practices funded lion (around US$2.7 billion) in 2010 (Economist by expanding state subsidies have contributed Intelligence Unit 2010b), which represents 8.76 to higher and more stable wheat yields. Since percent of the total budget allocations in Ka- around 2002, government support for agricul- zakhstan for 2010 and 36 percent of agricultural ture has increased significantly in the form of GDP (World Bank 2010a). The most important subsidies on the prices of fuel, seed, fertilizer, budget allocations are for (a) development of ru- and agricultural chemicals. ral finance associated with agricultural machin- ery and (b) support for crop production, such Government Policy for as marketing, phytosanitary, and soil fertility Agriculture activities. GRK expenditures are much smaller for livestock production than for crop produc- 1.10. The Government of the Republic of Ka- tion, even though livestock contributes about 44 zakhstan (GRK) is encouraging diversification percent of total agricultural output. The types of of the country’s economy to reduce its depen- GRK expenditures in support of the agriculture dence on oil, whose price volatility and result- sector are varied and can be classified as fol- ing fluctuations in revenues make budget man- lows: (a) direct input subsidies, (b) subsidized agement challenging. The GRK recognizes the credit programs (that is, with reduced interest importance of the agriculture sector for diversi- rates), (c) market price support schemes (such fying economic growth, reducing rural poverty, as the one carried out by the Food Contract Cor- and improving food security. The GRK’s main poration to stabilize grain prices and maintain strategies for agriculture are set out in a strate- state reserves), (d) expenditure on developing gic plan, which is elaborated every three years. public goods (such as market information, land The Three-Year Strategic Plan for 2009–11 de- title and registration, disease control, and seeds fines three strategic objectives for the agricul- and grain quality analysis and grading), and (e) ture sector in Kazakhstan. The first objective is expenditure on forestry and fisheries. to achieve the sustainable development of the agro-industrial sectors, increasing their compet- itiveness, ensuring food security, and adapting Exposure of Agriculture to agrarian production to the conditions for World Natural and Climatic Disasters Trade Organization (WTO) accession. The sec- ond objective is to preserve, use rationally, and 1.12. Agricultural production in Kazakhstan rehabilitate forest, fauna, and natural resources is an extremely risky economic endeavor. Many as well as to establish conditions for sustain- risks associated with agricultural production in able water supply and efficient water manage- Kazakhstan are due to climate events. Drought ment. The third objective is to establish normal is the most pervasive peril affecting crop pro- 44  Kazakhstan Figure 1.1 Agricultural GDP and Occurrence of Droughts in Kazakhstan, 1991–2010 Kazakhstan: Agriculture Value added (annual % growth) 29% 21% 17% 13% 3% 2% 7% 6% 9% -1% 0% -6% -5% -7% -3% -21% -12% -23% -19% -24% 1991 1992 1993 1994 1995 1996 1997 1999 2000 2001 2002 2003 2004 2005 2006 2007 2009 2010 1998 2008 Year reported with drought Source: Authors from World Bank 2010b; Economist Intelligence Unit 2011a; KHM data. duction in Kazakhstan. Reasonably higher lev- Government Objectives for els of agricultural productivity can be achieved during years of adequate rainfall, but the region Crop Insurance is subject to frequent drought and is considered a zone of risky agriculture. Historically, Ka- 1.13. The GRK has three objectives for crop zakhstan’s agricultural production has suffered insurance. The first objective is to protect farm- from serious drought events. As a result, the ag- ers against loss of their crop production due to ricultural value added in Kazakhstan is marked the effects of adverse natural, climatic, and bio- by frequent and sharp year-to-year fluctuations. logical hazards. Kazakhstan is one of the most Besides droughts, hailstorms and autumn early exposed countries in the world to drought losses frost are important perils affecting crop produc- in spring wheat, followed by late spring and tion, and outbreaks of pests and diseases can lead to severe crop losses. Figure 1.1 shows the relation between fluctuations in annual agricul- tural GDP growth and the occurrence of major sharply. Local agricultural officials began to set pro- ductivity thresholds for individual fields. Fields that drought years (highlighted in yellow). 7 consistently failed to meet the threshold—typically 0.6 to 0.7 ton per hectare against a national average of about 0.9 ton per hectare—were taken out of grain production and converted to permanent pasture. The 7  Caution should be exercised in interpreting the decline in grain area accelerated in the mid-1990s, GDP shortfalls in 1993, 1994, and 1995, as the de- when shrinking livestock inventories caused demand cline in GDP in these years was not related only to for feed grain to plummet, leading to a 75 percent drought. In the early 1990s, following the breakup of drop in area planted with barley between 1993 and the Soviet Union and the loss of massive government 1999. During these six years, total grain area in Ka- subsidies for state and collective farms and live- zakhstan contracted at a rate of nearly 2 million hect- stock enterprises, agricultural productivity declined ares per year. Agricultural Insurance Feasibility Study  45 harvest frosts, hail, and pests and diseases. The lished the legal, financial, and organizational GRK uses crop insurance to protect small and basis for the implementation of crop insurance. medium farmers from bankruptcy following a The implementing agencies are the Ministry of major crop loss. The second objective is to help Agriculture (through the Direction of Strategic farmers to gain access to rural finance by provid- Planning), the Fund for Financial Support for ing insurance collateral that farmers can use to Agriculture (FFSA), the private commercial in- protect their loans against default in the event of surance companies, the farmer mutual crop in- a major crop loss. The third objective is to en- surance associations, and the local authorities in hance the effectiveness of government support each oblast and rayon. Under the PPP, the GRK programs for crop production. provides financial contributions to the crop in- surance scheme through an indemnity fund (the 1.14. In accordance with its objectives for FFSA). The GRK supports agricultural insur- agricultural insurance, the GRK enacted the ance by paying 50 percent of the insurance in- Law on Compulsory Crop Insurance, which is demnities reported by the insurance companies. dated March 10, 2004, and became effective on This is similar to a free 50 percent quota-share April 1, 2004. The law established the terms and reinsurance facility. Government payouts are conditions for implementation of a mandatory made both to private insurance companies and salvage-based loss of investment costs (LIC) to farmer mutual associations. crop insurance scheme for all farmers growing a range of strategic grain, oilseed, and other field 1.16. Crop insurance has reached high lev- crops in Kazakhstan.88 Under the law, the in- els of uptake and penetration in Kazakhstan, surance indemnity is determined in accordance which is a function of its obligatory nature; with the normative production cost per hectare however, the system is experiencing serious fi- for each type of crop, multiplied by the planted nancial and operational drawbacks. Because it area of the insured crop. The insured loss is de- is mandatory, crop insurance uptake in Kazakh- stan averaged 74 percent of the cropped area for termined as “a positive difference between the the period 2005–10. Notwithstanding the high normative cost per 1 hectare of crop production, level of penetration, from 2005 to 2010 the av- which is set at the moment of concluding an in- erage annual loss ratio after 50 percent govern- surance contract, and the revenue from 1 hectare ment reimbursement of claims was 75 percent, of crop production in the area affected by the indicating that the program is not financially sus- adverse natural events, multiplied by the ex- tainable. In addition to the financial drawbacks, act area on which the given crop was produced the crop insurance system also has experienced and which was affected by an adverse weather operational problems, as evidenced by the low event.” level of sum insured chosen by farmers to insure their crops: average of US$25.7 per hectare in 1.15. The Kazakhstan crop insurance sys- 2010. From a financial point of view, the insur- tem is based on a public-private partnership ers and mutual associations are very exposed to (PPP). The Law on Compulsory Crop Insurance catastrophic drought losses because the current created the institutional framework and estab- system is not reinsured again excess losses. Objectives and Scope of the 8  The salvage-based LIC crop insurance policy (termed “loss of investment cost” policy for short) Study protects farmers against adverse natural, climatic, and biological phenomena resulting in production 1.17. The overall objective of the current study shortfalls that cause farmers’ expected revenues from the insured crop in the insured unit to fall short is to assist the GRK in improving the existing of the investments made in growing the crop in that mandatory crop insurance program. The spe- unit. cific objectives of this study include (a) to per- 46  Kazakhstan form a crop and weather risk assessment for key perience and the best insurance and reinsurance crops; (b) to review the current crop insurance industry practices for agricultural insurance. scheme; (c) to identify the potential gaps in the current scheme and to provide recommendations 1.19. The study follows the principles estab- for its improvement, based on international ex- lished in the agriculture risk management perience and best practice; (d) to identify oppor- framework developed by the World Bank. The tunities for the development of alternative crop development of market-based agricultural insur- insurance products, including area-yield index ance risk transfer solutions in the region implies insurance (AYII) and weather index insurance the promotion of several practices, including es- (WII), as well as hail named-peril crop insur- tablishment of (a) an adequate agricultural risk ance; and (e) to identify ways to tailor the provi- management framework for countries in the re- sion of crop insurance to small farmers and the gion, which would include farmer segmentation, rural poor (particularly in the southern part of an accurate assessment of the risks faced by the the country). agriculture sector, an adequate risk financing strategy for agricultural risks, and proper institu- 1.18. The study aims to identify sustainable tional arrangements; (b) well-identified roles of market-based alternatives to the current crop the public sector, insurance industry, and farm- insurance system in Kazakhstan. In this regard, ers with regard to agricultural insurance; and (c) all of the options for improving the current sys- possible actions to be taken by governments in tem that were developed under this study are order to support agricultural insurance. market based and take into account global ex- Map 1.3 Oblasts Examined under the Agricultural Insurance Feasibility Study and Rayons Selected for Development of Agricultural Insurance Prototypes Source: Authors. Agricultural Insurance Feasibility Study  47 1.20. The study focuses mainly on spring Outline of the Report wheat crop production in the principal grow- ing regions of Kazakhstan. Owing to the geo- 1.21. The study is set out in six chapters. graphic size of the country, the study is limited Chapter 2 presents an overview of agricultural to spring wheat grown in the eight main produc- production systems and markets in Kazakhstan, tion oblasts for spring wheat: NKO, Akmola, followed by an assessment of the climatic haz- Kostanay, Aktobe, Pavlodar, West Kazakhstan ards and other risks affecting spring wheat in the (WKO), East Kazakhstan (EKO), and Kara- country’s main crop areas. Chapter 3 reviews the ganda. The study also explores opportunities for structure and performance of the current manda- developing crop insurance for small farmers in tory crop insurance system in Kazakhstan and Tole-Bi rayon in South Kazakhstan (SKO). For identifies a series of institutional, operational, those activities related to the identification of technical, and financial drawbacks of the current opportunities for developing new crop insurance system. Chapter 4 presents a phased strategy and products (crop hail, AYII, and WII), the study a series of options and recommendations for the is limited to a prefeasibility analysis and the GRK to consider for the introduction of market- eventual development of prototype products for based solutions that aim to strengthen the cur- spring wheat in a few rayons in the north and rent scheme. Chapter 5 explores the opportuni- east (including Altynsarinski and Auliyekolski ties for developing new crop insurance products rayons in Kostanay; Aktogayskiy and Zhelezin- in Kazakhstan, including prefeasibility analyses ski rayons in Pavlodar; Bulandinski and Enbek- for AYII, WII, and named-peril hail insurance shilderski rayons in Akmola) and Tole-bi rayon for selected rayons. Finally, chapter 6 deals with in SKO. The oblasts and rayons selected for the the challenges of tailoring crop insurance to the analysis are highlighted in red in map 1.3. needs of lower-income smaller farmers. 48  Kazakhstan Chapter 2: Crop and Weather Risk Assessment Kazakhstan and then an analysis of spring wheat Objectives and Scope of production and yields and the climatic con- Agricultural Crop and Weather straints to production, including an analysis of Risk Assessment rainfall data and the relationship to national and rayon-level spring wheat crop production and 2.1. To date, in Kazakhstan there has been yields. The final part of this chapter presents the little formal risk assessment for crop insur- results of a crop risk assessment model (CRAM) ance purposes of the key climatic, biological, that uses time-series rayon-level production and and natural perils and their impact on crop yield data to estimate values at risk, expected production, yields, and farm incomes. The risk losses, and expected claims costs for spring assessment presented in this chapter aims (a) to wheat in the eight selected oblasts in Kazakh- aid policy makers and planners in Kazakhstan stan. This analysis is very relevant to crop in- in understanding the major climatic and natural- surers’ understanding of risk accumulation and peril risk exposures in spring wheat, which is the maximum expected losses in spring wheat. main food and export crop grown in the country, (b) to quantify wherever possible the value of Data Availability for Crop and expected spring wheat crop losses in normal and Weather Risk Assessment catastrophic loss years, and (c) to discuss the im- plications of these findings for any modification 2.3. Three types of data are commonly used of the country’s current crop insurance program. to assess climatic risk in crop production. They Specifically, it aims (a) to identify and quantify are (a) time-series weather data, (b) crop dam- the key natural, climatic, and biological perils af- age and production loss data by cause of loss for fecting spring wheat production in Kazakhstan, each crop, which may include estimates of the (b) to quantify the frequency and severity of the financial value of the damage or losses, and (c) perils affecting spring wheat production, and (c) time-series crop area, production, and yield data. to define and map homogeneous risk zones for The analysis of variance in annual crop produc- spring wheat crops. tion and yield data is commonly used to design and rate multiple-peril crop insurance (MPCI) 2.2. This chapter presents a preliminary risk programs. assessment of weather risks and their impact on spring wheat crop production and yields in 2.4. Kazakhstan has a modern, efficient na- northern Kazakhstan. This chapter starts with tional meteorological service known as the a review of data availability in Kazakhstan for National Hydro Meteorological Service or Ka- spring wheat risk assessment purposes. This zhydromet (KHM). The surface meteorological is followed by an overview of climate and the network managed by the KHM includes 260 agro-ecological regions and spring wheat crop weather stations (50 automated) and 71 agro- production systems in the selected oblasts of meteorological measurement points. The vari- Agricultural Insurance Feasibility Study  49 ables recorded at most meteorological stations to have approximately 1,600 weather stations in include air temperature, precipitation rate, snow, order to achieve an optimum weather station soil moisture, and evaporation. The data commu- density. The KHM continuously upgrades its nication system is well structured, but somewhat weather station network. The geographic distri- obsolete in some of the more remote locations. bution of weather stations in Kazakhstan is pre- Each set of observations undergoes a strict qual- sented in map 2.1. ity check, according to World Meteorological Organization (WMO) standards. Weather data 2.6. The weather risk assessment performed are transferred to each oblast center at the ap- under this study was based on information pro- propriate frequency (three hours for temperature vided by the KHM. Under this feasibility study, to one dekad for soil moisture). From the oblast the KHM provided the World Bank with access center, a communication specialist transfers in- to daily rainfall and minimum and maximum dividual weather messages into the KHM’s cen- daily temperature data for 10 weather stations tralized web server in Astana. The web server located in seven rayons selected for the develop- also shares weather data with other profession- ment of agricultural insurance prototypes. The als in the country (synoptic specialists, agro-me- selected weather stations for which the daily teorologists, climatologists, and Gismeteo). data were provided are located in the following seven rayons: Auliyekolski and Altynsarinski in 2.5. Given the huge size of Kazakhstan, the Kostanay; Aktogayskiy and Zhelezinski in Pav- weather station density is currently not able to lodar; Bulandinski and Enbekshilderski in Ak- provide adequate coverage for all the country. mola; and Tolebi in South Kazakhstan (SKO). In According to the WMO, Kazakhstan would need addition to the daily data, the KHM also provid- Map 2.1 Weather Station Network in Kazakhstan, 2011 Source: KHM. 50  Kazakhstan ed monthly rainfall and average minimum and are carried out by the rural county administra- maximum daily temperature data for the follow- tions (akimats). The farm accounting surveys ing 18 weather stations located in nine oblasts: contain complete information on the area sown Bulaevo and Saumalkol in North Kazakhstan by each type of farmer. To determine the volume (NKO); Diyevskaya and Mikhailovka in Ko- of agricultural production and average yields per stanay; Yegendykol and Schuchinsk in Akmola; hectare of the commercial farms and household Mikhailovka and Aktogay in Pavlodar; Kamen- farms, sample statistical surveys (sample size is ka and Chingirlau in West Kazakhstan (WKO); 30 and 5 percent, respectively, of all farms) are Komsomolskoe and Novoalekseevka in Aktobe; conducted at semiannual and annual intervals. Korneevka and Karaganda Agricultural Experi- mental Station in Karaganda; Dmitriyevka and 2.8. A major statistical analysis at the rayon Samarka in East Kazakhstan (EKO); and Shim- level for spring wheat was performed under kent City and Kazygurt in SKO. this World Bank study. The World Bank ana- lyzed ARKS 17-year (1994–2010) time-series 2.7. Kazakhstan has very good records for data on annual sown area, harvested area, crop spring wheat crop production. The National production, and yield for spring wheat, with a Agency of Statistics (ARKS) is responsible for breakdown into production enterprises and com- collecting, recording, and managing national mercial farms, for each of the 118 rayons located crop production data and statistics in Kazakh- in the eight oblasts selected for the analysis (Ak- stan. It is charged with recording seasonal crop mola, Kostanay, NKO, Aktobe, WKO, Pavlodar, acreage, production, and yield data for all major EKO, and Karaganda). The quality of the data food crops and also for horticultural crops in Ka- is, in general, very good. Only 4.6 percent of the zakhstan. In addition to ARKS, the Department data entries are missing. These time-series data of Statistics, under the Ministry of Agriculture enabled a series of useful analyses to be con- (MoA) and the Land Resources Management ducted. To date, preliminary analyses have been Agency, is also involved in the collection of ag- made of (a) the rayon-level, oblast-level, and na- ricultural statistics. A third important source of tional-level distribution of crop exposure (value agricultural statistics is the “household account- at risk) for spring wheat by type of farmer and ing” system, which is carried out by the regional overall, (b) the annual variation in spring wheat administrations (akimats) of townships, villages crop production and yields at the rayon level to (auls), and rural counties (rayons). There is a identify the areas of higher yield-variability risk, close interaction between all state bodies that and (c) to conduct simple correlation analysis record agricultural statistics in Kazakhstan. The with rainfall variables. Furthermore, these series data are collected from all categories of farms were used to develop the risk assessment mod- engaged in agricultural production and services. els at the rayon level to assess the risk exposure Among them there are agricultural enterprises for spring wheat as well as to establish expected (production enterprises), peasant farms (also yields and illustrative premium rates for (a) an called commercial farms), and household farms. individual grower MPCI program for spring Statistical observations are conducted on the wheat (see chapter 4 for further details) and (b) basis of the Statistical Register of Agricultural an area-yield index insurance (AYII) program Organizations, which includes all of these cat- for spring wheat (see chapter 5). egories of agricultural producers. Production enterprises are required to complete and submit 2.9. In Kazakhstan, there is no systematic a general statistical survey on a monthly, quar- monitoring and recording of loss or damage to terly, and annual basis. To obtain data on com- spring wheat production arising from natural mercial farms and household farms, general perils, including floods and droughts. In some farm accounting surveys (household accounting countries, public sector organizations (minis- books and commercial farm accounting books) tries of agriculture or the agencies responsible Agricultural Insurance Feasibility Study  51 for natural disaster management) systematically cause of loss; therefore, there is no national da- record crop damage (area damaged and percent- tabase of historical damages by cause of loss in age loss of crop production and yields) arising crops. Without such a database, it was not pos- out of major natural or climatic events and the sible to use crop damage data in this report. In cause of loss; these estimates of damages often order to overcome this problem, the World Bank are used to determine compensation payments team sought to establish the relationship be- for farmers and rural households in the affected tween shortfalls in spring wheat yield and rain- areas. Time-series crop damage data are there- fall, which made it possible to identify drought fore very useful for analyzing the frequency and years. However, it was not possible to identify severity of major events. In Kazakhstan, neither years with losses due to other perils such as hail the MoA nor the Ministry of Emergency Situ- or freeze. ations 9 systematically records crop damage by Climate and Agro-Ecological Regions 9  According the Ministry of Emergency Situations (written communiqué, March 5, 2011), between 1996 2.10. Kazakhstan experiences a marked conti- and 2010 no government financial disaster relief as- nental and dry climate. Kazakhstan experiences sistance was provided to farmers affected by natural a continental climate, with long cold winters disasters in Kazakhstan. Map 2.2 Monthly Rainfall Distribution for Selected Oblasts in Kazakhstan Source: Authors based on KHM rainfall data. 52  Kazakhstan and very short hot summers and a short grow- dry season during the months of June, July, and ing season that varies from as little as 105 days August. In southern Kazakhstan, the total annual in the north to 165 days in the south (Nomura rainfall is higher than in northern Kazakhstan: 2008). Seasonal temperatures are polarized and for example, in SKO the average annual rainfall vary depending on the region. Average winter for the period 1990 to 2010 was 556 millime- temperatures during the day are −16°C to −18°C ters. Although precipitation is higher in southern in the far north and about −6°C in the south; Kazakhstan on account of very high evapotrans- summer temperatures average 21°C in the north piration levels, most agriculture is dependent on and 27°C in the south. Snow starts to fall around irrigation. Map 2.2 shows the monthly distribu- November, and the mountain passes are snow- tion of rainfall for eight oblasts in Kazakhstan. bound until April and sometimes even into May. Precipitation in the spring wheat crop production 2.11. In Kazakhstan, drought events are deter- areas is very low. Total annual rainfall in north- mined by two important patterns of atmospher- ern Kazakhstan averages from 280 millimeters in ic circulation that affect interannual rainfall Aktobe to 400 millimeters in NKO. The rainfall variability. The so-called “Azores high” affects in northern Kazakhstan is distributed throughout rather homogeneously the entire country by gen- the year, with a peak during the months of June, erating anticyclones (usually associated with dry July, and August. In southern Kazakhstan, rain- conditions) that move from west to east, while fall is also distributed throughout the year, but the so-called “Siberian low” generates anoma- with peak rainfall in winter and a pronounced lies of opposite signs in the east and west of Map 2.3 Agro-Ecological Zones in Kazakhstan Source: Arka Consulting from KHM Agricultural Insurance Feasibility Study  53 Kazakhstan. A consequence of the interplay of resources for agricultural production. The these patterns of atmospheric circulation is that country has 76.5 million hectares of agricultural western and central Kazakhstan are usually dry, land. According to the 2006 agricultural cen- whereas more favorable conditions for agricul- sus, 61 percent of the agricultural land in Ka- ture are found in the northeast. 10 How wide- zakhstan is permanent pasture, and 32 percent spread or localized a drought event may be is is arable land (systematically cultivated for the influenced by the interaction of such circulation production of row crops). Of the remainder, 3 patterns. percent is used for hay production and 4 percent is “long-term fallow” (indicating potentially ar- 2.12. The country is divided into 16 agro- able land that has remained uncultivated for at ecological zones according to temperature and least several consecutive years). Of the 24 mil- availability of water for plant growth. The flat lion hectares of arable land, about two-thirds, areas of the country can be divided according approximately 18 million hectares, are devoted to the cumulative temperature suitable for grow- to grain production. Total sown area, including ing crops and the humidity factor measured by grains, forage crops (mostly perennial grasses), the hydrothermal ratio (HTR). Under this crite- technical crops (chiefly oilseeds and cotton), and rion, the territory of Kazakhstan can be divided food crops (potatoes, vegetables, and melons) into nine agro-climatic zones: from moderately decreased sharply during the late 1990s due to warm (zone I) to very dry and hot (zone IX). The the contraction of grain and forage-crop areas hilly and mountainous areas of Kazakhstan can (USDA, FAS 2010). be divided into seven additional natural land- scape zones (zones X to XVI). Temperature is 2.14. Kazakhstan privatized its agricultural higher in the south than in the north, with cu- land in 2003, and today three types of legal en- mulative temperatures of 4,000оС in the south tities are recognized. The first are agricultural and 2,000оС in the north. Soil moisture content enterprises (also termed production enterpris- also varies from north to south. During the warm es), many of which are former state collective summer months, HTRs vary from 0.2 in the farms that have been privatized. They include south to as much as 1.1 in the north (map 2.2). joint-stock companies, limited-liability partner- An HTR of 1.0–1.3 indicates a wet zone (forest ships, and cooperatives and are located mainly steppe), 0.7–1.0 indicates an arid zone (steppe), in northern and central Kazakhstan. They are 0.5–0.7 indicates a very arid zone (dry steppe), typically large-scale commercial grain-produc- 0.3–0.5 indicates a dry zone (semidesert), less ing companies. The second are peasant or indi- than 0.3 indicates a very dry zone (desert). Map vidual farms (also termed commercial farms), 2.3 summarizes the agro-ecological zones of which are generally less than 1,000 hectares and Kazakhstan. involved in commercial crop and livestock pro- duction. The third are household plots, which Overview of Spring Wheat Crop are not registered and consist of small family vegetable plots and livestock holdings that pro- Production in Kazakhstan duce mainly for self-consumption. As shown in table 2.1, in 2004 there were 2.2 million farms 2.13. Kazakhstan is well endowed with land in Kazakhstan, the bulk of which, 93 percent, were household plots accounting for less than 1 percent of all arable land, but contributing to 10  “Anticyclone” conditions imply subsidence of 50 percent of the value of agricultural output. At moist air from top layers in the atmosphere—hence the other extreme, in 2004, the 4,600 production less condensation (dry conditions). “Cyclone” condi- tions imply convergence of moist air from surface at- enterprises controlled nearly 13 million hect- mospheric layers—hence more cloud formation (wet ares or 59 percent of the total arable area, with conditions). an average size of farm of slightly greater than 54  Kazakhstan 2,800 hectares and with some farms as large as 2 million tons of barley and a small amount of 400,000 hectares. Between 2004 and 2007, the oats, corn, and rice, but wheat is by far the coun- total number of farms increased to 2.4 million. try’s most important commodity. The production In spite of privatization in 2003, a freely func- of oilseeds (sunflower seed and rapeseed) is in- tioning land market has been slow to develop, creasing, but total oilseed output remains well and much of the former state-owned arable land below 1.0 million tons per year. The country has not been privatized and instead is leased to also grows a small amount of cotton in southern private corporate farms under 49-year leases Kazakhstan, with annual lint output at around (World Bank 2010; Nomura 2008). 100,000 tons (USDA, FAS 2010). 2.15. Kazakhstan is an important producer 2.16. The main spring wheat crop production and exporter of high-quality wheat. Average areas are situated in northern Kazakhstan. annual production is about 13 million tons, but About 83 percent of the country’s spring wheat output is highly dependent on weather and in is produced in four oblasts located in north- recent years has fluctuated between 10 million central Kazakhstan: Kostanay, Akmola, Pavlo- and 17 million tons per year. Between 2 million dar, and NKO; 3 percent of the country’s wheat and 8 million tons are exported annually, mainly is produced in EKO, 4 percent is produced in to destinations in Europe (including the Russian Karaganda, and 7 percent is produced in western Federation and Ukraine), northern Africa, and Kazakhstan (Aktobe and WKO). Spring wheat Central Asia). Kazakhstan also produces around occupies 95 percent of the total wheat area in Table 2.1 Farm Ownership Structure in Kazakhstan, 2004 (or 2007) Commercial agriculture Subsistence Production Commercial Household plots Total Item enterprises farms (individual (individual (multiple ownership) ownership ownership) Number % Number % Number % Number % Number of farms 7,340 0.3 194,550 8.1 2,206,870 91.6 2,408,760 100 (2007) Number of farms 4,600 0.2 156,000 7.2 2,000,000 92.6 2,160,600 100 (2004) Labor force 326 14 280 12 1,782 75 2,388 100 (thousands) Agricultural land (thousands of 43,420 56 34,228 44 325 0.4 77,973 100 hectares) Arable land (thousands of 12,921 59 8,816 40 231 1 21,968 100 hectares) Average arable farm 2,809 57 0.1 10 size (hectares) Gross agricultural 171 24 178 26 349 50 698 100 output (KZT billions) Source: World Bank 2010a; Nomura 2008. Agricultural Insurance Feasibility Study  55 Kazakhstan and virtually all of the wheat in the 2.17. In northern Kazakhstan, on account of four north-central oblasts. Minor grains include the extreme winter climate, all wheat is spring spring barley and oats (which are grown in the sown and depends on a combination of snow- same region as spring wheat), winter wheat melt and summer rainfall. Planting of spring (southern Kazakhstan), and rice (southern Ka- wheat commences in mid-May, once the winter zakhstan, mostly in Kzylorda). Out of the total snows have melted and average soil tempera- area planted with spring wheat in northern Ka- tures have achieved the minimum temperatures zakhstan, 68 percent is planted by large agribusi- (12˚C–15˚C) for seed germination and crop ness enterprises (farms with more than 1,000 growth. In northern Kazakhstan there is a very hectares of wheat), while the remaining 32 per- narrow fortnight’s window for sowing spring cent is planted by commercial farms (farms with wheat of between May 15 and May 30, accord- between 100 hectares and 1,000 hectares). Map ing to the region. Wheat that is planted beyond 2.4 shows the area planted with spring wheat (in the end of May is exposed to early autumn frosts hectares) in the eight oblasts of northern and from the beginning of September. Given the central Kazakhstan at the rayon level. The most very low spring and summer average rainfall in important spring wheat rayons—with between much of northern Kazakhstan, the level of win- 320,000 hectares and 640,000 hectares of spring ter snowfall and thus snowmelt at the time of wheat—are located in Kostanay, NKO, and Ak- sowing is a critical factor in determining the suc- mola. Conversely in many rayons in EKO and cess of the spring wheat crop: indeed, farmers WKO, the area planted with spring wheat is less and scientists who were spoken to as part of this than 2,500 hectares. study advised that they could predict in April or Map 2.4 Area Planted with Spring Wheat at the Rayon Level in Northern Kazakhstan Source: Authors from ARKS data. 56  Kazakhstan May whether the harvest would be successful, tract Corporation, a state-owned enterprise, is according to the quantity of accumulated winter a major player in the Kazakhstan grain market snowfall. The main varieties of spring wheat are and was established by government to maintain 90-day to 110-day varieties, and harvesting nor- grain reserves and to stabilize grain prices for mally starts in late August and runs through to the benefit of producers and consumers. Annu- mid-September. ally it purchases some 10 to 15 percent of the wheat market and appears to have a strong influ- 2.18. Most spring wheat in Kazakhstan is ence on farm-gate wheat prices paid by the rest grown under extensive farming systems us- of the market. In view of the wide fluctuations ing low levels of technology and requiring low in wheat prices, returns to wheat production are production costs. Spring wheat production is highly variable: in times of high wheat prices, mechanized, using technical practices that date farmers who incur low average costs of produc- back to Soviet Union times. Technology levels tion can make high profits if they achieve aver- are generally low, and the use of chemical fer- age yields of about 10 centners (1 metric ton) tilizers is extremely low for most farmers. In per hectare.12 order to produce 1 metric ton of spring wheat, recommended fertilizer rates are on the order of 2.20. Spring wheat average yields in Kazakh- 35–45 kilograms of nitrogen, 8–12 kilograms stan are low. On account of the marginal climate, of phosphorous, and 17–27 kilograms of potas- long-term soil degradation, and low technology sium. However, average fertilizer use is much use, average yields for spring wheat were about lower than recommended. In general, produc- 10.1 centners (1.1 metric tons) per hectare over tion enterprises have better access to production the past five years. This compares unfavorably credit and use higher levels of purchased inputs with, for instance, Canada, which has similar of seed, fertilizers, and equipment than commer- climatic conditions and where yields reach 27 cial farms. In 2010 average costs of production centners (2.7 metric tons) per hectare, or Aus- for wheat varied widely, from a low of about tralia, which has a similar extensive crop system KZT 5,000 (US$35) per hectare11 for the lowest- and harvests an average of about 18 centners (1.8 technology producers to a high of about KZT metric tons) per hectare of wheat. Spring wheat 20,000 (US$140) per hectare and occasionally yield performance is more variable for com- as high as KZT 25,000 (US$170) per hectare for mercial farms than for larger production enter- the highest-technology producers. prises. While the spring wheat average yield for commercial farms was 9.6 centners per hectare 2.19. Wheat prices in Kazakhstan tend to be for the most recent five years, the spring wheat very volatile. Average farm-gate prices paid average yield for production enterprises was 7.5 to farmers between 2006 and 2010 varied sig- percent higher, or 10.3 centners per hectare, for nificantly: from a low price of KZT 12,600 the same period. The main reason that produc- (US$104) per metric ton in September 2006 to tion enterprises achieve higher average yields as high as KZT 31,000 (US$261) per metric ton for spring wheat than commercial farms is that in September 2008, which was a severe drought they use improved seeds and fertilizers and im- year when total production was significantly re- proved tillage practices. Spring wheat yield per- duced. Prices averaged KZT 25,000 (US$172) formance is also uneven throughout the north- per metric ton in September 2009, which was a ern areas of the country. The best performance good year for spring wheat, and rose sharply to for spring wheat crops is in the rayons situated about KZT 35,000 per metric ton in 2010, which in the northern areas of NKO and Kostanay. was a very severe drought year. The Food Con- In these areas, average spring wheat yields are 11  At a 2010 exchange rate of KZT 145 = US$1.00 12  1 centner is equal to 100 kilograms. Agricultural Insurance Feasibility Study  57 above 12 centners per hectare. The worst per- centers per hectare. Crop management practices formance for spring wheat crops is observed in fueled by expanding state subsidies contributed WKO, Aktobe,13 southern regions of Karaganda, to higher and more stable wheat yields. Begin- and southwest of Pavlodar. In these areas, spring ning in 2002, government support for agriculture wheat rayon-level five-year average yields are increased significantly in the form of reduced below 6 centners per hectare. Map 2.5 summa- (subsidized) prices for fuel, seeds, fertilizer, and rizes the geographic distribution of spring wheat agricultural chemicals. The average wheat yield yields at the rayon level throughout the eight for 2005 through 2009 was 13 percent higher selected oblasts in northern Kazakhstan. Further than the average yield for 1986 through 1990, details of this analysis are provided in annex 1. which was the peak of the so-called intensive technology movement in the Soviet Union. One 2.21. Spring wheat crop production and yields of the most interesting recent developments in improved significantly during the last decade. Kazakhstan agriculture has been the introduc- tion and spread of reduced-tillage technology. Average spring wheat yields in the main produc- According to MoA figures, reduced tillage was tion areas in northern Kazakhstan reached 10.1 employed on almost 60 percent of the sown grain area in 2009, including 1.3 million hectares un- der zero tillage. The sector also increased the use 13  Average spring wheat yields are somewhat high- of fertilizers and certified seeds. The application er in two of the most westerly rayons of WKO, and this also applies to two rayons located in northern rates for mineral fertilizer increased nearly six- Aktobe. fold between 1999 and 2010 and continue to in- Map 2.5 Average Spring Wheat Yields per Rayon in Northern Kazakhstan, 2006–10 Source: Authors based on ARKS spring wheat yield data. 58  Kazakhstan crease due, in part, to the subsidies on fertilizer efficient of variation (CoV), in national average prices. Perhaps the most important technologi- spring wheat yields: 29 percent, compared to 5 cal factor contributing to the improvement in percent in the European Union and 8 percent in Kazakhstan grain yield has been the increase in Canada.14 The northernmost areas of the country the use of certified seeds. The government has are less risky for spring wheat production than been providing support to agricultural research the southern, western, and eastern areas. In the facilities, paying 40 percent of the research and rayons situated in NKO, the north of Kostanay, development costs for breeder and foundation and the northwestern areas of Akmola, the CoV seeds. Most spring wheat enterprises use only for spring wheat yields is less than 40 percent. first-reproduction seed. However, the inventory In the rayons situated in the south of Kostanay of machinery is becoming outdated (particularly and the eastern parts of Akmola, the CoV for among commercial farmers) and has declined spring wheat yields is between 40 and 50 per- significantly over the past 20 years. Figure 2.1 shows the evolution between 1994 and 2010 in the sown area and yields of spring wheat in the eight selected oblasts in northern Kazakhstan. 14  The CoV is the standard deviation about mean annual yield divided by the mean yield and expressed 2.22. Spring wheat crop production is particu- as a percentage. A CoV of more than 100 percent shows that the standard deviation is larger than the larly risky in Kazakhstan. Kazakhstan has a very mean yield—in other words, crop yields are highly high variation in yield, as expressed by the co- variable. Figure 2.1 Spring Wheat Sown Area and Yields in Northern Kazakhstan, 1994–2010 North Kazakhstan Region: Evolution of Spring Wheat Sown Area and Yields 14.0 16.0 12.4 12.7 11.6 11.3 14.0 12.0 10.8 10.1 9.5 9.4 12.0 10.0 8.7 8.4 7.7 10.0 8.0 7.3 7.5 6.3 6.4 8.0 6.0 5.0 6.0 y = 1.6032ln(x) + 5.6195 4.0 4.0 R = 0.25203 4.0 2.0 2.0 0.0 0.0 1994 1995 1996 1997 1999 2000 2001 2002 2003 2004 2005 2006 2007 2009 2010 1998 2008 Sown Area (million hes) Yield (Centner/he) Log.(Yield (Centner/he)) Source: Authors based on ARKS data. Agricultural Insurance Feasibility Study  59 cent. In Karaganda and western areas of EKO, both higher and less variable (and therefore less the CoV for spring wheat is mostly between 50 risky) average yields in spring wheat over time and 60 percent, except for the rayons situated in than the commercial farms. The analysis of vari- the mountainous areas of EKO, where the CoV ance in spring wheat yields for the 17-year se- is between 40 and 60 percent. Pavlodar shows a ries from 1994 up to and including 2010 found high CoV for spring wheat production at the ray- that the observed CoV is, on average, 10 percent on level: on average, between 50 and 70 percent. lower for production enterprises than for com- The oblasts situated in the western areas of the mercial farms. This difference is accentuated in country show the highest level of risk for spring the main spring wheat production areas of Ko- wheat production. For instance, in WKO the av- stanay, NKO, and Akmola, where the observed erage CoV at the rayon level for spring wheat is CoV of rayon-level spring wheat yields is 16, between 70 and 100 percent. Map 2.6 summariz- es the distribution of the CoV for spring wheat 27, and 25 percent lower, respectively, for pro- at the rayon level in the eight selected oblasts in duction enterprises than for commercial farms. north-central Kazakhstan. The main reason for the differences between production enterprises and commercial farms is 2.23. Spring wheat crops produced by produc- the introduction of technologies that conserve tion enterprises are less risky than spring wheat soil moisture, such as zero tillage, which enables crops produced by commercial farms. The pro- the crop to perform better during the recurrent duction enterprises, through their use of higher droughts that effect Kazakhstan’s grain-produc- levels of inputs and technology, tend to achieve ing regions. Map 2.6 Coefficient of Variation of Spring Wheat Yields at the Rayon Level in Northern Kazakhstan Source: Authors based on ARKS spring wheat data. 60  Kazakhstan Key Climatic Perils and Impact nual average spring wheat yields and cumulated rainfall and snowfall between January and Sep- on Crop Production and Yields tember of each year that are above 0.65 (65 per- cent). Figure 2.2 shows the relationship between 2.24. In northern and central Kazakhstan, spring wheat annual average yields and the total spring wheat production and yields are highly cumulated snowfall and rainfall from January to influenced by climatic and biological factors. September for the period 1994 to 2010. Drought is the most pervasive peril affecting rain-fed crop production in northern Kazakh- 2.26. Spring wheat farmers in Kazakhstan stan. Spring wheat crops can also be damaged suffer severe losses due to drought. Between by hailstorms and autumn early frost. Pests and 1994 and 2010 crop years, spring wheat suf- diseases, mainly fungal diseases like leaf blotch fered significant crop losses on six occasions: caused by Septoria tritici and rust caused by 1995, 1996, 1997, 1998, 2004, and 2010. This Puccinia tritici, are also common. study estimated the value of the historical spring wheat production losses in each year (see the 2.25. Spring wheat yields in Kazakhstan summary in figure 2.3 and annex 1 for further are highly influenced by the occurrence of details). In 1995 a drought affected Kostanay droughts.15 Reasonably high yields can be and Karaganda, causing estimated production achieved during years with adequate rainfall, but losses in the spring wheat crop valued at KZT the country is subject to frequent droughts and is 74.4 billion (equivalent to a 26 percent reduction considered a zone of risky agriculture. Histori- in the value of production).17 In 1996 a severe cally, Kazakhstan grain production suffers from drought affected the western part of the country serious drought two out of every five years. As a (WKO and Aktobe), and a moderate drought af- result, crop production and yields are marked by fected the eastern oblasts (Pavlodar and EKO). frequent and sharp year-to-year fluctuations. The The total losses due to the 1996 event amounted aggregate annual average spring wheat yields to KZT 46.4 billion (or a 15 percent reduction for the eight selected oblasts in the northern and in the value of production). In 1997 the spring central regions of Kazakhstan are highly corre- production areas in Kazakhstan were affected lated with the cumulated rainfall and snowfall by drought again. On this occasion, the reduc- index from January to September, as shown by tion in the total value of production was only 4 the overall correlation coefficient (aggregate for percent of the expected value for that year. The the eight oblasts) of 73 percent.16 This strong re- year 1998 was one of the worst years for drought lationship between spring wheat average yields damage in spring wheat crop production. The and cumulated rainfall between January and 1998 drought was particularly severe in WKO, September is also evidenced at the oblast level. Aktobe, Kostanay, and Akmola, but also affect- In this regard, all of the selected oblasts in north- ed NKO, Pavlodar, and Karaganda. The event ern and central Kazakhstan, except NKO and caused a 51 percent reduction in total expected Karaganda, show correlation coefficients for an- value of spring wheat crop production (a loss of KZT 179.4 billion). The years 2004 and 2005 were also dry. The total estimated value of losses amounted to KZT 74 billion (18 percent) in 2004 15  For a more detailed indication of how the cor- and KZT 49.1 billion (12 percent) in 2005. The relation coefficients were determined, see the note to figure 2.2. 16  The correlation coefficient or R value in this case is interpreted to mean that 73 percent of the variation 17  Losses calculated in terms of gross value of pro- in oblast annual yields is explained by the variation duction lost due to yield shortfalls in respect of the in total January to September rainfall each year. This expected yield for each of the years, assuming sown means that the remaining 27 percent of yield varia- area and prices are equal to the most recent five-year tion is due to variables other than rainfall. average. Agricultural Insurance Feasibility Study  61 Figure 2.2 Spring Wheat Yields and Total Cumulated Snowfall or Rainfall in Northern and Central Kazakhstan from January to September, 1994–2010 Source: Authors from KHM and ARKS data. Note: Correlations are computed by (a) cumulating for the January–September period precipitation measured at each of the weather stations for which monthly data were made available by the KHM, (b) taking the simple average of each pair of stations belonging to the same oblast, and (c) relating the cumulated precipitation index with oblast-level spring wheat yield data. The northern-central Kazakhstan rainfall index is calculated by aggregating the oblast cumulative precipitation indexes as a share of each of the oblasts on the average spring wheat planted area for the period 2006–10. 62  Kazakhstan Figure 2.3 Gross Value of Production of Spring Wheat Lost due to Droughts in Kazakhstan, 1994–2010 North Kazakhstan : Evolution of Spring Wheat Gross Value of Production (GVP) and Yields 600 14.0 500 12.0 Yield (Centner/hectare) 10.0 GVP (KZT billion) 400 8.0 300 6.0 200 4.0 100 2.0 0 0.0 1994 1995 1996 1997 1999 2000 2001 2002 2003 2004 2005 2006 2007 2009 2010 1998 2008 Crop Year Actual GVP (KZT billion) Actual Yield (Centner/he) GVP Losses (KZT billion) Expected Yield (Centner/he)/ Expected GVP (KZT billion) Source: Authors from ARKS data. dry conditions in Pavlodar and eastern Kazakh- tions. For example, at Kostanay weather station, stan during 2008 also caused losses in spring the return period for a hail event in the month of crop production. Most recently, in 2010, a dev- May is one in three years; in the month of Au- astating drought affected the main spring wheat gust, it is one in eight years (table 2.2). Hail is a crop production areas throughout much of the localized peril that tends to cause severe damage country (Kostanay, Akmola, Karaganda, NKO, in wheat at the time of crop maturity and harvest and Pavlodar), causing losses amounting to KZT in August and early September; these months 158.5 billion, equivalent to a shortfall of 58 per- tend to have a lower hail exposure. cent compared with the expected gross value of spring wheat production of KZT 274.4 billion. 2.28. Early autumn frost can be a problem Figure 2.3 shows the spring wheat crop produc- for spring wheat crop production in some ar- tion losses from 1994 to 2010. eas of northern Kazakhstan. The occurrence of early autumn frosts during late August and 2.27. Hail is reported to be a moderate to se- the beginning of September may damage wheat vere problem in spring wheat in some parts of crops that were sown late in the season. Early the country. Many parts of Kazakhstan experi- frost damage affects wheat crops when they are ence hail in early and mid-summer associated in milk grain or dough phenology stages prior to with major rainstorms: the months of peak hail harvest. The damage occurs when temperatures activity are May to July, as shown by the data on fall below −2°C for more than two hours. Loss- monthly hail incidence for selected weather sta- es due to early autumn frost can be moderate to Agricultural Insurance Feasibility Study  63 Table 2.2 Monthly Return Period for the Occurrence of Hailstorms in Kazakhstan number of years Recurrence period for hailstorms for each month throughout the year Oblast I II III IV V VI VII VIII IX X XI XII Kostanay 42 5 3 6 8 21 84 Akmola 5 8 5 13 63 NKO 7 5 11 5 7 Pavlodar 21 8 9 7 30 10 63 SKO 11 4 3 8 11 21 42 42 42 Source: KHM hail frequency statistics for 1990–2010. severe. Historical records of monthly absolute of up to one-third of the former wheat lands. The minimum temperatures indicate that the return resulting mosaic of weedy fields, pastures, and period for frost below −2°C during the last week bare ground provides ideal breeding grounds for of September is once in 20 years for some loca- locusts. In 1999 more than 7 million hectares tions in Akmola, Karaganda, Kostanay, and Pav- were invaded by Italian locusts, and 220,000 lodar. The probability of having an early frost hectares of crops were destroyed, causing total increases dramatically for each week beyond the damage equivalent to US$15 million. In 2008, last week of September. again, more than 200,000 hectares of crops were destroyed in SKO (Latchininsky and Sivanpillai, 2.29. Spring wheat is susceptible to locust at- 2010). The government was forced to conduct a tacks in northern Kazakhstan. Locust attacks are massive chemical control campaign throughout relatively common in northern Kazakhstan, and the country with assistance from the Food and the country has suffered recurrent crop and pas- Agriculture Organization. ture damages from locusts. There are two main species of locusts in Kazakhstan: (a) the “Asian” 2.30. Spring wheat fungal leaf diseases are locust, which is not considered a major problem, also a problem for spring wheat production in and (b) the Italian locust (Callitamus italicus), Kazakhstan. Despite the dry climate, cultivation which is the most common and dangerous pest of susceptible varieties results in epidemics of in nearly all of Kazakhstan. In 1999 the country leaf rust on average in one year out of four, af- experienced a severe outbreak of Italian locusts. fecting over 1 million hectares, with yield losses This species has a peak cycle of 10 to 12 years, so of up to 25–30 percent. Excess moisture and the next outbreak is expected in 2010 or 2011.18 high humidity during the month of July gener- The scale of the problem increased dramatically ate conditions conducive to the development of after independence (during 1996–2001), when rust. Most of the wheat cultivars planted in the cessation of state subsidies for wheat production region are susceptible to leaf rust, although sev- in the northern steppes led to the abandonment eral resistant lines and new varieties have been tested in recent trials (Morgounov, Rosseeva, and Koyshibayev 2007). Stripe rust, caused by 18  “Kazakhstan: Locust Invasion in West under Puccinia striiformis f. sp. tritici, is considered Control, Officials Say.” IRIN Asia, July 10, 2007. http:// the most important disease of wheat in Central www.irinnews.org/report.aspx?reportid=73155. Asia and the Caucasus. Although stripe rust has 64  Kazakhstan been present in the region for a long time, it has spring wheat production in Kazakhstan is con- become a serious constraint to wheat production centrated in a relatively small area. Out of the in the past 10 years. Rust attacks were observed KZT 452 billion in total spring wheat VaR, 87 in northern areas of Kazakhstan in 2007 and in percent (KZT 392 billion) is concentrated in a 2009. relatively small area of approximately 240,000 square kilometers comprising NKO, the north- Assessment of Crop ern rayons of Kostanay, and the northern and western rayons of Akmola. The remaining 13 Production Risk Exposures percent of spring wheat VaR, or KZT 56 bil- lion, is distributed throughout a vast area that Spring Wheat Values at Risk comprises EKO, Pavlodar, Karaganda, Aktobe, WKO, rayons located in the south of Kostanay, 2.31. The total spring wheat values at risk and rayons located in the east of Akmola. Map (VaR) in northern Kazakhstan are KZT 452 2.7 summarizes the geographic distribution of billion (about US$3.0 billion).19 The bulk of spring wheat VaRs at the rayon level throughout the eight selected oblasts in northern Kazakh- stan. Further details of this analysis are provided 19  For the purpose of evaluating spring wheat pro- in annex 1. duction, an average price of KZT 3,210 per centner was considered. This price is the result of the aver- age spring wheat farm-gate price for the month of 2.32. Because 87 percent of spring wheat risk harvest (September) for the three most recent crop exposure is concentrated in a relatively small seasons (2008, 2009, and 2010). geographic area of northern Kazakhstan, the Map 2.7 Spring Wheat Risk Exposures in Northern Kazakhstan Source: Authors from ARKS data. Agricultural Insurance Feasibility Study  65 chances of experiencing catastrophic losses in Expected Value of Spring Wheat Crop spring wheat crop production are high. There Losses is a huge accumulation of risk in the area com- posed of NKO, the northern rayons of Kostanay, 2.34. The expected value of losses for spring and the northern and western rayons of Akmo- wheat crops at the rayon level for each of the la, and the chances of experiencing a systemic eight selected oblasts in northern Kazakhstan event affecting the main part of the spring wheat was estimated. The estimation was based on an crop portfolio are high. analysis of variance in time-series average wheat yields in each zone under the CRAM model (see 2.33. The major spatial differences in spring annex 1 for full details of the model’s assump- tions). wheat VaRs will need to be addressed carefully in the redesign and strengthening of the obliga- 2.35. The analysis of expected losses shows tory crop scheme for spring wheat production. that spring wheat production in Kazakhstan The purpose of any crop insurance scheme is to is extremely risky, with annual average ex- spread risk optimally both spatially and tempo- pected losses valued at KZT 66.5 billion (about rally. The concentration of VaRs in NKO, the US$443 million). The annual average expected northern rayons of Kostanay, and the northern losses for spring wheat are valued at KZT 66.5 and western rayons of Akmola will need to be billion per crop year (US$443 million), equiva- assessed closely under the redesign and strength- lent to 14.7 percent of the total spring wheat VaR ening of the exisiting obligatory crop insurance of KZT 452 billion. However, within northern scheme. Kazakhstan, some oblasts and rayons are more Map 2.8 Expected Losses for Spring Wheat in Northern Kazakhstan Source: Authors from CRAM. 66  Kazakhstan risky than others. Map 2.8 shows the average ex- stan than the annual expected losses for spring pected losses for spring wheat crop production, wheat produced by commercial farms. While expressed as a percentage of the VaR in each of the annual average expected losses for spring the rayons in northern Kazakhstan. wheat produced by commercial farms are 17.24 percent of the total spring wheat VaR produced 2.36. The highest average annual expected by them, the average expected losses for spring losses in spring wheat are in Aktobe and WKO wheat produced by production enterprises are located in western Kazakhstan. Within these only 13.39 percent. The main reason for these two oblasts, only two rayons in western WKO differences is that production enterprises man- and two rayons in northern Aktobe have annual age their production risks better than commer- average expected losses of less than 25 percent. cial farms (they have more working capital and In all other rayons, the average annual expect- better machinery) and use soil moisture conser- ed losses for spring wheat are extremely high, vation technologies (such as zero tillage). The at between 25 percent to more than 38 percent fact that the expected losses of spring wheat of the total values at risk. These annual average crops are lower for production enterprises than expected losses are equivalent to a total loss of for commercial farms is an important finding spring wheat production one in every three to that should be taken into consideration in the de- four years. This indicates that the western zone sign of any individual grower loss of yield crop of Kazakhstan is very marginal for spring wheat insurance scheme. The eventual modification crop production. of the current scheme should recognize the risk management efforts implemented by the differ- 2.37. Conversely, spring wheat production ent types of farms insured. is much less risky in the northern oblasts of NKO, Kostanay, and Akmole. In rayons such 2.39. The analysis of 17-year (1994–2010) as those situated in NKO, the northeastern and rayon-level yields for spring wheat in NKO southern areas of Akmola, and the northern area shows that 1998 was the worst loss year in this of Kostanay, the annual average expected loss series, with total production losses of 7.7 mil- for spring wheat ranges between 10 to 15 per- lion metric tons of spring wheat, equivalent cent of total VaR. The rayons of Pavlodar and to a financial loss of 41.6 percent of the total the western areas of EKO can be considered as expected value of spring crops in northern Ka- intermediate with regard to the risks for spring zakhstan. Although 1998 was a year of severe wheat production. In these rayons, the average loss for spring wheat production in northern expected losses range between 18 and 27 per- Kazakhstan, even worse crop losses could oc- cent. Finally, a vast geographic area compris- cur in the future. From an insurance viewpoint, ing the center and south of Kostanay, the center underwriters need to know with a high degree and east of Akmola, most of Karaganda, and the of confidence the maximum losses that they eastern rayons of EKO has acceptable levels of might incur (termed the probable maximum expected losses. In these areas, the average ex- loss, PML)20 either one in 100 years or, to be pected loss per hectare averages from 15 to 18 even more conservative, one in 250 years. This percent of the total VaR. information is an invaluable aid to structuring an insurance and reinsurance program and to 2.38. The analysis of expected losses also found that spring wheat produced by com- mercial farms is much more risky than spring 20  The probable maximum loss is defined as “an wheat produced by production enterprises. An- estimate of the maximum loss that is likely to arise on the occurrence of a single event considered to nual expected losses for spring wheat produced be within the realms of probability, remote coinci- by production enterprises are 22 percent lower, dences and possible but unlikely catastrophes being on average, for the whole of northern Kazakh- ignored.” Agricultural Insurance Feasibility Study  67 determining how much capital must be reserved in the design of a risk financing and risk reten- to cover the PML year. tion and risk transfer or reinsurance strategy for the Kazakhstan insurance market. 2.40. The World Bank’s PML cost analysis at 100 percent yield coverage found, for northern 2.41. The PML cost analysis at 100 percent Kazakhstan, the 1-in-10-year expected PML is yield coverage found that commercial farms equivalent to a loss of 34.0 percent of the total are more exposed than production enterprises value at risk for spring crop or a loss of KZT to catastrophic losses in northern Kazakhstan. 153.6 billion (US$1.02 billion), while the 1-in- The 1-in-100-year expected PML is equivalent 100-year PML is equivalent to a loss of 54.61 to a loss of 58.0 percent of VaR for commercial percent of the total VaR for spring crop or a loss farms and a loss of 55.7 percent of VaR for pro- of KZT 246.8 billion (US$1.6 billion). These duction enterprises. PML estimates show that the spring wheat crop in northern Kazakhstan is very exposed to cata- Conclusions of the Spring Wheat Risk strophic (mainly drought) losses and that these Assessment losses greatly exceed the retention capability of local insurance companies (figure 2.4). The 2.42. The analysis of rayon-level crop produc- PML presented in this report is preliminary and tion and yields for spring wheat in northern will need to be developed further to (a) establish Kazakhstan shows that this crop is heavily ex- the catastrophe loading that must prudently be posed to losses caused by droughts. This is evi- added to the calculated base rates and (b) assist denced by the average loss cost estimated by the Figure 2.4 Probable Maximum Losses for the Spring Wheat Portfolio in Northern Kazakhstan 70% Percentage of Loss over the total VAR 60% 61% - 277 billion 60% - KZT 271 billion 57% - KZT 263 billion 50% 54% - KZT 247 billion 50% - KZT!225 billion 40% 34% - KZT 154 billion 30% 20% 10% 0% 0 50 100 150 200 250 Return Period (years) Source: Authors from CRAM. 68  Kazakhstan CRAM for a 17-year period, 1994 up to 2010, of should also take into account the differences 14.71 percent of the total value at risk in spring in expected yields and risk between production wheat production and a calculated 1-in-100-year enterprises and commercial farms. In Kazakh- PML of 54 percent of the national gross value of stan there are important differences in the level spring wheat production. of technology used by different types of farmers to grow spring wheat. Even when they are situ- 2.43. The design of any crop insurance pro- ated in the same region, their yields are different, gram for spring wheat in northern Kazakhstan and their crops perform differently under similar should take into account the differences in ex- crop stress (for example, drought) situations. In pected yields and yield variability by rayon. The the case of spring wheat production in northern analysis of expected yields and expected yield Kazakhstan, production enterprises perform variability for the different rayons in north- better than commercial farms, in terms of both ern Kazakhstan showed different levels of risk quantity and stability of yield. The reasons for for spring wheat and, therefore, different rates these differences in the spring wheat yields of across the different rayons in the region. production enterprises and commercial farms should be taken into consideration when design- 2.44. The design of any crop insurance pro- ing future crop insurance products and schemes gram for spring wheat in northern Kazakhstan in Kazakhstan. Agricultural Insurance Feasibility Study  69 Chapter 3: Review of Kazakhstan Crop Insurance Program 3.1. This chapter reviews the compulsory (MoA), the Fund for Financial Support for Ag- crop insurance program, including its key fea- riculture (FFSA), the private and mutual insur- tures, results, and financial performance, and ance companies, and the local authorities in each highlights its issues and drawbacks. Chapter oblast and rayon. 4 considers options for strengthening the pro- gram. 3.4. The crop insurance law in Kazakhstan is aimed at meeting three objectives. The first Policy and Regulatory objective is to protect farmers against loss of Framework for Crop Insurance their crop production due to the effects of ad- verse weather events. Kazakhstan is one of the most exposed countries in the world to losses 3.2. Kazakhstan has a history of state-sup- in spring wheat caused by drought, followed ported compulsory agricultural insurance. by late spring and harvest frosts, hail, and, to a During the Soviet period, a national compulsory crop insurance scheme operated from 1970 to lesser extent, excessive rain and flooding as well 1991, and all state and collective farms partici- as crop pests and diseases. The Government pated in it. The scheme was based on a multiple- of the Republic of Kazakhstan (GRK) seeks to peril crop insurance (MPCI) cover that provided protect small and medium farmers from bank- limited indemnity for loss of production costs ruptcy following major crop losses by provid- invested in growing the crop. Following inde- ing subsidized crop insurance for the main crops pendence in 1991, there was no crop insurance grown in Kazakhstan. The second objective is to in Kazakhstan for the next decade. assist farmers in gaining access to rural finance by protecting their crop loans against default due 3.3. The current crop insurance system in to weather-induced crop failure. The third objec- Kazakhstan is a public-private partnership tive is to enhance the effectiveness of govern- (PPP) that was created by law in 2004. Law no. ment support programs for crop production. 533-II of March 10, 2004, established the legal and regulatory framework and the financial and 3.5. The current crop insurance system in Ka- organizational basis for implementation of a na- zakhstan is compulsory by law. The law states tional crop insurance program in Kazakhstan. that all farmers in Kazakhstan who produce a A copy of the law is attached in annex 2. The series of strategic crops, including cereals, oil- law sets out the terms and conditions for imple- seeds, sugar beets, and cotton, are obliged to mentation of a compulsory salvage-based loss of purchase crop insurance; otherwise, they cannot investment costs (LIC) crop insurance program obtain access to other government-subsidized for farmers growing a range of strategic cereal, programs supporting agriculture and are sub- oilseed, and other field crops in Kazakhstan. ject to the application of financial penalties by The implementing agencies are the Direction of the GRK. Under the compulsory crop insurance Strategic Planning of the Ministry of Agriculture program, all private commercial insurance com- 70  Kazakhstan panies and farmer mutual associations licensed policy is sometimes referred to as a salvage- to operate crop insurance are equally obliged to based loss of yield policy because it only indem- offer crop insurance and to insure all of the insur- nifies crop production losses at the point when ance proposals received regardless of the quality the sales value (revenue) of any residual har- of the risk. The only reason that an insurer may vestable crop production (termed the “salvage”) refuse to conclude a compulsory contract of in- is inadequate to cover the costs of production surance cover is if the insured farmer has failed invested in growing the crop up to the time of to submit his or her application and to complete loss. Alternatively, this policy is called a loss of the insurance contract, including the payment of investment costs crop insurance policy (“loss of premium, by the agreed cutoff date. investment cost” policy for short). 3.6. The Law of Compulsory Crop Insurance 3.9. A key potential advantage of the LIC is very comprehensive and specifies the terms policy is that it can be used in situations where and conditions of the standard crop insurance there are no accurate historical records of indi- policy that is offered throughout Kazakhstan. vidual farmer crop production and yield. A con- The law prescribes the terms and conditions of ventional individual grower loss of yield policy the standard LIC crop insurance policy that all requires each grower to provide data on between insurance companies are obliged to adhere to, five and 10 years of their actual crop produc- including the insured crops and insured perils, tion and yields in order to establish an average the amount of the sum insured for each crop or normal yield and then to establish an insured grown in each oblast and rayon, which is based yield against which to measure yield reduction on three optional levels of crop production in- or loss due to an insured peril. In Kazakhstan, vestment costs, and the minimum and maximum few farmers can provide an accurate history of premium rates that can be charged for each crop their crop yield. The LIC policy can, however, in each oblast in the country. The law also states be offered to individual growers in situations the basis of indemnity and loss assessment pro- where data on yield history are not available, cedures that apply on the standard LIC policy. because the basis of indemnity does not depend on measuring yield loss against a preestablished 3.7. The law specifies the GRK’s financial insured yield. The LIC policy establishes a sum support to the compulsory crop insurance pro- insured based on the production costs invested in gram. In addition to its statutory and regulatory growing the crop; in the event of a loss due to an roles in setting the terms and conditions of the insured peril(s), the policy makes an indemnity program, the GRK also provides financial sup- payment when the estimated sales value of the port in the form of a 50 percent reimbursement salvage is inadequate to cover the costs invested of paid claims to the private and mutual insur- in the crop at the time of loss. ance companies each year. The settlement of the 50 percent claims is administered through the FFSA. Under the law, the GRK is also respon- 3.10. The main drawback of the LIC policy is sible for funding the FFSA’s operating costs. that, in the event of partial crop area or produc- tion losses, the remaining or salvageable crop yield has to be measured on a field-by-field ba- Compulsory Crop Insurance sis and the value of the salvage estimated. This Policy Terms and Conditions method of loss adjustment is a time-consuming and costly exercise. 3.8. The Kazakhstan compulsory crop insur- ance policy is a loss of yield policy that indemni- 3.11. The LIC policy has been extensively pro- fies the insured when the value of the harvested moted in various countries for more than 30 production falls short of the costs invested in years. It is usually linked to bank lending in the growing the crop due to an insured peril. The form of seasonal crop loans and is found in Mex- Agricultural Insurance Feasibility Study  71 ico, in other parts of Central and South America, Compulsion of Cover and in several Eastern European countries. 3.13. Kazakhstan is one of very few coun- 3.12. This section reviews the key features of tries in the world where crop insurance cover the Kazakhstan compulsory LIC policy. A sum- is compulsory for all farmers who grow cere- mary of the main terms and conditions of the als, oilseeds, and other strategic crops. Because policy wording, which is termed the Standard crop insurance cover is legally enforceable, a Form of Compulsory Crop Insurance Contract, farmer who deliberately avoids purchasing crop is contained in box 3.1, and full details are pre- insurance is penalized by heavy fines. Currently sented in annex 2. about 100 countries offer some form of pub- Box 3.1 Summary of Terms and Conditions of Cover of the Kazakhstan Standard Compulsory LIC Crop Insurance Policy Type of cover. The underlying basis of insurance and indemnity is a salvage-based loss of yield policy. Insured interest. The main crops insured are spring and winter wheat, spring and winter barley, winter rye, buckwheat, oats, millet, maize (grain), chickpeas, peas, brassica, rice, sunflower, safflower, soybeans, sugar beets, and cotton. Location. All crop-growing regions in Kazakhstan are covered. Criteria for acceptance of risk or compulsion of cover. Crop insurance is compulsory for all producers of the above insurable crops throughout Kazakhstan. Insured perils. Cover is provided against loss or damage to crop production due to “adverse weather events,” defined as (a) long-lasting natural phenomena (drought, frost, low temperatures, excess moisture in the soil, excess moisture in the air, flooding, and shallow dry wind) and (b) short-lasting natural phenomena (hail, excessive rain, frost, strong wind, and mud flow). Cover period. Cover is provided from the time of sowing of the insured crop through to completion of harvest. Insured unit. For each farmer, the insured unit is defined as a “field.” The insured is obliged to declare and insure each and every separate field of the insurable crop(s) and to submit a map of the field locations. Sum insured. The sum insured is based on the normative costs of production for 1 hectare of the insured crop, multiplied by the insured area. The insured is permitted to select from three optional levels of normative costs of production: (a) science-based agricultural technology, (b) simplified agricultural technology, and (c) costs of fuel and lubricants, seeds, and wages. Deductible (franchise). The compulsory insurance law prohibits the use of franchises or deductibles. Basis of indemnity and claims settlement. Where the action of insured perils causes the actual value of harvested production (salvage) to fall short of the costs of production invested in growing the crop (the sum insured), the policy indemnifies the amount of shortfall. In the event of a total loss, the indemnity is the sum insured for 100 percent of the damaged area. In the case of a partial loss of area or yield, the salvage (harvestable production) from the affected area is estimated in field at the time of harvest and valued at the prevailing sales price of the crop. Where the value of salvage (crop revenue) falls short of the investment costs, the shortfall is indemnified. Exclusions. Any losses that occur due to causes other than “adverse weather events” are excluded. Other conditions. Cover is only binding from the time of payment of premium by the insured. Source: Authors based on the Standard Form of Compulsory Crop Insurance Contract (see annex 2). 72  Kazakhstan lic, private, or PPP agricultural insurance (Ma- diseases in cereals are a major cause of loss in hul and Stutley 2010), and agricultural insur- adverse climatic years in Kazakhstan. In most ance is compulsory in China (compulsory only other countries that operate similar LIC policies, for swine epidemic diseases), Cyprus (for all coverage is usually “all risk” and includes all crops), Japan (for wheat and rice only), Kazakh- natural, climatic, and biological perils (unavoid- stan (for major strategic crops, but voluntary for able and uncontrollable pests and diseases) that livestock), the Democratic People’s Republic of result in loss of expected crop production and Korea (for rice and maize), the Netherlands (for crop revenue. In Kazakhstan, it is likely that loss livestock epidemic diseases), Switzerland (for assessment is more complicated because the ad- livestock epidemic diseases), and the Windward justment needs to take into account losses due to Islands (windstorm cover for export bananas). insured adverse climatic events and to separate In all other countries, either agricultural insur- these from other uninsured causes of loss. This ance is purely voluntary or borrowers of credit theme is discussed further in chapter 4. are obliged by the lender to purchase crop credit insurance protection. The potential advantages Definition of Insured Unit and disadvantages of compulsory national crop insurance schemes are reviewed later in this 3.17. The definition of the insured unit is chapter. critical for adjusting and indemnifying losses under any crop insurance scheme. In Kazakh- Insured Crops stan, the insured unit is the separate or indi- vidual “field.” The policy requires farmers to 3.14. Under the 2004 Law on Compulsory declare and insure all of their fields sown with Crop Insurance, insurance is mandatory for a the insurable crop(s) and to provide maps and list of 17 strategically important food and ex- schedules showing the location and area of each port crops, including spring wheat and other field and crop(s). With the lack of field boundary grains, oilseeds, sugar beets, and cotton. A full fences and roads, the definition of what consti- list of the compulsory insurable crops is given tutes a separate field may, however, be very open in box 3.1. Spring wheat is the most important to interpretation at the time of loss assessment. crop, accounting for more than 90 percent of the program’s total sum insured (TSI) over the past Cover Period six years (2005 to 2010). 3.18. Most crop insurance programs provide Insured Perils protection during the growing season and ter- minate cover on completion of the harvest. 3.15. The compulsory LIC policy insures a There are, however, a great many variants on the wide range of perils. The insured perils include a inception dates of cover, and some MPCI loss broad range of climatic perils, including drought, of yield programs (for example, in Spain and winter freeze, frost, low temperature, hail, ex- Portugal) only begin cover once the sown crop cessive rain, flooding, waterlogging (excess soil has germinated and emerged and a full stand has moisture), excessive humidity, wind (hot dry been established, while other programs insure winds and strong winds), and one natural peril, against loss of the sown seeds due to perils that mudflow, that may or may not be associated with cause germination failure. The LIC program in excessive rain. Mexico insures crops from the time of sowing against natural and climatic events as well as 3.16. The compulsory LIC policy excludes all biological perils that lead to germination failure. other causes of crop loss, including the natu- In a few cases, including the Federal Crop In- ral peril of fire and all biological perils (pests surance Program in the United States, coverage and diseases). Plant pests and especially fungal may even extend to “prevented sowing,” for ex- Agricultural Insurance Feasibility Study  73 ample, due to extreme drought or excessive rain surance and subsequent amendments. In 2003 that prevents the farmer from gaining accessto the GRK through the MoA commissioned a crop the field with machinery to sow the crop. insurance premium rating study for the LIC crop insurance policy through the Kazakh Actuarial 3.19. The cover period of the compulsory LIC Center (CJSC). The CJSC used 11-year (1991– policy is not clearly defined. Article 9 of the LIC 2001) MoA crop-area damage data at the oblast policy wording states that the contract is deemed level to design a system of regional or oblast-lev- valid and binding for all parties from the moment el technical premium rates for the major insured of payment of the insurance premium until an crops (grains, oilseeds, sugar beets, and cotton). agreed termination date. The Law on Compul- For each major type of crop in each oblast, the sory Crop Insurance stipulates that the contract average loss cost rates were calculated accord- of insurance must be concluded and be in place ing to the percentage of total sown area dam- no later than 15 days after completion of sow- aged and adjusted by spreading catastrophic loss ing: it is, however, unclear whether the intention years in particular oblasts—for example, West of the policy is to provide back-dated coverage Kazakhstan (WKO)—over the whole portfolio. from the time of sowing of the crop and whether The technical rates were reduced by 50 per- losses that occur within the 15-day period would cent on account of the government’s 50 percent be deemed insured or not. Under a voluntary claims subsidies and then loaded by factors of crop insurance program, it would be unaccept- 10–25 percent for insurers’ administrative and able to permit farmers to purchase cover up to operating (A&O) expenses and profit margins.22 15 days after completion of sowing because of Further details of the CJSC’s premium rating the potential for antiselection. Such polices nor- methodology are presented in annex 2. The fi- mally carry a sales cutoff date of at least 30 days nal commercial premium rates were published prior to the start of sowing.21 In Kazakhstan it in the 2004 law and last updated in 2008, when might be argued that antiselection is minimized a system of minimum and maximum reference because the program is compulsory for all farm- rates was introduced for grain crops. They have ers. However, allowing farmers to purchase remained unchanged since then. cover up to 15 days after completion of sowing may lead to moral hazard—for example, where 3.21. For each crop and group of oblasts, the crop germination is poor, the farmer may lower GRK premium rates are presented in terms of his or her husbandry and management standards, a minimum and a maximum rate that may be thereby accentuating the loss of crop production charged by the insurance companies. A list of in the knowledge that the policy will pay out a the current reference premium rates that apply claim. in 2011 are contained in table 3.1. In the higher- rainfall growing areas of northern and eastern Premium Rates Kazakhstan, the premium rates for grains are be- tween 1.78 and 3.48 percent, rising to between 3.20. Government is responsible for setting 5.21 and 9.15 percent in the most drought- the reference premium rates on the compulso- prone areas of western Kazakhstan, including ry crop insurance scheme, and these rates are the oblasts of Aktobe and WKO. For all other declared in the Law on Compulsory Crop In- classes of crops, including oilseeds, sugar beets, and cotton, the minimum and maximum refer- ence rates are the same for all oblasts; in other words, there is no regional differentiation of the 21  Examples where antiselection could arise under premium rates. a voluntary program include where preexisting ad- verse climatic conditions are developing at the time of sowing (for example, a major freeze or drought) and farmers purchase cover knowing there is a high 22  Full details of the rating methodology are set out probability of crop failure. in CJSC (2003). 74  Kazakhstan 3.22. Since the GRK reimburses the insur- the GRK and then published by law. ers for 50 percent of the claims free of cost, the MoA- determined technical premium rates 3.24. In each oblast, farmers are permit- were adjusted downward to reflect only 50 per- ted to choose between three optional levels of cent of the expected claims costs. These reduced sum insured for each insured crop according technical rates were then loaded for acquisition to technology levels. The three optional levels costs, insurers’ operating expenses, and reason- of sum insured correspond to high-technology/ able profit margins to derive the MoA-approved high-production-costs cultivation (science- minimum and maximum commercial premium based agro-technology), medium-technology rates or reference rates (table 3.1). As such, cultivation (simplified agro-technology), and farmers receive an implicit premium subsidy of low-technology cultivation (based on three com- 50 percent of the amount of premiums they have ponents of production input costs: wages, fuel, to pay. and seeds). Farmers are free to elect to insure their crop at any level, irrespective of their ac- Sum Insured tual use of technology and production costs. Ta- ble 3.2 presents a summary of the three levels of 3.23. In Kazakhstan, the sum insured of the average normative costs or sum insured in KZT compulsory LIC policy is determined by a per hectare for each type of crop that applied in “normative cost of production” per hectare. 2009 and 2010. This cost is established by MoA per crop, oblast, and agro-ecological zone in the country. The 3.25. The per hectare sums insured are in normative costs of production are approved by general very low, reflecting both the low aver- Table 3.1 Crop Insurance Commercial Premium Reference Rates Set by Lawtan Premium rate (%) Crop and oblast Minimum Maximum Grains (cereals) Аkmola, Аlmaty, East Kazakhstan (EKO), Zhambyl, Kostanay, North 1.78 3.48 Kazakhstan (NKO) Karagandy, Kyzylorda, Pavlodar, South Kazakhstan (SKO) 3.17 5.83 Аktobe, WKO 5.21 9.15 Oilseeds National (applicable in all oblasts where the crop is grown) 2.01 3.44 Sugar beets National (applicable in all oblasts where the crop is grown) 5.76 8.39 Cotton National (applicable in all oblasts where the crop is grown) 0.92 Source: Law on Compulsory Crop Insurance 2004. Agricultural Insurance Feasibility Study  75 age technology levels adopted by Kazakhstan’s gram by capping the maximum permitted sums farmers, the low costs of production for most insured. Table 3.2 shows that in 2010 for the crops, and the government’s desire to manage highest level of science-based agro-technology, the financial exposure on this compulsory pro- the average sum insured for all crops was only Table 3.2 Average Normative Costs of Production in 2009–10, by Type of Crop KZT per hectare unless otherwise noted Costs based on Science-based Simplified Crop salary, fuels and agro-technology agro-technology lubricants, and seeds Brassica napus 8,409 6,230 2,353 Buckwheat 8,482 5,848 3,180 Chick peas 7,990 6,101 3,699 Cotton 24,060 14,211 9,706 Grain maize 25,804 16,836 8,640 Millet 4,830 3,392 1,731 Oats 6,904 4,682 2,987 Peas 8,064 5,780 3,935 Rice 33,524 21,459 5,130 Safflower 8,323 6,407 3,311 Soybeans 10,624 6,822 4,803 Spring barley 7,676 5,111 3,104 Spring rye 6,518 3,874 2,449 Spring wheat 8,829 5,831 3,407 Sugar beets 40,646 23,469 7,583 Sunflower 9,026 5,661 3,824 Winter wheat 9,361 6,143 3,699 Average all crops 9,909 6,540 3,674 Average all crops (US$ per hectare) 9,909 6,540 3,674 Source: Authors based on Regulation March 25, 2009, no. 410. 76  Kazakhstan KZT 9,908 (US$68)23 per hectare, with a range 3.28. Where a total crop loss occurs, the loss from a low for millet of KZT 4,830 (US$33) per assessment and indemnity procedures are rela- hectare to a high for sugar beets of KZT 40,646 tively simple. In the case of total crop loss, the (about US$280) per hectare. For the lowest- main task at the time of in-field crop loss assess- sum-insured option, based on the costs of wag- ment is to measure the area (in hectares) of total- es, fuel, and seeds, the average sum insured for ly (100 percent) damaged crop due to an insured all crops was a very low KZT 3,674 (US$25) per peril. The damaged area is then multiplied by hectare in 2010. For spring wheat, which is the the normative costs of production (sum insured) most important insured crop grown in Kazakh- per hectare selected by the insured to derive the stan, 2010 average sums insured ranged from a total value of the claim. No deductible is ap- maximum of KZT 8,829 (US$61) per hectare to plied to the claim. The LIC policy wording is, a minimum of KZT 3,407 (US$23) per hectare. however, ambiguous as to whether the total loss Details of 2010 normative costs for wheat by must apply over the entire area of the declared oblast and agro-ecological risk zone are includ- insured unit (field) in order to qualify for a claim ed in annex 2. The normative costs per hectare or whether a total loss in any hectare of the in- are multiplied by the number of insured hectares sured crop will open the policy for a claim. This in each field to derive the TSI for each crop in lack of clarity over the definition of the insured each insured unit. unit may lead to misunderstandings between the insurer and the insured at the time of loss assess- Basis of Indemnity ment. Also the policy does not clearly state what constitutes a total loss and does not appear to carry a constructive total loss clause.24 3.26. The policy indemnifies the farmer when the expected revenue (actual remaining crop production times the prevailing farm-gate 3.29. In the case of partial losses due to in- sales price) derived from the insured crop in sured perils, the loss assessment procedure the insured unit (which in Kazakhstan is de- and basis of indemnity are considerably more fined as the individual “field”) falls short of the complicated. In the case of partial losses to crop sum insured (investment cost) by the farmer. production and yields, in-field assessment is re- In such cases, the insurance policy indemnifies quired using 1 meter square crop-cut samples the farmer by the difference between the sum located at random in the field to estimate the insured (investment cost) per hectare minus the actual remaining harvestable yield (salvage) in expected revenue per hectare times the area of the affected area. Loss assessment is a costly the insured unit. and time-consuming process, and the law stipu- lates that a committee of five persons (including the local rayon executive authority, the FFSA 3.27. Damage is assessed by in-field sampling representative, insurance agent, insurance com- by a committee of up to five persons (organiza- pany, and insured farmer) must inspect the af- tions). The field loss assessment exercise is de- fected field(s) and determine the damaged area signed to assess (a) the cause of loss to verify and, within this area, the amount of harvestable whether this is due to an insured (uninsured) peril(s), (b) the affected or damaged area of the crop in each insured unit, and (c) whether dam- age to the crop is a partial or total loss. The pro- 24  Under the Mexican LIC policy there is a total cedures for indemnifying total and partial losses constructive loss clause under which a total loss is are explained below. defined as “where more than 90 percent of the sown crop stand and expected production has been lost and where it is not economic to harvest (salvage) the remaining proportion of the crop because the harvest costs would exceed the sale value of the salvage.” It 23  A 2010 exchange rate of KZT 145 = US$1.00 is not clear whether this definition applies under the was used in this analysis. Kazakhstan compulsory LIC policy. Agricultural Insurance Feasibility Study  77 yield (salvage). The value of salvage must then sating insurers for 50 percent of all the claims be calculated in one of two ways: (a) according incurred, and the second is by funding the A&O to the actual sales value of the crop received by expenses of the FFSA. Insurance companies and the farmer or (b) prior to sale of the crop, by mutual societies are not required to pay pro rata the estimated local sales price for the crop. If the premiums for this cover, which is similar to a value of salvage is lower than the sum insured free quota-share (50-50 percent) reinsurance (investment costs), the difference is indemnified. agreement. A worked example is given in box 3.2. However, this procedure has a major drawback: because 3.31. The 50 percent claims compensation an insured price for valuing the salvage is not fund is administered by the FFSA, which is preagreed at the time of binding the insurance responsible for monitoring and managing the contract, neither the insurer nor the insured has financial transactions of the scheme on behalf a clear idea of the amount of indemnity that is of government and for approving the claims due in the event of a partial crop loss (this issue reimbursements to individual insurance com- is discussed further in chapter 4). panies. Since 2005, the FFSA has maintained a database on each and every insured farmer, Government Financial showing data by oblast and rayon on the crops grown, premium rates, and claims. Between Support to Crop Insurance in 2004 and 2010, the FFSA received total (A&O) Kazakhstan subsidies from government of KTZ 319 mil- lion (US$2.5 million) or an average of KZT 3.30. The GRK provides financial support for 53 million (US$0.32 million) per year (table the scheme in two ways. The first is by compen- 3.3). Apart from the direct financial support of Box 3.2 Example of Indemnity Calculations for Partial Crop Losses Sum insured details Insured crop. Sunflower Insured field number. 53 Area of insured field. 300 hectares Sum insured per hectare. KZT 3,390 per hectare Total sum insured field number 53. KZT 1,017,000 Cause of loss. Drought Partial loss: Claims calculation Affected area. 300 hectares, partial losses Harvestable yield (salvage). 5.5472 metric tons Sales price for sunflower. KZT 55,000 per metric tons Value of salvage. KZT 305,100 Value of salvage per hectare. KZT 1,107 per hectare (305,100 ÷ 300) Difference between sum insured and salvage value. −KZT 2,373 per hectare (1,107 – 3,390) Insurance payment: KZT 711,900 (300 hectares x KZT 2,373 per hectare). Source: Authors based on FFSA data. 78  Kazakhstan the FFSA’s A&O expenses, the GRK also indi- of KZT 4.4 billion (US$28.9 million) from gov- rectly subsidizes the costs of field-level loss as- ernment. sessment activities (government staff, vehicles, equipment, and so forth). 3.33. Over the past six years (2005–10), the FFSA reimbursed insurance companies a to- 3.32. Over the past six years, the GRK pro- tal of KZT 3.84 billion (US$25.6 million), vided KZT 4.7 billion to the FFSA, of which 93 equivalent to 46.7 percent of total paid claims. percent was allocated to settling the 50 percent Over this period, the FFSA settled an average of of claims and 9 percent was allocated to pay- KTZ 641 million (US$4.3 million) per year for ing the A&O expenses of the FFSA. Table 3.3 50 percent claims to the insurance companies. shows that, in the start-up phase of this scheme, However, over the past three years the trend was the GRK provided KZT 2 billion per year for for increased claims, and in 2010 the cost of the the first two years and since then has provided 50 percent claims reimbursement to the FFSA smaller payments, such that by the end of 2010 was KZT 1.2 billion (US$8.0 million). At end- the scheme had received total financial subsi- 2010, the claims fund had KZT 538 million in dies valued at KZT 4.7 billion (about US$31 reserves, which would be inadequate to cover million).25 Over the six years to 2010, the 50 per- the average claims cost of the past three years. cent claims compensation fund received a total The level of GRK budgetary support for crop in- surance in 2011 is not known. Issues relating to the whether it is more cost-effective for govern- 25  At the current 2011 exchange rate of KZT 150 ment to provide 50 percent quota-share coinsur- = US$1.00. ance of claims or to switch its financial support Table 3.3 State Budget Program 050: “Support for Crop Insurance” KZT (millions) Allocation Disbursements Budget Agent services 50% claims Year Actual 50% Balance on allocation payment for the compensation claims payments claims fund FFSA fund 2004 2,000.0 10.0 1,990.0 2005 2,000.0 10.0 1,990.0 520.1 3,459.9 2006 100.0 60.0 40.0 236.0 3,263.9 2007 300.0 68.4 231.6 350.2 3,145.3 2008 100.0 68.4 31.6 819.1 2,357.8 2009 100.0 49.9 50.1 693.1 1,714.7 2010 100.0 52.2 47.8 1,224.5 538.0 Total 4,700.0 318.9 4,381.1 3,843.1 Average 783 53 730 641 Source: Authors adapted from FFSA data. Agricultural Insurance Feasibility Study  79 to subsidizing premiums or to purchasing rein- 2010, resulting in a major reduction in the num- surance or some form of catastrophe excess of ber of insured polices and a shift in the portfolio loss reinsurance protection are considered fur- toward larger farms. ther in chapter 4. 3.38. It is difficult to report the type of benefi- Performance Assessment: ciary and percentage of insured farms in Ka- zakhstan. In 2010 there were 7,441 registered Technical Results, Liabilities, agricultural enterprises (production enterprises) Reinsurance and 193,435 peasant farms (commercial farms) or a total of 200,876 farms (ARKS). With a to- 3.34. This section presents a review of the tal of 16,766 insured crop policies in 2010, this compulsory crop insurance results from 2005 suggests an insurance penetration level of only to 2010. Full details are presented in annex 3. 8.3 percent of all farms. However, these figures must be interpreted with caution. Practically all Coverage (Insured Crops, Insured (100 percent) of the 7,441 production enterpris- Farmers, and Insured Area) es purchase crop insurance because of their high profile and the major financial penalties (fines) 3.35. Crop insurance has reached high lev- they are likely to incur if they do not purchase els of penetration in Kazakhstan. Owing to its compulsory cover. Conversely, crop insurance compulsory nature, crop insurance has reached uptake by small commercial farms is very low high levels of penetration both in terms of the across the country. Finally, data on the number number of insured farmers and the amount of in- of farm holdings distort the true penetration rates sured area. for crop insurance because they include large numbers of small peasant livestock holdings, 3.36. The crop insurance portfolio is com- which are outside the scope of the scheme. posed mainly of spring sown cereals, of which spring wheat is the most important insured 3.39. The crop insurance scheme is highly crop. In the first six years of operation (2005 to concentrated in the main spring cereal belt of 2010), spring crops accounted, on average, for northern Kazakhstan. Over the past six years, a 98 percent of the insured area, and spring wheat total of 73.4 million hectares of crops were in- accounted for 86 percent of the insured area of sured under the compulsory LIC scheme. Over- spring crops (annex 3). all, 77 percent of this total area was underwrit- ten in the three northern oblasts of Kostanay (29 3.37. Over the period 2005 to 2010, the com- percent of insured area), Akmola (26 percent), pulsory crop insurance scheme issued an aver- and NKO (22 percent). Conversely, crop insur- age of 23,494 crop insurance policies per year, ance is relatively unimportant in the other 11 with an average of 523 hectares of insured oblasts, none of which accounts for more than 5 crops per policy. Coverage peaked in 2008, with percent of total insured area to date (figure 3.1). a total of 33,957 insured policies, but in 2010 the number of policies was halved to only 16,766. 3.40. The insured area has expanded over In 2010 the average size of insured farm was time, with a peak of 15 million insured hectares significantly larger than in any other year, with in 2009, representing 82 percent of the total eli- an average of 756 hectares per policy, or 144 gible acreage of the insurable crops grown in percent of the five-year average of 523 hectares Kazakhstan in 2009. In 2005, the first year of per policy. According to the insurance industry, the compulsory scheme, the insured area was the major losses experienced in 2009 dispropor- 10.5 million hectares (70 percent of total sown tionately affected smaller farmers: insurers were area); by 2008 this had increased to 14.5 million very reluctant to insure these small farmers in hectares (84 percent of total sown area), and in 80  Kazakhstan Figure 3.1 Distribution of Insured Area in Kazakhstan, by Oblast, 2005–10 35% % of total insured area (hectares) 29% 30% 26% 25% 22% 20% 15% 10% 5% 4% 5% 5% 3% 3% 2% 1% 1% 0% 0% be y O KO O KO l a by a ay a ar at nd EK SK ol rd to od an am m N W m lo ga Ak Al vl st Ak zy Zh ra Pa Ko Ky Ka Source: Authors based on FFSA data. 2009 it increased again to 15.0 million hectares cover for this group of farmers. Many crop (82 percent of total sown area). In 2010, how- producers interviewed under this study do not ever, the insured area declined significantly to perceive any benefits from the compulsory crop 12.7 million hectares or only 68 percent of total insurance scheme, and some prefer to pay the eligible acreage. fines for failing to contract insurance than to pay the premiums. 3.41. Over the first six years of the scheme, the insured area of crops averaged 73 percent of 3.42. The penetration of crop insurance, as the total sown area; this implies that, in spite measured by the ratio of insured area to total of the compulsory nature of the scheme, more planted crop area, is very uneven across the than a quarter of the cropped area was not country. In the northern and eastern regions insured. The fact that, on average, nearly one- of Kazakhstan, where large-scale spring wheat quarter of the insurable area remained uninsured production is concentrated, over the five-year shows that the compulsory scheme is failing to period from 2006 to 2010, the percentage of achieve one of its core objectives—to ensure total insured area exceeded 70 percent of total that all farmers are insured against catastrophic planted area in the oblasts of Kostanay (90 per- crop loss. There are several reasons for the fail- cent), Pavlodar (79 percent), NKO (76 percent) ure to achieve a higher level of insurance pen- Akmola (75 percent), and EKO (72 percent). etration. Over the past three years, most insur- Conversely, uptake was much lower in the very ance companies incurred negative underwriting drought-prone region of western Kazakhstan in results, particularly in the drought-prone regions the oblasts of WKO (58 percent) and Aktobe (52 of WKO and Aktobe, and they are increasingly percent). Even lower penetration rates were re- reluctant to operate in these high-risk areas. Also corded in the southern region of Kazakhstan, as their results were, on average, worse for small evidenced by the rates in Zhambyl (41 percent), peasant farmers, and, as evidenced in 2010, in- Almaty (41 percent), and SKO (27 percent). surance companies were not willing to renew See figure 3.2 and map a1 in annex 3. Given the Agricultural Insurance Feasibility Study  81 Figure 3.2 Penetration of Crop Insurance, by Oblast, 2006–10 100% 90% 90% % total planted area insured 79% 80% 75% 74% 76% 68% 70% 61% 58% 60% 55% 50% 41% 41% 40% 27% 30% 20% 10% 0% O KO O KO l be a y by a ay a ar at nd EK SK ol rd to od an am m N W m lo ga Ak Al vl st Ak zy Zh ra Pa Ko Ky Ka Source: Authors based on FFSA data. compulsory nature of the scheme, the extremely in the number of insured farmers and insured low uptake in southern Kazakhstan is unexpect- crop area, the TSI was correspondingly reduced ed. The main reasons for this are (a) insurers’ to KZT 47.3 billion (US$325 million). See an- reluctance to insure the often very small farms nex 3 for details. and (b) the fact that a high proportion of agricul- tural cropping is irrigated and farmers do not see 3.44. The financial liability of the compul- any value in purchasing the catastrophe drought sory crop insurance scheme is distributed very insurance cover. unevenly throughout Kazakhstan. Figure 3.3 shows that, over the past six years, the scheme Scheme Liability (Total Sum Insured) liability was highly concentrated in the three and Levels of Protection northern oblasts of Akmola, NKO, and Kostanay, accounting for 78 percent of total sum insured, 3.43. Since inception of the scheme in 2005, which reflects the fact that grain production is average total liability has been about KZT 40 highly concentrated in these regions. This rep- billion (US$267 million) per year. Over the past resents a very large accumulation of risk in the six years, the TSI of the scheme increased from event of a major drought affecting the northern KTZ 34.4 million (US$229 million) in 2005 to region. Conversely, no other oblast accounted a peak of KZT 52.9 billion (US$353 million) in for more than 5 percent of scheme liability over 2009, reflecting the gradual increase in coverage the past six years, and liability was very low in over time as the program matured in terms of the southern part of the country (SKO, Almaty, the number of insured farmers and amount of in- Zhambyl, and Kyzylorda). sured area. In 2010, however, with the reduction 82  Kazakhstan Figure 3.3 Distribution of Total Sum Insured per Oblast, 2005–10 35% 31% 30% 25% 25% 22% 20% % of TSI 15% 10% 5% 4% 5% 3% 2% 3% 1% 2% 1% 1% 0% be O KO O KO l a y by a ay a ar at nd EK SK ol rd to od an am m N W m ga lo Ak Al vl st Ak zy Zh ra Pa Ko Ky Ka Source: Authors based on FFSA data. Levels of Sum Insured Purchased by est permitted level of sum insured, which cov- Farmers ers only three production costs (wages, fuel and lubricants, and seeds): over the six-year period, 3.45. The level of sum insured elected by the the average sum insured was KZT 3,289 (about US$23) per hectare, with a range from an av- farmers to protect their crops is extremely low, erage low in 2007 of KZT 2,871 (US$20) per and over the past six years the average sum in- hectare to an average high in 2010 of KZT 3,728 sured for all crops was only KZT 3,289 (US$23) (US$26) per hectare. See annex 5. per hectare.26 As noted, the MoA, in conjunc- tion with the local authorities in each region, is 3.46. On the basis of this study, it appears that responsible for setting the sums insured for each many farmers elect the lowest permitted level of type of crop in each oblast and agro-ecological sum insured per hectare in order to minimize zone based on three levels of technology pack- the amount of premium they have to pay. Farm- age (high, medium, and low) and the associated ers’ reasons for selecting the lowest permitted costs of production. Farmers are then permitted level typically are that the policy only insures to choose which level of production costs they a small proportion of their production costs and wish to insure for crops grown in their fields. expected yield and that they would have to incur An analysis of the actual average sums insured a near total loss in order to trigger an indemnity. elected by farmers since inception of the scheme They do not see the coverage provided as benefi- found that most farmers are purchasing the low- cial to them and so generally opt for the lowest permitted level of sum insured. 26  At a 2010 average exchange rate of KZT 145 = 3.47. The current levels of sum insured do not US$1.00. afford farmers adequate protection against the Agricultural Insurance Feasibility Study  83 Figure 3.4 Sums Insured and Average Farm-Gate Prices for Wheat and Effective Yield Coverage Levels of LIC Policy, 2004–10 4500 3.50 Average sum insured (KZT per hectare) Wheat yield equivalent (centner per 4000 Wheat price (KZT per centner) 3.00 2.86 3500 2.50 3000 2.30 2.00 hectare) 2500 2000 1.50 1.29 1.38 1500 1.05 1.00 0.93 1000 0.50 500 0 - 2004 2005 2006 2007 2008 2009 2010 2011 Average Sum Insured (KZT/Ha) Wheat price (KZT/Centner) Wheat Yield Equivalent (Centner/Ha) Source: Authors based on FFSA data on sum insured and September wheat price data from Arka Consulting. loss of the costs they have invested in growing hectare (2005) to KZT 3,728 per hectare (2010). the crop. For spring wheat, where average to- Over this period, however, the average Septem- tal costs of production were about KZT 12,000 ber farm-gate price for wheat increased 350 per- to KZT 17,500 per hectare in 2010, the current cent, from KZT 1,149 per centner (2005) to KZT levels of sum insured of about KZT 3,500 per 3,988 per centner (2010).27 With the increase in hectare (minimum) and KZT 9,000 per hectare wheat prices, which are used to value salvage, (maximum) only represent between 20 and 50 there has been a major reduction in the underly- percent of the actual total costs invested in pro- ing insured wheat yield. In 2005, with an aver- ducing the crop. In the event of a total loss, the age sum insured of KZT 3,288 per hectare when indemnity payment is therefore inadequate to the average farm-gate sales price for wheat was cover the farmers’ incurred costs and may not be very low at KZT 1,149 per centner, the policy adequate to cover the costs of production loans would start to indemnify losses when the re- from banks. maining yield fell below 2.85 centners per hect- are; however, by 2010, with a high average sales 3.48. Over the past five years, the effective price of KZT 3,988 per centner, a farmer would amount of yield protection afforded to farmers only receive an indemnity when his actual yield by the LIC policy gradually eroded. Under the fell below 0.93 centner per hectare. LIC policy, an indemnity is payable when the value of any harvestable production (salvage) 3.49. The current LIC crop insurance policy is inadequate to cover the costs of production provides one of the lowest levels of yield protec- invested in growing the crop. The analysis pre- sented in figure 3.4 shows that, over the past six years, the average sum insured per hectare 27  Wheat prices based on data provided by Arka only increased 13 percent, from KZT 3,288 per Consulting in 2011. 84  Kazakhstan tion of any major individual grower national (99 percent after government claims subsidies). crop insurance scheme in the world. The cur- In the worst-loss year of 2010, it incurred a 261 rent LIC policy, with its very low sums insured, percent loss ratio: underwriters paid out claims equates to a level of insured yield coverage of of KTZ 2.61 for every KTZ 1.00 they received about 1 centner per hectare or on average only in premiums prior to recoveries from the FFSA. 14 percent of the national average spring wheat On account of the very poor results, all but three yield. Most individual grower loss of yield in- of the seven registered commercial insurance surance policies in the world offer a minimum companies no longer participate in the scheme. level of insured yield coverage of about 40 to 50 percent of average yield, and in some coun- Regional Performance (by Oblasts) tries the maximum coverage that farmers can purchase is 75 to 85 percent of the normal aver- 3.52. Scheme performance varies widely age yield. There is a need to review the levels across different regions, and the very poor of insured yield coverage provided under the underwriting results in Aktobe and WKO are Kazakhstan compulsory crop insurance scheme, making the scheme financially unviable. The and this subject is discussed in chapter 4. pattern of claims varies widely by geographic re- gion. The best-performing oblast is NKO, which over the past six years contributed 22 percent Financial Performance: Premiums, of total scheme liability but only 3 percent of Claims, and Claims Ratios claims and had a long-term loss ratio of only 24 percent. The next best-performing oblast is Ko- Overall Results stanay, which had the largest share or 31 percent of total scheme liability but only accounted for 3.50. The compulsory crop insurance pro- 11 percent of total claims and had a long-term gram in Kazakhstan experienced poor overall loss ratio of 73 percent. At the other extreme are underwriting results over the period 2005 to Aktobe and WKO in western Kazakhstan, which 2010. The long-term average loss ratio for the collectively accounted for only 4.8 percent of to- six-year period (2005–10) was 140 percent, and tal scheme liability over the past six years but in- in four of the six years the scheme operated at a curred 41 percent of all claims and had six-year financial loss, with the gross claims to premium, long-term loss ratios of 381 and 507 percent, re- or loss ratio, exceeding 100 percent. The average spectively. These two oblasts are severely preju- net loss ratio to the insurance companies after dicing the financial viability of the program, and government reimbursement of 50 percent insur- measures of controlling the claims costs in these ance losses (claims) was 75 percent. Assuming two oblasts are urgently needed (figure 3.5). See average administrative and acquisition expenses annex 3 for full details. of 25 to 30 percent of gross premium for the in- dustry, the average net loss ratio of 75 percent Causes of Loss indicates that, at best, the insurance companies and farmer mutual insurance associations are 3.53. Drought was the main cause of loss un- operating on a breakeven basis, but that most of der the compulsory crop insurance scheme. the companies are operating at a financial loss Drought events accounted for 91 percent of the (table 3.4). total reported damaged area in the five-year pe- riod 2006–10; with the inclusion of losses due to 3.51. Underwriting results deteriorated badly drought in combination with other perils such as over the past three years (2008–10), which co- hail or freeze, total drought losses rose to 97 per- incided with drought loss years, especially in cent of the total damaged area. The second most 2010. Over the past three years, the program important cause of loss was hailstorms, which incurred negative underwriting results in all accounted for 2.5 percent of the total area lost years and an average loss ratio of 182 percent due to insured perils. See annex 3. Agricultural Insurance Feasibility Study  85 Table 3.4 Summary of Kazakhstan Crop Insurance Financial Results, 2005–10 Item 2005 2006 2007 2008 2009 2010 Total Number of policies 19.0 13.6 25.4 34.0 32.2 16.8 141.0 (thousands) Total insured area 10.5 9.1 12.1 14.5 15.0 12.7 73.8 (millions of hectares) Sum insured (KZT 34,372 26,650 34,796 46,645 52,903 47,266 242,631 millions) Premiums (KZT mil- 899 685 997 1,093 1,114 1,074 5,862 lions) Average premium 2.61 2.57 2.87 2.34 2.11 2.27 2.42 rate (%) Claim payments 1,065 478 701 1,710 1,465 2,805 8,223 (KZT millions) Loss ratio (%) 119 70 70 156 131 261 140 Loss cost (%) 3.1 1.8 2.0 3.7 2.8 5.9 3.4 FFSA compensation 520 236 350 819 693 1,225 3,843 (KZT millions) FFSA compensation 49 49 50 48 47 44 47 (% of claims) Loss ratio (% net of 61 35 35 81 69 147 75 FFSA support) Source: Authors based on FFSA data. Adequacy of Premium Rates and Claims and maximum premium rates, and the minimum Costs premium rates were lower than the original single fixed premium rates. In 2008 the scheme 3.54. Over the past six years, average premium was opened up to competition from the newly rates tended to fall, although this is not justified formed mutual crop insurance associations. For by the claims experience. At the time of incep- the past three years (2008–10) the premium rates tion in 2005, a system of single fixed premium charged by the insurers decreased to an aver- rates for each crop in each oblast was adopted, age of 2.24 percent, representing a reduction of and these rates applied during the three-year pe- nearly 17 percent from the previous three-year riod from 2005 to 2007. During this three-year average. According to the commercial insurance period, the average premium rate was 2.68 per- companies, the mutual insurers generally charge cent. Rates varied little from year to year, rang- their members the minimum premium rates set ing from a low of 2.57 percent in 2006 to an av- by government, and the resulting price competi- erage high of 2.87 percent in 2007. Prior to the tion reduced the overall average premium rates 2008 renewal, the rating system was reviewed charged on the scheme. At the same time that and amended to introduce a system of minimum average premium rates were reduced, the claims 86  Kazakhstan Figure 3.5 Scheme Liability, Claims, and Loss Ratio, by Oblast, 2005–10 35% 600% 30% 507% 500% 25% 412% 381% 400% 20% 300% 15% 199% 200% 10% 176% 132% 5% 105% 100% 100% 73% 24% 30% 0% 2% 0% be O KO O KO y a l by a ay a ar at nd EK SK ol rd to od an am m N W m ga lo Ak Al vl st Ak zy Zh ra Pa Ko Ky Ka % of TSI % of Claims Payments Loss Ratio Source: Authors based on FFSA data. on the scheme rose significantly (figure 3.6). urgent need to review and adjust the premium rates on this scheme in order to maintain its fi- 3.55. The combination of higher claims and nancial viability. lower average premium rates for the past three years means that the scheme is now seriously 3.56. The claims experience between 2005 underrated. Figure 3.6 compares the annual and 2010 shows that the scheme is severely average premium rate charged on the scheme underrated in WKO and Akmole in western with the annual average cost of losses, which is Kazakhstan, and this applies to a lesser extent equivalent to the value of claims divided by the in several other oblasts. The six-year long-term total sum insured and expressed as a percent- average loss cost on the scheme currently stands age. The loss cost is a useful ratio, as it shows at 3.4 percent, but the pattern of losses varies the minimum premium rate that would need to widely across geographic regions. As shown in be charged on an insurance scheme to cover the figure 3.7, over this period average premium paid claims. Figure 3.6 shows that, in the three rates were highest in WKO, with average premi- years to 2007, average premium rates were ad- um rates of 7.8 percent; however, this premium equate to cover the average loss cost, but, after rate was totally inadequate to cover the actual 2008, the average loss cost exceeded the aver- claims, as evidenced by the loss cost of 39.6 age premium rates in all three years, culminat- percent. This is followed by Aktobe, with aver- ing in a peak loss cost of 5.93 percent of total age premium rates of 5.8 percent, but again the sum insured in 2010 compared with an average program was very underrated in this oblast, as premium rate of only 2.27 percent. There is an the breakeven rate to cover actual claims should Agricultural Insurance Feasibility Study  87 Figure 3.6 Average Premium Rate and Loss Cost, 2005–10 6.50% Average premium rate & loss cost 5.93% 5.50% 4.50% (% of TSI) 3.67% 3.50% 3.10% 2.87% 2.77% 2.50% 2.61% 2.57% 2.34% 2.27% 2.01% 2.11% 1.79% 1.50% 2004 2005 2006 2007 2008 2009 2010 2011 Average Premium Rate % Loss cost (%) Source: Authors based on FFSA data. have been 22.2 percent. The scheme was also part of the oblast, which is drier, rayon-level underrated in Pavlodar, Zhambyl, EKO, and loss costs are higher, with a maximum of 0.5– Almaty. In contrast, the scheme performed very 1.0 percent loss cost. Similar differences in well in NKO, with a six-year long-term average drought risk exposure and loss costs between loss cost of only 0.5 percent compared to aver- rayons also apply to other oblasts, such as Pav- age premium rates of 2.0 percent. In effect, this lodar, Akmola, Kostanay, and WKO (see map oblast cross-subsidized the very poor results in 3.1). This analysis clearly demonstrates a need WKO and Aktobe for the past six years. Substan- to consider introducing a system of rayon-level tial premium rate increases would be required in rates in the future. WKO and Aktobe if the scheme were actuarial- ly rated in these oblasts; however, the minimum 3.58. The analysis of variation in yields at the average rates of at least 40 percent (WKO) and rayon level also shows significant differences 22 percent (Aktobe) would not be commercially acceptable to farmers in these oblasts. in the yields obtained by different types of farm- ers, and these differences need to be addressed 3.57. The variation in claims experience be- in rating. Specifically, the analysis of time-se- tween rayons in the same oblast clearly indi- ries rayon-level spring wheat yields obtained by cates the need to introduce a system of rayon- the large-scale production enterprises and the level actuarial rating. The five-year claims smaller commercial farms shows that yields are experience also varies significantly between much more variable for the commercial farmers rayons in each oblast. For example, in rayons on account of their use of lower levels of tech- located in northern NKO, the average loss costs nology. This group of farmers should be charged are less than 0.25 percent, but in the southern higher premium rates. 88  Kazakhstan Figure 3.7 Adequacy of Premium Rates, by Oblast, 2005–10 Zhambyl 3.5% WKO 7.8% SKO 2.7% Pavlodar 3.6% NKO 2.0% Kyzylorda 5.2% Kostanay 1.7% Karagand 4.7% a EKO 3.4% Almaty 3.5% Aktobe 5.8% Akmola 1.9% 0.0% 5.0% 10.0% 15.0% 20.0% 25.0% 30.0% 35.0% 40.0% Average premium rate and loss cost (% of TSI) Loss cost Avg Premium Rate Source: Authors based on FFSA data. Financial and Reinsurance Issues ward increasing claims over the past three years, which coincided with three very bad drought 3.59. The claims results were very volatile in years (table 3.4). the past six years, with a trend over the past three years for increased claims. Over the past 3.60. Over the past six years, the scheme paid six years, the pattern of claims on the compul- total claims of KZT 8.22 billion (US$62 mil- sory crop insurance scheme ranged from a low lion), of which private and mutual insurers of KZT 478 million (US$3.9 million) in 2006 to paid 53 percent of total claims and govern- a high of KZT 2,805 million (US$8.4 million) ment, through the FFSA, reimbursed the insur- in 2010.28 There also was a marked trend to- ers for 47 percent of the total value of claims. As shown in table 3.4, the government provides 50 percent proportional or quota-share reinsur- ance protection to the insurance companies free 28  Actual annual average exchange rates are KZT of cost. 122 = US$1.00 for 2006 and KZT 145 = US$1.00 for 2010. Agricultural Insurance Feasibility Study  89 Map 3.1 Variation in Loss Costs, by Rayon, 2006–10 Source: Authors based on FFSA data. 3.61. Under the compulsory crop insurance panies that are fully capitalized and maintain scheme, insurance companies have little scope claims reserves would not be able to sustain such for managing the liability on their 50 percent losses in the medium term. of risk retention and are very exposed to major losses. The insurance companies have very little 3.62. Currently, private commercial insur- ability to manage their exposure to losses on this ers are unable to obtain commercial rein- compulsory scheme because they are not able to surance protection to limit their exposure to exercise decisions over whether to accept indi- catastrophic losses on the obligatory crop in- vidual risks or to decline to underwrite the port- surance scheme. Several private commercial folio in regions they consider to be too risky. In insurers have, in the past, sought quotations spite of the government’s reinsurance protection, from international reinsurers to provide either the insurance companies remain very exposed to quota-share or nonproportional reinsurance on loss, as they retain unlimited liability on their 50 their portfolios. International reinsurers, how- percent share of claims. Any company that wrote ever, are reluctant to provide capacity support an unbalanced portfolio that was overly con- for the scheme for various reasons, including centrated in the western oblasts of Kazakhstan (a) the compulsory nature of the scheme, which would have incurred unsustainable underwriting prevents them from selecting which risks they losses over the past six years, especially in 2010, are willing to underwrite in which region(s), as evidenced by gross loss ratios of 900 percent (b) the fact that the scheme is underrated, and in Aktobe (450 percent net of FFSA reinsurance) (c) the fact that indemnity and loss assessment and 1,075 percent in WKO (537 percent net of procedures would require strengthening to meet FFSA reinsurance). Even those insurance com- international standards. 90  Kazakhstan 3.63. The farmer mutual crop insurance as- 3.65. Government is also exposed to cata- sociations are not required to maintain claims strophic losses on this scheme, which may not reserves and are unable to meet their liabilities have been budgeted for. To date, government if claims exceed the net premiums they col- has funded the FFSA to the tune of KZT 4.4 bil- lect from their members. The private registered lion and has paid out KZT 3.8 billion in 50 per- insurance companies are obliged by insurance cent reinsured claims. Government’s liability legislation to meet minimum capital require- over the past six years was very volatile, vary- ments for underwriting any class of business, ing from total payments of KTZ 250 million in but also retain full liability for settling any and 2006 to KZT 1.2 billion in 2010. In common all legitimate claims. In contrast, the farmer with the problem faced by insurance compa- mutual crop insurance associations, which have nies, government’s liability cannot be budgeted been formed since 2008, are not required to ex ante, and the claims fund may be inadequate meet minimum capital requirements or to meet to cover catastrophic claims that may only be claims that exceed the net premiums (premium incurred every 100 or more years (figure 3.8). net of management charges, which are typically on the order of 25 percent of gross premium) 3.66. Due to ambiguities in the indemnity for- paid into their mutual funds by their members. mula used to settle losses on the LIC policy, On the basis of this study, it is apparent that sev- both the insurance companies and government eral of the mutuals interviewed were not able to face considerable uncertainty over their finan- pay their 50 percent share of claims liabilities cial liability in the event of claims. The LIC in full in 2010 and were therefore forced to pro policy indemnity formula for partial losses val- rata down the claims paid to all farmers who ues any remaining harvestable production (sal- incurred losses. Nevertheless, these companies vage) according to the actual local sales price were able to collect the government 50 percent for wheat at the time of loss. This means that of claims payments in full and to return these neither the insurance companies nor the govern- amounts to their members. It is not possible to ment can predict their financial liabilities with report on the extent of this problem in both 2009 any accuracy at the time of renewing cover each and 2010, as the number of mutual associations year. Instead, they have to wait until the time of that could not meet their liabilities on their 50 harvest to discover their liability. This is illus- percent retained claims is not known. trated in table 3.5 for a 1 hectare field of wheat with an average normative cost of production 3.64. In Kazakhstan, there is a concern that, sum insured of KZT 3,500 per hectare and three if the mutual crop insurers are not able to meet different wheat prices that apply at harvest, in- their claims liabilities in full, this will rapidly cluding a low of KZT 2,000 per centner (as en- undermine farmers’ confidence in mutual in- countered in 2007), KZT 2,500 per centner, and surance. International experience shows that, where mutual crop insurers have not been able a high of 3,500 per centner for 2010 (similar to to meet their financial liabilities in catastrophe the average for 2010). When average sales pric- years, this rapidly undermines the confidence of es for wheat are low (KZT 2,000 per centner), their members and usually leads to the collapse insurers are more exposed to first loss because, of the mutual(s). There is an urgent need in Ka- once the actual average yield falls below 1.75 zakhstan to (a) review the extent of this prob- centners per hectare, the revenue from the crop lem among the mutual insurers, (b) seek ways will be less than the breakeven sum insured val- of strengthening their claims reserves, and (c) ue of KZT 3,500 per hectare. Conversely, when consider alternative ways of reinsuring the mu- wheat prices are high (KZT 3,500 per centner), tuals in order to cap their liability in the event of the insurers are much less exposed to loss, as catastrophic losses—for example, through some they will only be liable for a claim once the ac- form of nonproportional reinsurance on their re- tual remaining yield (salvage) falls below 1.0 tained risk. This subject is reviewed further in centner per hectare. Insurers’ and government’s chapter 4. uncertainties over their financial exposure will Agricultural Insurance Feasibility Study  91 Figure 3.8 Claims and Share of Claims Paid by Insurers and Government, 2005–10 3000 Paid claims (KZT millions) 2500 2000 y = 362.05x + 103.28 1500 R = 0.6515 1000 500 0 2005 2006 2007 2008 2009 2010 Total Claims Net Claims to Insurers Government Reinsurance Linear(Total Claims) Source: Authors based on FFSA data. only be removed by agreeing on a valuation assuming a wheat indemnity sales price of (b) price for salvage at the time of policy inception. KZT 2,000 per centner, (c) KZT 2,500 per cent- This valuation price could be based either on an ner, and (d) KZT 3,500 per hectare. Figure 3.9 average of the recent historical farm-gate prices shows that, for option a—the actual portfolio— for each crop in each oblast or on the MoA’s the 1-in-100-year PML is 9.4 percent of the TSI, estimated average prices for the current season. which would be equivalent to a loss of KZT 4.4 billion (US$29.3 million) on the 2010 portfo- 3.67. Insurers and government face major lio TSI of KZT 46.48 billion. This PML would exposures on this scheme in the event of a equate to a gross loss ratio of nearly 396 percent catastrophe year, as defined by the probable at 2010 premium terms. The analysis also shows maximum loss (PML) expected once in 100 that, if the actual average prices of wheat were years. To date, the worst loss on this scheme to fall to levels of between KZT 2,000 and KZT over the six-year history occurred in 2010 and 2,500 per centner, the PML exposures would be was equivalent to a loss of 5.9 percent of the very much higher on this scheme. total liability (TSI), valued at KZT 2.81 billion (US$19.3 million). This loss was equivalent to a Assessment of the Technical, return period of about one in 45 to 50 years, but Operational, and Institutional it is not the worst loss that insurers and their re- insurer (government) may expect in the future. Features of the Compulsory Figure 3.9 and table 3.6 show the results of a Crop Insurance Program modeling exercise to calculate the PML of the scheme for return periods of up to 250 years and 3.68. This subsection reviews the technical, op- four different scenarios: (a) one based on the ac- erational, institutional, and financial features tual average three-year portfolio TSI and three of the compulsory crop insurance program and 92  Kazakhstan Table 3.5 Examples of Indemnity Calculations According to Different Wheat Prices at Harvest Price 1: KZT 2,000 Price 2: KZT 2,500 Price 3: KZT 3,500 Indemnity calculation per centner per centner per centner Sum insured (KZT per hectare) 3,500 3,500 3,500 Breakeven yield (centner per hectare) 1.75 1.40 1.00 Sales value for wheat (KZT per centner) 2,000 2,500 3,500 Revenue value for salvage (KZT) 3,500 3,500 3,500 Source: Authors. Figure 39 Estimated PML Assuming 2010 Average Sum Insured per Hectare and Three Scenarios of Wheat Prices 25% % loss cost minus % TSI 20% 15% 10% 5% 0% 0 50 100 150 200 250 Return Period (years) 3-year Average Price Price= KZT 3500/centner Price = KZT 2000/center Price= KZT 2500/centner Source: Authors. highlights some of the key advantages, issues, vides comprehensive MPCI protection to the and drawbacks of the current program. farmer against the loss of production costs in- vested in growing the crop and (b) it can be used Technical Features in situations where there is inadequate or no in- formation on the historical crop yields of indi- 3.69. The salvage-based LIC policy imple- vidual farmers. This is in contrast to a conven- mented in Kazakhstan has several advantages, tional loss of yield MPCI policy, which relies but also a series of drawbacks. The main ad- on time-series farm-level crop production and vantages of the LIC policy are that (a) it pro- yield data to establish a normal average yield Agricultural Insurance Feasibility Study  93 Table 3.6 Estimated PML According to Different Price Assumptionsa Actual portfolio Wheat price 1: Wheat price 2: Wheat price 3: Return period (years) (average last KZT 3,500 per KZT 2,500 per KZT 2,000 per three years) centner centner center 10 785,966 682,553 1,159,043 1,707,702 25 1,778,278 1,565,370 2,520,147 3,484,647 50 2,867,899 2,543,651 3,929,562 5,430,554 100 4,372,797 3,960,971 6,195,513 7,950,115 250 7,354,656 6,847,274 9,111,056 11,736,143 Source:Authors based on FFSA data. a. PML values calculated based on portfolio TSI of KZT 46.48 billion. and then an insured yield for each farmer. There do not take into account differences in risk at are, however, several potential drawbacks to the the level of either the local rayon or the indi- LIC policy, including the need for in-field yield- vidual farmer. Ideally crop insurance premium based loss assessment where partial losses are rates should be established for homogeneous involved and the difficulty of establishing ob- agro-climatic risk zones, but in most countries jectively the salvageable amount of the crop and such risk zoning and mapping are not available; its sales value and determining whether this sal- rates are typically established for administra- vage value exceeds the insured investment costs tive regions and then smoothed to ensure con- leading to a claim. sistency with adjoining regions. In Kazakhstan, the oblast represents a very large geographic 3.70. The coverage levels provided by the crop area, and there is evidence that agro-climatic insurance scheme in Kazakhstan are extremely and soil conditions vary widely across most of low. The sum insured levels that MoA is respon- the oblasts. This variation is reflected in major sible for fixing each year by crop and by oblast differences in crop yields within the rayons in according to three technology levels—high, each oblast.29 A technical study is needed to re- medium, and low—are extremely low for most view the possibility of introducing a system of crops, as shown in table 3.2. In practice, almost rayon-level crop insurance premium rates in the all farmers in Kazakhstan elect the cheapest or lowest option of sum insured coverage, which short term, while more detailed risk zoning and is about KZT 3,500 per hectare (slightly less risk mapping are carried out. For this reason, than US$25 per hectare nationally for spring under this World Bank study, an actuarial rating wheat) because they will pay the least amount of premium for this option. However, this means that, on average, farmers are only in- 29  The CJSC identified this issue in its original rat- suring between 20 and 30 percent of their total ing study, which noted that oblasts are very large ter- ritories and agricultural conditions within the oblast production costs. The very low levels of crop may vary significantly, leading to a situation in which insurance cover are, therefore, often inadequate farms in low-risk areas end up subsidizing farmers to put farmers back into production in the event in high-risk areas because they both pay the same of a major crop loss. tariff averaged for the oblast. The CJSC, however, played down the need for rayon-level and possibly farmer-level premium rates classified by high or low 3.71. The crop insurance premium rates are risk because of the compulsory nature of the scheme calculated for each crop at the oblast level and (CJSC 2003). 94  Kazakhstan exercise was carried out at the rayon level for risk exposures—for example, by adopting snow spring wheat, and the results of this analysis are conservation measures in winter and minimum presented in chapter 4. tillage practices and soil moisture conservation practices during the growing season. Chapter 3.72. Crop insurance premium rates are cur- 2 shows that the large production enterprises rently fixed by law for each type of crop and achieve higher average yields and less variable group of oblasts according to minimum and yields in spring wheat and therefore represent a maximum rates, and insurers have little in- lower-risk group than commercial farmers. The fluence over rating decisions. The private and private insurers reported that they aim to in- mutual insurance companies have no say in set- centivize the best farmers by offering them the ting premium rates or in selecting risk. They minimum rates and to penalize low-technology are, however, permitted to charge an individual users in riskier areas by charging them the maxi- farmer a premium rate that falls between the mum permitted rates. MoA fixed minimum and maximum rates. Pri- vate insurance companies use the minimum and 3.74. Crop insurance premium rates do not in- maximum premium rates to manage, in part, clude specific allocation for the costs of in-field their risk: in the most drought-prone regions loss assessment or loss adjustment expenses. of the country, they charge the maximum pre- Under the current system, each of the five or- mium rates to try to avoid writing too much crop ganizations involved in the in-field assessment business, and when they consider an individual of total and partial crop losses is responsible for farmer to be a poor risk, they again quote the bearing its own costs of this exercise. In 2010 maximum premium rates. While private insur- the average size of a crop insurance policy was ance companies use this difference in minimum 756 hectares, with an average premium of KZT and maximum premiums to perform some sort 64,037 (US$442) or KZT 85 (US$0.58) per of risk selection, farmer mutual associations ad- hectare. Because of the combination of very vised that they nearly always have to operate at low average sums insured per hectare and low the level of minimum premiums, because their premium rates, the average premium per policy members are not willing to pay the higher rates is very low for a cover requiring such intensive permitted by law. For spring wheat grown in the in-field loss assessment, and it is likely that the more favorable, higher-rainfall regions of Ak- total costs of adjusting losses will exceed the to- mola, Almaty, Kostanay, EKO, and NKO, the tal premium generated per policy. This evidence current commercial premium rates are very low, suggests a need to review (a) the adequacy of at between 1.78 percent (minimum) and 3.48 the premium rating methodology for covering percent (maximum); where mutuals are forced loss assessment expenses and (b) the ways of to charge the minimum rate of 1.78 percent due overhauling loss assessment procedures in order to competition, it is unlikely that this rate will be to make them more efficient and to reduce loss sustainable in the medium to long term. There is assessment expenses. See chapter 4 for recom- a need to reconsider the system of minimum and mendations regarding strengthening of loss as- maximum permitted premium rates. sessment procedures. 3.73. The current rates for crop insurance pre- Operational Features miums do not differentiate between farmers in terms of technology use or implementation of 3.75. The obligatory nature of the scheme risk management practices. The current rating means that there is a major potential problem system does not differentiate between farmers of moral hazard in the most drought-prone in terms of their technology levels or risk man- regions of the country. In some regions of Ka- agement practices and therefore does not pro- zakhstan, seasonal rainfall is extremely mar- vide any incentive for farmers to manage their ginal for spring wheat production and, even in Agricultural Insurance Feasibility Study  95 years of normal rainfall, is very dependent on tant for insurers to attempt to manage their crop the amount of winter snowfall and soil moisture insurance portfolios in extreme drought years, at the start of the cropping season. Western Ka- and they do this mainly by setting underwrit- zakhstan (WKO and Akmola) experiences major ing limits in each region according to the rela- droughts every two or three years, and it is possi- tive rainfall patterns and drought exposures. In ble for farmers to predict a drought year accord- Kazakhstan the huge size of the country means ing to the amount of winter snowfall. However, that, even in extreme drought years, not all the because of the obligatory nature of the scheme, spring wheat grown in the country is equally af- insurers cannot decline to insure farmers in these fected by drought. If insurers were able to set areas even when they can foresee that they will underwriting capacity limits in each region, they incur major drought-induced crop losses. In a would be able to manage their drought exposure predicted severe drought year, some farmers in to a major extent. This is, however, currently im- these regions are likely to modify their behavior possible to achieve because of the compulsory in two ways: (a) by electing to buy the maxi- nature of the crop insurance program. mum level of “normative costs” sum insured in the expectation of receiving a claims payment 3.78. Crop insurance is currently marketed and (b) by incurring less than their normal level exclusively through local sales agents in each of expenditures on crop husbandry and inputs oblast and rayon, and this represents an ex- because they know they are likely to lose their pensive delivery channel. All private companies crops and be able to submit a claim on their in- market their crop insurance policies through lo- surance policies (this is termed moral hazard). cal sales agents who receive a commission for The private insurance companies have incurred their services that is paid out of the premium major losses in western Kazakhstan in the past collected from the farmer. Many of the mutual and are now very reluctant to insure farmers in insurance companies also rely on sales agents to this region. This is one of the major reasons that promote and market their crop insurance policies most private companies have ceased underwrit- to farmers who then sign up to become a member ing crop insurance. of the mutual. Current legislation caps the com- mission paid to agents and brokers at 10 percent 3.76. Private insurance companies do not of the commercial premium (also termed origi- have their own networks of locally based quali- nal gross premium).30 This represents a signifi- fied agronomists to conduct preinspections on cant cost, which is paid by the farmer. In many the insured farms at the time of sowing in order countries, insurers use rural service organiza- to confirm whether the farmer has complied tions such as agricultural banks, microfinance with the correct sowing practices, seed applica- institutions, input suppliers, or farmer associa- tion rates, and so forth. As such, cover is open tions to deliver and administer crop insurance to to moral hazard. The costs of establishing such their clients. In such cases, the rural organization a network and inspecting each and every farm acts as an agent, but because of its much lower would be prohibitively high. Under the current operating overhead and economies of scale in dealing with large numbers of clients or mem- rating system, insurers are not able to increase bers, both the commission payments as well as rates to cover their A&O expenses. the company’s own administrative expenses can be significantly reduced. This offers the poten- 3.77. The obligatory nature of the scheme means that it is very difficult for insurance companies to exercise any form of accumu- lation control. Drought and frost are systemic 30  In early years of the program, the agent’s fees were as high as 30 percent of the premium, but the or covariate risks that have a potential to cor- regulators subsequently imposed a 10 percent ceil- relate over very wide geographic areas. In the ing in order to make the scheme more acceptable to context of any MPCI program, it is very impor- insurers. 96  Kazakhstan tial to pass on these cost savings to the insured in Institutional Features the form of lower premium rates. The subject of developing alternative and cheaper ways to de- 3.81. The Kazakhstan compulsory crop insur- liver crop insurance to small and medium farm- ance scheme is a PPP that is highly regulated ers in southern Kazakhstan is reviewed further by government and underwritten by the private in chapter 6. commercial (and mutual) insurance sectors, with financial claims subsidy support provided 3.79. Loss adjustment requires the participa- by government through the FFSA. The regula- tion of several parties, is expensive, and some- tory, operational, and financial roles assumed by times lacks transparency. In Kazakhstan up to the GRK include (a) to approve the standard LIC five persons are involved in adjusting the crop policy for crop insurance, including the basis of losses at the local level, and this is a very time- coverage and insured perils and other policy consuming and costly exercise. The main draw- terms and conditions; (b) to approve each year back of any LIC policy is the requirement, es- the sums insured based on the normative costs of pecially where partial crop losses are involved, production of the crops that are subject to com- for individual farmer and field-by-field loss as- pulsory insurance; (c) to develop the minimum sessment to determine the damaged area and the and maximum commercial premium rates to be remaining amount of harvestable crop (termed used for the different crops and regions under salvage) in the field. Such assessments are the the crop insurance scheme; (d) to approve the basis for calculating the expected value of the budget for supporting crop insurance; (e) to es- salvage. It is then necessary to determine wheth- tablish and regulate the commissions payable to er the value of salvage exceeds the costs invested sales agents; (f) to set the norms and procedures in the crop up to the time of the loss and whether for the adjustment of crop losses at the individual it is economic for the farmer to continue man- insured farmer and field level; and finally (g) to aging the remaining undamaged portion of the provide financial subsidy support both through crop through to harvest or whether these costs contributions to the A&O costs of the FFSA and will exceed the salvage value. This process is in the form of a 50 percent share in the value often subject to conflict over whether the value of paid claims. In most countries where govern- of salvage exceeds the sum insured investment ments support agricultural insurance PPPs, they costs, in which case there would be no claim, or usually assume a much smaller degree of opera- whether the expected value of salvage falls short tional and underwriting control, and their main of the sum insured costs of production, in which function is to provide legal and regulatory sup- case there would be a claim. port and financial subsidies. 3.80. Crop output valuation prices are deter- 3.82. At the inception of the compulsory crop mined at the time of harvest as opposed to be- insurance scheme, seven private insurance ing preagreed and specified in the policy word- companies registered their interest in under- ing. This section has shown that, because actual writing this product, but on account of poor farm-gate wheat prices can vary widely accord- underwriting results only three companies sup- ing to demand and supply, this leads to major ported the scheme in 2010. The level of partici- variations in the amount of physical crop pro- pation of the private insurance companies in crop duction and yield that is required to cover the in- insurance in Kazakhstan is very low. Only seven sured costs of production. This situation, where out of 37 private nonlife insurance companies in neither the farmer nor the insurer can quantify the market are currently licensed to operate crop his indemnity exposures (the amount of crop insurance, and only three of them actively un- production and yield loss that needs to occur be- derwrote the obligatory crop insurance scheme fore an indemnity is paid) is very unsatisfactory in 2010. The main reason companies have and needs to be reviewed. ceased underwriting this class of business is that Agricultural Insurance Feasibility Study  97 they view it as unprofitable; crop insurance is sociates without having to establish any capital the least profitable of 10 classes of compulsory or claims reserves. According to the interviews insurance in Kazakhstan. There is a danger that, held with farmer mutual insurance associations unless the scheme can be returned to profitabil- during the mission, several mutuals did not col- ity, the private insurance sector companies may lect enough premiums to pay the full amount of cease to provide their support in the future. claims incurred during the 2010 crop season. In order to remain solvent, these mutual insurers 3.83. Due to the lack of interest of the pri- were forced to reduce the 2010 paid claims on a vate insurance industry in participating in the proportional basis according to the total amount mandatory crop insurance scheme, the GRK of premiums they had collected during the year. decided in 2006 to license farmer mutual crop insurance associations to write crop insurance. Under Law no. 163-III ZRK of the Republic of Evaluation of Crop Insurance Kazakhstan on Mutual Insurance, dated July 5, Effectiveness for Key 2007, government authorized farmers to form Stakeholders mutual insurance associations to underwrite the compulsory crop insurance policy. As of 2011, 3.85. This final section briefly summarizes the more than 38 farmer mutual insurance asso- main issues and concerns that face the three ciations are offering crop insurance in Kazakh- principal stakeholder groups involved in the stan. Kazakhstan obligatory crop insurance scheme. These stakeholders are farmers, insurers, and 3.84. Private insurance companies and farmer government (box 3.3). mutual associations providing crop insurance are not equally regulated. While private insur- Farmers ance companies are subject to the regulation of the Agency for Financial Market and Financial Regulation and Control, the farmer mutual crop 3.86. The compulsory nature of the crop in- insurance associations are regulated separately surance scheme is very unpopular with many by the Law on Mutual Insurance. The Agency farmers. With very few exceptions, farmers met for Financial Market and Financial Regulation during the conduct of this study expressed their and Control exercises tight control over insur- major dissatisfaction with the obligatory nature ance activities in the country, including crop of the scheme. As a result, some farmers refuse insurance. Nonlife private insurance companies to purchase crop insurance and prefer to pay the are required to have a minimum capital of KZT accompanying fines. In addition, the majority of 1.2 billion (US$8.3 million) in order to operate. farmers purchase the minimum permitted level On top of the minimum capital requirements, of protection in order to minimize their costs. private insurance companies are frequently Although most farmers appear to understand monitored on their solvency, net retentions, and that the very low levels of sum insured cover- implementation of risk management procedures. age they are purchasing do not afford them with Conversely, farmer mutual associations offering adequate financial risk protection in the event of crop insurance are not regulated or supervised a major crop loss, they have such low expecta- by the Agency for Financial Market and Finan- tions of the scheme that they prefer to purchase cial Regulation and Control. Therefore, they are minimum cover in the knowledge that they will not subject to minimum capital requirements only receive an indemnity in the extreme case of or controls over net retentions and solvency re- near total crop failure. Moreover, farmers do not quirements. In fact, an association with more understand the operation of the policy and are than 250 farmers can constitute a farmer mutual dissatisfied with the cumbersome indemnity and association and provide crop insurance to its as- loss assessment procedures. 98  Kazakhstan Box 3.3 Key Issues Facing Main Stakeholders in the Compulsory Crop Insurance Scheme Farmers • Dissatisfaction with compulsory nature of crop insurance. • Reluctance to purchase insurance in spite of being compulsory. • Extremely low levels of coverage. • Difficulties in understanding the insurance agreement especially the basis of indemnity and loss assessment procedures. Government Insurers: • Very good social intentions, but compulsory nature of the • Limitations to perform risks selection/underwriting. program is very unpopular with most farmers. • No incentives to invest in underwriting/claims • Intention of obligatory insurance in early years to develop crop management infraestructure. insurance market has not been achieved. • Uncertainties about financial exposure. • Uncertainties about financial exposure. • Difficulties to gain access to reinsurance. • Uncertainties about budgeting claims participation in • Concern by private commercial insurers that they do not face a level catastrophic events. playing field with the Farmers’ Mutual Crop Insurance Associations. • Crop Insurance has been unprofitable to date. Source: Authors. Insurance Companies operating systems and procedures. In particu- lar, they are concerned about their exposures in 3.87. In recent years, the private insurance western Kazakhstan, where they have incurred companies have lost money underwriting the major losses over the past six years. In spite of obligatory crop insurance scheme, and they government financial support in the form of 50 have increasingly withdrawn from underwrit- percent reimbursement of all claims, the insur- ing this class of business. Originally, seven ance companies face major financial exposures insurance companies supported the obligatory because they are unable to access commercial crop insurance scheme, but on account of severe quota-share and or nonproportional reinsurance financial underwriting losses, today only three to cap or limit their liability to loss. companies are actively underwriting it. The in- surance industry’s main concerns are that, be- 3.88. The private commercial insurers are cause of the compulsory nature of the scheme, concerned about what they perceive to be un- they can only exert very limited control over fair competition from the farmer mutual in- individual risk selection and underwriting. They surance associations. The private insurers are therefore lack any incentives to invest in crop concerned that the mutual insurers are not sub- underwriting and claims management staff or in ject to insurance regulations and do not need to Agricultural Insurance Feasibility Study  99 maintain minimum capital or claims reserves. insurance sector to develop a sound and stable They also noted that the mutuals are not subject crop insurance market based on a partnership to the minimum crop insurance premium rates between the private and public sectors, while at and therefore undercut their rates, engaging in the same time providing time to educate farm- unfair competition. Finally, they expressed con- ers about the benefits of crop insurance so that cerns about the reputational risks that the insur- they would continue to purchase cover once the ance industry faces when the mutuals are not scheme was made voluntary. After six years, the able to meet their claims liabilities, as occurred scheme has failed to develop a strong crop in- in 2010. surance market or to educate farmers. The fact that the program is very unpopular with many Government farmers also suggests that it is failing to meet its social objectives. Finally, government faces 3.89. The Kazakhstan compulsory crop insur- major uncertainties over its financial exposure to ance scheme was launched with very well-in- claims. tentioned social objectives, but it is failing to achieve its financial and institutional objec- tives. The program was originally conceived as 3.90. In summary, the current obligatory crop a mechanism to ensure that all farm workers and insurance scheme is failing to meet the re- small peasant farmers would receive a minimum quirements of all three major stakeholders. For indemnity in the event of crop failure due to these reasons, government has asked the World drought or other natural and climatic perils. The Bank to provide technical support to review obligatory nature of the scheme was intended to the scheme and to identify practical options for be a short-term measure that would enable the strengthening it in the future. 100  Kazakhstan Chapter 4: Strategy and Options for Strengthening the Current Crop Insurance Program 4.1. This section presents a series of options ing crop insurance to large numbers of small to and recommendations for strengthening and medium farmers who are widely scattered over improving the current national compulsory very large geographic regions. The obligatory crop insurance scheme for the Government of crop insurance scheme has, however, suffered the Republic of Kazakhstan (GRK) to consider. major financial losses in the past three years due These options and recommendations are made to very severe droughts in parts of Kazakhstan, on the basis of the detailed diagnostic technical, and unless the scheme is reformed and strength- institutional, financial, and operational review of ened, it is in danger of failing. Any changes to the obligatory crop insurance scheme that was the existing scheme will need to be introduced presented in chapter 3. Where relevant, it draws gradually and phased in over a three- to five- on international practice and experience. year period, with the central aim of moving this program onto a sounder commercial basis and 4.2. A phased approach is recommended toward a more market-oriented approach. for strengthening and improving the current scheme and for gradually converting it into a 4.3. This study identified three major phases fully market-based system. Agriculture in Ka- and strategic options for moving toward a mar- zakhstan has had to adjust to the major struc- ket-based national crop insurance scheme. The tural changes that occurred at independence 20 phased approach offers government a sequential years ago, including the switch from collective set of measures that can be adopted over the next and state farming systems to individual house- three to five years to move this program onto a hold farms, small to medium commercial farms, much more sound commercial footing. If taken, and large production enterprises, along with the such measures should help the scheme to attract financial challenges this brought to individual international reinsurance support. The three farmers. These challenges include the need to phases include the following: invest in crop production machinery and equip- ment and to purchase seeds, fertilizers, and plant 1. Strengthening the current obligatory insur- protection chemicals. Since 2004, the GRK, the ance scheme and achieving financial stabil- Ministry of Agriculture (MoA), insurance com- ity (short term, one to three years) panies, and other stakeholders (including, since 2008, the farmer mutual insurance associations) 2. Transitioning toward a market-based crop have invested in the promotion and implemen- insurance system (short term, one to three tation of a mandatory crop insurance scheme. years) These investments laid the foundations for a na- tional scheme, and this is a considerable achieve- 3. Transforming the obligatory scheme into ment, given the logistical challenges of provid- a national commercial crop insurance pool Agricultural Insurance Feasibility Study  101 backed up by international reinsurance un- levels. Immediate changes in legislation that der a suitable public-private partnership should be considered include removing the mini- (medium term, three to five years). mum and maximum premium rates from the law and permitting insurance companies to set their Within each of these strategic options, a series own rates for each crop in each oblast and rayon. of legal and regulatory, technical, operational, Similarly, more flexibility should be given to in- institutional, financial, and reinsurance recom- surers to determine the sum insured they offer to mendations are made. each farmer under the loss of investment costs (LIC) policy. Phase 1: Returning the Obligatory Crop Insurance Scheme to Profitability and 4.7. Options for gradually phasing out oblig- Financial Stability atory crop insurance and replacing it with a sys- tem of voluntary insurance need to be consid- ered. In Kazakhstan, the GRK originally planned 4.4. This section sets out a series of short- to make crop insurance compulsory for the first term measures designed to return the crop in- two or three years only. It was thought that, once surance scheme to profitability, to ensure the farmers gained knowledge and experience with continued participation of private commercial it, crop insurance would become voluntary. To insurers, and to ensure financial stability of the date, however, the law has not been amended to system. permit voluntary insurance. Some of the major problems associated with obligatory insurance Legal and Regulatory Considerations are highlighted in chapter 3, including farmers’ dissatisfaction with crop insurance, which some 4.5. It is recommended that the current law regard as a tax, and the fact that farmers in low- be reviewed and reformed. There is a clear need risk regions who adopt high levels of risk man- to review and amend the Law on Compulsory agement practices and technology are currently Crop Insurance of 2004 to make the scheme cross-subsidizing farmers in the most risk-prone more market oriented and to ensure its viability areas, who perhaps should not be producing an- in Kazakhstan. International experience shows nual crops because of the very high frequency that crop insurance is most successful when it is and severity of drought and other causes of loss. implemented by the private insurance sector on The obligatory nature of the program also causes a strictly commercial basis and when insurance major problems for insurers because they are un- companies have responsibility for selecting risk, able to exercise any control over risk selection underwriting, and setting their own terms and or accumulation, which are key functions for un- conditions. Under a public-private partnership derwriting a commercially viable crop insurance (PPP), agricultural insurance legislation should program. As evidenced over the past six years, define the types of public sector support that most private insurers have lost a lot of money will be available and set the overall framework writing obligatory crop insurance policies, and for agricultural insurance. An agricultural insur- it is very unlikely that any new companies will ance law should not, however, prescribe specific register to underwrite this business so long as terms and conditions of cover, including sums cover is obligatory. While government has been insured, rates, and loss assessment procedures. very successful in promoting mutual insurers, These specifics should be the responsibility of some mutualsexperienced major losses in 2010, underwriters. threatening their future viability. Options for gradually phasing out obligatory crop insurance 4.6. In the short term, it is recommended that and replacing it with a combination of volun- the crop insurance law be amended to permit tary insurance and crop credit–linked insurance insurance companies to set their own policy should be explored further by the various work- terms and conditions, rates, and sum insured ing groups. 102  Kazakhstan 4.8. In the medium term, the crop insurance ricultural insurance legislation. Such an expert law should be amended to provide the basis for should be familiar with the legal and regulatory a fully market-based national crop insurance frameworks of major PPP agricultural insurance scheme under an appropriate public-private schemes in countries such as Spain and Turkey partnership. Under this strategic option, the that may have applications to Kazakhstan. GRK may wish to consider establishing some form of crop insurance pool, drawing on the ex- Technical Strengthening of the Loss of perience of countries, such as Spain, Turkey, and Investment Costs Policy the Republic of Korea, that operate commercial agricultural insurance pool schemes under strong 4.11. On the assumption that, in the short PPP relationships. The potential benefits of coin- term, government and the insurance sector will surance pools in agriculture are reviewed in this need to continue to offer the uniform LIC pol- chapter. The law would need to be amended to icy, a series of recommendations are identified reflect changes under a national pool program. for improving and strengthening this product. It is assumed that there will not be time prior 4.9. Any changes in the crop insurance leg- to the 2012 crop season to introduce any new islation need to consider both private commer- products or programs and that the standard LIC cial insurers and private farmer mutual crop policy will continue to be marketed for all in- insurance associations. As indicated in chapter sured crops. Some simple, practical changes for 3, the mutuals are regulated separately from the strengthening the design of the LIC policy are private insurance companies under the Law of identified in box 4.1, and these measures are re- the Republic of Kazakhstan on Mutual Insur- viewed further below. There should be time to ance, dated July 5, 2007 [[. Under this act, the introduce these amendments to the existing LIC farmer mutual crop insurance associations do policy for implementation in the 2012 season not have to meet any minimum capital reserves starting in May 2012. requirements to cover catastrophic losses, they have considerable flexibility in setting the pre- 4.12. In the short term, even if government mium rates charged their members, including the elects to maintain LIC crop insurance as an ability to reduce rates or suspend premium pay- obligatory class of insurance, special consider- ments the following year if they have achieved ation will need to be given to farmers located in an underwriting profit, and they are not legally Aktobe and West Kazakhstan (WKO). As shown obliged to settle claims that exceed their col- in chapter 3, Aktobe and WKO are so exposed lected premiums. The insurance companies are to drought losses that conventional crop insur- regulated both by the insurance act and by the ance cannot continue to be provided to farmers obligatory crop insurance law; they have to meet in these two oblasts. Alternative solutions need both minimum capital requirements and claims to be considered if the overall program is to be liabilities in full. In order to create a level play- stabilized and returned to profitability. The GRK ing field, it is important to bring mutuals under therefore needs to consider short-term amend- the same regulations as private commercial in- ments to the crop insurance law to reflect this surers. situation. 4.10. The obligatory crop insurance law needs 4.13. There is a need to introduce a sales pe- to be reviewed and redrafted to support the riod and final sales cutoff date(s) at least one introduction of market-based crop insurance. month prior to sowing to avoid situations of The law requires significant modifications and antiselection and moral hazard. Currently the amendments in order to move to a market-based policy is very exposed to antiselection because system. This task needs to be conducted by a le- the final date for binding cover is two weeks af- gal expert with knowledge of international ag- ter the completion of sowing. Farmers can there- Agricultural Insurance Feasibility Study  103 fore monitor the development of the season, and implications for premium rating under the cur- if preexisting drought conditions are develop- rent LIC basis of insurance and indemnity. See ing, they can purchase the highest level of sum the section on rating for further discussion. insured cover in the expectation of receiving a higher indemnity. Most multiple-peril crop in- 4.17. Government should amend the law to surance (MPCI) programs introduce final sales permit insurance companies to set their own cutoff dates for each crop in each region at least premium rates for each crop in each zone. one month prior to the recommended sowing Actuarial rating should be introduced to reflect dates. It is recommended that policy makers in differences in risk exposures between rayons in Kazakhstan introduce a similar sales cutoff date each oblast and possibly differences in technol- for the obligatory crop insurance scheme. ogy levels and risk exposures between different types of farmers.31 In conjunction with revisions 4.14. It is recommended that the cover period to the rating procedures, policy makers should be clearly stated in the policy wording and spe- study the merits of introducing an individual cifically that cover only incept following germi- farmer bonus-malus system in the future. Farm- nation of the crop and full stand establishment. ers who have not submitted claims in the past Full stand establishment is defined as more than would receive a premium discount at renewal, 10 centimeters in height for grasses (Graminae) while farmers who have submitted frequent and two-leaf stage for dicotyledons. This recom- claims would receive loaded premium rates. mendation is intended to avoid moral hazard, Since the Fund for Financial Support for Agri- whereby farmers fail to adopt the recommended culture (FFSA) maintains a database of claims seed varieties, sowing densities, and sowing dating back to 2005, it would be a relatively sim- dates. ple task to determine which farmers are eligible for a rate discount and which are subject to a rate 4.15. The insured unit for the purposes of loss increase. adjustment should be strengthened. Currently the insured unit (IU) is understood to be the 4.18. For the purposes of valuing actual rev- individual “field” of each crop. In northern Ka- enue in cases of partial crop losses, it is recom- zakhstan, fields may be extremely large (greater mended that a preagreed unit value be specified than 1,000 hectares), and field boundaries are in the policy wording at the time of contracting very poorly demarcated on the ground. In order insurance. The reason for this proposal is to to avoid conflict at the time of loss assessment, it overcome the drawbacks of the current scheme, is recommended that this definition be tightened where partial losses are valued at the current lo- up. For small units of less than 250 hectares, un- cal market price of the crop at the time of loss. derwriters may wish to consider redefining the Chapter 3 highlights the uncertainty this causes IU as “the sown area of all fields of the same for insurers and government alike because they crop type, grown in the same farm location.” cannot assess their financial liability until losses are incurred. It is normal on any MPCI policy, 4.16. It is recommended that crop insurers be given greater flexibility to set their own levels of sum insured with each insured, according 31  To adjust rates due to local farm-level factors, to the insured’s requirements for cover. It is underwriters in Argentina typically select several fac- recommended that government modify current tors that influence risk exposures and are easy to legislation to permit insurance companies to ne- verify in the event of a claim, including sowing dates, gotiate the levels of production cost-based sum previous crop, type of tillage, type and texture of soil, and evidence of waterlogging. These factors are duly insured with their clients—in other words, to elicited in the application form, and each is given a provide farmers with more choice. If sums in- score, which is then used to adjust premium rates sured are significantly increased, this will have up or down. 104  Kazakhstan Box 4.1 Recommendations for Strengthening Compulsory LIC Crop Insurance Policy Criteria for acceptance of risk or compulsion of cover. Even if government decides to maintain compulsory crop insurance for all producers in the short term, special consideration will need to be given to farmers located in Aktobe and WKO. Insured perils. Current coverage should be maintained to include loss or damage to crop production due to adverse weather events. Sales cutoff date. A policy sales cutoff date should be set, for example, April 1 or a date to be announced. Cover period. The cover period should be from the time of crop emergence and full stand establishment (for example, wheat at least 10 centimeters tall) through to completion of the harvest. Insured unit. The definition of the insured unit, which is currently defined as the “individual field,” should be strengthened. For small farms of less than 250 hectares, the IU could be defined as “the total area of all fields of the same crop grown in the same location or farm.” Sum insured. Government should amend the law to permit insurance companies to have the option to establish an agreed sum insured with each farmer according to the farmer’s own circumstances and crop insurance requirements. Premium rates. Government should amend the law to permit insurance companies to set their own premium rates for each crop in each zone. Actuarial rating should be introduced to reflect differences in risk exposures between rayons in each oblast and possibly differences in technology and risk exposures between different types of farmers. Bonus-malus system. Underwriters should introduce a bonus-malus system with the following objectives: • Introduce the rating of individual farmers within each rayon. Farmers who do not submit claims would be rewarded with reduced premiums over time; farmers who submit frequent claims would be penalized with higher premiums to reflect their higher risk exposure. • Reduce the tendency of farmers to submit speculative claims notices in the hope of receiving an indemnity. Farmers would only submit claims in cases where a major insured cause of loss has occurred that is likely to give rise to a settlement. This would reduce the costs of in-field loss assessment for the insurance companies. Basis of indemnity and claims settlement. In the case of a partial loss of area or yield, salvage (harvestable production) from the affected area should continue to be estimated in the field immediately prior to harvest. However, the crop sales price that is used to value the salvage should be preagreed based on an historical sales price for each crop in each rayon and stated in the policy wording. Where the value of salvage (crop revenue) falls short of the investment costs, the shortfall is indemnified. Source: Authors. including LIC policies, to preagree and specify crop production and yield they have to lose be- in the policy special conditions affecting the unit fore the value of salvage falls short of the costs of insured price that will be used to value crop out- production invested in growing the crop, thereby put, which is usually based on historical aver- opening the policy for a claim. A preagreed sales age farm-gate prices for each crop in each zone. price means that the farmer can calculate exactly Similarly, the current system is very unclear to how much yield he needs to lose in order to trig- farmers, who do not know how much of their ger a claims settlement. Agricultural Insurance Feasibility Study  105 Increasing the Sum Insured Levels scheme has operated on a breakeven basis. The premium rates were last revised in 2008, and the 4.19. The current LIC policy does not provide scheme results deteriorated badly over the past adequate levels of protection to the majority of three years, with an average loss ratio of 182 per- farmers. In most cases, farmers only purchase cent (91 percent net of the government 50 per- the cheapest “nominal costs” coverage, which cent claims subsidy). In some oblasts, including for spring wheat is equivalent to about 10 to 15 NKO and Kostanay, the scheme performed very percent of the value of the expected crop rev- well, but in others, including Aktobe, WKO, and enue. Increasing the levels of sum insured cov- Zhambyl, it performed very poorly. The scheme erage would have implications for all stakehold- is now severely underrated in these oblasts, and ers. For farmers, higher coverage would mean rates require adjusting on an actuarial basis. that the likelihood of receiving a claims pay- ment would be significantly increased and, in 4.21. The law specifies a system of minimum the event of a claim, the indemnity would be ad- and maximum premium rates per oblast, but equate to cover their production costs in full and given the differences in risk exposures and to ensure that they can repay any credit due. The claims experience between rayons, it is also downside would be that they would have to pay recommended that a rayon-level rating system higher premiums (see below for further discus- be introduced for the scheme. With up to six sion). For the private insurance companies and years of actual claims experience for each crop mutuals, increased coverage levels would have in each rayon or oblast, it is now possible to re- to be accompanied by actuarial rate increases, vise the premium rates and update them on an and, with the increased sum insured liability, actuarial basis in each rayon. A simple actuarial it would be essential for the insurers to have a analysis was conducted on the six-year actual comprehensive reinsurance protection program premiums and claims at the rayon and oblast in place. For government, the higher coverage level to examine the rate changes that would levels would mean that budgetary allocations be required to achieve a target loss ratio of 60 would have to be correspondingly higher to cov- percent (after application of the government 50 er the 50 percent claims reimbursements. There percent claims subsidies) on the scheme. The re- is a need to set the sums insured in each rayon sults of this analysis at the oblast level are sum- according to (a) the actual production costs of marized below, and the full results are presented different types of farmers in each rayon and (b) by rayon in annex 4. The six-year actual results the risk exposure in each rayon. For example, a at the oblast level indicate that, on average, the sum insured of KZT 10,000 per hectare could current premium rates are adequate to generate easily be covered in North Kazakhstan (NKO) a 60 percent loss ratio net of the government 50 at a reasonable premium rate, but to provide the percent claims reimbursement in Akmola, Ko- same high sum insured in Pavlodar or Aktobe stanay, Kyslorda, and NKO. (However, caution would require extremely high premium rates. should be exercised in interpreting the calcu- lated rates in Kyslorda because very little busi- Revising the Crop Premium Rating ness has been underwritten in this oblast to date, Methodology and the results may not be representative of the real pattern of claims.) In order to achieve a 60 4.20. There is a need to revise the oblast-level percent target loss ratio, very significant rate crop premium rating methodology and to up- increases would be required in Aktobe (aver- date the premium rates on an actuarial basis. age premium rate increase from 5.8 to 18.5 per- The six-year long-term gross loss ratio at end- cent), in WKO (from 7.8 to 33.0 percent), and 2010 was 140 percent (equal to a 70 percent loss in Zhambyl (from 3.5 to 12.1 percent). Smaller ratio net of the government 50 percent claims actuarial rate increases would also be required reimbursement), suggesting that, on average, the in East Kazakhstan (EKO) and Pavlodar in or- 106  Kazakhstan der to achieve a net loss ratio of 60 percent after efit from lower premium rates, high claimants government 50 percent claims subsidy (table 4.1 would be penalized through higher rates. The and figure 4.1). bonus-malus system would also help to reduce unnecessary loss assessment expenses for insur- 4.22. Scheme managers should also examine ers.32 The crop insurance law would need to be the merits of introducing a bonus-malus sys- tem to adjust the premium rates charged to individual farmers according to their own in- 32  The bonus-malus system could also be de- dividual claims experience for each crop over signed to reduce the number of false claims and the costs of unnecessary loss assessment visits. In Bra- the past six years. Since the system of obliga- zil, Allianca do Brasil has successfully implemented tory crop insurance has been in operation for six a bonus-malus system to avoid this problem. Farm- years, it would be possible to introduce a bonus- ers automatically lose a 5% bonus at the renewal of malus system for adjusting the premium rates of the policy if they submit a loss notice on their policy (regardless of whether or not the loss qualifies for a individual farmers according to their own claims claims settlement). Claims personnel must be trained experience over time: low claimants would ben- to explain this to the insured. Table 4.1 Adjustments in Average Premium Rates to Achieve an Average 60 Percent Target Loss Ratio, by Oblast Adjusted gross Adjusted net Six-year premium rate premium rate Six-year Implied % actual average for 60% target for 60% target Oblast average loss change in rate premium rate loss ratio loss ratio cost (2005–10) required (2005–10) (before FFSA (after FFSA intervention ) intervention) Akmola 1.9 2.0 3.38 1.69 −13 Aktobe 5.8 22.2 36.97 18.48 217 Almaty 3.5 4.6 7.64 3.82 10 EKO 3.4 6.0 9.96 4.98 47 Karaganda 4.7 4.7 7.86 3.93 −16 Kostanay 1.7 1.3 2.10 1.05 −40 Kysylorda 5.2 0.1 0.15 0.08 −99 NKO 2.0 0.5 0.79 0.40 −80 Pavlodar 3.6 7.1 11.81 5.90 66 South Kazakhstan 2.7 0.8 1.34 0.67 −75 (SKO) WKO 7.8 39.6 65.94 32.97 323 Zhambyl 3.5 14.6 24.27 12.14 243 Total 2.4 3.4 5.65 2.82 17 Source: Authors based on crop insurance results for 2005–10. Agricultural Insurance Feasibility Study  107 Figure 4.1 Average Actual Premium Rates and Premium Rates Required to Achieve a 60 Percent Loss Ratio (Net of Government 50 Percent Claims Subsidies), by Oblast 35.00% 30.00% Premium Rate (%) 25.00% 20.00% 15.00% 10.00% 5.00% 0.00% be y O KO O KO l al a ay a ar a by at rd ol nd t EK SK od To an to am m m N W lo ga Ak vl st Al Ak sy Zh Pa ra Ko Ky Ka Actual Average Premium Rate Adjusted Premium Rate 60% Target Loss Ratio (with government 50% claims subsidy) Source: Authors based on crop insurance results for 2005–10. amended to permit the introduction of a bonus- accompany higher sums insured is through an malus system. analysis of yield variation at the rayon and, pref- erably, individual farmer levels. Such an analy- 4.23. If in the future farmers elect higher lev- sis cannot be conducted on the past six years of els of sum insured under the salvage-based LIC claims experience. policy, this would have important implications for rating. Under the LIC policy, if a farmer Measures to Improve Scheme elects to insure a higher level of production costs, Profitability the threshold or breakeven yield for a given sales price of the crop increases. The risk of incurring 4.24. The six-year results show that the finan- an insured loss also increases with higher levels cial viability of the crop insurance scheme is of insured or breakeven yield, and this means being severely prejudiced by the inclusion of that higher premium rates would have to be ap- Aktobe and WKO, and the options for these two plied at higher levels of sum insured. Under the oblasts must be considered now if the scheme is current rating system, the setting of maximum to return to profitability. Chapter 3 shows that and minimum rates for each crop in each oblast over the past six years Aktobe and WKO in west- may reflect the range of sum insured, from the ern Kazakhstan collectively accounted for only low-technology, minimum costs of production 4.8 percent of total scheme liability but contrib- package to the high-cost scientific technology uted 41 percent of total claims and had six-year package. However, the only systematic way to long-term loss ratios of 381 and 507 percent, calculate the rate increases that would need to respectively. These two oblasts are severely af- 108  Kazakhstan fecting the financial viability of the national crop 3.3 billion or US$22.0 million) and a reduction insurance program, and measures of controlling in the six-year long-term loss ratio from 140 to the claims costs in these two oblasts are urgently 96 percent. Following application of the govern- needed. Options presented below include (a) in- ment 50 percent reimbursement of claims, the creasing government’s share of claims subsidies “as if” loss ratio for the insurance would have from 50 to 75 percent or even 100 percent in been reduced to about 48 percent (table 4.2). As WKO and Aktobe and (b) removing these two shown in figure 4.2, if Aktobe and WKO had not oblasts from the crop insurance scheme and es- been included in the scheme, the performance of tablishing a separate government compensation the scheme would have been significantly bet- mechanism for farmers in them. ter in 2005 (loss ratio reduced from 119 to 71 percent), in 2009 (loss ratio reduced from 131 to “As If” Analysis with and without Aktobe 46 percent), and in 2010 (loss ratio halved from and WKO 261 to 136 percent). Notably in 2008 the “as if” loss ratio would have increased slightly from 4.25. An “as if” analysis of the six-year re- 156 to 180 percent because in this year the ac- sults shows that the actual long-term loss ratio tual claims in Aktobe and WKO were very low. would be reduced from 140 to 96 percent with These two oblasts (in common with much of the elimination of Aktobe and WKO and that, the northern and central regions of Kazakhstan) following the government’s 50 percent par- were traditionally livestock-grazing regions that ticipation in claims, the program would have were converted to cereal (mainly spring wheat) generated sound underwriting profits for the production during Soviet times. The underly- insurance industry. If Aktobe and WKO had ing problem in most of Aktobe and WKO is that not been included in the scheme since 2005, this soils are mostly poor and average annual rainfall would have reduced total premium earnings by is very low. In a normal year, rainfall is barely 13 percent, but would have led to a huge sav- adequate for growing spring wheat in most of ings in claims costs for insurers of 40 percent of the rayons in these two oblasts. As such, these actual claims (or a reduction in claims of KZT oblasts are very marginal for spring wheat crop Table 4.2 Crop Insurance Accumulated Six-Year Results with and without Aktobe and WKO, 2005–10 Whole portfolio Modified portfolio Item % reduction (with Aktobe and WKO) (without Aktobe and WKO) Number of policies 140,961 134,193 −5 Total insured area (hectares) 73,770,915 69,453,803 −6 Sum insured (KZT thousands ) 242,631,438 231,048,598 −5 Premiums (KZT thousands ) 5,861,958 5,102,670 −3 Average premium rate (%) 2 2 −9 Claim payments (KZT thousands ) 8,222,776 4,910,314 −40 Loss ratio (%) 140 96 −31 Loss cost (%) 3.39 2.13 −37 Source: Authors based on FFSA crop insurance data for 2005–10. Agricultural Insurance Feasibility Study  109 Figure 4.2 Annual Loss Ratio with and without Aktobe and WKO, 2005–10 300% 261% 250% 200% 180% Loss ratio % 156% 136% 150% 131% 119% 100% 71% 70% 68% 70% 65% 46% 50% 0% 2005 2006 2007 2008 2009 2010 WITH Aktobe / WKO WITHOUT Aktobe / WKO Source: Authors based on FFSA crop insurance data for 2005–10. production, and no matter how a crop insurance in Aktobe and WKO on the understanding policy is structured, it will always be very heav- that the GRK will reimburse them for a higher ily exposed to loss and the technical and com- percentage of claims in these two oblasts. Cur- mercial premium rates that would have to be rently government reimburses insurers for 50 charged will be so high (an average commercial percent of the value of all paid claims. Under rate prior to application of government claims this option, the GRK would agree to indem- subsidies of between 37 percent in Aktobe and nify a higher percentage of claims incurred in 66 percent in WKO) as to make the scheme Aktobe and WKO only. According to the insur- commercially unviable in these two oblasts. ance industry, insurers have already asked gov- There are, however, a few areas where spring ernment to consider raising its share to 70 or 75 wheat production is more stable and where crop percent of total claims in these two oblasts in insurance might possibly be considered in the 2012. This study calculated the additional costs future, albeit at very high rates. These include to government of financing (a) 75 percent and Taskalisky and Kaztalisky rayons, located in the (b) 100 percent of claims in Aktobe and WKO southwest of WKO, and Karagaly and Martuk based on an analysis of the past six-year actual rayons, located in northern Aktobe. claims and the assumption that government re- imburses exactly 50 percent of all claims in the Future Options for Aktobe and WKO other oblasts. Over the past six years, govern- ment’s 50 percent contribution to claims costs 4.26. The first option is for insurance com- would have amounted to KZT 4.1 billion or an paies to continue underwriting the LIC scheme annual average of KZT 0.685 billion per year. If 110  Kazakhstan Table 4.3 Budgetary Implications to Government of Increasing the Claims Compensation Level from 50 Percent to 75 and 100 Percent of Claims in Aktobe and WKO, 2005–10 KZT thousands Option A: 75% government Option B: 100% government share of claims share of claims Government Total paid Cost to Cost to Year 50% share of % increase % increase claims government government claims 2005 1,064,870 532,435 656,273 23 780,110 47 2006 477,670 238,835 244,427 2 250,018 5 2007 700,538 350,269 377,709 8 405,149 16 2008 1,709,623 854,812 869,927 2 885,042 4 2009 1,465,129 732,565 996,650 36 1,260,735 72 2010 2,804,945 1,402,473 1,794,519 28 2,186,565 56 Total 8,222,776 4,111,388 4,939,504 20 5,767,619 40 Average 1,370,463 685,231 823,251 20 961,270 40 Source: Authors based on FFSA crop insurance data for 2005–10. Note: This analysis assumes that government’s share of total paid claims is exactly 50%. As discussed in chapter 3, over the past six years government’s share was slightly lower, at 47% of total claims. government would agree to increase its share of ing capacity of the insurance industry and in- claims in Aktobe and WKO, the “as if” annual creases the profitability of the program after the cost to government would increase to (a) option government’s increased claims reimbursements, A, 75 percent claims share in Aktobe and WKO: in the medium to long term it does not tackle the KZT 0.823 billion per year or an increase of 20 fundamental issue that commercial spring wheat percent, and (b) option B, 100 percent claims crop insurance is not financially viable in most share in Aktobe and WKO: KZT 0.961 billion of the rayons located in Aktobe and WKO. per year or an increase of 40 percent. Option B shows the maximum expected costs to govern- 4.27. The second option for government to ment if it wishes to continue including Aktobe consider is whether to take Aktobe and WKO and WKO in the LIC scheme and assuming out of the crop insurance scheme altogeth- that the insurance companies act purely as ad- er and to establish a separate disaster relief ministrators in these two oblasts, but do not ac- scheme for producers of annual crops in these cept any claims liability. The “as if” 100 percent two oblasts. The argument for taking Aktobe claims cost to government in these two oblasts and WKO out of the scheme altogether centers would be an average of about KZT 550 million on the following: these areas (a) have marginal or US$3.7 million per year (see table 4.3). While rainfall and are not suitable for growing annual this option may provide government a short-term grain crops and (b) are effectively uninsurable, solution that retains the support and underwrit- with six-year long-term loss costs of 22 and 40 Agricultural Insurance Feasibility Study  111 Box 4.2 Options for Streamlining and Reducing the Costs of Policy Marketing, Preinspections, and Loss Assessment Systems and Procedures Introduce preinspections. Because the scheme is compulsory, preinspections should reduce potential antiselection and moral hazard, which are often associated with MPCI schemes. Moreover, few of the insurance companies have regional networks of trained field agronomists to conduct preinspections and loss assessment, and this task is likely to be prohibitively costly for small farm units. Finally, a system of sample preinspections should be considered for the large farms in northern Kazakhstan. Streamline loss notification and loss assessment procedures to make them more cost-effective. The law stipulates that five individuals (organizations) must be involved in field-level loss assessment, and this is a very costly exercise for each party. In addition, the composition of this committee is weighted against the insurer. It is recommended that loss assessment be simplified. In most countries, loss assessment functions are carried out exclusively by the insurer’s own trained loss assessors or by a third-party qualified loss adjuster, and the farmer attends the loss assessment exercise. The loss assessor adopts standardized procedures for estimating the amount of yield reduction or loss, and the farmer is required to approve or refute the assessed loss, in which case a second assessment is conducted. No other parties are required. In Kazakhstan, farmers are required by law to obtain a meteorological report from the nearest meteorological weather station to prove that an insured event has occurred. This requirement does not seem relevant in the case of drought (which accounts for 95% of all losses to date), which is a progressive peril as opposed to being a specific event. It is recommended that this requirement be discontinued. Farmers should, however, be required to submit a loss notification report to the insurer within a specified period—for example, 48 hours for specific events (such as hail, flood, and autumn frost)—and to notify drought losses when these become apparent. Finally, a bonus-malus system should be introduced to dissuade farmers from submitting claims save where a major loss has occurred and is likely to lead to an indemnity. Use remote sensing to support in-field loss assessment. The NSA already uses remote sensing to estimate the sown area, production, and yield for the MoA and to monitor crop status during the growing season. It is recommended that managers of the crop insurance scheme review potential supporting roles by the NSA at the time of loss assessment. Review marketing and distribution channels with a view to reducing costs. Most policies are sold through local agents located in each oblast and subregion. The agents are currently paid 10% brokerage by law. Alternative crop insurance marketing and distribution channels should be promoted in order to reduce costs, including sales though cooperatives and farmer associations, input dealers, rural banks, and grain merchants. Support farmer education and training programs on crop insurance. In conjunction with the proposed improvements in operating systems and procedures, greater emphasis needs to be placed on farmer education and training programs that explain the basis of insurance and indemnity in the LIC program. Source: Authors. percent, respectively, and catastrophic losses ev- The cost of this program for Aktobe and WKO, ery other year, with loss ratios as high as 1,000 assuming the same 100 percent original claims percent in major drought years. It is therefore costs and compensation levels as under the cur- necessary for government to consider whether rent scheme, would average about KZT 550 mil- to continue promoting annual crops in these lion (US$3.7 million) per year, with a peak of oblasts and to use a separate disaster relief fund about KZT 1.6 billion (US$10.7 million) in a to compensate farmers in severe drought years. very severe drought year, such as 2010. 112  Kazakhstan Measures to Strengthen Operating and assessing the loss, and the process is very Systems and Procedures time-consuming, costly, and weighted against the insurer. In most countries, loss assessment 4.28. In conjunction with the changes iden- is carried out exclusively by the insurer’s own tified to strengthen the LIC policy, a series of trained loss assessors or by a third-party quali- potential measures should be considered to im- fied loss adjuster, and the farmer attends the prove underwriting and claims operating sys- loss assessment exercise. No other parties are tems and procedures and to reduce the costs of involved. The loss assessor adopts standardized these operations. Box 4.2 summarizes some of procedures for estimating the amount of yield the key operational areas that require strength- reduction or loss, and the farmer is required to ening. approve or refute the assessed loss, in which case a second assessment is made; if the insured 4.29. Loss notification procedures are cum- is still not satisfied with the outcome of the loss bersome and require the insured to obtain a assessment, he is entitled to request arbitration. weather report from the nearest meteorological It is recommended that in Kazakhstan scheme station to validate that an insured event has oc- management review the in-field loss assessment curred. While an event report might be used to procedures and reduce the number of persons prove that a specific peril such as frost or exces- required to attend the loss assessment. In some sive rain leading to flooding has occurred (but countries, crop loss assessment functions are not hail, as this is usually very localized), such a contracted out to specialist companies of loss report is usually irrelevant in the case of a pro- adjusters. Although Kazakhstan does have certi- gressive peril, such as drought, which is the ma- fied and approved companies of loss adjusters, jor cause of loss in Kazakhstan, accounting for at present none specializes in crop loss assess- more than 95 percent of claims. It is recommend- ment; in the medium term, options for creating ed that the loss notification procedure be contin- such specialist entities should be explored. ued, but that the requirement for the insured to submit a meteorological report be dropped. This 4.31. In the future, it may be possible to use will require an amendment to the crop insurance remote sensing to aid the crop loss assessment law. The introduction of a bonus-malus system process. The National Space Agency (NSA) has is also recommended to reduce the propensity a very well-developed remote-sensing capability of farmers to declare small losses and to declare to provide crop monitoring at a resolution of the only major losses that are likely to give rise to a “individual field” in northern Kazakhstan, where claims settlement. This measure should reduce fields tend to be much larger than in the south. the number of in-field loss assessment exercises There are many potential applications of the that need to be conducted and lead to cost sav- NSA’s crop-monitoring services for the existing ings. crop insurance scheme, especially to determine yield at the farm level. It is recommended that 4.30. There is a need to review the in-field loss management of the scheme meet with the NSA assessment procedures with a view to stream- to review remote-sensing services that could po- lining this activity and making it more cost-ef- tentially be provided for monitoring major loss fective. Currently the law stipulates that once a events. loss has been notified by the insured farmer, rep- resentatives from five organizations must form a 4.32. It is also recommended that policy mar- committee to assess the loss in-field. This com- keting and distribution channels be reviewed mittee comprises one representative from the lo- with a view to reducing administrative costs. cal administration, the FFSA, the local agent, the Currently most policies are promoted and sold insurer, and the insured farmer. Each party is re- by individual sales agents in each oblast and sponsible for covering its own costs of attending rayon, but potentially costs could be reduced if Agricultural Insurance Feasibility Study  113 these functions were channeled through rural ers. Failure to do so creates biased competition aggregators such as farmer cooperatives, input and raises the chances that mutuals will fail to dealers, and rural banks, including microfinance meet their financial obligations. institutions (MFIs). Financial and Reinsurance 4.33. There is a need to accompany improve- Considerations ments in the design and implementation of the crop insurance scheme with education and 4.36. Crop insurance companies in Kazakh- training programs to raise the level of farmer stan are very exposed to catastrophic losses on awareness of the role of crop insurance in risk their retentions, and options for enhanced rein- management. Many farmers do not appear to surance protection need to be considered. The understand how the LIC policy works and are GRK currently provides free proportional rein- skeptical of the benefits of crop insurance. In surance protection equal to 50 percent of claims parallel to the improvements in policy design, to the private insurance companies and mutuals rating, and operational procedures, it is recom- in Kazakhstan. However, neither the private in- mended that government provide support to surers nor the mutuals have any reinsurance pro- commercial insurance companies and mutual in- tection on their 50 percent retentions, and they surers to conduct crop insurance education and are therefore very exposed to major systemic training programs. drought losses. Private insurance companies are better able to absorb and settle catastrophic loss- es than the mutuals because of their much larger Measures to Strengthen Institutions size, formal requirements for capital and claims reserves, and diversified nonlife insurance port- 4.34. In the short term, finding ways to encour- folios, in which crop insurance only represents age more private commercial insurance com- a very small fraction of their overall premium panies to support the crop insurance scheme earnings and overall liability. However, there is a major challenge. Several measures were is still a need to design affordable reinsurance identified that could be attractive to local insur- protection for the private companies on their 50 ers, including (a) the introduction of actuarially percent retentions through some form of stop- determined premium rates, (b) the implementa- loss reinsurance protection. See figure 4.3).. tion of measures to reduce insurers’ liability in Aktobe and WKO, and (c) the implementation 4.37. The farmer mutual crop insurance asso- of measures to strengthen loss assessment pro- ciations are very exposed to losses that exceed cedures and give insurers more direct control their members’ premium contributions, and over this important function. However, in the ways of capping their exposure to catastrophic medium term, it is also probable that insurers losses through some form of nonproportional will insist on being given greater control over or stop-loss reinsurance protection need to be risk selection and underwriting if they are to join developed. The mutuals are usually small, do not the scheme. have reserves, and only underwrite crop insur- ance business. They have to settle claims out of 4.35. There is also a need to create a level the net premiums they receive from their mem- playing field for commercial insurers and bers (which are typically on the order of 70 to farmer mutual insurance associations. Mu- 75 percent of gross premium after payment of tual insurance associations should in the future acquisition costs and the mutual’s own adminis- be regulated under the Agency for Financial trative and operating expenses); in years where Market and Financial Institutions Regulation the actual claims exceed the average, they are and Control and be required to follow the same unlikely to be able to meet their liabilities in full. guidelines with regard to capital requirements If the mutuals are to become a financially stable and insurance reserves as the commercial insur- and viable alternative to private commercial 114  Kazakhstan Figure 4.3 Structure of Proportional (Quota-Share) and Nonproportional (Stop-Loss) Reinsurance Proportional quota-share Loss ratio = ∞ Nonproportional stop-loss retention 50% insurance company 50% government reinsurance Stop-loss reinsurance layer Loss ratio = 100% Insurer primary retention Loss ratio = 0% Source: Authors. crop insurance, it is essential to put in place a Simple Burning Cost Analysis for formal risk layering and proportional or nonpro- Government Excess of Loss Claims Cover portional risk transfer (reinsurance) program in order to cap their losses at somewhere between 4.39. An “as if” analysis was carried out on 70 and 100 percent of their premiums.33 the six-year actual claims under the assump- tion that, instead of providing 50 percent quo- 4.38. In the short term, it is unlikely that the ta-share reinsurance protection, government current scheme will be able to meet the stan- provided aggregate nonproportional excess of dards required by international reinsurers, and loss claims compensation for the crop insur- any nonproportional reinsurance solutions ance industry. This analysis assumes that, if will probably have to be provided by the GRK. government had not provided free 50 percent To date, the GRK has paid crop insurers a total proportional claims compensation over the past of KZT 3.8 billion in 50 percent quota-share re- six years, the insurers would have had to have insured claims or an average of KZT 630 million charged premium rates that were twice as high per year. This section briefly reviews whether it to cover their 100 percent claims liability. There- would be more cost-effective for the govern- fore, the actual six-year annual average premi- ment to support an excess of loss program or to ums were doubled to represent the original gross finance premium subsidies or both. premiums the insured would have had to have charged. Two “as if” claims scenarios were ana- lyzed. The first is that insurers retain claims up 33  In insurance terms this is referred to as gross to 100 percent of the value of the original gross net premium income, which is the original gross pre- mium the company has earned net of any policy can- premiums (that is, up to 100 percent loss ratio) cellations and returns of premium. and that government provides unlimited protec- Agricultural Insurance Feasibility Study  115 Table 4.4 “As If” Analysis of Costs to Government of Aggregate Insurance Industry Excess of Loss Protection and Full Premium Rates Charged to Farmers, 2005–10 KZT (thousands) “As if” cost to “As if” cost to Actual premium “As if” 100% gross government for government for Scenario and (with govern- premium (without Actual total losses in excess losses in excess year ment 50% claims government 50% paid claims of 100% gross 70% gross pre- reinsurance) claims reinsurance) premium mium With Aktobe and WKO 2005 898,607 1,797,214 1,064,870 0 0 2006 684,722 1,369,444 477,670 0 0 2007 997,392 1,994,785 700,538 0 0 2008 1,093,232 2,186,463 1,709,623 0 179,099 2009 1,114,366 2,228,732 1,465,129 0 0 2010 1,073,639 2,147,278 2,804,945 657,668 1,301,851 Total 5,861,958 11,723,915 8,222,776 657,668 1,480,950 Average 976,993 1,953,986 1,370,463 109,611 246,825 Without Aktobe and WKO 2005 801,657 1,603,315 569,520 0 0 2006 673,999 1,347,999 455,304 0 0 2007 902,616 1,805,232 590,778 0 0 2008 916,678 1,833,355 1,649,162 0 365,814 2009 897,503 1,795,005 408,788 0 0 2010 910,218 1,820,435 1,236,762 0 0 Total 5,102,670 10,205,340 4,910,314 0 365,814 Average 850,445 1,700,890 818,386 0 60,969 Source: Authors based on FFSA insurance data. tion for any losses in excess of 100 percent of ment claims compensation program cuts in for gross net premium income (GNPI). The second any losses in excess of 70 percent of GNPI. A is that, in recognition that insurers have to pay further assumption is that initially government business acquisition costs (brokerage) and ad- excess of loss claims protection is provided free ministrative and operating (A&O) costs out of of charge to the insurers. Finally, the “as if” their original gross premiums, which reduces the analysis was carried out with and without Ak- premium they retain to cover claims, the govern- tobe and WKO. The main caveat of this analysis 116  Kazakhstan is that the government excess of loss compensa- curred any claims in the past six years; a priority tion cover can only be analyzed on an annual of 75 percent would have incurred only one small aggregate industry-level basis and not by indi- claim in 2008 of KZT 0.27 billion. However, vidual private commercial insurance company great caution must be exercised in interpreting or farmer mutual insurance association. In prac- this simple six-year aggregate “as if” analysis. tice, individual companies underwrite different First, it is based on only six years of data and, portfolios in different regions of the country although it includes 2010, which was a very se- and incur different ratios of premium to claims; vere 1-in-50-year drought and biases the results, the claims to the government stop-loss program even more severe 1-in-100-year losses could oc- would inevitably be higher than can be modeled cur. Second, the aggregate analysis excluding under this simple “as if” analysis applied to the Aktobe and WKO is unrealistic and masks the historical claims. reality that individual insurance companies and mutuals would have incurred losses in excess 4.40. With Aktobe and WKO included, the of 100 percent of their premium on their own government-funded aggregate industry excess regional crop insurance portfolios over the past of loss compensation with a priority of 100 per- years. The analysis is presented to illustrate the cent of GNPI would have incurred only one option of capping industry losses in catastrophe claim in the six-year period—KZT 0.66 bil- years through some form of government catas- lion in 2010 (average of KZT 110 million per trophe claims compensation. year); with a priority of 70 percent of GNPI, it would have incurred two claims—a small one Indicative Rating for Aggregate Stop-Loss in 2008 and another in 2010—with total stop- Reinsurance Protection loss claims of KZT 1.5 billion (average of KZT 347 million per year). As shown in table 4.4, if 4.41. As an extension of the simple six-year “as Aktobe and WKO had been excluded, a priority if” analysis, some preliminary modeling was of 100 percent of premium “aggregate” excess conducted for spring wheat (which accounts for of loss compensation cover would not have in- more than 95 percent of the obligatory crop in- Table 4.5 Rating Analysis for Aggregate Stop-Loss Protection for Obligatory LIC Crop Insurance Scheme: Spring Wheat Cover Only Stop-loss premium Amount (KZT Option Limit (% of TSI) XS. Priority (%TSI) (%TSI) billions) With Aktobe and WKO (total sum insured: KZT 41.8 billion) Priority 70% of 96.56 xs 3.44 0.75 0.313 GNPI Priority 100% of 95.09 xs 4.91 0.64 0.266 GNPI Without Aktobe and WKO (total sum insured: KZT 39.2 billion) Priority 70% of 96.65 xs 3.35 0.66 0.271 GNPI Priority 100% of 95.22 Xs 4.78 0.57 0.225 GNPI Source: Authors. Agricultural Insurance Feasibility Study  117 surance portfolio) to estimate the layering and the “as if” average annual costs to the GRK of pricing on an aggregate stop-loss protection stop-loss protection and 50 percent premium cover for the existing LIC scheme. This analy- subsidies of between KZT 850 million and KZT sis used simulation techniques to calculate more 911 million per year (table 4.6). than 5,000 iterations of the expected aggregate losses (loss costs or loss ratios) over the entire 4.43. It is recommended that the GRK study scheme. The results of the simulation exercise very carefully the issues surrounding premium were then used to estimate the value of claims subsidies before deciding whether to switch in excess of the specified priorities (assuming 70 from the current system of claims subsidies to and 100 percent priorities and the provision of premium subsidies. In the current system, gov- full-value reinsurance protection for claims in ernment compensates 50 percent of the claims excess of these priorities). Appropriate loading costs and then caps premium rates at approxi- was then added to the calculated technical rates mately 50 percent of the technically required to cover reinsurers’ expected costs and profits. rates. In some regions of the country, current The results of this stop-loss rating analysis are premium rates are above the actuarially required presented in table 4.5 for the two priorities with rates, and in other parts of the country actual and without Aktobe and WKO. While caution rates are far too low. On the one hand, this re- must be exercised because this is a preliminary sults in distorted crop insurance price signals rating analysis, it is considered robust and sug- in the market; on the other hand, the 50 percent gests that the costs of providing full-value (that claims compensation does not provide local in- is, up to 100 percent of the total sum insured) surers with the required catastrophe protection aggregate stop-loss protection in excess of 100 on their retained claims. Finally, international and 70 percent priorities would be on the order reinsurers are not willing to support an under- of KZT 0.27 billion and KZT 0.31 billion per priced scheme. While the authors are very cau- year, respectively. tious about recommending premium subsidies, it would be preferable in Kazakhstan to have an Premium Subsidy Considerations actuarially rated and commercially priced pro- gram and for government first to decide whether 4.42. With the switch from 50 percent propor- tional reinsurance to nonproportional reinsur- to provide financial support in the form of pre- ance and the parallel need for insurers to dou- mium subsidies or nonproportional stop-loss re- ble their premium rates, the GRK may need to insurance protection (in the short term) and then consider whether farmers can afford these rate to promote the participation of international re- increases or whether it will be necessary to in- insurers (in the medium term). troduce premium subsidies. In order to illustrate the potential costs of premium subsidies, the six- 4.44. In conclusion, while further analysis year analysis was expanded to include both the is required to study the options for introduc- aggregate industry-level stop-loss reinsurance ing a government-financed nonproportional protection and 50 percent premium subsidies. stop-loss reinsurance program for individual Over the past six years, government’s 50 percent private and mutual insurers, this option has of claims liability averaged KZT 685 million per potentially large benefits. The most important year, but in severe drought years, such as 2010, benefit would be the guarantee that, if a mutual government’s 50 percent claims share was much or indeed a private commercial insurer incurred higher, at KZT 1.4 billion. If Aktobe and WKO catastrophic losses in excess of its premium and had been excluded from the scheme, govern- reserves, the government would indemnify these ment’s 50 percent quota-share claims reimburse- losses rather than having the insurer default on ment would have been reduced to an average of its obligations to farmers. However, the impli- KZT 409 million per year. This compares with cations for insurers of moving to a nonpropor- 118  Kazakhstan Table 4.6 “As If” Analysis of Costs to Government of Aggregate Insurance Industry Excess of Loss Compensation Scheme and Assumed 50 Percent Premium Subsidies, 2005–10 KZT (thousands) “As if” cost “As if” 100% “As if” cost to Total cost to Total cost to to govern- gross premium government Government government for government for ment for Scenario (without for losses 50% losses in excess losses in excess losses in and year government in excess of premium of 100% premium of 70% premium excess of 50% claims 70% gross subsidies + 50% premium + 50% premium 100% gross reinsurance) premium subsidies subsidies premium With Aktobe and WKO 2005 1,797,214 0 0 898,607 898,607 898,607 2006 1,369,444 0 0 684,722 684,722 684,722 2007 1,994,785 0 0 997,392 997,392 997,392 2008 2,186,463 0 179,099 1,093,232 1,093,232 1,272,331 2009 2,228,732 0 0 1,114,366 1,114,366 1,114,366 2010 2,147,278 657,668 1,301,851 1,073,639 1,731,307 2,375,490 Total 11,723,915 657,668 1,480,950 5,861,958 6,519,625 7,342,908 Average 1,953,986 109,611 246,825 976,993 1,086,604 1,223,818 Without Aktobe and WKO 2005 1,603,315 0 0 801,657 801,657 801,657 2006 1,347,999 0 0 673,999 673,999 673,999 2007 1,805,232 0 0 902,616 902,616 902,616 2008 1,833,355 0 365,814 916,678 916,678 1,282,491 2009 1,795,005 0 0 897,503 897,503 897,503 2010 1,820,435 0 0 910,218 910,218 910,218 Total 10,205,340 0 365,814 5,102,670 5,102,670 5,468,484 Average 1,700,890 0 60,969 850,445 850,445 911,414 Source: Authors based on FFSA insurance data. tional reinsurance program would be that they very unpopular with farmers unless government would have to charge actuarially determined were to subsidize premiums. Issues relating to rates (rates would have to be roughly doubled on the provision of premium subsidies are reviewed average), and this move would likely prove to be further at the end of this chapter. Agricultural Insurance Feasibility Study  119 Phase 2: Transition toward a Market- and support in the form of premium subsidies Based Crop Insurance System and nonproportional reinsurance by government (and possibly by international reinsurers). It is 4.45. This section explores a set of issues and expected that these market-oriented measures options for the GRK to consider during the will encourage insurers to sign up for the crop transition to a market-based crop insurance insurance program. system. These center on the introduction of in- dividual grower MPCI, either as a complement 4.47. In addition, measures must be taken to to or as a substitute for the current LIC policy, ensure that the farmer mutual crop insurance and the introduction of commercial international associations will remain in the Kazakhstan reinsurance. It is assumed that crop insurance crop insurance scheme, but only if they are able to meet their financial liabilities in full. will continue to be obligatory for farmers during As noted in the previous section, in 2010 some this interim phase. This is a decision that GRK mutual crop insurers were not able to meet their will have to review with the local stakeholders. financial liabilities in full and, where claims Moreover, it is assumed that Aktobe and WKO exceeded their net premiums, were forced to will no longer be included in the crop insurance pro rata down each claimant’s paid claim. Such scheme because spring wheat cannot be com- actions contravene the purpose of contracting mercially insured in these two oblasts or in any crop insurance and erode farmers’ confidence in of the individual rayons. It is assumed that farm- the program. Therefore, measures are needed to ers in these two oblasts will be protected by a ensure that suitable excess of loss reinsurance is separate disaster relief mechanism, but details of in place. this separate program were not considered fur- ther under the current feasibility study. Introduction of New Crop Insurance Products and Programs Interim Measures to Crowd in the Private Commercial Insurance 4.48. There is a need to offer more choice of Companies crop insurance products in order to satisfy the risk transfer requirements of different types 4.46. Government is keen to reform the com- of producers growing different types of crops pulsory crop insurance program to encourage located in different agro-climatic regions of greater participation by private commercial in- Kazakhstan. For the past six years, Kazakhstan surers. The past six years have seen an exodus has implemented a single LIC multiple-peril of private commercial insurers from the Kazakh- crop insurance policy for grains, oilseeds, fibers stan insurance scheme, and if the scheme is to (cotton), and some leguminous and root crops. remain viable, this exodus must be halted now. While the LIC product is suitable for grains, oil- Indeed, measures need to be taken to crowd in seeds, and other field row crops, it is not suitable as many of the private commercial insurers as for most horticultural and fruit crops, especially possible if the scheme is to diversify its range of those that have multiple or staggered harvests. crop insurance products and to offer other class- Furthermore, the product is poorly suited to ir- es of agricultural insurance cover, including for rigated crops. If Kazakhstan is to develop a vol- livestock, forestry, and possibly aquaculture. In untary crop insurance market that is demand the short term, it is predicted that excluding Ak- driven and meets farmers’ risk management and tobe and WKO will encourage nonlife insurance risk transfer needs, it will be necessary to start companies to join the crop insurance scheme. investing in the design and rating of new crop This section discusses market-oriented mea- insurance products. sures, including the introduction of new crop in- surance products, starting with individual grow- 4.49. Under this World Bank feasibility study, er MPCI, actuarially determined premium rates, risk assessment and product design and rating 120  Kazakhstan analyses were conducted to assess the potential tant respects. First, a preagreed insured yield for introducing four new types of crop insur- is valued at the time of policy subscription (the ance products into Kazakhstan, including both insured yield is usually calculated as a percent- traditional indemnity-based products and new age of the individual farmer’s historical average index-based products. These products include or normal crop yield or the local area average individual grower MPCI, which is reviewed in yield). Second, a preagreed unit valuation price this chapter, and three products that are covered is applied to the insured yield to calculate the in chapter 5: traditional crop hail insurance, sum insured. Third, loss assessment involves area-yield index insurance (AYII), and weather measuring the actual yield and comparing it to index insurance (WII). the insured yield; the amount of shortfall is then indemnified at the preagreed valuation price. Individual Grower MPCI for Kazakhstan Basing insurance and indemnity on loss of yield is potentially much more transparent and under- Features and Challenges for Introducing standable for farmers, and loss assessment is MPCI also much more objective, as it measures yield loss rather than comparing the expected shortfall 4.50. There are a number of drawbacks of in production costs to the estimated value of the the LIC multiple-peril yield-based crop insur- remaining crop (salvage revenue). ance product that Kazakhstan’s insurers have underwritten for six full years. Although one 4.53. The international experience with MPCI of the principal attractions of the LIC policy is is that the product is very popular with farmers, that it does not require accurate yield data on in- but on account of the high premiums associated dividual growers, there are several drawbacks, with it, most schemes are dependent on govern- which are highlighted in chapter 3: (a) farmers ment support in the form of premium subsidies. dislike the product because it does not provide The international literature on MPCI highlights them with a clearly established loss of yield several drawbacks of voluntary schemes, includ- guarantee that they can understand, (b) the basis ing antiselection and moral hazard, the difficulty of indemnity is complicated and highly depen- of establishing average farmer yields and corre- dent on the prevailing market price for the crop sponding premium rates, the high costs of pre- at the time of loss, and (c) the product is com- miums, which require government support in the plicated to rate accurately for different levels of form of subsidies, and the often very high costs LIC sum insured. of individual grower in-field loss inspection and assessment. While these drawbacks are indeed 4.51. Given the insurance industry’s experi- very valid, they apply as equally to the existing ence with underwriting the LIC yield-based LIC policy in Kazakhstan. Currently issues of policy for the past six years, it would be rela- antiselection are less of a problem because the tively simple to make the transition from this scheme is obligatory for all farmers, but because cover to a more standardized individual grower it is a loss of yield multiple-peril scheme, it MPCI policy. The crop insurers of the LIC pol- shares the drawbacks of other MPCI schemes. icy have gained considerable experience in un- derwriting loss of yield MPCI and in conducting 4.54. In Kazakhstan, the main challenges for in-field loss assessment to establish actual yields introducing individual grower MPCI concern and the amount of loss. This experience would the procedures for establishing an individual enable them relatively easily to design, rate, and grower “normal average yield” for a given implement individual grower MPCI. crop and the basis on which to establish an in- sured yield as a percentage of the average yield. 4.52. An individual grower MPCI product is With the possible exception of the United States, different from the LIC policy in several impor- very few countries maintain accurate time-series Agricultural Insurance Feasibility Study  121 databases on the crop production and yields of MPCI cover or an area-yield index cover. individual farmers. MPCI is, however, the most popular crop insurance product with farmers, 4.56. A further challenge pertains to the rat- and it is widely underwritten by commercial ing procedures for MPCI cover when indi- insurance companies throughout the world, in vidual grower yield data are not available and both developed and developing countries. In rayon-level yield data have to be used. MPCI most countries, insurers use regional (county, rating is based on a statistical analysis of varia- district) historical crop production and yield data tion in time-series yields. The individual grower to establish an average yield and then an insured MPCI rating exercise for Kazakhstan was based yield or optional insured yield as a percentage on 17 years of rayon-level spring wheat annual of the regional average yield. Where high lev- yields for production enterprises, commercial els of insured yield coverage are offered—for farmers, and the rayon for both groups of farm- example, 75 percent or greater of the regional ers. The rating methodology involved cleaning yield—this may lead to major antiselection by and detrending the rayon average yields and farmers whose average yields are normally be- estimating the average expected yield for each low the regional average, usually on account of rayon based on the average yield for the last five their below-average technology and lower use years. Insured yield coverage levels of between of inputs. In Kazakhstan, most farmers except a minimum of 10 percent and a maximum of 40 for the very big production enterprises are un- percent of the rayon level’s expected yield were likely to be able to provide data on their histori- then selected, and the losses for each coverage cal individual crop yields by field and by farm level were simulated 5,000 times to establish for the past 10 years or more. It is therefore pro- the pure loss cost rates. In order to estimate the posed that MPCI be developed on the historical higher variability in yield between farmers, the rayon-level yield data that, in the case of spring coefficients of variation around mean yield were wheat, are available for up to 17 years and can increased by a factor of 15 percent. These were be disaggregated by type of farmer into produc- then adjusted to include catastrophe loadings, tion enterprises and commercial farms, if re- to derive the technical rates, and to determine quired. In Kazakhstan antiselection is less of an the indicative commercial premium rates for a issue because the maximum insured yields that target 60 percent loss ratio that would need to are offered do not exceed 30 to 40 percent of the be charged to individual spring wheat farmers in rayon average yield; only in exceptional cases each rayon. where average yields are very stable within a rayon might 50 percent coverage be provided. Insured Yields and Sum Insured 4.55. In Kazakhstan the methodology adopted 4.57. Under an MPCI policy, the sum insured by the National Agency of Statistics (ARKS) contains two elements: an insured yield and for sampling and recording crop sown, harvest an agreed unit valuation price. The analysis of area, and yields at the rayon level is consid- variation in 17-year spring wheat yields found ered to be accurate and could form the basis these to be so variable that the maximum insured on which to design either an individual grow- yields that could be offered to farmers in individ- er MPCI cover or AYII, at least in the case ual rayons were no more than about 40 percent of spring wheat. Under this study, a detailed of the expected yield. For the purpose of this ex- review was conducted in the eight main grain- ercise, spring wheat was valued on the basis of producing oblasts of the rayon-level crop area, the average national wheat prices for the period production, and yield data for spring wheat for 2008 to 2010: KZT 3,120 per centner. However, the past 17 years. It was concluded that these any unit valuation price could be used, based on data are accurate and can form the basis for the the costs of production per centner and the final design and rating of either an individual grower sales price. 122  Kazakhstan Table 4.7 Example of the Calculation of the Sum Insured for an MPCI Policy: Spring Wheat Grown in Akkol Rayon, Akmole Oblast Item and type of farmer Production enterprise Commercial farm Overall Expected yield (centner per hectare) 8.9 8.1 8.7 Insured yield coverage options (centner per hectare) 10% coverage 0.9 0.8 0.9 20% coverage 1.8 1.6 1.7 30% coverage 2.7 2.4 2.6 40% coverage 3.5 3.2 3.5 50% coverage 4.5 4..0 4.3 Unit insured value (KZT per centner) 3,120 3,120 3,120 Sum insured (KZT per hectare) 10% coverage 2,761 2,515 2,702 20% coverage 5,522 5,029 5,404 30% coverage 8,284 7,544 8,106 40% coverage 11,045 10,059 10,808 50% coverage 13,806 12,574 13,510 Source: Authors. Table 4.8 Estimated Total Sum Insured for the Compulsory MPCI Crop Insurance Scheme, by Coverage Levels and Oblast Insured yield coverage level Oblast 10% 20% 30% 40% 50% Amount insured (KZT millions) Akmola 10,985 21,970 32,956 43,941 54,926 EKO 1,181 2,361 3,542 4,723 5,904 Karaganda 1,346 2,691 4,037 5,383 6,729 Kostanay 14,786 29,573 44,359 59,145 73,931 NKO 13,216 26,432 39,648 52,864 66,080 Pavlodar 910 1,821 2,731 3,641 4,552 Total 42,424 84,849 127,273 169,697 212,121 Amount insured (US$, millions) 293 585 878 1,170 1,463 Source: Authors based on ARKS rayon-level spring wheat production data. Agricultural Insurance Feasibility Study  123 4.58. Under the MPCI policy, there is more estimates of the total sums insured that would flexibility in offering farmers optional cov- apply across the six eligible oblasts (Aktobe and erage according to their circumstances and WKO being eliminated from the scheme).34 The higher levels of coverage than under the cur- actual 2010 TSI for the obligatory scheme was rent LIC policy. Chapter 3 showed that, under KZT 47.3 billion (US$326 million). Under an the LIC policy, farmers elected the lowest nor- MPCI program, at 20 percent coverage across mative cost level of sum insured of about KZT all rayons and an average unit valuation price of 3,500 per hectare for spring wheat in 2010, KZT 3,120 per centner, the TSI would increase equivalent at the prevailing sales price of wheat to KZT 84.8 billion (US$585 million). At the to an underlying insured yield of about 1 centner highest level of coverage—50 percent—the to- per hectare or 10 to 15 percent of the average tal scheme liability would rise further to KTZ yield of the crop. One of the major objectives 212 billion (US$1.46 billion). With the proposed of the MPCI policy would be to offer farmers higher levels of sum insured and total scheme the option to purchase higher levels of coverage liability, expected claims would also rise; there- up to 40 percent of the expected rayon yield. An fore, it is very important to examine the expected example of the calculation of the insured yield losses that might occur. and sum insured is given in table 4.7 for spring wheat in Akkol rayon, which has an overall MPCI Rating Methodology and Indicative expected yield of 8.7 centners per hectare. At Premium Rates the maximum 40 percent insured yield cover, the farmer would have an insured yield of 3.5 4.60. The statistical rating methodology used centners per hectare; at the given unit insured in this study to establish individual grower price of KZT 3,120 per centner, the sum insured MPCI rates conforms to the MPCI rating pro- would be equivalent to KZT 10,808 per hectare. cedures that are adopted by the insurance in- The 40 percent coverage level would afford the dustry. The rating procedures were based on an farmer about three times the current LIC protec- analysis of variance in the rayon-level 17-year tion. Conversely, in those rayons where farmers yields and were adjusted to reflect the higher achieve much higher average yields of spring variation of disaggregated individual farmer wheat, the levels of insured yield coverage and yields in spring wheat by multiplying the coef- sums insured would be correspondingly higher. ficient of variation by 1.15. Full details of the The maximum expected yield in any one rayon individual grower MPCI rating procedures are is 17.29 centners for commercial producers in contained in annex 2 and annex 4. Mendikara rayon in Kostanay, and in this case the 40 percent coverage level is equivalent to 4.61. The rates presented in this report are in- 6.9 centners per hectare, with a sum insured of dicative of commercial premium rates for a 60 KZT 21,578 per hectare. The spring wheat in- percent target loss ratio, but the final decisions sured yields and sums insured were calculated over rates will be taken by insurers and their separately for production enterprises, commer- reinsurers. The burning cost or pure premium cial farms, and overall in all the rayons and are rates were calculated for each level of insured presented in annex 4. yield coverage, and a very conservative catas- trophe or reserve load was added to derive the 4.59. The total sum insured (TSI) of the technical rates for each coverage level from 10 scheme would be significantly higher if these proposals were adopted, assuming that all farmers continued to purchase MPCI cover— 34  For the purposes of this analysis, the average that is, if the individual grower MPCI program sown area of spring wheat per rayon for 2008 to 2010 continued to be obligatory for all spring wheat was used to establish the TSI in each rayon, each farmers. Table 4.8 provides some illustrative oblast, and in total. 124  Kazakhstan to 50 percent of expected yield. The technical premium rates are, in effect, subsidized by a fac- rates were grossed up by 40 percent to achieve a tor of 50 percent because of the government 50 conservative target loss ratio of 60 percent. The percent claims reinsurance program. These rates indicative commercial rates for spring wheat by were doubled to reflect the six-year average full rayon and coverage level are presented in an- (100 percent) commercial premium rates that in- nex 4, and the MPCI rating tool and database surers would have had to have charged if there are available to the key stakeholders on request. had been no government financial support for Finally, the commercial premium rates derived claims. The average full rates per oblast would under this study are high not only because of the therefore range from an average low of 3.5 per- underlying yield variability, but also because of cent in Kostanay to a high of 9.4 percent in Kara- ganda (table 4.9). The rayon-level MPCI indica- the conservative loadings applied; if this scheme tive commercial premium rates for spring wheat passes to the detailed design and planning stage, with a 60 percent target loss ratio for 20 percent there should be room to analyze, for example, coverage would range from an average low of the actual A&O costs of insurers. If these are 3.6 percent for NKO (and are in fact lower than found to be lower than assumed under this study, the existing LIC average rates in this oblast) to it will be possible to reduce the 40 percent gross- a high of 7.1 percent in Pavlodar on account of up to a more reasonable level. the much higher yield variability in the Pavlodar rayons. At the higher level of 30 percent insured 4.62. A comparison was made of the aver- yield coverage, the commercial premium rates age premium rates that are actually charged in most rayons would exceed 8 percent. The on the obligatory LIC scheme and the com- one exception is NKO, where 40 percent cov- mercial premium rates that would be charged erage could be offered at relatively affordable under the proposed MPCI program. The LIC commercial premium rates of about 10 percent. Table 4.9 Actual Average Rates and Indicative Average Commercial Premium MPCI Rates for a 60% Target Loss Ratio premium rates (%) New MPCI commercial rates for spring wheat LIC scheme rates (insured yield coverage level) Six-year “As if” 100% average LIC full LIC Oblast 10% 20% 30% 40% 50% premium rates premium (50% rates) ratesa Akmola 1.9 3.9 1.5 4.8 8.9 13.6 18.4 EKO 3.4 6.8 2.5 6.9 12.0 17.3 22.7 Karaganda 4.7 9.4 1.8 5.6 10.2 15.2 20.4 Kostanay 1.7 3.5 1.4 4.6 8.7 13.2 17.9 NKO 2.0 4.0 1.1 3.6 6.7 10.3 14.1 Pavlodar 3.6 7.1 2.1 6.2 11.1 16.3 21.5 Total 2.2 4.4 1.4 4.5 8.3 12.6 17.1 Source: Authors based on FFSA insurance data and ARKS rayon-level wheat yield data. a. “As if” full premium rates assuming no government 50% claims payment subsidies. Agricultural Insurance Feasibility Study  125 At the highest level of 50 percent insured yield of switching its current 50 percent of claims re- coverage, the commercial premium rates in most insurance to 50 percent premium subsidies are rayons would exceed 15 percent, being in some shown in table 4.11, assuming that the scheme cases (such as in Karaganda, EKO, and Pavlo- continues to be underwritten on 100 percent of dar) above 20 percent of TSI. spring wheat area. The fiscal implications for government are that, if the levels of insured 4.63. If the scheme continues to be obligatory yield protection and sums insured are increased, for all farmers, the estimated total commercial the corresponding premium subsidies will also premium income generated by the individual increase significantly, as shown by the illustra- grower MPCI cover for spring wheat would tive cost of KTZ 1.9 billion for 50 percent pre- increase significantly. Applying the commer- mium subsidies at the 20 percent coverage level. cial premium rates to the TSI corresponding to See further discussion at the end of this chapter. each coverage level indicates that 100 percent premium for a scheme that insures all eligible MPCI Probable Maximum Loss Estimates spring wheat grown in the six oblasts may be on the order of KZT 3.8 billion (US$26.1 million) 4.65. Spring wheat production in Kazakhstan at the 20 percent coverage level, rising to a total is very exposed to drought losses, and, with the premium of KZT 36 billion (US$251 million) at increase in insured yield coverage and sum the 50 percent coverage level (table 4.10). insured under the proposed individual grower MPCI scheme for spring wheat, it is very im- 4.64. This analysis also suggests that the gov- portant that the probable maximum loss (PML) ernment may need to consider introducing pre- exposures be quantified and a risk financing mium subsidies as an alternative to its current and reinsurance strategy identified. Under the practice of subsidizing 50 percent of claims in current risk modeling exercise for spring wheat order to make crop insurance more affordable MPCI cover, the PML associated with return pe- to farmers. The potential costs to government riods of up to one in 250 years was calculated for Table 4.10 Estimated Commercial MPCI Premium for Coverage of 10 to 50 Percent for Compulsory Scheme Covering All Eligible Insured Acreage Insured yield coverage level Oblast 10% 20% 30% 40% 50% Premium (KZT millions) Akmola 165 1,056 2,949 5,956 10,088 EKO 30 163 424 819 1,341 Karaganda 24 149 411 819 1,370 Kostanay 210 1,366 3,844 7,803 13,267 NKO 142 943 2,659 5,420 9,290 Pavlodar 19 113 302 593 978 Total 589 3,790 10,590 21,410 36,334 Premium (US$ millions) 4.1 26.1 73.0 147.7 250.6 Source: Authors based on ARKS rayon-level spring wheat production data 126  Kazakhstan Table 4.11 Fiscal Cost to Government of 50% Premium Subsidies on MPCI Program Insured yield coverage level Oblast 10% 20% 30% 40% 50% Cost (KZT millions) Akmola 82 528 1,475 2,978 5,044 EKO 15 81 212 410 670 Karaganda 12 75 205 409 685 Kostanay 105 683 1,922 3,902 6,634 NKO 71 472 1,330 2,710 4,645 Pavlodar 10 56 151 296 489 Total 295 1,895 5,295 10,705 18,167 Cost (US$ millions) 2.0 13.1 36.5 73.8 125.3 Source: Authors based on ARKS rayon-level spring wheat production data. Table 4.12 Estimated Probable Maximum Loss for MPCI Wheat Scheme, by Level of Coverage Coverage Item 10% 20% 30% 40% 50% Total sum insured (KZT 42,424 84,849 127,273 169,697 212,121 millions) PML (% of TSI) 1 in 10 years 0.41 2.09 5.62 10.85 17.13 1 in 50 years 3.13 10.83 19.79 29.14 37.93 1 in 100 years 5.43 16.80 28.70 39.17 46.98 1 in 250 years 10.80 25.99 38.71 46.96 54.42 PML (KZT millions) 1 in 10 years 175 1,777 7,153 18,414 36,336 1 in 50 years 1,329 9,191 25,191 49,444 80,456 1 in 100 years 2,304 14,251 36,524 66,463 99,652 1 in 250 years 4,581 22,051 49,268 79,688 115,428 1-in-100-year PML (%) 391 376 345 310 274 1-in-100-year PML (US$) 15.9 98.3 251.9 458.4 687.3 Source: Authors. Agricultural Insurance Feasibility Study  127 Figure 4.4 Estimated PML for MPCI Wheat Scheme, by Level of Coverage PML Loss of Investment coverage under three different price scenarios (in Loss Cost) 60.00% 50.00% 40.00% 30.00% 20.00% 10.00% 0.00% 1 50 100 150 200 250 10% Coverage 20% Coverage 30% Coverage 40% Coverage 50% Coverage Source: Authors. each coverage option from 10 to 50 percent of ance support for this initiative. Currently gov- average rayon yield; the results are summarized ernment provides 50:50 quota-share reinsurance for an obligatory scheme in table 4.12 and figure protection to the private commercial and mutual 4.4. The analysis suggests that, for a maximum crop insurers, but this protection does not cap 50 percent coverage level, the expected losses their exposure to catastrophic losses. Therefore, could be on the order of KZT 36.3 billion (17 for the purposes of this study, some preliminary percent of the value of the TSI) for a 1-in-10- analyses were conducted for nonproportional year loss, and KZT 99.7 billion (47 percent of stop-loss reinsurance protection for the spring TSI) for a 1-in-100-year loss, equivalent to a wheat MPCI program. loss ratio of about 274 percent. 4.67. Some preliminary modeling was con- Implications for MPCI Reinsurance ducted to establish the indicative pricing for aggregate stop-loss reinsurance protection for 4.66. Given the exposure of spring wheat pro- the spring wheat MPCI program. The modeling duction to catastrophic drought in Kazakhstan, was conducted assuming full-value protection it is extremely unlikely that the insurance sector and priority levels of 70, 100, and 150 percent will be willing to assume the increased liabili- of GNPI for the four levels of MPCI insured ties implied under the proposed MPCI program yield coverage. The results of this analysis are unless government is willing to provide reinsur- summarized in table 4.13, and further details of 128  Kazakhstan the methodology used, which draws on accepted tional reinsurance. In order to gain access to in- reinsurance industry stop-loss pricing methods, ternational reinsurance capacity, the individual are presented in annex 4. The analysis shows that insurance companies would need to demonstrate for losses in excess of 100 percent of GNPI, the that they have (a) financial ability and adequate full-value stop-loss reinsurance pricing would be solvency margins to sustain losses, (b) a core on the order of KZT 991 million (26.14 percent crop insurance underwriting and claims man- of GNPI) for 20 percent coverage, rising to KZT agement team, (c) a balanced MPCI portfolio 5.8 billion (16 percent of GNPI) for 50 percent with adequate geographic spread (and indeed coverage. If the lower 70 percent of GNPI prior- reinsurers may insist on sum insured limits per ity were adopted, the aggregate stop-loss pricing oblast or region to ensure accumulation control); would be correspondingly higher. (d) commercial premium rates for each level of insured yield coverage that are technically de- 4.68. In practice, unless the crop insurers rived and contain both catastrophe loading and form a single pool, the pricing of stop-loss rein- margin to cover A&O costs and profit margin; surance cover would have to be conducted for and (e) the necessary in-field preinspection and each individual insurance company and mutu- crop loss assessment staffing systems and proce- al insurer. This would have to be done accord- dures to control risk and adjust losses in a timely ing to the size and distribution of each insurer’s and accurate fashion. portfolio and the level of insured yield coverage in each region. 4.70. While some commercial insurers may be able to meet the requirements of international 4.69. During the transition to a more market- reinsurers, it is unlikely that, at present, any of based crop insurance program, it is possible the farmer mutual crop insurance companies that the larger commercial crop insurance will be able to meet these requirements. In the companies may be able to seek reinsurance short term, it is likely that government will need support from international reinsurers, either to provide nonproportional reinsurance cover. on a nonproportional stop-loss basis or as a combination of proportional and nonpropor- 4.71. Finally, while the nonproportional stop- Table 4.13 Indicative Pricing for Aggregate Stop-Loss Reinsurance for the Wheat Scheme, by Coverage Level Insured yield coverage level Item 10% 20% 30% 40% 50% Stop-loss reinsurance premium (KZT millions) Priority 70% of GNPI 242 1,115 2,675 4,948 7,930 Priority 100% of GNPI 227 991 2,243 3,895 5,845 Priority 150% of GNPI 208 829 1,711 2,664 3,515 Stop-loss reinsurance premium (% of commercial premium GNPI) Priority 70% of GNPI 41.14 29.42 25.26 23.11 21.82 Priority 100% of GNPI 38.59 26.14 21.18 18.19 16.09 Priority 150% of GNPI 35.24 21.88 16.15 12.44 9.67 Source: Authors. Agricultural Insurance Feasibility Study  129 loss reinsurance model illustrated in table 4.13 insurance through the private commercial insur- assumes a single excess layer providing full- ance sector. value cover, in practice very few international reinsurers would be willing to offer unlimited 4.74. In the short term, if crop insurance is protection. They will normally insist on a lay- made voluntary, it is likely that there will be ered reinsurance program with capped liability, a major reduction in demand while the farm- and for higher levels of catastrophe risk, it may ing sectors adjust to the realities of a demand- be necessary for government to assume liability driven voluntary crop insurance system. At this (see the end of this chapter for further discus- stage, it is not possible to predict how great the sion). contraction in demand by farmers will be, but it is likely to be significant. Under a voluntary Phase 3: Transformation into a Fully system, crop insurers will be free to select which Commercial Crop Insurance Scheme types of farmers, which crops, and which regions they are willing to underwrite. 4.72. This section presents proposals for a vol- untary commercial public-private agricultural 4.75. The Law on Compulsory Crop Insur- insurance partnership for Kazakhstan, a cen- ance will need to be amended to reflect the new tral feature of which would be the formation of requirements of a voluntary PPP crop insur- ance scheme in Kazakhstan. According to the a crop coinsurance pool system, backed up by roles that government plays in supporting this international reinsurance. This section presents scheme, the law will need amending, for exam- a series of options and recommendations for the ple, to focus more specifically on government GRK and key stakeholders to consider for the financial support in the form of proportional or introduction of voluntary agricultural insurance nonproportional reinsurance and potentially pre- in Kazakhstan, the creation of a crop insurance mium subsidies. pool system designed to encourage participation by the private commercial insurance companies, 4.76. Under a future voluntary crop insurance and the introduction of a specialist agricultural program, Kazakhstan insurers will need to link insurance entity to act on behalf of the pool that crop insurance with other rural services—for is capable of developing and implementing new example, input supplies (seeds, fertilizers)— classes of agricultural insurance. The pool would and with seasonal crop credit provided through be supported by a commercial reinsurance pro- the banking system. In many countries, public gram that would be placed with international re- sector provision of crop credit is closely linked insurers. The roles of government and the FFSR with crop insurance, and banks make their lend- are also explored, and it is likely that their roles ing conditional on the farmer having crop insur- will center on financial support in the form of ance in place at the time of receiving his loan. premium subsidies and possibly catastrophe re- In other words, crop insurance is mandatory insurance protection. for credit recipients and voluntary for noncredit recipients. Normally such crop credit–linked Transition from Obligatory to Voluntary programs make the bank the first beneficiary on Crop Insurance the crop insurance policy in order to ensure that any indemnity is used to repay the outstanding 4.73. As part of the transition to a market- amount of loan; any balance on the insurance based crop insurance system, policy makers in settlement is then paid to the farmer. Countries Kazakhstan will need to consider making crop that operate compulsory crop credit insurance insurance voluntary. As previously noted, Ka- programs include Mexico (where both commer- zakhstan is one of a small minority of countries cial crop loans and small farmer group loans are to adopt obligatory crop insurance and almost insured on a compulsory basis), India (where the unique in trying to implement obligatory crop world’s largest national area-yield crop credit in- 130  Kazakhstan surance scheme, which is implemented through 4.79. Coinsurance pools for agricultural in- the public sector Agricultural Insurance Compa- surance have proved to be very popular with ny of India, AIC, covers about 25 million small- private and mutual insurers in many countries. holder farmers per year), the Philippines (where Notable among these are the Agroseguro pool in public sector crop credit provision is again Spain,35 the TARSIM pool in Turkey, the live- conditional on the farmer having an individual stock insurance pool in the Philippines, the hail grower MPCI cover in place through the paras- insurance pool in Austria, and various other pool tastal agricultural insurer, the Philippines Crop arrangements in China, Malawi, Mongolia, and Insurance Corporation). See Mahul and Stutley Ukraine. (2010) for further details on countries with link- ages between crop credit and crop insurance. 4.80. The rationale for forming a coinsurance pool in Kazakhstan centers on several key fac- Rationale for Creating an Agricultural tors. These include the following: Insurance Pool • A very small number of private commer- 4.77. In Kazakhstan under the third phase cial companies are currently supporting this of transformation to a market-based crop in- scheme, and government needs to attract surance system, it is recommended that policy commercial insurers if crop insurance is to makers review the potential to form a coinsur- remain a viable proposition in Kazakhstan. ance pool specifically dedicated to underwrit- LIC/MPCI crop insurance is a catastrophe ing a national agricultural insurance portfolio class of business, and many insurance com- under a suitable private-public partnership. In countries where insurance markets are develop- panies are reluctant to risk their capital on ing and there is little or no tradition of crop or it. Under a pool agreement, individual com- livestock insurance or rural insurance infrastruc- panies can choose to participate with very ture, a pool coinsurance program may be a much small shares of the overall risk. more attractive and cost-effective proposition for commercial insurance companies than trying • The start-up investment costs are prohibi- to operate independently. tively high for individual insurance compa- nies, which must create their own internal 4.78. The potential benefits of an insurance crop underwriting and claims departments pool include the ability to underwrite a much and then develop regional networks of mar- broader and larger book of business and the keting and sales agents, crop inspectors, potential to achieve a much better geographic and loss assessors to administer the scheme. spread of risk than if each company were op- Pools offer the opportunity to create a sin- erating independently. Other benefits include gle centralized insurance underwriting and economies of scale in the costs of developing claims management and loss assessment new products and programs, underwriting risks, capability (often termed a managing under- and adjusting claims when a single lead coin- writer company, MUC) and, for individual surer is appointed (or a separate management members of the pool, to contribute to the underwriting unit is created) to implement the costs of running the MUC, while benefiting business on behalf of the pool members. There from the advantages of economies of scale are also major potential cost savings in the pur- in fixed and variable costs. chase of reinsurance protection for a pooled co- insurance program. Potential drawbacks include a possible reduction in competition and in the 35  Agroseguro stands for the Agrupación Española range of products, services, and premium rates de Entidades Aseguradoras de los Seguros Agrarios offered by the pool. Details on the benefits and Combinados (Spanish Group of Insurance Entities of limitations of pools are contained in box 4.3. the Combined Agrarian Insurance). Agricultural Insurance Feasibility Study  131 Box 4.3 Benefits and Limitations of Coinsurance Pool Arrangements Benefits. Economies of scale can be realized by operating as a single entity with shared (pooled) A&O functions, leading to cost savings due to the following: • Reduced staffing requirements (fixed costs) • Shared costs of product research and development, actuarial, and rating • Reduced costs of underwriting and claims control and loss adjustment. Cost advantages are realized because the pool is able to purchase common account (pooled) reinsurance protection rather than each company trying to place its own reinsurance program. This leads to advantages due to the following: • Stronger negotiating position with reinsurers • Larger and more balanced portfolio and better spread of risk • Reduced costs of reinsurance due to pooled risk exposure • Reduced transaction costs (reinsurance brokerage). Companies do not have to compete on rates in a soft market, and the pool has the ability to maintain technically set rates. Most pools operate as the sole insurance provider or monopoly (for example, Austria, Senegal, Spain, and Turkey), so there is no competition on pricing. The pool has the ability to maintain underwriting and loss adjustment standards. Under a pool monopoly arrangement, the MUC can ensure that common and high standards are maintained in the underwriting of crop and livestock insurance and in the adjustment of claims. Where companies compete against each other for standard crop insurance business, loss adjustment standards often vary between companies. Limitations. The pool may act as the sole agricultural insurer, resulting in lack of competition in the market in terms of the following: • Range of products and services offered • Range of perils insured • Regions where agricultural insurance is offered or the type of farmer insured • Premium rates charged by the pool. Source: Authors. • Common standards are lacking at the re- rates down to unsustainable levels. Under gional level in the underwriting of crop risks, a pool agreement, all insurers would issue especially with regard to the in-field loss standard crop insurance policies and would assessment capabilities of individual insur- adopt the same premium rates for each crop ance companies and farmer mutual insurance in each zone and region. associations. Under a pool agreement, the managing underwriter would be responsible • It is difficult to arrange commercial interna- for coordinating all underwriting and loss tional reinsurance protection for individual adjustment activities and for ensuring that Kazakhstan insurance companies with very common standards are adopted throughout different underwriting standards and portfoli- the country. os. Under a pool agreement, the MUC would purchase a single reinsurance program, and it • A lack of consistency in crop rating and com- would be much cheaper to transact a single petition is driving crop insurance premium reinsurance contract for the pool. 132  Kazakhstan Institutional Framework for a PPP claims, in maintaining a database of individual Agricultural Insurance Pool grower crop insurance and claims, and in set- tling the government’s 50 percent share of crop 4.81. The proposed agricultural insurance claims to the insurance companies. In the future, pool would involve the active participation of it is recommended that the FFSA’s roles no lon- the public and private sectors. The institutional ger include participation in field-level loss as- framework proposed for the pool is outlined in sessment, but that its functions be expanded to figure 4.5, which draws on the experiences of include the following: the Spanish and Turkish agricultural insurance pools (see annex 5). The functions of each of the • Coordination with the crop insurance pool’s key stakeholders are discussed below. Managing Underwriting Company in the development of the technical studies re- 4.82. The Agency for Financial Market and quired for the design of new crop, livestock, Financial Institutions Regulation and Control forestry, and aquaculture insurance policies would play an important legal and regulatory and programs role in this scheme. The agency would be re- sponsible for amending the current law to make • Management of the government’s financial crop insurance a voluntary class of business and fund for the support of agricultural insur- to permit insurance companies to design and rate ance and disbursement of funds (including, their own agricultural insurance policies. The as appropriate, premium subsidies and ca- law should be amended to create an enhancing tastrophe reinsured claims payments) to the framework for agricultural insurance that per- MUC on behalf of the pool coinsurers mits new classes and products such as livestock and forestry insurance to be added in the future. • Maintenance of databases on crop insurance The legislation should also be modified to reflect underwriting and claims government’s supporting roles, including possi- bly financial subsidies on premiums and catas- • Provision of information and advice to trophe claims reinsurance. farmers. 4.83. The MoA would continue to play a cen- 4.85. The central feature of the proposed sys- tral role in the policy and planning for agri- tem would be to create an agricultural coinsur- cultural insurance in Kazakhstan and would ance pool for nonlife insurance companies that represent government’s fiscal interests in sup- would underwrite all classes of agricultural porting the scheme. It is proposed that the MoA, insurance business. The agricultural insurance through its Department of Strategic Planning, pool would be a legally constituted insurance would represent the GRK’s interests in the agri- company with paid-up capital contributed by cultural insurance scheme and be responsible for each participating nonlife insurer. It would re- planning the three-year and annual agricultural port to and be supervised by a board of direc- insurance plans and budget in conjunction with tors, which would coordinate the activities of the the FFSA and the private insurance industry. pool coinsurers with government departments. The MoA could also play support research and development into new agricultural insurance 4.86. It is proposed that the coinsurers also products and programs. create a separate MUC. This MUC would be re- sponsible for underwriting the scheme on behalf 4.84. The FFSA would continue to act as the of the coinsurers and for handling premiums and main public sector implementing agency. Cur- claims on behalf of the pool coinsurers. It would rently the FFSA is responsible for participating also negotiate reinsurance on behalf of the pool in the adjustment of field-level crop insurance coinsurers. Agricultural Insurance Feasibility Study  133 Figure 4.5 Organizational Framework for the Proposed Crop Insurance Pool Find for Financial Support Agency for Financial Market to Agriculture (FFSA) Ministry of Agriculture and Financial Institutions Regulation and Control Financial Support: Policy, planning, premium subsidies, research and Insurance legal catastrophe, reinsurance development and regulatory Pool Management Board Kazakhstan Non-Life Insurance Companies International Reinsurers Agricultural Insurance Pool Farmer Mutual Crop Insurance Associations Managing Underwriting Company Farmers Associations, Cooperatives, Large Farmer production Enterprises Rural Banks and other Aggregators Small and Medium Farmers Source: Authors. Functions of the Pool Coinsurers and norms and conditions of the general insurance the Managing Underwriting Company act and amended agricultural insurance law. Agricultural Insurance Pool 4.88. Spain has one of Europe’s oldest and largest agricultural insurance pools, formed 4.87. It is proposed that participation in the in 1980. There are currently 28 coinsurers in agricultural insurance pool be open to all non- the Agroseguro pool, which includes both pri- life private insurance companies in Kazakh- vate and mutual insurance companies, includ- stan. The local stakeholders would also need to ing Mapfre Insurance and Reinsurance Com- decide whether to open up participation to the pany, Spain’s largest insurance company, and farmer mutual crop insurance associations, so the Spanish public sector catastrophe reinsurer, long as they could meet the minimum capital Consorcio de Compensación de Seguros. The requirements of the pool and comply with other largest shareholder in the pool is Mapfre, with a 134  Kazakhstan shareholding of more than 15 percent; the small- is for the coinsurers to appoint a lead coinsurer est coinsurer has less than a 1 percent share in to manage the business on behalf of the mem- the pool. Size of shareholding is therefore not bers. The second is for the coinsurers to create a bar to participation. Each company’s share of a separate limited-liability company owned and annual agricultural insurance premiums and li- funded by the pool members, with its own core ability is determined according to its percentage team of underwriting and claims management share in the pool during that underwriting year. staff, an equipped office, and regional capability Participation in the pool is completely volun- to administer and implement loss assessment. tary, and insurance companies are permitted to join and leave the pool after completion of an 4.92. In Spain and Turkey, agricultural in- underwriting campaign (year). In order to main- surance programs are implemented on behalf tain continuity, in practice companies agree to of the pool coinsurers by specialist managing join the pool for a three-year period. In Turkey companies, established by and reporting to the TARSIM agricultural insurance pool has 22 the pool coinsurers. In Spain, Agroseguro is a members, but each insurance company has equal limited-liability company formed by the pool shares in the scheme. insurers in 1980 to transact agricultural insur- ance and to manage claims and indemnify losses 4.89. It is recommended that the Kazakhstan on behalf of the coinsurers. Agroseguro started agricultural insurance pool be constituted on with a very small team of agricultural underwrit- lines similar to those of the Spanish model. In ers, claims managers, loss assessors, and office Spain, participation in the pool is voluntary, in- support staff; today it has grown into Europe’s dividual insurance companies are permitted to largest agricultural insurance management com- hold different shares in the company, and each pany, underwriting more than 260,000 agricul- company’s share of premiums and claims is pro- tural insurance policies and a further 30,000 portional to its shareholding, as opposed to any livestock, forestry, and aquaculture policies other criteria (for example, the amount of premi- generating total premiums of US$864 million in um the companies individually collect and cede 2010. In 2010 Agroseguro had a full-time com- to the pool). Finally, the concept of requesting plement of about 75 permanent staff based in its companies to sign up for a period of three years headquarters in Madrid and in each of the 14 au- should also be considered to ensure stability. tonomous regions. It has a general management unit, a legal department, and regional branches 4.90. Under the market-based pool for crop as well as core operational departments respon- insurance, it is hoped that the participation of sible for (a) product research and development, private insurance companies will be high and (b) production and communication (underwrit- that shares in the pool will be fully subscribed. ing), (c) claims administration and loss assess- The principle of a coinsurance pool is that 100 ment, (d) administration and accounting, and (e) percent of the liability is divided between the organization and information technology sys- subscribing members according to preagreed tems. As such, it functions as a very professional underwriting limits and risk retention limits of commercial managing company on behalf of its each company. If, however, the scheme cannot coinsurers. Agroseguro’s internal A&O costs be fully subscribed in the start-up phase, govern- are financed out of earned premiums on the ag- ment could consider becoming a coinsurer for a ricultural insurance business it writes on behalf limited period of time. of the pool: in 2010 its internal A&O expenses amounted to 3.55 percent of total earned premi- Managing Underwriting Company ums (Agroseguro 2011). In Turkey, the agricul- tural insurance pool is managed by TARSIM, a 4.91. There are two main options for manag- private corporation established in 2005 by the ing the agricultural insurance pool. The first insurance companies to carry out all the un- Agricultural Insurance Feasibility Study  135 derwriting and claims adjusting and settlement ther their own networks of sales agents or farmer tasks of the pool in accordance with the Agricul- associations, cooperatives, individual brokers, tural Insurance Act of 2005 (Bora 2010; Uçak banks, and other channels of distribution. In and Berk 2009). Spain, the companies market the Agroseguro policies on their own paper, and they receive a 4.93. In Kazakhstan, it is recommended that, preagreed commission to cover policy market- in conjunction with the formation of an agricul- ing. All policies are placed with Agroseguro, tural insurance pool scheme, the participating and all premium net of commissions is pooled coinsurers also approve the establishment of a by Agroseguro. A key principle of the Spanish managing underwriting company. Initially this system is that the shares of premium and liabili- company would only require a small full-time ty are determined by each company’s sharehold- staff, including a general manager (who should ing in the pool at the start of the underwriting be an experienced nonlife insurance expert), an campaign and not by its actual share of policy agricultural insurance underwriter (who should sales and volume of premium generated. It have experience under the current crop insur- is proposed that the same procedures apply to ance scheme), a claims manager, a crop loss as- the Kazakhstan pool—namely, pool members sessment manager (who would be responsible would be responsible for all policy marketing for coordinating the field-based loss assessment and sales, they would receive an agreed com- activities), a small number of junior underwrit- mission for transacting this business on behalf ers, data analysts, and database specialists, and of the pool, premiums net of commission would accounting and other back-up support staff. The be paid over to the MUC, and each member’s MUC should also have a small permanent staff share of liability would be based on its share capable of managing the program, especially in the coinsurance pool and not its volume of claims notification and loss assessment proce- sales. In this way, all pool members would share dures in each region or oblast of Kazakhstan. It equally in the pool of risks in different regions of is recommended that the MUC assume responsi- Kazakhstan. bility for all product design and rating, crop un- derwriting and policy issuance, and management 4.95. The MUC would be responsible for un- of premiums, through to being responsible for dertaking product design and rating, underwrit- claims, loss assessment, and claims settlement. ing and risk acceptance, claims administration, The MUC should report regularly and submit and loss assessment and for negotiating com- premium and claims bordereau accounts to the mon account reinsurance protection on behalf pool coinsurers’ management committee. of and reporting to the pool coinsurers. The MUC would design and rate standard crop in- Operating Systems and Procedures surance products and policies, which would then be approved by the insurance regulator. Once 4.94. Under the proposed agricultural insur- approved, the policies would be marketed by ance pool, the participating insurance com- the pool coinsurers at the MUC agreed premium panies (coinsurers) would be responsible for rates for each crop and risk zone. As such, in- marketing the MUC-approved crop insurance surance would not compete on the basis of pre- policies at agreed premium rates. The proposed mium rates, but rather on the quality of services operating systems and procedures are shown in to farmers. The MUC would be responsible for figure 4.6. Both in Spain and in Turkey, Agrose- handling premiums and for settling claims on guro and TARSIM, respectively, are responsible behalf of the pool. for the design and rating of crop (and livestock and other) insurance policies and in setting com- 4.96. Loss notification and loss assessment mercial premium rates for each region. The pool procedures would be streamlined and strength- coinsurers are exclusively responsible for all ened under the proposed system and would fall policy promotion, marketing, and sales using ei- under the direct control of the MUC. The MUC 136  Kazakhstan would be responsible for establishing a regional sessors would report to and be controlled by the network of part-time crop loss assessors in each MUC regional representative. oblast and rayon. These part-time loss assessors would receive intensive training in crop loss as- 4.97. Farmers or their agents would be re- sessment procedures and in the use of standard- sponsible for communicating losses (potential ized loss assessment procedural manuals, which claims) to the MUC local or regional office. would be designed by the MUC (possibly with Wherever possible, they would use web-based or support from the MoA and the FFSA). In the cell phone reporting systems. In order to reduce start-up phase of the program, it is likely that a the propensity for farmers to submit claims for national network of about 40 to 50 trained loss minor losses that fall within their self-retention, assessors would be adequate to manage the ex- it is recommended that a bonus-malus system be pected volume of crop losses, so long as these introduced (see chapter 3 for further details). teams were properly funded and had access to transport and communications.36 The loss as- 4.98. The losses would be adjusted by the MUC-approved loss assessors and would no longer involve the formation of a five-person 36  This recommendation is based on the Spanish system, where Agroseguro maintained a network of about 350 crop loss assessors in 2010 who adjust- ed nearly 95,000 claims or an average of about 275 claims per adjuster (Agroseguro 2010). Figure 4.6 Operating Systems and Procedures for the Proposed Crop Insurance Pool Kazakhstan: Proposed Operating Procedures Flow Chart Payment of Premium Cooperatives Association Managing Underwritng Companiey (MUC) Insured Farmers, Pool Insurance Intermediaries Policy Companies promotion marketing & sales Communication of Losses Loss Assessment and Payment of Claims Source: Authors adapted from Burgas 2007. Agricultural Insurance Feasibility Study  137 rayon-level committee. The only persons who these roles are considered below in the context would be required to attend the loss assessment of Kazakhstan. would be the approved loss assessor, the farmer (insured), and, at the farmer’s request, his lo- Legal and Regulatory Issues cal agent. The loss assessor would be required to complete a standard loss assessment form in 4.100. The GRK can facilitate a review and the field, to estimate the amount of crop loss amendments of the crop insurance law to sup- or yield reduction, and to confirm whether this port the introduction of the voluntary agri- loss is above or below the threshold that would cultural insurance pool into Kazakhstan. As give rise to a claim. The farmer would then be previously reported, the current law will require required to countersign the field loss assessment a comprehensive review and amendments to fa- form and to confirm his agreement or not with cilitate the introduction of voluntary commercial the assessed result. Where the farmer does not crop insurance. In this process, the GRK may agree with the assessment, he would first be en- wish to review the Turkish Agricultural Insur- titled to a second assessment conducted in the ance Act of 2005, which led to the introduction presence of a third party; if at this stage there is of the subsidized TARSIM agricultural insur- still no agreement, arbitration procedures would ance pool. apply. If the farmer agrees with the assessed results, the loss assessment would then be for- Data and Information Systems warded to the MUC for processing, to be either 4.101. There may be important roles for the closed (losses below the threshold for a claim) or GRK to play in enhancing data and informa- adjusted and settled (losses above the threshold tion systems for crop insurance in Kazakhstan. for a claim). The MUC would be responsible for If the insurance sector is to design new crop in- settling claims to each claimant either directly or surance products and programs, it will increas- through local distributers, as agreed. ingly need to have access to time-series crop production and yield data, weather data, and Government Support remote-sensing data. Government can facilitate the access to these data. In addition, the GRK 4.99. As part of the switch to a market-based has already identified a need to increase the den- agricultural insurance pool system in Kazakh- sity of weather stations if it is to introduce WII. stan, the role of public sector support for this The Second Agricultural Post Privatization As- scheme should be reviewed. To date, the GRK’s sistance Project (APPAP II) contains a subcom- main support for the obligatory crop insurance ponent with funding to invest in weather sta- scheme has been in the form of free quota-share tions. Furthermore, if AYII is to be introduced in reinsurance of 50 percent of the incurred claims, the future, the insurance sector will need to col- but this is likely to change under the transition laborate closely with the MoA and the ARKS to to a market-based scheme. There is now a con- obtain accurate estimates of area (rayon) yields siderable body of literature on the different ways at the time of harvest. See chapter 5. in which governments around the world support crop insurance, and this section briefly reviews Product Research and Development and some of the roles the GRK might play in the fu- Access to Data ture.37 A summary of the potential roles that gov- ernment might play is presented in box 4.4, and 4.102. Governments can assist the private insur- ance sector by conducting research into the de- mand for new agricultural insurance products 37  For a recent review of government support for and providing access to crop and weather data agricultural insurance in more than 65 countries, see and statistics. Such data are essential if insur- Mahul and Stutley (2010). ers are to design and rate new products and poli- 138  Kazakhstan Box 4.4 Roles for Government in Supporting Agricultural Insurance Legal and regulatory framework. One of the most important functions for government in facilitating agricultural insurance markets is to establish an appropriate legal and regulatory framework and, where necessary, to enact specific agricultural insurance legislation. Data and information systems. Time-series data and information on crop production, yields, and climate are essential for the design and rating of any traditional crop insurance product or new weather index product. Governments can provide an invaluable service by creating national databases and then making these databases available to all interested private commercial insurers, either free of cost or at concessionary rates. Product research and development. Among the major start-up costs for any new crop or livestock insurance program are the design (including the design of loss assessment procedures), rating, and pilot testing of new products and programs. Such costs may be prohibitive for individual private commercial insurers, especially in developing countries. In such situations, there is justification for government to provide financial support for product design and rating, especially where the products and rates are then made available to all interested insurers. Education, training, and capacity building. Governments can play an important role in new agricultural insurance programs by supporting (a) farmer awareness and education programs and (b) capacity building, workshops, and technical training programs for key agricultural insurance staff. Catastrophe risk financing. Agricultural insurance often has to protect against catastrophic perils of flooding, drought, and windstorm in crops and outbreaks of epidemic diseases in livestock. Most insurance companies do not have adequate capital to retain their catastrophe risk exposures, and they typically purchase some form of contingency financing or reinsurance protection. For new companies, which typically do not have large amounts of capital and have not yet built up claims reserves, the ability to retain risk is usually low; they typically need to purchase quota-share treaty reinsurance and then to seek nonproportional reinsurance protection on their retention. In start-up situations where the insurance company does not have an established track record and loss history, the costs of reinsurance protection may be very high. In such situations, government support for the reinsurance program may be highly cost- effective. Public sector premium subsidies. Premium subsidies are the most widely practiced form of government support for agricultural insurance and are used by more than two-thirds of countries that have some form of agricultural insurance. Governments justify the provision of agricultural insurance premium subsidies on the grounds that they make insurance more affordable for farmers, particularly small and marginal farmers, and thereby increase the rate of adoption and uptake of agricultural insurance. However, premium subsidies have major drawbacks: they disproportionately benefit larger farmers to the detriment of small and marginal farmers, they tend to promote moral hazard by encouraging crop production in high-risk regions, they are very difficult to reduce or withdraw once introduced, and they represent a major cost to government. Source: Authors. cies. In Spain, the State Agricultural Insurance sion of needed data and information and in the Body and the MoA plays a central role in as- future development of WII and remote-sensing- sisting Agroseguro with the development of new based insurance products and programs. products and programs. In Kazakhstan, the MoA and the FFRS could perform a similar and very Education, Training, and Capacity Building important role going forward. Also the Hydro Meteorological Service (KHM) and the NSA are 4.103. On the basis of this study, there appears likely to have very important roles in the provi- to be an important role for the GRK to play in Agricultural Insurance Feasibility Study  139 strengthening insurance awareness and edu- percent of the full cost of premium, but in some cation training programs for farmers. Govern- countries (including Italy, Portugal, Spain, and ments can play a key role in supporting aware- the United States), governments provide subsi- ness and education programs and providing dies as high as 75 to 80 percent of the premium. capacity building for farmers as well as work- Premium subsidies are, however, very contro- shops and technical training programs for key versial for several reasons. The provision of non- agricultural insurance staff. In Kazakhstan, crop discriminatory premium subsidies is regressive insurers have given relatively little priority to because subsidies disproportionately benefit the explaining the role of insurance and the opera- larger farmers to the detriment of small and mar- tion of the LIC policy. Government, through the ginal farmers. Also subsidies that cover a large MoA and other rural institutions, could provide part of the overall premium tend to promote financial and logistical support for farmer-level moral hazard, whereby farmers grow high-risk crop insurance education. Insurance company crops that attract high premium subsidies in re- staff will also need specialist training in product gions that are not technically suited to the crop. design, actuarial and rating, underwriting and Once premium subsidies have been introduced, claims administration, and loss assessment sys- it is politically very difficult to reduce or with- tems and procedures. Similar training also needs draw them, and in many of the countries that op- to be provided to staff in the banks, MFIs, and erate nondiscriminatory premium subsidies, the suppliers if these organizations get involved as fiscal costs to government are extremely high. delivery channels or agents. 4.106. In Kazakhstan, it is unlikely that volun- Catastrophe Risk Financing tary crop insurance will take off without gov- ernment support for premium subsidies. Since 4.104. In Kazakhstan under the PPP, it is likely the introduction of obligatory crop insurance in that government will continue to play an im- 2005, the insurance companies have, in effect, portant role in supporting the risk-financing received 50 percent premium subsidies through insurance and reinsurance program. In many the government’s 50 percent share in claims. countries, government is actively involved in the Under the proposal to increase the level of crop reinsurance of agriculture. Government acts as insurance coverage and to introduce actuarially a catastrophe reinsurer (either directly or indi- determined rates, the crop insurance premiums rectly through a national reinsurance company) that farmers will have to pay will have to in- in Brazil, Canada, China, India, Kazakhstan, crease significantly going forward. It is likely, Korea, Spain, and the United States. In Kazakh- however, that, under a voluntary crop insurance stan, it is proposed that commercial international program, very few farmers will continue to buy reinsurance of the pool scheme be introduced. crop insurance unless this is accompanied by It is, however, likely that the GRK will need to premium subsidies. It is therefore suggested that continue supporting the reinsurance program on government switch the bulk of its financial sup- a catastrophe basis, at least in the short term. port out of 50 percent quota-share reinsurance This subject is reviewed further below. and into crop insurance premium subsidies. The cost implications are considered below. Premium Subsidies Financial and Reinsurance Options 4.105. Premium subsidies are the most widely practiced form of government support for agri- 4.107. The proposal assumes that the pool man- cultural insurance and are used by more than agers will purchase international reinsurance two-thirds of countries that have some form protection (nonproportional and possibly pro- of agricultural insurance. Globally, average portional as required) at fully commercial re- premium subsidy levels are on the order of 50 insurance rates and that the GRK will provide 140  Kazakhstan Figure 4.7 Financial Flows of the Proposed Crop Insurance Pool Kazakhstan Financial Flows Catastrophe Fund for Financial Support to Reinsurance International Agriculture, FFSA Reinsurers Premium subsidies Commercial Reinsurance KAZAKHSTAN Farmers Payment of Premium AGRICULTURAL INSIRANCE POOL Payment of Claims Source: Authors. financial support in the form of crop insurance products that are introduced, the commercial premium subsidies and possibly catastrophe premium costs, farmers’ ability or willingness reinsurance. The possible financial and reinsur- to pay for crop insurance, the levels of premium ance options for the Kazakhstan agricultural in- subsidies offered by government, and decisions surance pool are presented in figure 4.7. Some over whether to link crop insurance to crop preliminary estimates are made for the volun- credit. tary MPCI program of the potential demand for cover and the associated premiums and liability 4.109. Some provisional financial estimates of based on the assumptions presented in this chap- total sum insured, premium, and costs of pre- ter, along with provisional estimates of the costs mium subsidies to government were calculated of premium subsidies and some comments on a for the spring wheat MPCI program over the structured risk financing and risk transfer pro- next five years. The model assumes an average gram for Kazakhstan. 40 percent coverage level and uptake rates over the next five years under a voluntary program Financial Estimates starting in year 1 with 10 percent uptake and ris- ing to 50 percent uptake after five years. These 4.108. It is currently very difficult to estimate are optimistic assumptions. On this basis, total the future demand for crop insurance under a scheme liability in year 1 might be on the order voluntary commercial crop insurance scheme. of KZT 17 billion (US$113 million), rising by There are considerable uncertainties over the fu- year 5 to KZT 85 billion (US$567 million), and ture demand for voluntary commercial crop in- commercial premium in year 1 might be nearly surance. In practice, demand will be influenced KZT 2.1 billion (US$14.0 million), rising to by farmers’ demand for and acceptance of the KZT 10.7 billion (US$71.3 million) by year 5. existing and new crop (and livestock) insurance The estimated costs to government of premium Agricultural Insurance Feasibility Study  141 Table 4.14 Five-Year Estimates of Voluntary MPCI Uptake, Total Sum Insured, Premium Income, and Costs of Premium Subsidies (40% Coverage Level) KZT millions MPCI 40 % MPCI uptake coverage and 100% Year 1 Year 2 Year 3 Year 4 Year 5 Item uptake 10% 20% 30% 40% 50% Total sum insured 169,697 16,970 33,939 50,909 67,879 84,849 Total commercial premium 21,410 2,141 4,282 6,423 8,564 10,705 Cost of GRK premium subsidies 25% of premium 5,353 535 1,071 1,606 2,141 2,676 50% of premium 10,705 1,071 2,141 3,212 4,282 5,353 65% of premium 13,917 1,392 2,783 4,175 5,567 6,958 Probable maximum loss 1 in 100 years 66,470 6,647 13,294 19,941 26,588 33,235 1 in 250 years 79,688 7,969 15,938 23,906 31,875 39,844 Source: Authors. subsidies assuming a 50 percent premium sub- protect through a risk financing and reinsurance sidy level would be nearly KZT 1.07 billion program; if a more conservative level of protec- (US$7.1 million) in year 1, rising to KZT 5.35 tion is required, management could consider the billion (US$35.7 million) by year 5. Estimates 1-in-250-year PML estimates (table 4.14). are also provided for subsidies of 25 and 65 per- cent of premium (table 4.14). 4.111. It is assumed that, in the future, the agri- cultural insurance pool will purchase common Risk Financing and Reinsurance account stop-loss protection from internation- al reinsurers in order to protect the program 4.110. The associated financial liability that the against catastrophic losses. It is likely in the pool coinsurers would need to protect through initial stages that international reinsurers will a combination of insurance and reinsurance only be willing to provide layered stop-loss re- was also modeled for 1-in-100-year and 1-in- insurance protection in order to limit their liabil- 250-year probable maximum loss scenarios. ity to catastrophe claims and that the GRK may In year 1 for 10 percent uptake of the MPCI therefore need to participate in the structured scheme, the 1-in-100-year PML liability was risk financing program by providing catastrophe estimated at KZT 6.6 billion (US$44 million), reinsurance for low-frequency but high-severity rising to KZT 33.2 billion (US$221 million) by losses. An example of layered insurance and re- year 5 (assuming 50 percent uptake). It is recom- insurance is presented in figure 4.8. mended that this be the minimum level of pro- tection that the pool scheme management should 4.112. For the proposed spring wheat MPCI 142  Kazakhstan Figure 4.8 Example of Agricultural Risk Layering 50 Goverments 45 40 Reinsurers Size of the Loss 35 Risk Transfer 30 25 Insurance Companies 20 15 Cooperatives/Mutuals Risk Pooling 10 5 0 Agricultural Producers Risk Retention m ic l e al or rg iu ph Sm in ed La tro M M as at C Type of Event Source: Mahul and Stutley 2010 Figure 4.9 Illustrative Stop-Loss Reinsurance Program for MPCI Wheat Program at 30 Percent Coverage Level Loss ratio % GNPI 300% Government of Kazakhstan Layer 3, 100% xs 200% GNPI 200% International reinsurers 150% Layer 2, 50% xs 150% GNPI International reinsurers 100% Layer 1, 50% xs 100% GNPI Agricultural insurance pool Primary retention 0% Source: Authors. Agricultural Insurance Feasibility Study  143 program and 30 percent coverage level, figure coverage level, the pricing would be 3.9 percent 4.9 presents an illustrative commercial insur- of GNPI for layer 1; 2.7 percent of GNPI for ance and stop-loss reinsurance program. This layer 2; and 3.4 percent of GNPI layer 3.38 program is intended to provide layered protec- tion to the pool coinsurers for losses in excess 4.113. It is unlikely that international reinsur- of 100 percent priority up to a maximum 300 ers agree to reduce the pool’s priority to less percent loss ratio (300 percent of GNPI). The than 100 percent of GNPI. In this instance, 300 percent loss ratio is equivalent to a loss of the pool may also elect to reduce its exposure about 35 percent of the total sum insured and is by reinsuring some of its primary retention on a equivalent to an expected 1-in-200-year PML. Under this scenario, the international stop-loss proportional or quota-share treaty basis. These reinsurers would provide layered protection for and other risk-layering options will be devel- two layers: layer 1, 50 percent in excess of 100 oped further in the design and planning stages percent of GNPI; layer 2, 50 percent in excess of of a commercial crop insurance program for Ka- 150 percent of GNPI, while government would zakhstan. come in with catastrophe reinsurance protec- tion for losses in excess of 200–300 percent of GNPI. In year 1, assuming 10 percent uptake of 38  At 10 percent uptake, GNPI would be equivalent the MPCI spring wheat program and 30 percent to KZT 1.492 billion (US$10 million). 144  Kazakhstan Chapter 5: Opportunities for New Crop Insurance Products 5.1. In Kazakhstan, there appear to be both Named-Peril Crop Insurance a need and opportunities for risk differentia- tion and product development. Under the ma- 5.3. Crop hail was the second most important ture agricultural insurance programs found in cause of insured claims on the obligatory crop countries such as Australia, Canada, Spain, and insurance scheme for the past six years. Hail is the United States, the markets are highly devel- a moderate-to-severe problem in many parts of oped and differentiated in terms of the range of Kazakhstan, with peak months of exposure oc- crop, livestock, forestry, and aquaculture insur- curring between May and July (chapter 2). Over ance products they provide to different segments of the farming community and in terms of the the past six years, hail was the second most im- perils that are underwritten. In the case of crop portant cause of loss after drought, accounting insurance, products generally range from simple for about 2.5 percent of the total area lost due named-peril crop hail or frost covers through to to insured perils. On the basis of field visits, it individual grower multiple-peril crop insurance appears that, under a voluntary scheme, farmers (MPCI) policies and new index-based products. in some regions may want hail-only insurance. Currently in Kazakhstan a single loss of invest- This section presents some preliminary recom- ment costs (LIC) policy providing a low level mendations for the design and rating of a spring of costs-based protection is compulsory for all wheat hail policy for Kazakhstan. peasant farmers and agribusiness enterprises, irrespective of whether the product meets their Features, Advantages, and risk management needs, and this probably ex- Disadvantages of Named-Peril Crop plains why many farmers are dissatisfied with Insurance the existing scheme. 5.4. Crop hail insurance is the world’s old- 5.2. As part of this World Bank study, an as- est form of crop insurance. The product is sessment was conducted of the potential to de- very standard and well understood and is exten- sign and implement new crop insurance prod- sively applied to the insurance of wheat and a ucts. These products include named-peril frost wide range of cereal, horticultural, and tree fruit and hail cover, area-yield index insurance (AYII), crops. Single-peril hail insurance is the simplest and weather index insurance (WII) for specific and best-known type of indemnity-based crop types of Kazakhstan farmers and for different re- insurance; it has operated for more than 100 gions according to the key risk exposures. This years in Argentina, Australia, Europe, New Zea- chapter presents the findings of this assessment land, and North America. Today there is a large for these new types of crop insurance products, body of accumulated experience with crop hail describing the features, indicative sums insured, insurance and indemnity products. Wordings are and premium rates for each product. Where pos- readily accessible through international associa- sible, it also offers provisional estimates of the tions of hail insurers, premium rates can initially potential demand for each product. be set based on international experience, and so Agricultural Insurance Feasibility Study  145 long as suitably high each and every loss (EEL) pected output. Finally, loss adjustment is based deductibles (or franchises) are maintained, the on estimated percentage damage to the crop ac- rates are generally not high. Finally, standard- cording to its growth stage, and this procedure ized damage-based loss assessment procedures is usually easier and cheaper to implement than can be accessed from the international hail as- yield-based loss assessment. sociations and training can be provided to local staff. 5.7. Insurers are generally willing to in- sure hail damage because it is considered a 5.5. Crop hail insurance is distinguished nonsystemic or noncatastrophe class of crop from the Kazakhstan LIC policy in that the insurance business and is not subject to anti- insurance and indemnity system is based not selection by farmers. Many crop insurers are on the “loss of crop yield” but rather on the very reluctant to offer MPCI loss of yield cover “percentage hail damage” caused to the crop. against drought, flood, windstorm, and flooding Under a damage-based indemnity system, phys- because of the systemic or catastrophic nature of ical loss or damage to the crop is measured in these perils. Conversely, hail is usually a rela- the field soon after a specific loss as a result of tively high-frequency but low-severity peril, and an insured peril, and the claim is usually settled as long as the insurer can achieve a geographic shortly after the time of loss. Normally the dam- spread of risk, it is not subject to catastrophic age is measured as a percentage loss, and this losses. Hail is usually an unforeseeable and un- percentage is applied to an agreed sum insured predictable event; unlike drought cover, it is not (for example, incurred production costs or oth- subject to antiselection and or moral hazard. er agreed value) for the crop. The sum insured may be adjusted downward if actual produc- 5.8. Damage-based crop insurance and in- tion is found to be below the normal produc- demnity policies also have several drawbacks. tion potential for reasons that are not insured, First, the product is best suited to specific perils for example, poor crop establishment. A de- that cause obvious and easily measured damage ductible is usually applied to the loss expressed to the crop, such as hail or wind and sometimes as “percentage damage,” although this can be frost or excessive rain, but it is not suitable for a fixed value. This method is most applicable progressive perils that affect the crop over time, to programs with a single or limited number of such as drought, and where losses can only be discrete-event perils (for example, hail, wind- measured objectively in terms of yield reduction storm, and frost). A worked example of the basis or loss. Second, the product is not suitable for of insurance and indemnity for hail insurance is other perils such as flood. Indeed, flood is not given in annex 6. included as a single peril on traditional indemni- ty-based crop insurance policies because of the 5.6. A named-peril (hail) damage-based in- problems of antiselection. demnity policy has key advantages. First, there is no need to collect time-series data on indi- Considerations for the Design of Spring vidual grower production and yield with which Wheat Cover in Kazakhstan to establish a normal average yield and then an insured yield because the policy uses damage- 5.9. Hail insurance is a very flexible class of based indemnity rather than loss of yield. Sec- insurance that can be designed to provide farm- ond, the sum insured can be set according to an ers with a wide range of options and choices. agreed value per acre based either on production Key features of a hail policy for spring wheat costs or on production costs plus an element of are reviewed below, and options for the design the expected gross profit margin; it can also be of this cover are presented in box 5.1. set according to a revenue valuation based on the farm-gate sales price of the crop times the ex- 5.10. Hail insurance can be designed to pro- 146  Kazakhstan Box 5.1 Possible Features of a Named-Peril Hail Policy in Kazakhstan Insured crop. Spring wheat is insured during the cover period against physical loss or damage due to the action of direct hail damage. This policy does not insure against loss of quality (price downgrading) in spring wheat. Basis of insurance and indemnity. Named-peril hail is a percentage damage–based policy. Insured perils. For hail, the insured peril is direct physical damage or loss to the insured crop. In addition, physical loss or damage caused by wind associated with hail may be considered an optional peril, subject to the payment of an additional premium. Cover period. The crop is covered from the time of emergence and full stand establishment (defined as 10 centimeters in wheat) through to the completion of harvest. Cover may be purchased at any time between the opening and closing dates of cover, subject to a waiting period of 24 hours from the time of payment of the premium up to the time of inception of hail cover. Insured unit. The insured is obliged to declare and insure all of his fields of spring wheat grown in the same rayon. The insured may choose to insure all of his fields of spring wheat as (a) a single insured unit (whole-farm basis) or (b) separately by field (field-by-field basis). Basis of sum insured. A fixed amount in KZT per hectare, which the insured may elect based on his coverage requirements. This amount may range from costs of production per hectare through to a maximum level based on the average crop revenue value for spring wheat in that rayon, as specified by the local department of the Ministry of Agriculture (MoA). The total sum insured (TSI) will be calculated by multiplying the sum insured per hectare by the area (in hectares) of each and every insured field declared by the insured and summing the total. Basis of indemnity. Indemnity is based on either the gross value of loss or policy excess. (a) Gross value of loss is equal to the sum insured times percentage hail damage. Where the percentage damage exceeds the policy excess, giving rise to a claim, the indemnity is equal to the sum insured times (percentage damage minus policy excess), or the net value of claim. Policy excess is applied on an EEL basis in each insured unit as defined. Excess options include percentage damage deductible (for example, 6 percent EEL), percentage damage franchise (for example, 6 percent EEL), or a fixed-value amount (deductible or franchise; for example KZT 10,000 EEL). Loss notification procedure. The insured is responsible for notifying the insurer of a hail event that is expected to exceed the policy excess (deductible or franchise) within 48 hours of its occurrence. Loss assessment procedure. In-field loss assessment uses standard procedures to measure the area damaged by hail and to assess the average percentage hail damage in each insured unit. Underinsurance and overinsurance (of spring wheat area). In the event of a loss, if it is discovered that the insured has underdeclared his or her spring wheat sown area by more than 5 percent of the total area, the insurer retains the right to apply the “law of average” to any claims settlement. In the case of overinsurance, the maximum amount payable in the event of a total loss is the actual assessed cultivated area times the agreed per hectare sum insured and subject to the policy excess. Exclusions. All perils apart from hail (and wind, which is optional) are excluded. Other key conditions. The insured must declare and insure all of his spring wheat grown in the same rayon; premium is payable prior to inception. Source: Authors. Agricultural Insurance Feasibility Study  147 vide a very high level of crop revenue protec- excess is applied. At the highest level of aggre- tion if requested by the insured. The sum in- gation, the IU can be defined on a whole-farm sured for crop hail insurance is very flexible, as basis as the entire area of the same crop grown in it is not dependent on an insured yield per se. the same farm or location. Conversely, insurers Insurers will generally permit farmers to insure may agree to offer insurance on a field-by-field their crops against a minimum level equivalent basis, in which case, the IU is the individual field. to the costs of production per hectare or amount Finally, some insurers agree to offer crop insur- of crop loan per hectare through to a maximum ance on an acre-by-acre basis, such that any area valuation based on gross revenue (yield times within a field subject to damage that exceeds the expected market sales price). In Kazakhstan, in policy excess is eligible for an indemnity. For principle, a spring wheat grower achieving an Kazakhstan spring wheat, it is recommended average yield of about 15 centners (1.5 metric that insurers offer optional covers for whole- tons) per hectare could insure 100 percent of his farm insurance and on a field-by-field basis. expected wheat revenue against hail, or a value of about KZT 45,000 per hectare (at the current 5.14. It is conventional on a hail policy to apply sales price for wheat of about KZT 3,000 per a first-loss excess (deductible) that is retained hectare). by the insured. The objective of the policy ex- cess is to eliminate small hail losses, which are 5.11. Hail insurance can be marketed both very time-consuming and costly for the insurer before and during the crop-growing season to adjust and would rapidly erode the premium right up to the time of harvest. Because hail is reserved to pay hail claims. It is intended to in- considered an unforeseeable and unpredictable demnify only severe hail losses that are of eco- event, many hail insurers will permit farmers to nomic consequence to the insured. The excess is purchase cover at any time during the growing usually applied on the basis of each and every season, subject to a waiting period of 24 or 48 hail loss event and can take several forms: hours. This is in contrast to loss of yield poli- cies that insure against drought, where a sales 1. A percentage damage deductible, which is cutoff period of usually a month prior to sowing deducted for the gross assessed percentage is required to avoid preexisting conditions and damage. An example would be a 6 percent potential antiselection. deductible that is applied on an EEL basis. If the assessed damage amounted to 5 per- 5.12. Hail insurers may allow the farmer to cent, this would fall below the deductible choose to insure only a part of his total fields and there would be no claim. If the assessed and total cultivated area. In the start-up phase damage was 15 percent, the net damage of a new crop hail insurance program for spring would be 9 percent (15 percent minus 6 per- wheat in Kazakhstan, it is, however, recom- cent deductible), which would be applied to mended that, in order to achieve volume and the sum insured. a spread of risk, farmers should be required to declare and insure all of their fields and farms 2. A percentage franchise, which is sometimes planted with the same crop located in any one termed a “qualifying franchise.” If the as- rayon. sessed damage amounted to 5 percent and a franchise of 6 percent applied, there would 5.13. The insured unit for the purposes of be no indemnity, as the damage would be loss adjustment can also be very flexible under below the 6 percent franchise. However, for a crop hail policy. There are many options for the example of 15 percent damage, which defining the insured unit (IU) for a crop hail pol- would exceed the 6 percent qualifying fran- icy—namely, the area of the insured crop over chise, the damage would be indemnified which the damage is estimated and the policy from the ground up or in full, and the in- 148  Kazakhstan sured would receive an indemnity of 15 per- sible to crop hail insurers in Kazakhstan, who cent damage applied to the sum insured. can adapt them to their own conditions. 3. A fixed-value excess. On some policies, Hail Exposure and Preliminary Rating rather than apply a percentage damage ex- Considerations cess, a fixed-value excess (for example, KZT 10,000 EEL) is applied either as a 5.18. Hail is usually a localized phenomenon franchise or as a deductible. In this case, that seldom accumulates over wide areas; as the gross value of the assessed hail damage such, it does not pose the same catastrophe ex- would have to exceed KZT 10,000 to give posure as systemic perils such as drought. Hail rise to a claim. is often a relatively high-frequency (that is, it oc- curs every season), but low-severity (it tends to 5.15. From a loss assessment perspective, be localized) class of crop insurance business. some hail insurers prefer to apply a deduct- ible as opposed to a franchise. The operation 5.19. In order to design and rate a crop hail of a franchise places very high demands on the insurance program, it is necessary to obtain need for accurate in-field assessment of per- data on three key parameters: hail exposure, centage damage. For example, with a 6 percent hail hazard, and hail severity. In some coun- franchise, where the percentage hail damage is tries, hail exposure data are available from me- light and where sampling error means that the teorological stations in the form of the number assessed damage could be anywhere from 4 to 8 of recorded hail days per month, but because percent over the IU, farmers tend to dispute any hail is a localized phenomenon, this may not assessment that falls below 6 percent and would be representative of hail occurrence at the re- give rise to a claim in full. Conversely, with a gional level. In Kazakhstan, information from 5 percent deductible that is deducted from the the Hydro Meteorological Service (KHM) for assessed damage, the level of precision required selected weather stations on the frequency of oc- when assessing low levels of hail damage is less currence of hail by month for the period 1990 to demanding. 2010 shows that the peak hail months are from May to July (figure 5.1). While exposure data 5.16. Crop hail deductibles or franchises are are readily available for the cultivated area and commonly on the order of 3 to 6 percent for ce- value of spring wheat per rayon and oblast, there reals, but in areas of high exposure to hail risk is very little recorded information in Kazakhstan they may need to be correspondingly higher. regarding hail severity and damage. For these In much of Europe, Latin America, and North reasons, a simple hail damage simulation model America, hail deductibles are between 3 and 5 was developed for spring wheat in Kazakhstan percent EEL. In Argentina, the industry has fol- in order to generate hail rates for each oblast. lowed a standard hail policy for wheat with 6 percent franchise for many years. In some parts 5.20. A simple hail rating simulation model of the world with a high hail risk exposure, de- was developed for spring wheat in Kazakhstan ductibles may, however, need to be higher, on using industry-approved hail-rating methodol- the order of 10 percent or greater per event. ogy. The model was developed for 13 sample weather stations in selected rayons in the oblasts 5.17. Crop hail damage assessment proce- of Akmola, Kostanay, North Kazakhstan (NKO), dures for wheat are very well developed. Stan- Pavlodar, and South Kazakhstan (SKO). The dard hail loss assessment procedures have been model combines exposure data for each rayon developed by the Argentine, European, South during each month of the growing season with African, and U.S. crop hail insurers, and these the hazard model (frequency of occurrence of manuals of procedures should be readily acces- hail by month), a hail severity index from low Agricultural Insurance Feasibility Study  149 Figure 5.1 Frequency of Occurrence of Hail in Kazakhstan, by Oblast, 1990–10 (21 Years) 45% Frequency of occurence (% of years) 40% 35% 30% 25% 20% 15% 10% 5% 0% ch ril ne ly ry st y ay er r r r be be be ar Ju Ap gu ua ob M ar Ju nu em m em br M Au ct ce Ja Fe ov O pt De Se N Kostanay Average Akmola Average Pavlodar Average SKO Average Source: Authors based on KHM data (figures taken from table 2.1). Figure 5.2 Average Rayon-Level Commercial Hail Rates for Spring Wheat in a Sample of Oblasts and Rayons 14.00% 12.9% Average premium rate (%) 12.00% 10.00% 7.5% 8.00% 6.7% 5.6% 5.5% 5.8% 6.00% 4.1% 4.4% 3.9% 4.00% 3.1% 2.8% 2.6% 2.0% 2.00% 0.00% l us nay KO ha n r A ino ay y Sh urt nt ka a - K aya sk ye sk a ko ld u k vk ke og in r - B ilov ur g dy .M bo sh go ta k ilo zy h im hm vs kt uc in os l no ha ho Ka Eg al ch ep ik ik Z Di O Pa da a- M K O -S ar St ay SK SK ol ay o Ko ay ar ay od a- an vl m a an od an an Pa ol ol vl Ak st N st m m st vl st Ko Ko Pa Ko Ak Ak Source: Authors based on crop hail incidence data provided by Arka Consulting. 150  Kazakhstan to very severe, and a hail vulnerability model. tion of this product, as hail risk exposure is not These models were then combined to simulate as widespread as drought risk exposure. There hail percentage damage with 10,000 iterations would be an important start-up cost—namely, and to estimate the average burning cost or pure the costs to design suitable crop hail loss as- loss cost for each station or rayon. The average sessment procedures for Kazakhstan and then loss cost was then adjusted for a 6 percent hail to identify a core of loss assessors who would franchise and grossed up by an assumed factor receive specialist training in assessing hail loss of 20 percent to derive average commercial hail in wheat. Finally, it is likely that there will be rates for each sample rayon. These rates are sum- demand for hail insurance for other crops, for marized in figure 5.2 and annex 6. These rates example, cotton and horticultural crops grown are preliminary and will require further analysis in southern Kazakhstan; over time, there should if a commercial crop hail scheme is launched in also be potential to develop and expand a crop Kazakhstan in the future. hail portfolio in Kazakhstan. 5.21. The preliminary hail rating for spring Area-Yield Index Crop wheat in selected rayons in Kazakhstan sug- gests that it should be possible to design hail Insurance cover at affordable rates to producers. For the 6 percent franchise, average rates vary from about 5.23. On the basis of this feasibility study, it 2.5 to 3.5 percent in the lowest hail-risk regions, appears that there may be considerable poten- up to 5 or 6 percent in the medium hail-risk re- tial in Kazakhstan to design and implement gions, and as high as 7.5 percent in Mikhailovka AYII as an alternative to or a complement of rayon in Kostanay and 12.9 percent in Kazgurt the existing individual grower LIC and pro- rayon in SKO. In the next stage, it would be use- posed individual grower MPCI programs. ful to validate these relative hail rates with local Outline proposals are presented below for a pro- agricultural specialists in each oblast and rayon totype area-yield index product and a program who are familiar with hail exposure and hail for spring wheat grown in Kazakhstan, but fur- damage in wheat. ther design work will be required if the insur- ance companies decide to proceed with the pilot Conclusions on Crop Hail testing and implementation of this product. It is assumed that AYII would be introduced as a 5.22. Crop hail insurance should be relatively voluntary program. Further details of the AYII easy to design and implement in Kazakhstan product are contained in annex 7. as a commercial crop insurance product. Since this is a noncatastrophe crop insurance product, 5.24. The prototype spring wheat AYII prod- it should be relatively easy for the crop insur- uct presented in this section excludes Aktobe ance companies and possibly the farmer mutual and West Kazakhstan (WKO) and their rayons insurance associations to underwrite this prod- because their risk exposures are commercially uct with limited access to reinsurance protec- uninsurable. The AYII program presented in tion. Because the average rates for single-peril this section therefore relates only to the six main hail insurance are potentially considerably lower wheat-growing oblasts of Akmola, East Ka- than the individual grower loss of yield covers zakhstan (EKO), Karaganda, Kostanay, NKO, (the LIC policy or MPCI and the AYII product), and Pavlodar. Although Aktobe and WKO are there is also more potential to market this cov- excluded, government may wish to use AYII er to Kazakhstan farmers on a voluntary basis to provide disaster compensation to farmers in and without the need for government premium these two oblasts. For this reason, coverage lev- subsidies. It is likely that the demand for single- els, insured yields, and indicative premium rates peril crop hail insurance for spring wheat will are presented separately for Aktobe and WKO in be quite low in the initial stages of implementa- annex 7. Agricultural Insurance Feasibility Study  151 Features, Advantages, and age. However, under the AYII cover, the policy Disadvantages of AYII does not indemnify each farmer according to his own losses, but rather according to the reduc- 5.25. AYII represents an alternative approach tion in actual average yield at the rayon level. to MPCI insurance that aims to overcome In this example, the rayon actual average yield many of the drawbacks of traditional individ- is 5 centners per hectare, and therefore the yield ual grower MPCI insurance. This product does loss is 2 centners per hectare (7 centners minus not indemnify crop yield losses at the individual 5 centners), with an indemnity payment of KZT field or grower level; rather, it makes indemnity 6,420 per hectare. Each of the insured farmers payments to growers according to yield loss or therefore receives the 2 centners per hectare shortfall against an average area yield (the in- indemnity over his 1,000-hectare farm, valued dex) in a defined geographic area (for example, at KZT 642,000 per farmer, in spite of the fact the total area sown with spring wheat in a single that farmer A does not suffer any yield shortfall rayon). An area-yield index policy establishes against the rayon-level 70 percent cover, while an insured yield, which is expressed as a per- farmer C incurs a shortfall of more than 2 cent- centage (termed the “coverage level”) of the his- ners per hectare. torical average yield for each crop in the defined geographic area, such as a rayon, that forms the 5.27. AYII works best where farmers’ crop insured unit. Farmers whose fields are located production systems, technology levels, output, within the IU may purchase optional coverage and yields in the defined IU are relatively ho- levels, which typically vary between a minimum mogeneous. It also responds best to systemic of 50 percent and a maximum of 80 percent of risks such as drought that tend to affect farmers’ the historical average yield. The actual aver- production and yields in the same way across age yield for the insured crop is established by wide geographic areas. In Kazakhstan, spring sample field measurement (usually involving wheat crop production systems are relatively crop cutting) in the IU, and an indemnity is paid homogeneous at the rayon level, but, as chap- by the amount that the actual average yield falls ter 4 shows, there are differences in the level of short of the coverage level purchased by each technology used and the average production and farmer. yields obtained by the two categories of wheat producers: production enterprises and commer- 5.26. An example of the basis of insurance cial farms. As AYII is intended as a catastrophe and indemnity for an AYII cover for spring product, in principle, the differences in technol- wheat is shown in figure 5.3. For this example, ogy and yield between these two categories of Bulandinski rayon in Akmola was selected. It is farmers should not negate the use of this crop assumed that the actual five-year area average insurance tool. See below for further discussion yield of spring wheat is 10 centners per hectare of homogeneity of spring wheat production at in Bulandinski rayon and that all farmers are of- the rayon level in Kazakhstan and the issue of fered the same coverage level of 70 percent of basis risk. the average yield or 7 centners per hectare. The unit sum insured is KZT 3,210 per centner, giv- 5.28. In Kazakhstan, a key potential advan- ing a standard sum insured of KZT 22,470 per tage of AYII over individual grower MPCI is hectare. Three farmers, each with total planted the ability to offer higher levels of insured yield area of 1,000 hectares, purchase cover. It is a coverage at lower rates because losses are ad- moderate drought season, and farmer A achieves justed against an area-yield index and not at an actual average yield on his land of 7 centners the level of the individual farmer. Chapter 4 per hectare; farmer B incurs more severe losses shows that, although individual grower MPCI is and achieves 5 centners per hectare; and farmer technically feasible in Kazakhstan, the very high C achieves only 3 centners per hectare on aver- variability in crop production and yields means 152  Kazakhstan Figure 5.1 Frequency of Occurrence of Hail in Kazakhstan, by Oblast, 1990–10 (21 Years) Insurance contract conditions Insured peril. All-risk policy Area-Yield Index Insurance Crop. Spring wheat` Payout Examples Rayon. Bulandinski in Akmola 11 Type of farmer. Production enterprise Production enterprise rayon-level expected yield 10 Production enterprise’s five-year rayon-level average expected yield (EY). 10 centners per hectare 9 Production enterprise’s rayon-level guaranteed yield at 8 PE Rayon Level Guaranteed Yield 70% coverage (GY). 7 centners per hectare 7 Yield (Mt./ha) Agreed price (AP). KZT 3,210 per centner 6 Insurance payouts = 2 centner per hectare Sum insured. KZT 22,470 per hectare 5 Production enterprise rayon-level actual yield Insured unit area (IUA). Farm A, B, and C, each with 1,000 4 hectares and identical TSI of KZT 2.25 million 3 Production enterprise’s rayon-level actual average yield 2 (AY). 5 centners per hectare, but farmer A’s actual yield (AYA) 1 = 7 centners per hectare; crop loss = 0 centners per hectare. 0 Farmer B’s actual yield (AYB) = 5 centners per hectare; crop loss = 2 centners per hectare. Farmer C’s actual yield (AYC) Farm A Actual Farm A Actual Farm A Actual = 3 centners per hectare; crop loss = 4 centners per hectare. Yield = 7 Yield = 5 Yield = 3 centner/ha centner/ha centner/ha Insurance payout calculation (IPC). If AY < GY, then IPC Actual Yield = (GY – AY) * AP* IUA. IPC = (7 centners per hectare minus Farmer Actual Loss 5 centners per hectare) times KZT 3,210 per hectare times Area-yield Index Pavouts 1,000 hectares = KZT 6,420 per hectare or KZT 642,000 per farmer. AYII provides payouts to all the farmers situated in the area selected as the insured unit Source: Authors. that the maximum coverage level that can be of- cy responds to yield loss at the level of the coun- fered to growers at affordable premium rates in ty or rayon and not at the level of the individual most rayons is only between 20 and 50 percent farmer, no farmer can influence the yield indem- of average yield. AYII indemnifies losses ac- nity payments, which minimizes antiselection cording to yield variation or loss at the area lev- and moral hazard. Administrative costs are also el—in this case the rayon—and, because the ag- greatly reduced because there is no need for pre- gregate rayon-level crop production and yields inspections on individual farms and loss assess- year-on-year are relatively stable, it should be ment is not conducted on an individual farmer possible to offer higher coverage levels of 30 to and field-by-field basis, but rather according to 70 percent and even higher in some regions and a preagreed random sampling of crop yields on at more affordable rates for AYII than for indi- plots within the IU. These cost savings can be vidual grower MPCI. passed on to farmers in the form of lower crop insurance premiums. 5.29. The AYII approach minimizes moral hazard and antiselection and lowers the costs 5.30. The main drawback of an AYII policy is of administration and loss assessment, which “basis risk” or the potential difference between offers the potential for insurers to reduce the the insured area-yield outcome and the actual premium costs charged to farmers. As the poli- yields achieved by individual insured farmers Agricultural Insurance Feasibility Study  153 within the insured area. Basis risk arises when grower product for medium to large wheat (or an individual grower incurs severe crop yield other cereal) producers and as a meso-level losses due to a localized peril, such as hail or product to protect the loan portfolios of coopera- flooding of a nearby river, but, because these lo- tives or microfinance institutions (MFIs) serving calized losses do not affect the average yield of large numbers of small rural households in in- the district or rayon, he does not receive an in- dividual rayons in southern Kazakhstan (further demnity. In addition, basis risk may arise where discussion in chapter 6). the crop production and yields of individual farmers are highly heterogeneous (different) in International Experience with AYII the same rayon, which will invalidate the use of an area-based approach. See below for further 5.32. AYII has been widely adopted for small- discussion of basis risk when farming systems, holder rice and wheat cropping in India and in production, and yields within the insured unit areas where crop insurance is linked to season- are not relatively homogeneous. al crop credit. India has operated a public sector AYII program for more than 30 years under its 5.31. AYII is potentially a flexible crop insur- public sector National Crop Insurance Scheme ance product that can be implemented at the (NAIS), which is implemented by the Agricul- micro level for individual farmers or, alterna- tural Insurance Company of India (AIC), a pub- tively, at the meso level for regional financial licly owned insurer that specializes in agricultur- institutions. In Kazakhstan, there may be scope al insurance. Crop insurance is compulsory for to design AYII both as a micro-level individual farmers who borrow seasonal production credit. Currently, this program insures about 25 million Indian farmers each year. In order to make crop Figure 5.4 Government Stop-Loss insurance widely available at affordable prices Protection for NAIS Scheme in India to small and marginal farmers, the government has capped premium rates at about one-third of Loss the actuarially required levels and then provided ratio AIC with free stop-loss reinsurance protection. (%) For food crops, reinsurance cuts in at a 100 per- Food Crops Commercial & cent loss ratio, while for commercial and horti- (70% of Horticultural cultural crops, the priority is at a 150 percent loss premium) Crops (30% of ratio (figure 5.4). The program therefore shares premium) several common features with the Kazakhstan Government of LIC scheme, including obligatory cover, capped India and State premium rates, and government support for rein- Government surance, although in Kazakhstan this is based on Stop-Loss a 50 percent quota-share protection as opposed Protection to the nonproportional stop-loss protection pro- vided in India. Although the NAIS has achieved 150% very high levels of insurance uptake by Indian AIC farmers, the scheme incurs major delays (often 100% Retention up to six months after harvest) in arriving at AIC the estimates of actual area yield and in settling Retention claims, and this delay is very unpopular with farmers. For the past five years, the World Bank has been working with AIC to strengthen and Source: Authors. reform the NAIS scheme, transforming it into a market-based system, including (a) introduc- 154  Kazakhstan Box 5.2 Main Features of India’s mNAIS Scheme for the Rabi Crop Season, 2010–11 Actuarial regime. The mNAIS scheme operates an “actuarial regime” in which the government’s financial liability is predominantly in the form of premium subsidies given to AIC and funded ex ante, thereby reducing the contingent and uncertain ex post fiscal exposure currently faced by the government under mNAIS and reducing delays in claims settlement. Up-front premium subsidies. AICI receives premiums (farmer collections plus premium subsidies from the government) and is responsible for managing the liability of the scheme through risk transfer to private reinsurance markets and risk retention through its reserves. It is financially able to operate on a sustainable basis. On-account partial payment. The mNAIS product continues to be based on an area-yield approach, with a provision for an early partial payment to farmers (in season) based on weather indexes. Small insurance units. Crop-cutting experiments to assess crop yield are lowered from the block level to the village level to reduce basis risk (that is, the mismatch between the individual farmer’s actual crop yield losses and the insurance indemnity). Cutoff dates. Adverse selection is reduced through the enforcement of early deadlines for purchase ahead of the crop season. Additional benefits. Additional benefits are offered for the prevention of sowing, replanting, postharvest losses, and localized risk, such as hail or landslides. Early settlement of claims. mNAIS combines weather-based indexes for on-account partial payment of claims in case of adverse mid-season conditions, whereas area-yield indexes are used for final payment of claims. The final estimation of loss is based on area-yield measurement at the time of harvest using crop- cutting experiments. Source: GFDRR 2010b. ing actuarial rating, (b) switching government 5.34. In Ukraine, a subsidized AYII scheme financial support from claims compensation to was launched for winter wheat in 2003, but crop insurance premium subsidies, and (c) open- this scheme was poorly designed and imple- ing up the market to competition from local and mented and has not taken off. In Ukraine, as in international reinsurers. This modified NAIS Kazakhstan, agriculture is an important sector, program was formally launched in the rabi crop especially for the production of winter wheat, season 2010/11 in about 10 percent of the total which is the leading export crop. Agriculture is NAIS command area; if successful, this market- very exposed to drought (for example, in 2003, based public-private partnership (PPP) system 2005, 2007), spring frosts, strong winds, hail, will be introduced gradually into all states of In- and, for winter crops, “winterkill” (low temper- dia. See box 5.2 for further details. ature and freeze that damage the crop), which in 2003 caused catastrophic losses in more than 5.33. Other countries that are operating AYII 70 percent of the winter crops. During the So- include, most notably, the United States (where viet period, Ukrainian state and collective farms the product is termed the Group Risk Plan, were insured under the standard MPCI program GRP). This product is also being researched that operated throughout the Soviet Union. Fol- in parts of Eastern Europe (Ukraine), Africa lowing independence, there was no crop insur- (Burkina Faso, Ghana, and Senegal), South ance until the early 2000s. Between 2001 and America (Brazil and Peru), and Asia (Bangla- 2003, a local insurer, with the assistance of an desh and Nepal). international agricultural reinsurer, analyzed the Agricultural Insurance Feasibility Study  155 Box 5.3 AYII Pilot Crop Insurance Scheme in Ukraine Crop area-yield insurance has been implemented poorly in Ukraine. A pilot hybrid MPCI and AYII scheme was launched in 2003 for all major field crops. The rayon was the insured unit for the AYII cover. Indemnities were paid based on regional (rayon) statistical records (but apparently not on official statistical reports) and farm-level inspections of actual yield; the farmer had to provide proof that the reduction in crop yield was due to an insured peril. This meant that farmers had to obtain reports from a local meteorological station. Complicated and unclear loss assessment procedures meant that payouts were usually delayed for up to six months. Recently producers have lost interest in the AYII product, and insurers have been looking for other ways to insure crops. Source: WFP and IFAD 2010. possibilities for designing and implementing Preconditions for the Design of AYII for both individual grower MPCI cover for cereals Spring Wheat in Kazakhstan and area-yield index insurance. There were ma- jor difficulties in designing and rating individual 5.35. There are several preconditions for the grower MPCI because, following the breakup of operation of AYII for spring wheat (and any the state and collective farms, it was impossible other crops) in Kazakhstan. These include the to establish normal average or expected yields following: for the newly formed large-scale commercial enterprises or for small to medium producers. • Homogeneous spring wheat–producing re- Similarly, there were difficulties in designing gions or zones (the insured unit), with low and rating AYII because, although relatively re- variation in yield between farmers in the in- liable oblast-level historical data were available, sured unit it was difficult to obtain accurate time-series in- formation at the rayon level. Further difficulties • For the defined IU, historical data on sown included the fact that after independence aver- area, production, and average yield in spring age crop yields at the rayon and oblast levels wheat for the past 15 years or more on which showed a major declining trend due to reduced to establish the insured yield and technical use of fertilizers, and the time-series data had to rates be detrended carefully. A voluntary pilot AYII scheme was launched in 2003 with a pool of lo- • An independent and statistically accurate cal insurers, backed by European reinsurers, but system of measuring average spring wheat the scheme was not popular with farmers and yields in the defined region or zone and on suffered from poor implementation, especially which to trigger claims payments. in the adjustment of losses. The scheme has been discontinued. See box 5.3 for further details of 5.36. AYII is only effective if individual farm- the problems of Ukraine’s AYII pilot scheme. In er spring wheat production systems and yields the design of any AYII scheme for Kazakhstan, within a defined insured unit (the rayon) are planners should learn from the lessons and expe- relatively homogeneous. Basis risk occurs un- riences of Ukraine.39 der an AYII program when spring wheat farm- ing systems are highly heterogeneous within the rayon in terms of type of soil, fertility, soil 39  Between 2003 and 2005, the private insurance sector in Ukraine also tried to develop crop weather index insurance, with limited success. For a very good review of the issues and challenges faced in Ukraine, see WFP and IFAD (2010). 156  Kazakhstan moisture retention, and farmers’ use of technol- notes that Kazakhstan has a comprehensive sys- ogy and inputs and when these differences trans- tem for measuring crop production and yields, late into highly variable yields of spring wheat and, in the case of spring wheat, it was possible among farmers in that rayon. Under an AYII to access 17 years of crop production and yield cover where all farmers in the IU are treated the data at the rayon level for production enterprises, same and losses are paid against shortfall on the commercial farms, and in total. This time-series area average yield, the danger is that farmers who production and yield data enabled the design and are located on the best soils and who use high rating of an AYII product for spring wheat that levels of technology may receive an indemnity, operates at the rayon level. The ARKS reports although their actual yields are well above the figures for sown area, production, and yields rayon’s insured yield. Conversely, farmers using for spring wheat separately for production en- very low levels of technology and whose aver- terprises, for commercial farms, and in total. age normal yields are well below the rayon av- The prototype AYII product presented in this erage yield may also receive an indemnity, even chapter is based on a single rayon-level AYII when they have not incurred any yield losses. product (that is, it is designed using combined This study sought to access data on individual crop production and yield data for both types of farmer yield for the six selected rayons in Ak- farms), but, if required, separate covers could mola, Kostanay, and Pavlodar to check for yield be designed for production enterprises and for variability and basis risk under an AYII program commercial farms, as the AYII rating tool has operating at the rayon level. To date, however, it been programmed to determine insured yields has only been possible to analyze spring wheat and premium rates for both types of farms in all data for small samples of individual farmers in rayons and oblasts. two rayons in Pavlodar and one in Kostanay. This analysis suggests that yield variability be- 5.38. For the operation of AYII, it is necessary tween farmers in the same year is usually low to have an independent, accurate, and timely enough to indicate that an AYII product could system for measuring and reporting actual av- operate effectively with an acceptable level of erage yields in each IU at the time of harvest basis risk. It is also notable from this analysis and to indemnify insured yield shortfall below that in a good year (for example, 2009), aver- the actual area average yield on that basis. In age yields are more similar between farmers in the same rayon, with coefficients of variation India, which has the oldest AYII program in im- (CoVs) of about 20 to 25 percent: however, in plementation, a national system of sample crop drought years (2008, 2010), individual grower cutting in randomly chosen fields is used to cal- yields are more variable (CoVs as high as 40 culate the actual average yield for each insured percent or greater), and this may be explained crop in each insured unit. While crop cutting is by differences between farmers in their manage- potentially a very accurate method of determin- ment of soil moisture through the use of mini- ing average yield in the IU, its accuracy depends mum tillage, for example. See annex 7 for full on having a statistically adequate number of details of this analysis of individual farmers’ sample crop cuts, which must be randomly sited yields and issues of basis risk in Kazakhstan. in selected fields and, if properly conducted, is a very time-consuming and costly exercise. In 5.37. In Kazakhstan, the National Agency of India, the system only works because the state Statistics (ARKS) has an accurate system for governments subsidize the costs of crop cutting measuring and reporting sown area, harvested for the insurance sector. In the United States, the area, production, and average yields for spring GRP area-yield plan operates at the county level: wheat at the regional (oblast) and zonal (ray- county-level actual average yields are estimated on) levels; 17 years of historical spring wheat by the National Agricultural Statistics Service data from 1994 to 2010 were available for the from grain-elevator area, production, and yield purposes of this study. Chapter 2 of this report reports, and no in-field crop cutting is conduct- Agricultural Insurance Feasibility Study  157 Table 5.1 Average Area of Spring Wheat per Rayon (Insured Unit) Number of Average area Minimum crop Maximum crop Number of rayons % of total Oblast rayons per per rayon area per rayon area per rayon with more than rayons oblast (hectares) (hectares) (hectares) 10,000 hectares Commercial farms Akmola 19 45,846 675 151,900 17 89 EKO 17 13,853 140 47,080 9 53 Karaganda 9 40,224 3,400 103,680 7 78 Kostanay 18 69,723 8,620 167,900 17 94 NKO 13 56,522 26,780 107,780 13 100 Pavlodar 11 21,664 520 108,560 7 64 Subtotal 87 41,305 140 167,900 70 80 Production enterprises Akmola 19 150,300 520 317,600 17 89 EKO 17 10,856 160 27,780 7 41 Karaganda 9 31,364 80 130,220 4 44 Kostanay 18 142,661 900 369,120 16 89 NKO 13 178,658 86,480 430,400 13 100 Pavlodar 11 16,850 520 67,500 5 45 Subtotal 87 96,533 80 430,400 62 71 Overall (commercial farms and production enterprises) Akmola 19 196,146 1,195 408,240 17 89 EKO 17 24,709 300 74,860 12 71 Karaganda 9 71,589 3,480 190,760 8 89 Kostanay 18 212,384 9,520 537,020 17 94 NKO 13 235,180 116,160 517,280 13 100 Pavlodar 11 38,513 1,300 176,060 6 55 Source: Authors. ed. This county-level yield estimation procedure at the rayon level at the time of harvest, and this is considered accurate and impartial by both the could form the basis for indemnity under any insurance sector and the insured farmers. rayon-level AYII scheme in the future. In Ka- zakhstan, the ARKS uses a multistage crop yield 5.39. Kazakhstan uses systematic sampling estimation procedure first to select villages and procedures for measuring actual average yields farmers who are differentiated into production 158  Kazakhstan enterprises and commercial farms and then to than 25,000 hectares, while NKO has the larg- select fields for in-field randomly selected crop est, with an average of about 235,000 hectares cutting using 1 meter by 1 meter squares. Crop per insured unit (table 5.1). cutting is conducted immediately prior to harvest for all major grain crops, including spring wheat, Insured Yield Coverage Levels the weight of the yield from each sample is ad- justed according to its moisture content, and the 5.41. AYII policies typically offer optional average yield in centners per hectare is calculat- levels of insured yield coverage of between a ed. The crop cut yields for each type of producer maximum of 90 percent and a minimum of 50 are then averaged to estimate the average yields percent of the average area yield. In India, the per rayon separately for production enterprises NAIS has traditionally offered three coverage and commercial farms and then in aggregate for levels, 60, 80, or 90 percent of the past five-year both types of farms in the rayon. Full details of average yield in the IU. The decision over which the yield estimation procedures are contained in coverage level will apply in an IU is based on the ARKS’s Guideline on Crop Yield Inspection Ar- coefficient of variation around the mean yield, rangements (ARKS 2004). This yield estimation such that in IUs with a low CoV, the maximum procedure at the rayon level is considered to be 90 percent coverage level is applied, and in IUs impartial and accurate and could form the basis with a high CoV only 60 percent coverage is of- for indemnifying yield loss under a rayon-level fered. Under the U.S. Group Risk Plan, farmers AYII program for spring wheat. may select from optional coverage levels of be- tween 50 and 90 percent of the county’s average Considerations for the Design of Spring yield. In Kazakhstan, it is proposed that the in- Wheat AYII Cover sured yield for an AYII program be set as a per- centage of the most recent five-year actual aver- Area Insured Unit age yields from 2006 to 2010. On account of the very high variability in yield in some oblasts and 5.40. For the operation of an AYII cover, it is rayons, it will, however, be necessary to offer necessary to have a minimum sown area of the coverage levels as low as 20 percent or as high as insured crop in each insured unit (rayon). The 80 percent of the rayon’s five-year average yield setting of a minimum sown area is to avoid mor- in some oblasts and rayons. For this reason, rates al hazard or, in other words, to ensure that indi- were calculated for a wider range of coverage vidual farmers are not able to influence the area- levels, from 10 percent through to 80 percent yield outcomes in the IU. In the United States, of the rayon five-year average yield (termed the the minimum area is 15,000 acres (about 6,000 expected yield). The rayon spring wheat insured hectares) per county (Skees, Black, and Barnet yield coverage levels for a minimum of 10 per- 1997). In northern Kazakhstan, where there are cent up to a maximum of 80 percent of the past some very large spring wheat production enter- five-year average yield are shown by oblast and prises, the criterion used under the current study by rayon in annex 7. was a minimum of 10,000 hectares per rayon based on the past five-year average sown area Insured Values and Sum Insured from 2006 to 2010 (combined sown area for production enterprises and commercial produc- 5.42. Under an AYII policy, the insured crop ers). The size of spring wheat insured unit varies yields can be valued either on a cost of produc- hugely across the six oblasts and the 73 (84 per- tion basis or on a farm-gate sales price basis. In cent of total) qualifying rayons, with more than India, the NAIS commonly sets the sum insured 10,000 hectares of spring wheat. Overall, EKO according to the amount of credit provided to the has the smallest insured units, with an average farmers. In the United States, the GRP permits area per rayon of spring wheat of slightly less farmers to insure their selected coverage level Agricultural Insurance Feasibility Study  159 at up to 150 percent of the reference sales price. production, and yield data and for coverage lev- In Kazakhstan, the sum insured could be based els from a minimum of 10 percent up to a maxi- on any valuation criteria requested by farmers, mum of 80 percent of rayon-level five-year av- from a costs of production valuation through to erage yields (2006–10). The procedure adopted an expected revenue valuation. However, unlike involved detrending the 17-year actual yields for the United States, it is recommended that the spring wheat in each rayon and simulating the maximum unit sum insured value not exceed 100 estimated yield shortfall at each level of insured percent of the expected farm-gate sales price for coverage from 10 to 80 percent over 5,000 it- the crop. For the purposes of this exercise and to erations (years) to derive the average pure loss maintain consistency with the MPCI policy out- cost rates for each rayon. The pure loss cost rates lined in the previous section, the past three-year were then smoothed, and a “security load” was average September sales price for spring wheat added based on the calculated probable maxi- of KZT 3,210 per centner was used. mum loss (PML) to derive the technical rates for each rayon and each coverage level. Finally, for 5.43. The estimated sums insured for a spring the purposes of this rating exercise, the techni- wheat AYII program assuming 100 percent in- cal rates were grossed up by 30 percent to derive surance uptake in the six oblasts and an aver- indicative commercial premium rates for a tar- age 50 percent coverage level would be on the get 70 percent loss ratio. The 70 percent target order of about KZT 212 billion (US$1.41 bil- loss ratio was designed to allow the participat- lion). The estimated sums insured for the spring ing insurers to cover their acquisition costs and wheat AYII program by oblast and in total for administrative and operating (A&O) expenses coverage levels of between 10 percent and a and to provide a reasonable profit margin. In maximum of 80 percent are shown in table 5.2. practice, the insurers and their reinsurers will be Under the assumption of 100 percent uptake of responsible for estimating their costs and profit the AYII product for spring wheat, at the 10 per- margins and for setting their target loss ratios ac- cent coverage level, the total sum insured (TSI) cordingly. Since the costs of loss assessment are would be about KZT 42.4 billion (US$283 mil- considerably lower for AYII than for individual lion), rising to a very significant KZT 339 bil- grower MPCI, the gross-up used in this rating lion (US$226 billion) at the maximum 80 per- exercise is correspondingly lower for AYII, at cent coverage level. However, under a voluntary 30 percent, than for MPCI, at 40 percent. See AYII program the actual uptake rates would be chapter 4. much lower, and the TSI would be correspond- ingly lower than the figures for 100 percent up- 5.45. Indicative commercial premium rates take. for AYII cover are presented in this report, but these rates are merely illustrative, and insurers Rating Methodology and Indicative and their reinsurers will make the final deci- Premium Rates sions about rates. The AYII commercial premi- um rates for spring wheat for a 70 percent target 5.44. A preliminary estimation of the techni- loss ratio are presented in annex 8 by rayon for cal and commercial premium rates for an AYII each level of insured yield coverage. The aver- program for spring wheat is presented in this age rayon rates in each of the six oblasts as well report, using internationally accepted AYII as the average percentage premium rates and crop rating methodology. Annex 7 presents corresponding value of the premiums are pre- full details of the rating methodology used to sented in table 5.2. The average rates for spring establish the rayon-level technical rates for an wheat vary from a low in NKO to a high in Pav- area-yield index policy for spring wheat. These lodar and EKO; in most oblasts, premium rates rates are based on an analysis of variance in the become very expensive for coverage levels that ARKS’s 17-year rayon-level annual sown area, are greater than 50 percent of the rayon five-year 160  Kazakhstan Table 5.2 Estimated Total Sum Insured and Indicative Commercial Premium for Spring Wheat AYII Program, by Level of Coverage Insured yield coverage level (% of rayon average expected yield) Oblast 10% 20% 30% 40% 50% 60% 70% 80% Total sum insured (KZT millions) Akmola 10,985 21,970 32,956 43,941 54,926 65,911 76,897 87,882 EKO 1,181 2,361 3,542 4,723 5,904 7,084 8,265 9,446 Karaganda 1,346 2,691 4,037 5,383 6,729 8,074 9,420 10,766 Kostanay 14,786 29,573 44,359 59,145 73,931 88,718 103,504 118,290 NKO 13,216 26,432 39,648 52,864 66,080 79,296 92,512 105,728 Pavlodar 910 1,821 2,731 3,641 4,552 5,462 6,372 7,283 Total 42,424 84,849 127,273 169,697 212,121 254,546 296,970 339,394 Indicative commercial premium for 70% loss ratio (%) Akmola 0.24 1.03 2.43 4.52 7.07 10.34 14.26 18.69 EKO 0.74 2.25 4.50 7.41 10.72 14.64 19.03 23.67 Karaganda 0.33 1.31 3.05 5.58 8.62 12.28 16.48 21.09 Kostanay 0.24 1.01 2.38 4.42 6.93 10.17 14.04 18.44 NKO 0.17 0.73 1.70 3.14 4.85 7.17 10.15 13.80 Pavlodar 0.38 1.50 3.36 5.96 8.97 12.62 16.75 21.14 Total 0.24 0.98 2.28 4.20 6.52 9.52 13.16 17.35 Indicative commercial premium for a 70% loss ratio (KZT millions) Akmola 26 226 799 1,987 3,885 6,813 10,968 16,429 EKO 9 53 159 350 633 1,037 1,573 2,236 Karaganda 4 35 123 301 580 991 1,552 2,271 Kostanay 36 299 1,055 2,615 5,124 9,021 14,537 21,817 NKO 23 192 675 1,662 3,206 5,682 9,390 14,594 Pavlodar 3 27 92 217 408 689 1,068 1,539 Total 101 833 2,903 7,131 13,837 24,234 39,087 58,885 Source: Authors. See annex 7 for full details. Agricultural Insurance Feasibility Study  161 average yield. Assuming 100 percent uptake, the On the basis of the feedback obtained during the corresponding commercial premium for an aver- focus group meetings in Kazakhstan, it appears age 50 percent coverage level would be about that most farmers consider the current premium KZT 14 billion (US$95 million). rates charged on the obligatory LIC scheme (av- erage of about 2.4 percent for the subsidized rate 5.46. In order to interpret these rates, it is use- and average of about 4.4 percent for the full rate, ful to consider a maximum premium rate that assuming no government 50 percent claims sub- farmers might be willing to pay assuming there sidies) to be too expensive. It was not possible in are no premium subsidies above 10 percent. the limited time available in the panel meetings The analysis clearly shows that in NKO, where to conduct any formal assessment of demand for crop yields are generally very stable, AYII insur- AYII insurance. ance could be offered for high coverage levels of up to 70 percent at affordable premium rates of 5.48. If a voluntary AYII program is to be less than 10 percent. In Akmola and Kostanay, launched in Kazakhstan, scheme management farmers could be offered coverage levels up to will first need to conduct a detailed study of about 60 percent of average rayon yields under demand in order to quantify farmers’ interest the assumption of a 10 percent maximum pre- in purchasing voluntary AYII and the cover- mium rate. However, in Pavlodar, where spring age levels and premiums they are willing to wheat yields are more variable, only a maximum pay. This demand assessment study should be of 50 percent coverage could be provided, and designed to provide a clear picture of farmers’ in EKO coverage as low as 40 percent would be ability and willingness to pay for voluntary AYII needed to avoid exceeding the 10 percent com- and the coverage levels and sum insured values mercial premium rate. Annex 7 shows which they wish to insure. This study may also provide maximum cover levels could be offered for less useful feedback on the need for government pre- than 10 percent commercial premium in each mium subsidy support. rayon in each oblast. Government Support for Premium Coverage Levels, Affordable Premium Subsidies Rates, and Demand for Voluntary AYII Insurance 5.49. If government were to elect to provide premium subsidy support, farmers would be 5.47. Under a spring wheat AYII program for able to choose either to insure at higher cover- Kazakhstan, the coverage level in each rayon age levels or to reduce their costs of purchasing should be set in accordance with (a) the under- AYII cover. If government were to provide 50 lying exposure to and frequency of risk and (b) percent premium subsidies, the AYII premium the commercial premium rate that the targeted rates would be much more affordable, as shown farmers can afford. In order for a crop insur- in table 5.3. The recommended study on AYII ance scheme to be both affordable for farmers demand would also enable estimates to be made (premium rates of no more than 5 to 10 per- of the TSI for a pilot AYII scheme, the estimated cent) and sustainable, the insured yield cover- premium income, and therefore the estimated age should be set at a level where payouts are costs to government of a 50 percent premium no more frequent than about one in seven to one subsidy (or other level of premium subsidy to be in 10 years. Where AYII coverage levels are set determined). too high and the commercial premium rates are also too high, farmers are discouraged from pur- Operational Considerations chasing crop insurance, and the scheme cannot achieve the economies of scale and premium 5.50. For the operation of a spring wheat volume necessary to be sustainable over time. AYII cover, the key requirement would be to 162  Kazakhstan Table 5.3 Average Indicative Commercial Premium Rates for Spring Wheat AYII Cover Net of Government 50 Percent Premium Subsidies, by Oblast and Level of Coverage Coverage level Oblast 10% 20% 30% 40% 50% 60% 70% 80% Total sum insured (KZT millions) Akmola 0.12 0.51 1.21 2.26 3.54 5.17 7.13 9.35 EKO 0.37 1.12 2.25 3.71 5.36 7.32 9.51 11.83 Karaganda 0.16 0.65 1.53 2.79 4.31 6.14 8.24 10.55 Kostanay 0.12 0.51 1.19 2.21 3.47 5.08 7.02 9.22 NKO 0.09 0.36 0.85 1.57 2.43 3.58 5.07 6.90 Pavlodar 0.19 0.75 1.68 2.98 4.48 6.31 8.38 10.57 Total 0.12 0.49 1.14 2.10 3.26 4.76 6.58 8.68 Source: Authors. formalize the procedures for estimating actual 5.51. It will also be necessary to provide farm- average yield in each rayon where insurance er education and training in the operation of is offered. It is recommended that the insur- AYII. Given the fact that AYII does not provide ance companies enter into a formal agreement insurance and indemnity at the individual farmer with the ARKS to provide the results of their level, but according to losses at the rayon level, crop-cutting yield estimates for each rayon. It it is essential that the principles of this cover be is essential to minimize the delay in publishing clearly explained to farmers under a program of these rayon-level area-yield estimates after the farmer education and training. harvest in order to settle claims to farmers in rayons where the actual average yield falls short 5.52. Distribution channels should be inves- of the insured yield. A common criticism of the tigated under any pilot AYII program in the AYII program in India is that it often takes be- future. Currently in northern Kazakhstan, most tween six and 12 months for state-level govern- crop insurance is marketed through sales agents ments to publish the results of their annual crop- located in each oblast and rayon. Alternative cutting experiments. In the meantime, farmers channels for marketing and administering AYII need to purchase seeds and fertilizers for the crop insurance should be investigated, including next season and to pay for land preparation and through farmer associations and cooperatives, sowing activities; they cannot wait 12 months rural banks, and input suppliers. The potential to receive their indemnities. In Kazakhstan, the for crop credit–linked insurance also merits in- ARKS should be able to make the results of its vestigation. crop-cutting surveys available prior to the start of the next spring wheat season. It is also likely 5.53. There may also be opportunities to that, under an AYII scheme, insurers and their market AYII not only to individual farmers reinsurers would wish to put in place indepen- and joint-stock companies and cooperatives, dent monitoring systems to verify the actual av- but also to banks and input suppliers in each erage crop yields in the insured rayons. rayon as meso-level financial protection. In Agricultural Insurance Feasibility Study  163 this instance, the cover would be designed to would be equivalent to a loss ratio of about 370 provide business interruption cover to the lend- percent. For the highest 80 percent coverage ing institution in the event that a catastrophic level, the 1-in-100-year PML would be about 45 drought in a named rayon(s) prevents farmers percent of TSI or a loss of KZT 152.7 billion from repaying their loans, forcing the lender to (US$1.02 billion), assuming 100 percent uptake. reschedule or write off its loans in that rayon(s). This would be equivalent to a loss ratio of about In addition, such a cover might be attractive to 259 percent. regional input suppliers who provide seeds, fer- tilizers, and plant protection chemicals to farm- 5.55. Some indicative reinsurance pricing was ers on credit against repayment at the time of the conducted for the spring wheat AYII program. wheat harvest. This analysis was conducted assuming (a) ag- gregate reinsurance protection over the entire Estimated PML and Implications for spring wheat insured area assuming 100 per- Reinsurance cent uptake, (b) three priority levels of 70, 100, and 150 percent of gross net premium income 5.54. Some preliminary estimates of the PML (GNPI), (c) reinsurance that would provide full- that might be expected under an area-yield in- value protection up to 100 percent of TSI in ex- dex program for spring wheat in Kazakhstan cess of these priority levels, and (d) insured yield are presented in figure 5.5 and table 5.4. For coverage levels of 10–80 percent of the rayon a 50 percent coverage level the 1-in-100-year five-year average yield. Under these assump- PML would be about 24 percent of TSI or a tions, table 5.5 shows that, for the 100 percent loss of KZT 51 billion (US$343 million), as- of GNPI stop-loss reinsurance option, full-value suming 100 percent uptake of the scheme. This protection, and 50 percent coverage level, the Figure 5.5 Estimated PML for the AYII Spring Wheat Scheme, by Level of Coverage 60.00% 50.00% 40.00% "10% Coverage" "20% Coverage" Loss cost (%) 30.00% "30% Coverage" "40% Coverage" 20.00% "50% Coverage" "60% Coverage" 10.00% "70% Coverage" "80% Coverage" 0.00% 1 100 150 50 200 250 Return period (years) Source: Authors. 164  Kazakhstan Table 5.4 Estimated 1-in-100-Year PML for the AYII Spring Wheat Scheme Assuming 100 Percent Uptake Coverage level Item 10% 20% 30% 40% 50% 60% 70% 80% PML (% of TSI) 1.11 4.53 10.04 17.40 24.29 31.84 39.15 44.99 PML (KZT billions) 0.5 3.8 12.8 29.5 51.5 81.0 116.3 152.7 PML (US$ millions) 3 25 85 197 343 540 775 1,018 1-in-100-year PML 466 461 440 414 372 334 297 259 loss ratio (%) Source: Authors. Table 5.5 Indicative Pricing for Aggregate Stop-Loss Reinsurance for a Full-Value AYII Spring Wheat Program Insured yield coverage level (% of rayon average expected yield) Item and priority 10% 20% 30% 40% 50% 60% 70% 80% Stop-loss reinsurance premium (KZT millions) 70% 54 349 1,046 2,295 4,297 7,112 10,733 15,157 100% 52 326 948 2,005 3,609 5,693 8,084 10,606 150% 49 295 820 1,634 2,764 4,008 5,078 5,731 Stop-loss reinsurance premium (% of GNPI) 70% 53.90 41.86 36.04 32.18 31.05 29.35 27.46 25.74 100% 51.43 39.13 32.66 28.12 26.09 23.49 20.68 18.01 150% 48.19 35.48 28.25 22.91 19.98 16.54 12.99 9.73 Source: Authors. stop-loss pricing would be on the order of 26.09 be considerably higher. percent of GNPI or KZT 3.6 billion (US$24.0 5.56. A layered risk financing program could million). The 100 percent of GNPI stop-loss re- also be developed for the AYII program involv- insurance option and full-value protection for ing both private insurers and reinsurers and the highest 80 percent coverage level would cost the GRK as a catastrophe reinsurer. Such a risk approximately 18.01 percent of GNPI or KZT financing program could be structured along 10. 6 billion (US$67 million). If the stop-loss exactly the same lines as the stop-loss reinsur- protection priority were lowered to 70 percent ance cover illustrated in chapter 4 for individual of GNPI, the aggregate stop-loss pricing would grower MPCI. Agricultural Insurance Feasibility Study  165 Financial Estimates for a Voluntary AYII Conclusions on AYII Portfolio 5.58. AYII for spring wheat is technically and 5.57. Some provisional financial estimates operationally feasible in Kazakhstan. However, were calculated for AYII cover for spring wheat until further research has been conducted on the assuming a voluntary program and 5 percent potential demand for this cover, it is very diffi- incremental uptake rates per year over the next cult to predict likely uptake rates under a volun- five years. Under the assumptions of a 50 per- tary crop insurance program. cent insured yield coverage level and 5 percent uptake per year over the next five years, the total 5.59. AYII for spring wheat could be under- sum insured might rise from KZT 10.6 billion written either as a micro-level individual farm- in year 1 with corresponding premium income er cover or as a meso-level product designed to of KZT 692 million, to KZT 53 billion after five protect the seasonal loan portfolio of agencies years, with premium income of KZT 3.5 billion. that are lending to cereal producers (banks, in- The costs to government of different levels of put suppliers, or MFIs) in Kazakhstan. There premium subsidies (ranging from 25 to 65 per- are two advantages in offering area-yield index cent of premium) are shown in table 5.6. These insurance at a meso level or aggregate product. uptake estimates are extremely ambitious for a The first is that basis risk would be much less voluntary insurance scheme and will need to be of a concern than under an individual grower refined following the recommended study on program. The second is that the transaction costs AYII demand. would be lower for this coverage than for indi- Table 5.6 Five-Year Estimates of Voluntary AYII Uptake, Total Sum Insured, Premium Income, and Costs of Premium Subsidies (50 Percent Coverage Level) AYII crop insurance uptake scenarios AYII 50% coverage level at 100% uptake Year 1 Year 2 Year 3 Year 4 Year 5 Item 5% 10% 15% 20% 25% Total sum insured (KZT millions) 212,121 10,606 21,212 31,818 42,424 53,030 Commercial premium (KZT 13,837 692 1,384 2,076 2,767 3,459 millions) Cost of GRK premium subsidies (% of premium) 25% 3,459 173 346 519 692 865 50% 6,918 346 692 1,038 1,384 1,730 65% 8,994 450 899 1,349 1,799 2,248 Probable maximum loss 1 in 100 years 51,516 2,576 5,152 7,727 10,303 12,879 1 in 250 years 70,634 3,532 7,063 10,595 14,127 17,659 Source: Authors. 166  Kazakhstan vidual farmer micro-level insurance. on time period. The parameters of the contract are set so as to correlate, as accurately as possi- 5.60. Meso-level AYII cover may be an effec- ble, with the loss of a specific type of crop by the tive tool for governments to operate for small policyholder.40 All policyholders within a defined family farms in southern Kazakhstan. This op- area receive payouts based on the same contract tion is reviewed further in chapter 6. and measurement at the same station, eliminat- ing the need for field loss assessment. WII is best 5.61. Farmers’ demand for and willingness to suited to weather hazards that are well correlated pay for AYII crop insurance will also have to be over a widespread area and where weather and studied further before any decisions are made crop yield are closely correlated. The strongest to proceed with the design of an AYII program. relationships typically involve a single crop, a This feasibility study identified a very low level marked rainy season, and no irrigation. WII is of interest in the obligatory LIC crop insurance less useful where more complex conditions ex- scheme by farmers, and it is probable that this ist. Localized risks (such as hail) or areas with would apply equally to voluntary crop insurance microclimates (for example, in mountainous in the future. It is recommended that a formal areas) are not suitable for WII. Similarly, the study be conducted to assess demand for AYII scope for WII is limited where crop production for crops. The experience of India, Ukraine, and is affected by many or complex causes of loss the United States may provide useful lessons for or where pests and diseases are major influences the design of AYII in Kazakhstan. on yields. For a given environment, other insur- ance products may be more appropriate (such as area-yield index insurance or named-peril crop Crop Weather Index Insurance insurance). These conditions are present in the north of Kazakhstan, making a feasibility analy- 5.62. The analysis carried out in this feasibility sis worth pursuing. study shows that developing WII contracts for hedging the drought exposure of spring wheat 5.64. As for AYII, basis risk is the key con- in the north of Kazakhstan is technically feasi- straint of WII. “Basis” can be defined as the ble. However, challenges in the possible scale of difference between the loss experienced by the implementation, in the commercial viability, and farmer and the payout triggered by the weather in farmers’ interest in WII may limit the scope of index. It could result in a farmer experiencing a application of this class of insurance products. yield loss without receiving a payout or receiv- The details of the analysis are presented in this ing a payout without experiencing a loss. WII section, but further research is needed if the in- works best where losses are homogeneous in surance companies decide to proceed with the the defined area and highly correlated with the pilot testing and implementation of this product. weather peril. A more detailed description of the It is assumed that WII will be introduced as a strengths and weaknesses of WII is provided in voluntary program. Further details of the WII annex 8. product are contained in annex 8. 5.65. WII can be retailed at different busi- Features, Advantages, and ness levels. At the micro level, the policyholders Disadvantages of WII (the insurer’s customers) are individual farmers, households, or small business owners who pur- 5.63. The essential feature of weather index- chase insurance to protect themselves against po- based insurance is that the insurance contract responds to an objective parameter (for exam- ple, measurement of rainfall or temperature) at 40  The material presented in this section was a defined weather station during an agreed-up- adapted from IFAD (forthcoming). Agricultural Insurance Feasibility Study  167 Table 5.7 International Experience with Weather Index Insurance at Different Levels of Aggregation Level of aggregation Description Example Weather-indexed insurance for China, Ethiopia, Ghana, India, Kenya, smallholder farmers; more than 30 Micro Malawi, Nicaragua, the Philippines, projects in about 25 countries; scale-up Thailand, Ukraine only in India Weather-indexed portfolio hedge for rural financial institutions that lend to poor Ghana, Peru, Vietnam (under Meso farmers; programs are too new to assess development) scale-up and sustainability Weather insurance or weather-indexed contingent credit line for governments or Ethiopia, Malawi, Mexico (both AYII Macro international organizations; major scale- and WII), Caribbean states risk pool up achieved in Mexico across most for hurricanes and earthquake states in the past decade Source: Dick 2009. tential losses caused by adverse weather events. of the total daily data set. In addition, reliable At the meso level, WII can be used to cover the and trustworthy ongoing daily collection and re- exposure of entities such as financial service porting procedures should be assured. providers, farmer associations, input suppliers, and processors from potential losses caused by 5.68. Beyond the quality of data, it is critical to adverse weather events. At the macro level, WII define the boundaries of the area(s) covered by can aid governments and relief agencies in de- the weather station(s) so that WII contracts can velopment and disaster management. be written for specific areas tied to a specific station. A general rule of thumb is to consider a International Experience with WII specific WII contract marketable within a 20 ki- lometer radius of the weather station; however, 5.66. The majority of WII experience has been the applicable area may be smaller or larger, and with micro-level applications and rainfall defi- case-specific evaluation must be carried out. In cit (drought). To date, many initiatives have been general terms, the more the terrain varies, the piloted, but only in India has a market-based more the acceptable distance from a station de- scale-up of WII taken place. Table 5.7 summa- creases. rizes the experience in the countries where WII has been piloted. 5.69. Agricultural data and information are the second part of the WII contract design Weather and Agricultural Data for WII equation. The most relevant information to be collected is data on productivity (yield), but a 5.67. WII relies on historical and current clear description of the agricultural production weather data that should adhere to specific practices carried out in the areas of interest is quality requirements. In order to meet require- also necessary. Unfortunately, yield data series ments for a commercial WII insurance and rein- of adequate length, sufficient quality, and at the surance transaction, a recommended guideline is appropriate level of disaggregation are generally that there be at least 20 years of historical daily not available. However, lack of quality yield data and that missing data not exceed 3 percent data does not pose as large a problem as lack of 168  Kazakhstan good weather data, since it is still possible to find payouts when losses are experienced, reduc- alternative approaches to estimating yield vari- ing basis risk as much as possible. The set of ability. One possibility is to simulate synthetic possible index combinations is unlimited, and yield data series through plant-growth models. numerous structures have been developed in the relatively short history of WII. One of the Design of WII Contracts most commonly adopted structures is that of a continuous payout triggered and limited by a 5.70. The objective of contract design is to cumulative measure of the weather variable (for capture the relationship between the weather example, rainfall) for each stage of crop growth. variable and the potential crop loss and to se- See box 5.4 and figure 5.6. lect the index that is most effective in providing Box 5.4 Payout Parameters in a WII Contract Using the drought coverage case represented in figure 5.5 as an example, the parameters that characterize an incremental payout structure can be defined as follows: • Trigger. Threshold above or below which payouts are due. Payments are due when the calculated value of the index is below the trigger level (300 millimeters). • Exit. Threshold above or below which no additional incremental payout is applied. The maximum pay- out is paid if the calculated value of the index is equal to or below the exit threshold (100 millimeters). • Tick. Incremental payout value per unit deviation increase from the trigger. With a maximum payout (the insured sum) of US$200, a trigger of 300 millimeters, and an exit of 100 millimeters, the monetary value of each deficit millimeter of rainfall below the trigger is US$200 divided by (300 millimeters–100 millimeters) or $1 per millimeter. Source: Authors. Figure 5.6 Payout Structure of a WII Contract for Drought 250 Exit: 100mm 200 Payout (US$) 150 100 Trigger: 300mm 50 0 1 51 101 151 201 251 301 351 401 Source: Authors. Agricultural Insurance Feasibility Study  169 Weather Data and Infrastructure in tions may be caused by general problems at the Kazakhstan weather station level. 5.71. As highlighted in chapter 2, the World 5.73. Over most of the country, the seasonal Bank feasibility study team carried out a de- precipitation cycle is of mid-range rainfall in- tailed review of the meteorological network tensity, with a cycle distributed relatively uni- managed by the KHM. The weather data pro- formly throughout the year. Therefore, rather vided by the KHM enabled an in-depth assess- than marked seasonal distribution and erratic ment to be conducted of weather risk in the patterns, as seen in many other regions of the selected spring wheat production areas and the world, average humidity and total seasonal cu- development of WII contract structures. mulated precipitation seem to be the main con- cerns for the farming environment. Given this 5.72. The continuity of historical data pro- hydrological precondition, the thermal regime vided by the KHM is generally of good qual- also plays a relevant role. Combined hydrother- ity. For the purpose of this feasibility study, the mal indexes are in common use in agro-mete- World Bank gained access to daily rainfall and orological monitoring in Kazakhstan and were minimum and maximum temperatures from 10 adopted in this study for the design of WII pro- weather stations located in seven rayons listed totypes. See the following sections and also ap- in table 5.8 and portrayed in map 5.1. The rate pendix C in annex 8. of missing data is generally very low (less than 1 percent). In particular, the amount of missing 5.74. An analysis of the last 25 years of data rainfall observations is in line with the record found no significant rainfall trend, but did find of missing temperature observations, and this a weak but statistically significant temperature indicates overall good management of weather trend. The analytical results of the trend analy- sensors, as events leading to missing observa- sis are presented in appendix D of annex 8. An Table 5.8 Selected Weather Stations and Statistics on Available Weather Data % of missing % of missing Number of Weather station Rayon Oblast daily rainfall daily temperature years observations observations Diyevskaya Auliyekolski Kostanay 26 0.0 0.0 Kushmurun Auliyekolski Kostanay 26 0.6 0.6 Kostanay Kostanayski Kostanay 26 0.0 0.0 Zholboldy Aktogayskiy Pavlodar 26 1.2 1.2 Aktogai Aktogayskiy Pavlodar 26 0.2 0.2 Mikhailovka Zhelezinski Pavlodar 26 0.0 0.0 Bolkashino Sanytausky Akmola 26 0.0 0.0 Schuchinskoye Enbekshildersky Akmola 26 0.0 0.0 Stepnogorsk Enbekshildersky Akmola 26 0.3 0.3 Tolebiysky Tasarik Sko 29 0.0 0.0 Source: Authors based on KHM data. 170  Kazakhstan Map 5.1 Weather Stations Analyzed in the Feasibility Study Source: Authors’ elaborations on Google Earth. Note: Weather stations are identified by the red markers. Rayons selected for the analysis under the feasibility study are highlighted in light gray shading (rayon names are in yellow). important step in the design of weather index in- lected, weather patterns may be comparable at surance programs is to check for the existence of low scale (the rayon level) and become more long-term tendencies in key atmospheric param- heterogeneous at the oblast level (see table 5.9). eters that would affect the price and effective- As expected, higher correlations are noted for ness of the contracts. A detailed trend analysis weather stations that are closer in distance and, for 10 weather stations in the target rayons was roughly, in the same rayon. In general, intra- carried out. This analysis, which is presented oblast variability of weather patterns seems to in appendix C of annex 8, found no significant be more pronounced. trend in annual rainfall. Instead, average annual temperature is increasing slightly in almost all of 5.76. Although from a climatological point of the analyzed weather stations. Since all indexes view correlations between weather data col- adopted for the contract design depend mainly lected at the rayon-level distance are very high, on cumulated rainfall and temperature trends do when referring to the potential coverage of the not seem to affect the developed structures, tem- territory for WII purposes, different evaluation perature detrending was not carried out. Nev- criteria must be applied. Even though Kazakh- ertheless, careful handling of issues related to stan’s climatic conditions seem appropriate for weather trends is recommended for the design relaxing the 20–25 kilometer radius rule as the of contracts for commercial use. criterion for identifying the scale at which WII can be retailed, the average distance between 5.75. Drought exhibits a systemic pattern weather stations is too great to assume that the across the north of Kazakhstan. The detailed current network could provide comprehensive correlation analysis carried out at the weather coverage of northern Kazakhstan. As illustrated station level indicates that, in the test areas se- in the maps presented in appendix A of annex 8, Agricultural Insurance Feasibility Study  171 Table 5.9 Correlation Matrix for Cumulated Precipitation (January to August) in Selected Weather Stations Kostanai Pavlodar Akmolinsk Schuchinskoye Stepnogorsk Baklkashino Mikhailovka Kushmurun Oblast and Diyevskaya Oblast and Oblast and Zholboldy Kostanay Aktogai rayon rayon rayon Kostanai Pavlodar Akmolink Diyevskaya — 72 78 Zholboldy — 54 27 Baklkashino — 78 59 Kushmurun 66 — 66 Aktogai 49 — 70 Schuchinskoye 61 — 59 Kostanay 60 59 — Mikhailovka 43 60 — Stepnogorsk 46 51 — Source: Authors based on KHM data. Note: In each oblast section of the table, the “shaded values” (top-right corner) correspond to correlations between cumulated seasonal precipitation, while “unshaded values” (bottom-left corner) correspond to correlations between de- viations of 10-day cumulated rainfall with respect to the average seasonal cycle. Subtracting the average seasonal cycle from the cumulated time series helps to identify more accurately the degree of homogeneity among weather patterns in an area. Similar seasonal cycles, typical of any region, may lead to overestimating weather pattern similarities. distances between contiguous weather stations Agricultural Data start from a minimum of 70 kilometers, which is probably too large for implementing a scheme 5.78. As discussed in chapter 2, the ARKS col- of full micro-level WII coverage of the entire lects annual average yield data for spring wheat territory. at the rayon level. Given the excellent quality of the historical yield reports, it was possible to 5.77. In conclusion, the excellent quality of use the official ARKS yield data for developing weather data provided by the KHM complies WII prototypes for the selected case studies.41 with all of the requirements for designing and The rayon annual average yield series that were operating WII. However, although this applies provided span the period from 1990 (or 1991 in to the areas surrounding the specific weather some cases) up to 2010. stations analyzed, full-scale implementation of a micro-level (farm-level) WII program may 5.79. Rayon yield reports provide average yield be hindered by the relatively low density of the series for two categories of farms: production weather station network. In the medium term, enterprises and commercial farms. While in it may be possible to overcome these structural the case of AYII, the analysis was carried out for constraints through efforts to improve the den- the two separate classes of farms, in the case of sity of the weather station network. In the short term, however, widespread full-scale implemen- tation of WII at the farm level does not seem 41  In WII this is not a common feature, as the de- to be realistic. Potential implementation of WII veloping countries in which these kinds of insurance contracts at the meso or macro levels, which may products are tested often do not have sufficiently good agricultural statistics. In such cases, informa- be less influenced by the density of the weather tion collected in the field and crop modeling ap- station network, could have greater chances of proaches are used to reconstruct the required crop being rapidly implemented. loss history. 172  Kazakhstan Table 5.10 Selected Rayons and Representative Weather Stations Station selected to represent the Rayon in which weather stations Oblast and selected rayon weather pattern of the rayon are actually located Akmola Bulandinsky WS Bolkashino Sandytausky Enbekshildersky WS Schuchinsk Burabaisky WS Stepnogorsk Stepnogorsk Kostanay Auliekolsky WS Diyevskaya Auliekolsky WS Kushmurun Auliekolsky Altynsarinsky WS Kostanay Kostanay Pavlodar Aktogay WS Zholboldy Aktogay WS Aktogai Aktogay Zhelezinsky WS Mihailovka Zhelesinsky Source: Authors. WII, in order to identify one specific relation- were considered representative of the selected ship between weather and yield, the two types of production environments. Since not all of the records were aggregated in a single rayon yield selected spring wheat–producing rayons host series. a main weather station, weather measurement points from neighboring rayons were adopted. 5.80. Drastic reductions in yield performance Table 5.10 provides a list of all production ray- were recorded during the early years of Ka- ons and the weather stations selected to repre- zakhstan independence. At the end of the 1990s, sent their weather patterns. The maps presented more intensive production technology again be- in appendix A of annex 8 provide a geographic came available and production levels improved representation of such information as well as a significantly. In order to avoid overestimating full list of weather stations in the north of Ka- the impact of weather elements on yield losses zakhstan. in the periods of low-production performance, data from 1990 (or 1991) to 1998 were recen- Contract Design tered at the mean level of the 1999–2010 period. See appendix D in annex 8 for more details. 5.82. The design of WII contracts for cov- ering the drought exposure of spring wheat 5.81. As discussed in preceding sections, in or- production in the north of Kazakhstan were der to capture the impact of weather variability designed on the basis of the methodology de- on spring wheat production in the six select- veloped by the Agriculture Risk Management ed rayons of northern Kazakhstan, the KHM Team (ARMT) of the Agricultural and Rural provided data from nine weather stations that Development (ARD) Department of the World Agricultural Insurance Feasibility Study  173 Bank.42 Specific additional indexing procedures precipitation and the contribution of tempera- were developed for the spring wheat environ- ture to plant growth. They are commonly used ment of northern Kazakhstan. In the contract in Kazakhstan to identify drought conditions design activities, three weather indexes were and therefore were selected for drought index- adopted: cumulative precipitation, hydrothermal ing purposes.45 Both indexes are structured as a ratio (HTR), and humidity factor (K). ratio between cumulated daily rainfall and cu- mulated average daily temperature. The differ- 5.83. The cumulative precipitation index can ences in the two indexes are related to the pe- be defined as the sum of all recorded precipita- riods across which the weather parameters are tion across a specific time period. It is one of the cumulated and to the weighting factors applied most common indexes adopted in the index ap- to the cumulated measures. The exact specifica- proach to insuring agricultural drought. In many tions of the HTR and K indexes are provided in contract structures for cereal crops, crop life is appendix C of annex 8. divided in phases (usually three) on the basis of the different water requirements of the differ- 5.85. Table 5.11 summarizes the findings of ent phenological phases. Accordingly, specific the contract design activities carried out for de- indexes are developed for each phase of crop veloping drought protection products for spring growth. Given the specific agroclimatic condi- wheat in northern Kazakhstan. As indicated in tions of Kazakhstan, breaking up the phases did preceding sections, daily precipitation and tem- not add value to the contract performance, and, perature data from nine weather stations were therefore, a single cumulative precipitation in- made available for capturing the variability in dex was adopted.43 An additional feature of the spring wheat yield in six rayons in three oblasts northern Kazakhstan production environment is of northern Kazakhstan. Accordingly, nine WII that precipitation preceding the sowing period prototypes, one for each of the available rayon– (mainly in the form of snow) may have a role in weather station combinations, were developed. crop growth.44 5.86. The case studies were useful for assess- 5.84. The two other indexes adopted in the ing the technical feasibility of WII for spring design of contract structures for northern Ka- wheat in Kazakhstan. In seven of the nine cases zakhstan (HTR and K) take into account both it was possible to model a drought-yield rela- tionship. Out of these seven cases, three struc- tures performed excellently, three performed 42  The concepts and operational details of the less well but were still considered acceptable, methodology are described in the web-based train- ing material available at www.agrisktaining.org. and one was not acceptable. In two of the nine cases, it was not possible to define a meaningful 43  This finding is quite intuitive and is related to the fact that, in the north of Kazakhstan, the demand for drought-yield index. water evaporation from the soil and plant transpira- tion (together defined as “evapotranspiration”) is low- 5.87. For the majority of the rayon yield– er than in many of the semiarid countries where WII weather station combinations, it was possible is usually implemented. Hence, water can be stored in the soil for longer periods of time, making crop to develop meaningful index structures. This growth less dependent on the conjectural precipita- indicates that using WII to cover spring wheat’s tion level. drought exposure in the north of Kazakhstan 44  This is again different from the case of semiarid may be technically feasible. countries where at the start of the crop life the soil water balance is approximately zero and the onset of the rainfall season replenishes soil water reserves. In Kazakhstan snowmelt provides significant water 45  The analysis of hydrothermal indexes draws ex- resources that can be stored by the soil. tensively on a detailed study by Baisholanov (2011). 174  Kazakhstan Table 5.11 Performance of Drought Index Contract Prototypes Farm location Reference Average (oblast and weather Overall yield Maximum Indicative Basis Selected rayon) station (rayon perfor- (centner payout (KZT premium risk index and weather mance per per hectare) rate index station) hectare) Kostanay, Auliekoski, High 9.2 HTR 19,000 13.7 18 Auliekoski Kushumurum Kostanay, Kostanay, Cumulative High 11.6 24,000 16.8 41 Altynsarin Kostanay precipitation Akmola, Stepnogorsk, High 10.8 K 23,000 15.2 50 Enbelshiderski Stepnogosk Akmola, Schuchinsk, Medium 10.8 K 23,000 14.9 86 Enbelshiderski Schuchinsk Pavlodar, Aktogay, Cumulative Medium 6.1 13,000 20.9 90 Aktogay Aktogay precipitation Pavlodar, Aktogay, Cumulative Medium 6.1 13,000 24.2 86 Aktogay Zholbody precipitation Akmola, Sandytauski, Medium- 9.1 K 13,000 3.1 67 Bualindiski Bodkashino low Pavlodar, Zhelenovsky, Low 8.4 — 26 0.0 0.0 Zhelesinski Mikhailovka Kostanay, Diyevkaya, Low 9.2 — 26 0.3 0.3 Auliekoski Diyevkaya Source: Authors. yellow = modeling successful and high-to-medium prototype per- Note: — = Not available. Color coding is as follows: yellow formance; brown brown = modeling successful and medium-to-low prototype performance; grey grey = modeling not successful. The prototypes were calibrated in order to trigger payouts for yield observations that fall below 65% of the average rayon yield level. For this reason, maximum payout levels vary according to rayon average yields. For the other simulations car- ried out in the study, the price reference used to convert yield into revenue losses was KZT 3,210, corresponding to the average of the reference wheat prices in 2008, 2009, and 2010. 5.88. Although it would be necessary to verify undertaken. In particular, developing synthetic the dynamics of each specific yield-drought in- rainfall and temperature data sets through a grid- dex relationship, the diverse performance re- ding analysis may provide the additional data corded in the various cases reflects the fact that reference points needed for carrying out the risk the different weather stations may be more or assessment and contract rating procedures.46 In less representative of the average meteorologi- cal conditions of the wheat production areas. If it becomes evident that the location of the 46  Such an approach has been piloted in Mexico, weather measurement point with respect to the where Agroasemex, a parastatal insurance and re- insurance company, has developed a methodology production area is the reason for poor contract using reanalysis techniques to obtain a simulated se- performance, specific corrective actions can be ries of weather variables. Similar activities are under Agricultural Insurance Feasibility Study  175 Figure 5.7 Historical Performance of Drought Index Contract for Spring Wheat in Altynsarin 25,000 20,000 15,000 KZT per hectare 10,000 5,000 0 -5,000 19 7 19 9 19 0 19 1 19 2 19 3 19 5 19 7 20 9 20 0 20 1 20 2 20 3 20 5 20 7 20 9 10 86 88 94 96 98 04 06 08 8 8 9 9 9 9 9 9 9 0 0 0 0 0 0 0 19 19 19 19 19 20 20 20 19 -10,000 -15,000 Cumulated Rain Payout Revenue Loss Source: Authors. Note: Rayon = Altinsarin; weather station = Kostanai; indicative yield level coverage = 65%; index = cumulative precipi- tation, March 1–July 20; trigger = 95 millimeters; exit = 30 millimeters; contract maximum payout = KZT 24,000; basis risk index = 41%; reference gross premium = 16.8%. combination with the gridding analysis, targeted Altinsarin rayon on the basis of data collected at network upgrading activities, like the installa- Kostanay weather station. The graph shows that tion of new automatic weather measurement the potential payouts that the index would have equipment, could then provide the means for provided in the last 25 years match very closely settling WII contracts. with the average spring wheat revenue losses suffered in the Altinsarin rayon. The contract 5.89. Two of the nine contract structures are presented in figure 5.8, which refers to the case presented in figures 5.6 and 5.7.47 The two of Aktogayskiy rayon, uses data collected at Ak- contract structures presented constitute the two togai weather station and shows a relatively less extremes of the range of technically accept- well-performing match between losses and pay- able structures. All of the contract structures outs. While all major drought losses are matched were developed and calibrated for a coverage by a payout (the 1999 revenue loss, in yellow on level of 65 percent of the average rayon yield.48 the graph, was attributable mainly to a locust at- The contract presented in figure 5.7 refers to tack), the size of the payouts does not cover the the drought-index structure developed for the corresponding losses exactly, and unnecessary payouts also would have been triggered by the contract. Accordingly, the lower performance of the contract for the Aktogayiskiy case is indicat- development in various other countries. ed by the significantly higher value of the basis 47  Graphical representations of the other structures risk index (90 versus 43 percent for the Altinsa- are presented in appendix F in annex 8. rin contract).49 48  In WII the yield coverage level is a purely indica- tive measure used only for calibration purposes. In fact, payouts are triggered only by specific realiza- tions of the weather indexes, regardless of the actual 49  The “basis risk index” was developed in order recorded yield level. to indicate the approximate mismatch between the 176  Kazakhstan Figure 5.8 Historical Performance of Drought Index Contract for Spring Wheat in Aktogayskiy 15,000 10,000 5,000 KZT per hectare 0 19 6 19 7 19 8 19 9 19 0 19 1 19 2 19 3 19 4 19 5 96 19 7 19 8 20 9 20 0 20 1 20 2 20 3 20 4 20 5 20 6 20 7 20 8 20 9 10 -5,000 8 8 8 8 9 9 9 9 9 9 9 9 9 0 0 0 0 0 0 0 0 0 0 19 19 -10,000 Yield reduction due to locust attack -15,000 Payout: K Payout: GTK Source: Authors. Note: Rayon = Aktogayskiy; weather station = Aktogai; indicative yield level coverage = 65%; index = cumulative precipitation, March 1–August 31; trigger = 150 millimeters; exit = 50 millimeters; contract maximum payout = KZT 13,000; basis risk index = 90%; reference gross premium = 20.9% 5.90. As indicated in table 5.11, indicative each of the contract structures indicates the gross premium rates of the technically accept- potential match between payouts and historical able contract structures range from 13.7 to 24.2 average yield losses in a specific area, basis risk percent. The potential costs of using WII con- also has a spatial component, generated by the deviations in individual farm yields from the tracts for insuring against drought may be very aggregate average yield on which the weather high, and the different case study areas have dif- index is calibrated. This dimension influences ferent exposures to drought risk.50 the ability of the contract to compensate the po- tential policyholder in the appropriate measure. 5.91. While the basis risk index calculated for In this respect, in order to provide some empirical indications of the degree of spatial basis risk, the local project team collected detailed information loss suffered and the potential payouts. It is defined on the 2008, 2009, and 2010 yield performance as the ratio between (a) the sum of the absolute val- of commercial farms and production enterprises ues of the difference between payouts and revenue located within a radius of approximately 20 to losses and (b) the sum of payouts. Such an index has no actuarial value or rating function and should be 25 kilometers from the Kostanay, Aktogayskiy, considered an approximate reference for comparing and Mikhailovka weather stations. The data are the performance of different index contracts. presented in appendix G of annex 8. 50  The premium rates suggest a progressively in- creasing exposure to drought moving from Kostanay to Akmola and then to Pavlodar. While this seems to 5.92. Figures 5.9 and 5.10 provide an “as if” be in line with the other findings of the study, con- scenario in which mismatches between 2010 tracts developed for Pavlodar (Aktogayskiy) may farm-level losses and the 2010 WII payouts have relatively lower performance and, consequently, are calculated for both commercial farms and the higher premium rates may also be factoring in the need to provide higher payouts in order to capture all production enterprises operating in the sur- potential losses. roundings of Kostanay weather station. Devia- Agricultural Insurance Feasibility Study  177 tions for both farm typologies oscillate between As indicated in chapter 4, crop insurance com- +14 percent and −30 percent of average yields, panies in Kazakhstan are very exposed to cata- indicating a relevant variability in the effective- strophic losses on their retentions, and options ness of compensating farm-level losses.51 While for enhanced reinsurance protection need to be it is clearly not possible to draw any statistically considered. The GRK currently provides free significant conclusions from this simple elabo- proportional reinsurance protection equal to 50 ration, the exercise indicates the potential pres- percent of the claims to the private insurance ence of a relevant dimension of basis risk. It also companies and mutuals in Kazakhstan. How- highlights the importance of carefully evaluat- ever, neither the private insurers nor the mutuals ing the spatial component of basis risk that is have any reinsurance protection on their 50 per- embedded in farm-level WII products.52 cent retentions; they are therefore very exposed to major systemic drought losses. Meso, Macro, and Reinsurance Applications of WII in Northern 5.95. In this respect, insurance companies Kazakhstan could use WII structures similar to the ones designed in this feasibility study to hedge the 5.93. As discussed, WII can be retailed at drought exposure of their agricultural under- different business levels apart from the indi- writing portfolios. The analysis carried out in vidual farmer or micro level. It can be retailed this study shows that it is possible to structure at the meso level to cover the exposure of enti- drought insurance products for spring wheat ties such as financial service providers, farmer production by calibrating weather indexes on associations, input suppliers, and processors rayon-level yield records. These structures could from potential losses caused by adverse weather technically form the basis of a reinsurance trans- events or at the macro level to aid governments action. and relief agencies in development and disaster management. The use of index insurance by na- 5.96. From the reinsurance industry’s point tional or regional governments (macro level) is of view, provided that the usual assurances on discussed in chapter 6. the reliability and independence of weather measurement are available, WII contracts are 5.94. One specific meso-level application of usually considered attractive because they are WII could be the use of index contracts as re- transparent, free of asymmetric information insurance coverage for insurance companies problems, and easy to manage. If the WII con- involved in agricultural insurance programs. tracts are structured appropriately, Kazakhstan insurance companies should be able to access international reinsurance markets for WII pro- tection. 51  It would have been interesting to replicate a sim- ilar analysis for the Aktogay and Mikhailovka weather stations, but the data collected in the surroundings 5.97. In order to test the performance of re- of Aktogay are incomplete, and it was not possible insurance applications of the WII contracts to develop an acceptable WII contract structure for developed in this study, a simulation scenario Mikhailovka. The farm-level data for Mikhailovka, also presented in figure A7 of annex 7, seem to be was carried out in which hypothetical WII re- less variable than the Kostanay farm-level data. insurance payouts and actual indemnities paid 52  As an additional caveat, what may be considered by the current LIC scheme for drought were acceptable in terms of basis risk may also depend compared. Two caveats are relevant for putting on the actual farmers’ experience with agricultural this exercise in the right perspective: insurance. Farmers who have been used to being compensated on the basis of in-field loss adjustment procedures may be more sensitive to the potential • Given the remarkable difference in coverage impact of basis risk. levels, in order to provide comparable pay- 178  Kazakhstan Figure 5.9 Mismatch for the 2010 Payout for Commercial Farms Operating in the Range of 20–25 Kilometers from Kostanay Weather Station Commercial Farms 20% 10% 0% 1 2 3 4 5 6 7 8 9 10 -10% -20% -30% Source: Authors. Note: The bars represent the positive or negative deviation between the 2010 actual loss and the 2010 WII payout divided by the average 2008–10 yield of each commercial farm. The total 2010 wheat-growing area covered by the 10 commercial farms amounted to 14,520 hectares. Figure 5.10 Mismatch for the 2010 Payout for Production Enterprises Operating in the Range of 20–25 Kilometers from Kostanay Weather Station Production Enterprises 15% 10% 5% 0% 1 4 -5% 2 3 5 6 7 -10% -15% -20% -25% -30% Source: Authors. Note: The bars represent the positive or negative deviation between the 2010 actual loss and the 2010 WII payout divided by the average 2008–10 yield of each production enterprise. The total 2010 wheat-growing area covered by the seven production enterprises amounted to 104,424 hectares. Agricultural Insurance Feasibility Study  179 outs, only a fraction of the LIC area covered 5.98. Figure 5.11 provides a graphical repre- should be reinsured with WII contracts. In sentation of LIC indemnities and potential cor- this respect, the share of area covered under responding WII reinsurance payouts. Although the LIC scheme to be reinsured was set at 5 payouts and indemnities do not match perfectly, percent.53 WII structures would have provided relevant payments in the main drought-driven loss • Indemnities provided by the LIC scheme are events.55 In addition, table 5.12 shows that the influenced by the level of wheat prices, and total amount of WII hypothetical payouts across this clearly weakens the link between the the four rayons and over the five-year history impact of drought on crop production and are in a comparable range, indicating that, for LIC indemnities.54 the area and time frame examined, the reinsur- ance coverage of WII would have yielded rea- sonable results. 53  In the large majority of policies sold in the LIC scheme, coverage is roughly equivalent to 10 percent of the yield level, while the drought index structures were calibrated to cover losses below 65 percent of nity amount for any specific drought event may be the average yield level. achieved. See chapter 3 for more details. 54  When wheat price levels are high, the yield 55  In this framework it should be considered that, threshold falls, and, conversely, when wheat prices within acceptable boundaries, insurance companies are low, the yield threshold rises. Hence, depend- may be less financially sensitive than individual farms ing on the price level, a different aggregate indem- to mismatches between loss and payouts. Figure 5.11 LIC Indemnities and Hypothetical WII Reinsurance Payouts, 2006–10 Aktogayskiy - WS Aktogai 2010 2008 2006 Enbekshilderskiy - WS Stepnogorsk 2010 2008 2006 Auliekolsky - WS Kushmurun 2010 2008 2006 Altinsarin - WS Kostanai 2010 2008 2006 - 10,000,000 20,000,000 30,000,000 40,000,000 50,000,000 60,000,000 70,000,000 Indemnities paid under current LIC scheme (KZT) Hypotetical WII payouts with 5% of LIC area reinsured (KZT) Source: Authors based on FFAS data. 180  Kazakhstan 5.99. Despite the reasonable performance in cantly higher than the amount of premium col- terms of covering the drought exposure of in- lected by the LIC scheme. Despite any potential surance companies, the financial dimension of difference in the pricing methodologies adopt- the WII approach to reinsuring the LIC scheme ed (including savings generated by adopting a seems to be out of scale and outside any accept- portfolio rating approach) and any discount of- able insurance logic. Table 5.13 compares the fered by willing reinsurance partners, the order amount of premium collected in the history of of magnitude of the two variables is difficult to the LIC scheme in the selected rayons and the reconcile. hypothetical cost of WII-based reinsurance cov- erage. The cost of the WII coverage is signifi- 5.100. In conclusion, the findings of this analy- Table 5.12 Aggregate LIC Indemnities and Hypothetical WII Reinsurance Payouts KZT (millions) Indemnities paid under current Hypothetical WII payouts with Year LIC scheme 5% of LIC area reinsured 2006 4,555,320 — 2007 — — 2008 78,545,211 70,351,578 2009 13,999,618 — 2010 98,577,822 88,897,508 Total, 2006–10 195,677,971 159,249,086 Source: Authors based on FFAS data. Note: — = Not available. Table 5.13 Indicative Cost of WII Reinsurance of LIC Drought Exposure Indicative gross Area insured Cost of 2.5% premium for WII under 2.5% of LIC of LIC area Premium Weather at 65% current LIC area insured WII Rayon collected in LIC station coverage level scheme (number of reinsurance scheme (KZT) (KZT per (number of hectares) coverage hectare) hectares) (KZT) Altinsarin Kostanai 4,042 174,922 4,373 17,675,868 10,930,646 Auliekolskiy Kushmurun 2,637 161,783 4,045 10,665,544 6,252,665 Enbek- Step- 3,439 47,811 1,195 4,110,551 3,657,782 shilderskiy nogorsk Aktogayskiy Aktogai 2,678 740 19 49,543 253,448 Source: Authors based on FFSA data. Note: Given that 50% of LIC indemnities are paid by the GRK, the percentage of the LIC area to be covered by WII rein- surance can be reduced by the same proportion and set at 2.5%. Agricultural Insurance Feasibility Study  181 sis show that adopting WII reinsurance in the ed to explore the feasibility of using remote- context of the current LIC framework does not sensing technology for insurance applications make commercial sense. At the same time, they that are currently adopted or under develop- reinforce the findings of chapter 3, which show ment in insurance schemes in other countries. that LIC is severely underrated. If the current Hence research exploring the potential role of LIC system is revised and made into an actuari- the NCRST in supporting the development of ally sound crop insurance scheme, a new evalu- index insurance is recommended. ation of the use of WII contracts for reinsurance purposes might lead to different conclusions. Operational Considerations Remote-Sensing Approaches to 5.104. Whether the GRK or the insurance in- Developing Index Insurance dustry decides to consider the adoption of WII contracts in the agricultural insurance pro- 5.101. In order to find new approaches to of- gram of Kazakhstan, it is recommended that fering crop insurance in Kazakhstan, solu- additional research be conducted to investi- tions based on satellite-measured data could gate the technical and commercial specificities be also investigated. The use of remote-sensing of WII. Depending on the targeted application approaches in agricultural risk management has level, the research agenda may have different become more popular in recent years. Satellite objectives. data can be used to estimate precipitation levels and reflection indexes, to measure energy bal- 5.105. For micro-level applications, a deeper ances, and even to run crop models. The bulk of understanding of the level of spatial basis risk the current commercial applications of remote- in spring wheat production is required. Such sensing indexes measure pasture growth through an understanding would indicate the effective- reflection indexes such as the normalized differ- ence vegetation index (NDVI) and related trans- ness of WII insurance products in compensating formations. However, research on the use of sat- farm-level losses, and this information would ellite information to estimate crop yield is under help insurers to understand the level to which way and could soon provide interesting results farmers would be prepared to accept payouts (see, for example, Rosema et al. 2010). that do not exactly match the losses they experi- ence in the field. 5.102. Kazakhstan capacity in remote-sensing technology is remarkable, and specific agri- 5.106. The most effective way to learn about cultural applications have already been devel- the potential of WII products is to develop pilot oped. The entity responsible for such activities tests in which a WII supply chain is set up and is the National Center of Space Research and prototype contracts are designed and retailed. Technologies (NCSRT) of the National Space Although, as mentioned, micro-level WII is not Agency (NSA). Among its various activities, the a realistic alternative to the current agricultural NCSRT operates a satellite-based crop monitor- insurance scheme in the short term, conducting ing system for the Ministry of Agriculture. The pilot tests could provide significant insights on activities of the NCRST in this field include es- its possible uses in the future. Even if not imple- timation of spring soil water content, estimation mented at full scale, WII may still prove to be of spring crop acreage, control of cereals sowing a useful risk management tool to be adopted on date, and empirical models for crop state assess- a case-by-case basis, where the required condi- ment and grain production forecasting. tions apply.56 5.103. So far, the work carried out by the NC- SRT has mainly sought to forecast production. 56  For example, in cases of large operations or co- However, the NCSRT holds the expertise need- operative units that have good historical data on yield 182  Kazakhstan 5.107. For a meso-level application, a useful in- • Fund agro-meteorological research leading vestigation would be to carry out a larger-scale to product design analysis aimed at determining the number of rayons in which the proposed cover would be • Provide technical assistance for training and technically feasible. In fact, the analysis carried product development out in the feasibility study showed that out of the nine cases examined in the north of Kazakh- • Facilitate the development of an enabling stan, the proposed cover was acceptable in six, legal and regulatory environment not acceptable in one, and impossible to model in two. • Facilitate access to reinsurance 5.108. The results of this analysis highlight the • Support regular monitoring, evaluation, and need to generate synthetic historical data sets impact studies. for rating purposes and to install additional weather measurement devices for contract Conclusions on WII monitoring and settling. 5.111. The analysis carried out in this feasibil- 5.109. In general terms, if Kazakhstan’s stake- ity study indicates that WII for spring wheat holders decide to promote the use of WII, a in the north of Kazakhstan is technically fea- suggested action would be to set up a technical sible. unit or a working group to address the techni- cal dimension of WII program development, 5.112. Despite the positive technical findings, which individual insurance companies gen- the actual density of the weather measurement erally cannot handle. Such a working group network does not make full-scale implementa- would need to combine the expertise of agrono- tion of farm-level WII a realistic option in the mists, agro-meteorologists, and insurance ex- short term. While potential action to address perts and would greatly benefit from the partici- this constraint may be undertaken, for the time pation of the GRK. Should the coinsurance pool being WII should not be considered a readily discussed in chapter 4 be developed, it would be implementable alternative to the current LIC in- the ideal seat for a technical unit on WII. surance scheme. Potential Roles of the Government in 5.113. In addition, the high potential costs of Enabling WII WII products developed in the analysis and a preliminary analysis of basis risk suggest that 5.110. From a WII perspective, it is possible to further research is needed to assess the poten- list what the development community has come tial interest of farmers in such an alternative to consider as useful public actions that govern- insurance approach. WII could also be applied ments and international donors can carry out as reinsurance to cover the drought exposure to support the development of WII programs of insurance companies. However, the analy- (see IFAD and WFP 2010). The following list sis carried out for this study found that adopt- summarizes these actions: ing WII reinsurance in the context of the cur- rent LIC framework does not make commercial • Improve weather station infrastructure and sense. Should the current LIC system be revised data systems to become an actuarially sound crop insurance scheme, a new evaluation of the use of WII con- tracts for reinsurance purposes might lead to dif- and have access to representative weather data. ferent conclusions. Agricultural Insurance Feasibility Study  183 Chapter 6: Tailoring Crop Insurance to the Needs of Lower- Income Smaller Farmers 6.1. The final section of this report presents ommends that in the future Kazakhstan’s crop some of the international lessons and experi- insurers should aim to develop and introduce ence on strategies and programs to address the several of these alternative crop insurance agricultural risk transfer and insurance needs products. The products that are recommended of small farmers in Kazakhstan. These lessons for individual farmers are described in table 6.1. may be applicable to the small household mixed They include named-peril (hail) crop insurance, crop and livestock farming sectors, which are multiple-peril crop insurance (MPCI), loss of mainly located in southern Kazakhstan. This yield insurance, and nontraditional index prod- section also draws, where possible, on the find- ucts, such as area-yield index insurance (AYII) ings of a short visit to one wheat-growing area and weather index insurance (WII). in Tole-bi rayon in South Kazakhstan (SKO), where many farmers work either small peasant 6.4. Named-peril crop hail insurance could farms or semisubsistence household plots. be offered for cash crops, such as cotton, if there is an appreciable exposure to hail risk. See also chapter 5 for comments on crop hail Identification of Appropriate opportunities in Kazakhstan. Crop Insurance Products 6.5. Individual grower MPCI is often not 6.2. To date, crop insurers in Kazakhstan considered suitable for small and marginal have offered a single loss of investment costs farmers because of the very high costs of ad- (LIC) crop insurance product mainly to medi- ministering such policies and the associated um and large cereal producers located in north- high costs of premiums. The very high costs of ern Kazakhstan, and there has been very little administering MPCI policies are associated with debate about the appropriateness of this prod- the need to conduct preinspections in order to uct for small and marginal farmers in southern minimize antiselection and moral hazard and Kazakhstan. While the LIC product is suitable the requirement to conduct time-consuming and for grains, oilseeds, and other field row crops, costly in-field loss assessments. While econo- it is not suitable for most horticultural and tree mies of scale can be achieved on large insured fruit crops, especially crops that have multiple units with several thousand hectares of insured or staggered harvests. Furthermore, the product crop, the administrative and operating (A&O) is designed to protect against catastrophic losses costs of insuring small and marginal farmers are equivalent to more than 75 or 80 percent of ex- prohibitively high.57 In addition, the very high pected crop revenue. Although this type of cover might be of interest to large cereal producers, it is not necessarily appropriate for small subsis- tence farmers, which require food security. 57  This point was made by the CEO of Kazakh- stan’s largest crop insurance company, who noted that, because of the very high costs of administering 6.3. A wide range of crop insurance products crop insurance, his company could only insure large is available internationally, and this report rec- production enterprises. 184  Kazakhstan Table 6.1 Classification of Agricultural Crop Insurance Products Type of product Payouts Availability Indemnity-based agricultural insurance (payouts based on the actual loss at the insured unit level) Named-peril % of damage Widespread Multiple-peril loss of investment Yield loss or loss of investment Eastern Europe, Kazakhstan, costs costs Mexico Multiple-peril loss of yield Yield loss guarantee Widespread Index-based agricultural insurance (payouts based on an index measurement) Area-yield index Area-yield loss Brazil, India, United States Canada, India, Malawi, Crop weather index insurance Weather index payout scale Mexico, United States Crop revenue insurance (payouts based on yield measurement and crop prices) Crop revenue insurance Yield and price loss Limited to the United States Source: Authors. premium rates on MPCI policies, which are of- the preconditions of AYII is that farming sys- ten on the order of 7.5 to 15 percent, make this tems in the defined area (insured unit) must be cover unattractive to small farmers unless ac- fairly homogeneous, conditions that are found companied by very high government subsidies in northern Kazakhstan where rain-fed spring of premiums. wheat is the main crop, is cultivated with very stable planting and harvest dates, and produces 6.6. AYII is considered a more appropriate relatively homogeneous yields. Agricultural product for small and marginal farmers be- production in southern Kazakhstan is, however, cause of the reduced costs of administering much more diversified within each rayon, with and adjusting this type of cover. AYII does not a mix of rain-fed and irrigated agriculture and require preinspections or individual farmer and much smaller and more scattered plots of main- field loss assessment and is therefore a less cost- ly winter sown cereals, horticultural crops, and ly product for an insurer to offer to small and cotton. AYII for winter wheat may not be ideally marginal farmers. India is the main country that suited in these regions due to the much less ho- has implemented AYII on a massive scale, link- mogeneous cropping systems and mix of rain- ing it to crop credit for small and marginal farm- fed and irrigated agriculture. ers. The program currently benefits more than 25 million producers each year. The Indian AYII 6.8. Crop weather-index insurance was first program is, however, more of a socioeconomic introduced into India 10 years ago, where it support program: it incurs very high financial was hailed as a micro-insurance product ideally losses and is only sustainable because of very suited to the needs of India’s poorest and most high levels of government subsidies. marginalized farmers. WII has been promoted for small and marginal farmers in developing 6.7. The suitability of AYII for small farmers countries because it has the potential to address located in southern Kazakhstan will need care- correlated risk affordably and is less challenging ful consideration. As noted in chapter 4, one of to operate than traditional individual farmer in- Agricultural Insurance Feasibility Study  185 demnity-based crop insurance products. WII has ducer who grows a crop(s) for sale and invests achieved some degree of scale-up by the private in purchased inputs and services, often using insurance sector in India and been introduced on formal credit. Where the farmer is a commercial a large scale as a compulsory crop credit insur- producer and uses credit to purchase seeds and ance cover by the Agricultural Insurance Com- fertilizers and plant protection chemicals, the pany of India (AIC), the public sector crop in- farmer faces a financial risk in the event of ad- surer. WII is now being pilot tested in about 25 verse climatic conditions leading to crop failure, countries in Asia, Africa, and Latin America, but and risk transfer through the purchase of crop to date none of the micro-level individual farmer insurance is often justified. For large producers crop insurance programs has achieved sustain- of commercial crops, individual grower MPCI ability and scale-up. or named-peril crop insurance may be a suitable product (figure 6.1). 6.9. In southern Kazakhstan, WII is only suitable for rain-fed agriculture. Given the fact 6.11. Individual farmer crop insurance is of- that much of agriculture in southern Kazakhstan ten appropriate for commercial and semicom- is irrigated, WII policies that insure against ex- mercial farmers. In Kazakhstan, private insurers cess or deficit rainfall are not suitable. For ex- are targeting the larger commercial agribusiness ample, in Tole-bi rayon, nearly 60 percent of enterprises because of the lower unit costs of the annual sown area of 758,000 hectares is ir- dealing with these farmers and the higher vol- rigated. High-value horticultural and vegetable ume of premium generated by each bound risk. crops and cotton tend to be grown on irrigated land, while cereals are grown on nonirrigated 6.12. The insurance companies, however, land. A further factor that will need to be studied face major challenges in implementing crop carefully is the drought risk exposure. In Tole- insurance for smaller semicommercial farm- bi rayon technical staff noted that drought is not ers on an individual farmer-by-farmer basis an economic risk exposure in rain-fed crops; because of the very high costs of delivering and other factors including pests and diseases that administering insurance on small farm units. are not indexable under a WII policy are more In Kazakhstan, the main clients of the farmer important.58 mutual crop insurance associations are small to medium farmers. Given their low overhead costs and proximity to their members, mutuals Farmer Segmentation and Crop are much better placed than private companies Insurance to offer crop insurance to smaller commercial and semicommercial farmers. For these smaller 6.10. Individual farmer crop insurance is most semicommercial farmers, index insurance (both effective when the farmer is a commercial pro- AYII and WII) may be a more viable solution than traditional MPCI (figure 6.1). 58  In this respect, the feasibility study team as- 6.13. Traditional individual farmer crop in- sessed the possibility of structuring a WII contract for surance is often appropriate for commercial winter wheat production in the neighborhood of the and semicommercial farmers, but not for sub- Tasaryk weather station in the Tole-bi rayon. However, information provided by local stakeholders seemed sistence farmers. There is much evidence today to overestimate the potential impact of drought on that traditional individual farmer MPCI does not wheat production in the area. On the basis of the data work for small and marginal farmers and usual- analyzed, the lowest yield recorded (drought related, ly ends up being heavily subsidized by govern- according to wheat producers) was higher than 70 percent of the average yield reference. With a drought ment. Few small subsistence farmers producing risk of such a small magnitude, drought index insur- food crops for on-farm family consumption can ance may not offer much protection. afford crop insurance—hence government inter- 186  Kazakhstan Figure 6.1 Suitability of Crop Insurance Products for Different Types of Farmers • Medium/large farm units • Mechanized production Commercial • Access to credir MCPI Farmers • High Levels inpit use Named • Product for sale Peril Index Insurance • Smallholder farmers Semicommercial • Some assets Farmers • Some access to credit • Part consumption, part sale • Very small, no land Subsistence • Ver few assets Social Programs Farmers • Subsistence farming Source: Adapted from Skees et al. 2009. vention to make crop insurance more affordable strument that can be designed to transfer risk through premium subsidies. In Kazakhstan, crop at various levels. It can transfer risk either at the insurance is unlikely to be a useful interven- individual farmer level, which is termed micro- tion for the very small rural households that are level insurance, or at an intermediate level of mainly subsistence producers. aggregation, as financial business interruption protection for banks and other lending organi- 6.14. For subsistence farmers, it may be much zations such as cooperatives and microfinance more cost-effective for governments to exam- institutions (MFIs), which is termed meso-level ine alternative food security mechanisms and insurance. It also is an instrument that region- social safety nets or, where they elect to use in- al and national governments can use to insure surance, to consider some form of macro-level against major systemic perils such as drought, weather index program to permit early pay- windstorm, freeze, and flooding (macro-level ments in the event of a major natural disaster. insurance). To date, several countries, including Ethiopia, Malawi, and Mexico, have designed macro-level 6.16. In the past decade, WII has been heavily rainfall deficit index cover that provides national promoted as an individual farmer micro-level or regional governments with immediate cash li- insurance product. However, relatively little re- quidity following a natural disaster and enables search and development effort has been conduct- the government to provide an early response. ed to find meso-level and macro-level solutions. Micro-level WII has been heavily promoted in developing countries as a low-cost (in terms Tailoring Crop Insurance for of administrative costs) product that is ideally Different Client Levels suited to the needs of resource-poor farmers, and it is being implemented or pilot tested in about Different Levels of Aggregation 25 countries in Asia, Africa, and Latin America (see table 5.7 in chapter 5). This product is typi- 6.15. Crop index insurance (including both cally designed to insure against systemic perils AYII and WII) is potentially a very flexible in- of “too much rainfall” (excessive rainfall and Agricultural Insurance Feasibility Study  187 flooding) and “too little rainfall” (drought), and pears to be an important role for linking disas- the premium rates associated with it are typi- ter risk management with an ex ante macro- cally on the order of 7.5 to 10 percent or higher, level weather index or AYII insurance policy.60 which is generally not affordable for the targeted To date, several countries, including Ethiopia, resource-poor farmers without government pre- Malawi, and Mexico, have designed macro-level mium subsidy support. The scale-up of private rainfall deficit index covers to provide national sector WII in India has been achieved in part or regional governments with immediate cash li- because of premium subsidies, while the public quidity following a natural disaster and to enable sector WII program has sold several million pol- the government to provide an early response. icies because it is linked on a compulsory basis Mexico is the only country that, to date, has used to seasonal credit. No other country that is op- AYII as a macro- or state-level crop insurance erating voluntary micro-level WII has achieved product for small and subsistence farmers (see commercial scale-up to date.59 below for further discussion). There appears to be considerable scope for using macro-level in- 6.17. Crop index insurance can also be used as dex products to provide small subsistence farm- a meso-level instrument to protect rural bank ers with a social safety net; commercial crop lending (loan portfolio or business interrup- insurance is not necessarily an appropriate or tion protection). From the bank’s perspective, cost-effective mechanism for them. farmers who have crop insurance protection are less likely to default on their loans in the event Key Organizational and Operational of a major weather-induced crop failure. It also Differences between Micro-, Meso-, and means that, in the event of a major regional loss, Macro-Level Insurance the bank’s loan portfolio is protected, enabling the bank to remain solvent, to reschedule farm- 6.19. There are some key structural and op- ers’ loans, and to continue lending. Claiming on erational differences between a micro-level in- a crop insurance policy and rescheduling loans dividual farmer crop insurance scheme, a me- are generally much more acceptable to a bank so-level crop insurance scheme purchased by a than having to resort to the courts to recover financial lending institution, and a macro-level their debts. Since 2010, meso-level WII pilot ini- regional or countrywide program purchased tiatives have been launched in Peru and Ghana, by a state or national government. Under a linked to regional bank lending to agriculture. micro-level program, the individual insured is The Peruvian product is a catastrophic El Niño– the farmer who pays a premium (which may or induced excessive rain leading to flood policy may not receive government subsidy support) that uses a sea surface temperature index to trig- and receives a claims payment in the event of a ger indemnities, while the Ghanaian product for loss. Where micro-level crop insurance is linked rural banks is linked to deficit or excessive rain- to crop credit, the lending institution is usually fall (Skees and Murphy 2009; Stutley 2010). In listed as the first beneficiary in order to ensure addition, in Vietnam, an innovative flood index that any claims settlement is used to repay the product linked to regional bank lending to rice amount of loan; the farmer then receives any farmers is awaiting launch (Skees and Hartell balance on the claim (figure 6.2). 2008). 6.20. Under a meso-level crop insurance pro- 6.18. At a regional or national level, there ap- gram, a financial lending institution would 59  For an up-to-date review of the current status of 60  For a comprehensive review of the links be- crop and livestock WII projects around the world, see tween disaster risk reduction and index insurance, WFP and IFAD (2010). see Warner et al. (2009). 188  Kazakhstan Figure 6.2 Organizational Structures for Micro- and Meso- or Macro-Level Crop Insurance Micro level insurance program Meso/Macro insurance program Insurer Insurer Policies, premiums, Policies, premiums, claims claims Policycholder is Distributor Aggregator (e.g. processor, bank) Policies, premiums, Aggregator sets the claims payout rules Policyholder is Farmer Farmers Source: Dick 2009. typically be the insured policyholder and would with the proposed area-yield index or WII prod- purchase crop insurance to protect its crop loan uct than individual farmers. portfolio against catastrophic climatic losses. Under the meso-level option, the lending insti- 6.21. Under the operation of a macro-level tution (for example, a rural bank or MFI) would policy, the insured policyholder is typically the declare its total agricultural loan acreage in each state (regional) government or national gov- region (municipality) and the level of cover- ernment. Under the Ethiopian and Malawian age it wishes to insure under the crop insurance macro-level WII rainfall deficit products, the policy in each municipality. The bank or MFI national governments are the policyholders, and lending to agriculture would purchase a single the covers are designed to trigger ex ante disas- aggregate policy to protect its loans and would ter relief payments to the national government. be responsible for paying the premium. In the In Mexico since 2003, the state-level govern- event of a claim in any or all insured regions, the ments have purchased state-level catastrophe losses would be computed and the bank or MFI drought WII cover and state-level AYII catastro- would receive an aggregate claims indemnity. phe insurance (seguros catastróficos) protection The bank or MFI would then decide whether or for specific target audiences of small subsistence not to pass on any of this indemnity payment to farmers located in specific regions and growing the individual farmers to whom it lends seasonal specific crops. These farmers are too small to crop production credit. If the bank or MFI elects participate in the private commercial crop in- to pass on some of the indemnities to farmers, it surance programs or public sector small farmer may also require them to share in the premium group loan and crop insurance programs. The costs. The major advantage from an insurance state governments are the policyholders, and viewpoint is that the bank or MFI would be able they pay 100 percent of the costs of premiums. to accept a higher degree of basis risk associated In the event of a claim, the state governments Agricultural Insurance Feasibility Study  189 Box 6.1 Macro-Level Catastrophic Climate Contingency Index-Based Insurance for State Governments in Mexico Objectives. To provide disaster relief assistance to small and marginal farmers affected by natural calamities Legal framework. The National Program for the Attention of Climatic Contingencies (PACC in Spanish) managed by the Ministry of Agriculture Main features. To provide disaster relief assistance to the farmers from a fund that is financed through the payouts generated by a macro-level agricultural insurance policy issued to the federal and state governments Target farmers. Small and marginal farmers (according to the definition provided by the Ministry of Agriculture) who suffer from the occurrence of nonrecurrent, unforeseen, and unpredictable climatic losses and have no access to any other type of financial insurance Covered perils. Drought, frost, hail, excessive rain, flooding, cyclones and tropical storms, and tornados Covered items. Rice, maize, beans, sorghum, barley, oats, wheat, rape, soybeans, peanuts, sugarcane, avocados, cacao, citrus, coco, passion fruit, coffee, apples, mangoes, nuts, pineapples, peaches, and plantains Mechanism of operation. Two types of assistance are offered: catastrophe insurance and direct support. • Catastrophe insurance. Federal and state government liabilities from the application of the disaster relief assistance to the farmers are transferred to the market through a macro-level in- dex-based insurance policy. The policy can be based on a weather index, an AYII, or an NDVI. The municipalities (rayons), the insured crops, and covered perils are selected by the federal and provincial governments ex ante of the inception of coverage and payment of the insur- ance premiums. The premiums are co-financed by the federal (75–90 percent) and the state (25–10 percent) governments. Assistance is equivalent to US$75 per hectare up to a maximum of US$1,500 for annual rain-fed crop (20 hectares) and US$750 per hectare for other crops (10 hectares). The waiting period to receive an indemnity is one to two months. The program is insured by commercial reinsurers and in the international market. • Direct support. Direct support is provided by the federal and state government when cata- strophic insurance is not available. The occurrence of the loss event is determined by the Na- tional Commission of Water (Conagua in Spanish). The direct support is co-financed in equal shares by the federal and state governments. Assistance is equivalent to US$75 per hectare up to a maximum of US$375 per hectare per farmer (5 hectares maximum). The waiting period to receive an indemnity is three to five months. Achievements. • 8 million hectares are insured. • 4.2 million head of livestock are insured. • 3.2 million farmers are protected by the program. • 30 of 38 states participate in the program. Source: Authors from SAGARPA. 190  Kazakhstan have a preagreed indemnity payment formula was sold to the state government, which was for distributing lump-sum indemnity payments then responsible for setting the indemnity rules from insurers and reinsurers to the target ben- for each farmer in the insured command area. eficiaries. See below for further discussion of Since 2003, the program has been massively Mexico’s macro-level crop index insurance pro- scaled up, with the development of index prod- grams. ucts based on (a) WII covers against drought, hurricane, and frost, (b) AYII covers providing Macro-Level Crop Index Insurance in all-risk loss of yield protection at a macro level, Mexico and (c) normalized difference vegetation index (NDVI) pasture drought cover for livestock pro- 6.22. In 2003 Agroasemex, the Mexican spe- ducers. Currently 30 of the 38 state governments cialist parastatal crop reinsurer, launched the in Mexico purchase climate contingency protec- first macro-level catastrophe drought crop in- tion. For crops, 8 million hectares are insured, surance index insurance program. Mexico is with 3.2 million small subsistence farmers pro- very exposed to catastrophic risk in agriculture, tected under the crop insurance programs; about including drought (80 percent of natural catas- 4.4 million head of livestock are insured under trophes), hurricanes (17 percent), excessive rain the NDVI pasture drought index program. Fur- or flooding (2 percent), and frost (1 percent). ther information on the Mexican macro-level in- Since 1995 the federal and state governments dex programs is presented in box 6.1. have operated a national natural disaster scheme under the FONDEN program, which is designed A Potential Disaster Relief Contingency to provide financial compensation to small rural Fund for WKO and Aktobe farming families who are not eligible for private crop and livestock insurance. Between 1995 and 6.24. Given the significant financial chal- 2003, the federal government and state govern- lenges of the current LIC insurance scheme in ments paid out US$212 million and US$74 mil- WKO and Aktobe, an alternative way for the lion, respectively, to small rural farmers under GRK to provide spring wheat producers with di- the program. In 2003 as part of the FAPRAC saster relief assistance could be to set up a pub- (Fund for the Care of Rural Population Affected licly supported contingency fund that would be by Weather Contingencies), government con- used to compensate farmers in case of extreme tracted Agroasemex to substitute the ex post di- loss events. Drawing on the recent international saster compensation programs with an ex ante experience—Mexico in particular—such a fund macro-level index insurance for catastrophic cli- could be ideally protected through a specific matic perils (Agroasemex 2007). reinsurance coverage that, on the basis of cus- tomized weather indexes, would trigger payouts 6.23. Agroasemex launched the first pilot in the event of adverse weather conditions and, in 2003 as a drought insurance index cover therefore, replenish the fund. Contingency fund for maize and sorghum grown in the state of resources would be also used to pay for the cost Guanajuato. The first product was based on a of the reinsurance coverage. drought index for five weather stations cover- ing an insured area of 75,000 hectares of rain- 6.25. The reference weather indexes could be fed maize and sorghum produced by several developed at the rayon level. This would make it thousand small subsistence farmers located in possible to have payouts triggered independent- Guanajuato. In recognition of the difficulties for ly by weather measurement carried out in each insurance companies to provide insurance cover rayon or at the oblast level, with an index based to individual farmers, Agroasemex specifically on a basket of stations that would trigger a pay- designed an aggregate or macro-level policy that out for the entire oblast (hence less frequently). Agricultural Insurance Feasibility Study  191 Table 6.2 Average Premium per Policy, by Size of Insured Farm KZT per hectare Farm insured area (hectares) Premium 1 100 500 1,000 2,500 10,000 Premium 80 8,000 40,000 80,000 200,000 800,000 (KZT) Premium 0.53 53 267 533 1,333 5,333 (US$) Source: Authors based on average premium rate of the obligatory crop insurance scheme for 2005–10. Box 6.2 Small Farmer Production Cooperatives in Southern Kazakhstan In Tole-Bi rayon, small farmers in one producer cooperative have devised their own solutions to crop production and purchase of crop insurance. The Berli Producers Cooperative in Tole-bi rayon was formed in 1993. It has 150 members, most of whom are active farmers owning a total of 2,100 hectares; of these, 600 hectares are cultivated with winter wheat. The average size of farm is about 5–10 hectares per member. Each farm is too small to produce wheat on its own. The cooperative has therefore formed a joint-stock company that rents the land from all the cooperative members and then farms it on a collective basis on behalf of the members, creating economies of scale in cultivating the land as a single aggregate farm. The members also receive a share of the profits at the end of the season, following the harvest and sale of the crops. The cooperative obtains obligatory LIC crop insurance each year through one of the farmer mutual crop insurance associations on the total area cultivated with wheat: for the mutual insurer, there are major advantages in dealing with a single entity rather than trying to insure 150 smallholder farmers, each with 5–10 hectares. Source: Authors. 6.26. A critical element to be assessed would be Organizational and Operational related to the actual cost of reinsuring such a Systems for Small Farmer Crop contingency fund. Given the very high drought exposure in Aktobe and WKO, reinsurance pay- Insurance outs would need to be frequent and therefore ex- pensive. Low-Cost Delivery Models 6.27. Depending on the estimated overall 6.28. Insurance companies throughout the financial exposure of this protection scheme world face major challenges in trying to iden- for WKO and Aktobe (see chapter 4) and the tify cost-effective ways to deliver and adminis- projected cost of dedicated reinsurance cover- ter agricultural crop and livestock insurance age, the GRK should evaluate whether it would programs for small farmers. In Kazakhstan this be more cost-effective to purchase reinsur- problem is accentuated by the very low sums in- ance on the international markets or to guar- sured adopted under the obligatory crop insur- antee replenishment of the fund with its own ance scheme. The average sum insured over the resources. past five years was only KZT 3,287 (US$22) per 192  Kazakhstan Box 6.3 Delivery Models for Small Farmer Insurance (Micro-Insurance) Full-service model. Commercial or public insurers provide the full range of insurance services, from initial development of the product, through distribution, to absorption of risk. Partner-agent model. Commercial or public insurers, together with MFIs or nongovernmental organizations (NGOs), collaboratively develop the product. The insurer absorbs the risk, and the agent markets the product through its established distribution network. This lowers the cost of distribution and promotes affordability. Community-based model. Local communities, MFIs, nongovernmental organizations, and cooperatives develop and distribute the product, manage the risk pool, and absorb the risk. As with insurance mutuals, there is no involvement on the part of commercial insurers. Provider model. Banks and other providers of microfinance can directly offer or require insurance contracts. These contracts are usually coupled with credit, for example, to insure against the risk of default. Sources: ProVention 2006; Cohen and McCord 2003 hectare, and the average premium rate was only to the cooperative management to farm on a 2.42 percent, generating an average premium joint-stock basis, and this brings economies of of about KZT 80 (US$0.53) per insured hectare scale in purchasing crop insurance. Features (table 6.2). Insurance companies need to reserve of the joint-stock approach are described in box premium to pay for normal expected claims, 6.2 for Berli Producers Cooperative. The land but also to pay for business acquisition costs, belonging to the 150 members is farmed as a sales agent fees, and their own A&O expenses. single unit, and the farm managers purchase a In Kazakhstan, sales agent fees are typically 10 single crop insurance policy for the cooperative. percent of original premium, and the insurer’s The joint-stock approach to farming is apparent- O&A expenses may be a further 10–15 percent ly adopted by between 50 and 60 percent of all of premium. Table 6.2 shows that, on very small the farmer cooperatives in SKO. The extent to farm units of up to 100 hectares, the total av- which this joint-stock approach to overcoming erage premium earnings generated by a single the scale problems for small farmer crop pro- policy are only about 53 hectares on average, duction and to contracting crop insurance could which is far too low to cover the fees of a sales be replicated in other parts of Kazakhstan is not agent, let alone enable a commercial insurer to known. cover its A&O costs. This problem also applies on small and medium commercial farms from 6.31. Insurance companies can deliver in- 100 to 1,000 hectares, where the total premi- surance products and services to small rural um of about KTZ 80,000 (US$533) per policy households and farmers in several ways. These would not be adequate to service the account on are listed in box 6.3. an individual farmer basis. Insurance companies therefore need to seek alternative, less costly Full-Service Model ways of marketing and administering their poli- cies to this group of farmers. 6.32. The traditional method—termed the “full-service” model—in which the insurance 6.30. In southern Kazakhstan, many of the company assumes full responsibility for all in- small household farmers are organized into surance functions would not be cost-effective production cooperatives and rent out their land for insurers of small farmers in Kazakhstan. Agricultural Insurance Feasibility Study  193 Figure 6.3 Partner-Agent Delivery Model for Crop (and Livestock) Micro-Level Insurance Partner-agent model for micro-level insurance Provide product, Develop claims, & product Mainstream product NGO/MFIs education Potencial insurance as insurance micro- Underwriting companies as agent insurance partner Premium clients Source: Al Hasan 2007 Under the full-service model, the insurer bears tion network and a large membership of farm- the full costs of delivering insurance to farmers, ers. Under a partner-agent model, the insurance including insurance awareness and education company enters into a formal contractual agree- and policy sales and marketing (through either ment with the agent, in which the agent assumes its own network of sales agents or commission responsibility for marketing and promoting the brokers), and relies predominantly on sales to insurer’s policies to its members, for collecting individual clients. Premiums are collected in- premiums from the insured, and paying these dividually by the company for the insured, and over to the insurer, for notifying claims to the in- claims notification and settlement are managed surer, and finally, in some cases, for distributing directly by the insurer. This model is typically claims settlement payments to the insured (fig- adopted by the private commercial crop insurers ure 6.3). Usually, the insurer agrees to pay the in Kazakhstan today under the obligatory crop agent a commission for its services. This model insurance scheme. With average crop insurance would potentially enable the private commercial premiums of about KZT 80 per hectare, this insurers in Kazakhstan to deliver crop insur- model is only feasible for large agribusiness en- ance more cost-effectively to large numbers of terprises with several thousand insured hectares. small and medium farmers. Such a model could It is not cost-effective for commercial compa- also be used to deliver livestock insurance to the nies seeking to serve smallholder agricultural small mixed cropping and livestock farmers lo- insurance in Kazakhstan. cated predominantly in southern Kazakhstan. Partner-Agent Model 6.34. The partner-agent approach has been successfully promoted for smallholder agricul- 6.33. In Kazakhstan, there may be consider- tural insurance in recent years in Africa, Asia, able potential for commercial insurers to en- and Latin America and offers a potential win- ter into a “partner-agent” relationship with win for both parties. For the insurance company, rural organizations (for example, cooperatives the distribution of its products through a rural or MFIs) that have an existing rural distribu- institution (agent) offers the potential to reach 194  Kazakhstan Box 6.4 Livestock Insurance Fund in Bangladesh Scope. The Livestock Insurance Fund, which is a component of the Community Livestock Development Program, compensates the owners of dairy cattle against mortality of their cows. Livestock mortality insurance is compulsory for dairy farmers who purchase cows or heifers on credit using microloans from the program. Insured animals include heifers, dairy cows, and beef cattle, but more than 70 percent of insured animals are dairy cows. The fund operates mainly in northwestern Bangladesh. Features. The LIF is a community-based program that covers animal mortality due to disease, accident, and any cause outside the control of the owner. Insurance is provided as part of an integrated package that includes credit, technical assistance, vaccines and veterinary services, concentrate feeds and fodder, and milk marketing services. The guarantee amount (sum insured) is the loan amount or replacement cost. The premium rate is 3 percent (previously 2.5 percent) of the loan money deducted at source. A service fee of 2.5 percent of the value of loan is charged to the Livestock Development Fund in order to contribute toward veterinary services (animal inspections, vaccinations) and to cover salaries of veterinary staff. Results. Between 2001 and 2005, 4,250 animals were insured, with a mortality rate of 3.8 percent and an associated loss ratio of 75 percent. Between 2006 and 008, 7,015 animals were insured, with a mortality rate of 1.1 percent and an overall loss ratio of about 45 percent. Key challenges. The Grameen livestock mortality product is not recognized under the Insurance Act 1938. It is not reinsured and is exposed to catastrophe claims (flood, cyclone, epidemic disease). Source: Authors based on information provided by Grameen Bank, March 2009. a large number of small farmers at low cost; for been heavily promoted by the Ministry of Agri- the rural institution, the agreement enables it to culture (MoA) in Kazakhstan since 2008 to un- expand the range of products and services of- derwrite the obligatory crop insurance scheme. fered to its membership and, where the organi- This model is potentially very relevant to devel- zation is involved in the provision of agricultural oping smallholder agricultural crop and livestock credit, the potential to protect its loan portfolio insurance, but only if mechanisms for providing with crop and livestock insurance. catastrophe reinsurance can be designed. This theme is explored further below using the Mexi- Community-Based Insurance Model can Agroasemex Fondos program as a possible model that could be adapted for Kazakhstan. 6.35. The community-based model includes all forms of informal agricultural insurance that Provider Model for Micro-Level Insurance is underwritten by local communities, MFIs, and NGOs. This model has proved very popular 6.37. In many parts of Asia, microfinance in- in southern Asia (Bangladesh, India, and Nepal), stitutions provide their micro-lenders with mi- where NGOs and MFIs have actively promoted cro-insurance products and services, which are livestock credit insurance for members over normally linked to credit. In Bangladesh, India, the past 20 years. This model is very similar and Nepal, MFIs are the main source of loans for to the provider model identified by ProVention resource-poor urban and rural households. Many (2006). of the MFIs have also introduced their own inter- nal insurance funds, providing limited life insur- 6.36. Farmer mutual insurance associations ance cover, health cover, maternity cover, and, in are a specific type of community-based private some cases, crop and livestock insurance. Most agricultural insurance organization and have of these internal micro-insurance programs are Agricultural Insurance Feasibility Study  195 linked to credit on a compulsory basis—in other tion; it is therefore very exposed to catastrophic words, members who borrow microfinance are losses (for example, flooding and epidemic dis- required to purchase life insurance cover for the eases). See box 6.4. for further details of the duration of the loan. Farmers who borrow live- GMPF livestock-credit insurance scheme. stock investment loans to purchase milk cows or buffalos are also required to insure the animals 6.39. In southern Kazakhstan, MFIs are very against death due to named perils or under all- active in lending to small urban and rural risk individual animal insurance covers. In lim- households, and there may be opportunities ited cases, the MFIs also offer crop insurance. to promote crop and livestock credit insurance through the MFIs. As of 2011, there are 1,769 6.38. Since 1999 in Bangladesh, the Grameen registered MFIs in Kazakhstan, of which 711 are Fisheries and Livestock Foundation (GMFP), actively lending to individuals and legal entities. which is part of the Grameen Bank (one of the The total loan portfolio of the MFIs amounts oldest MFIs in the world), has operated a live- to almost KZT 15 billion in personal loans to stock mortality compensation scheme, termed 85,379 persons and *KZT 425 million in legal the Livestock Insurance Fund (LIF).61 The LIF loans to 110 legal entitles. Of the total of 711 ac- program insures against death of the dairy cow tive MFIs, 570 (80 percent) are classified as ur- where this is “outside the control of the owner” ban and 141 (20 percent) are classified as rural, and, in effect, is an all-risks livestock mortality and their main clients are small farmers. It is not policy. Insurance is provided as part of an inte- known what proportion of the total loan portfo- grated package under which Grameen veterinary lio of MFIs is to farmers. There may, however, and extension staff assist in the preinspection of be potential to examine the role of protecting the the dairy cow or heifer and certify its health sta- MFI loans to small farmers and marginal crop tus. The animal is then routinely inspected and and possibly livestock producers through credit vaccinated by Grameen-trained veterinary staff, insurance covers similar to those developed by who also verify the cause of loss in the event of MFIs in southern Asia. death. These measures lead to greatly reduced livestock mortality rates and the ability to levy Mutual Agricultural Insurance as a very low premium rates for individual animal Solution for Small and Marginal Farmers mortality cover. The sum insured is equivalent in Kazakhstan to the amount of loan taken out to purchase the cow, and the current premium is charged at a rate 6.40. The farmer mutual insurance associa- of 3 percent of the value of the loan. Coverage tions have been heavily promoted by the GRK terminates once the loan has been repaid (usu- since 2008, and if their financial status could ally over a maximum of two years). In addition, be strengthened, they might be the ideal vehicle a fee of 2.5 percent of the value of the loan is for underwriting Kazakhstan’s small and mar- levied to cover the cost of veterinary services, ginal crop and livestock producers. As shown vaccinations, and technical assistance. The pro- in chapters 3 and 4, about 38 farmer mutual gram has operated for eight full years, during crop insurance associations are underwriting which slightly more than 7,000 dairy cows have the obligatory LIC crop insurance scheme. Cur- been insured, with an average mortality rate of rently, these mutuals have very limited finan- 2.8 percent. The LIF liability is totally retained cial reserves, and none is formally protected by within the GMPF, and the program does not car- any form of insurance or reinsurance. The in- ry any form of catastrophe reinsurance protec- dividual mutuals are therefore very exposed to catastrophic losses that exceed their reserves. In the event that claims exceed their reserves, as 61  See http://www.grameen-info.org/grameen/ happened in 2010, the mutuals have to pro rata GrameenMotsho/index.html. down each claim settlement made to their mem- 196  Kazakhstan bers who incur losses. International experience vides technical assistance and training on the shows that, when catastrophe claims occur that design and rating of crop and livestock insur- cannot be paid by the mutual, this often leads to ance policies and on loss assessment procedures. the collapse of the mutual. Agroasemex closely monitors the activities of the SIFs on a seasonal basis. 6.41. If the mutuals are to remain solvent and to underwrite crop or livestock insurance for 6.44. The SIF program has proved extremely small and marginal farmers in Kazakhstan, popular with farmers in Mexico. In 2005 there ways of providing some form of catastrophe re- were 176 functioning SIFs in 24 Mexican states, insurance protection must be developed. In the of which 159 were crop SIFs and 17 were live- short term, it is unlikely that the private com- stock SIFs. Table 6.3 shows that, in 2007 the SIFs mercial insurance sector in Kazakhstan or inter- insured more than 1 million hectares of crops and national reinsurers would be willing to provide more than 4 million head of livestock (mainly excess of loss reinsurance protection to the mu- cattle and pigs), generating Mex$647 million of tuals. It is likely that such a program would have premium (US$60 million; Agroasemex 2008). to be offered through the public sector. Given For crops, the policy is an individual grower the experience of the Fund for Financial Support MPCI cover that insures against either loss of for Agriculture (FFSA) with administering the the costs invested in the crop or loss of yield; for financial claims subsidies on the obligatory LIC livestock, the policy is a herd-based catastrophe program, the FFSA is well placed to adminis- mortality and disease cover that carries very low ter some form of excess of loss program for the average rates. The crop and livestock products mutual crop insurance assoc0iations in Kazakh- underwritten by the SIFs are eligible for federal stan. government premium subsidies, which average about 33 percent of the full premium. After 18 6.42. Mexico has many years of experience years of operation, the Mexican SIF program is with the operation of small farmer mutual crop a major agricultural insurance program for small and livestock insurance through the “fondos de and marginal farmers. aseguramiento” (self-insurance funds, SIFs) program, which is reinsured by Agroasemex, 6.45. A key feature of the Mexican SIFs is the the national agricultural reinsurance company. stop-loss reinsurance protection provided by The SIFs are legally registered, small-scale crop Agroasemex to each SIF. Under the agreement and livestock producer mutual companies whose between Agroasemex and the SIFs, Agroasemex primary function is to access group crop and is responsible for setting the premium rates for livestock credit. The SIF program was originally each crop and livestock program and for provid- conceived as a vehicle to provide small and mar- ing stop-loss reinsurance protection. The SIF is ginal farmers with access to credit; individually entitled to deduct 25 percent of the original pre- these farmers were too small to be eligible for mium to cover its A&O expenses. The SIF then credit, but collectively they could access group retains an average 70 percent of the premium net credit. As part of the agreement with the lending of A&O expenses, equivalent to 52.5 percent of banks, the provision of crop and livestock credit gross premium, to settle retained claims. The re- is linked on a mandatory basis to crop and live- maining average 22.5 percent of gross premium stock insurance. is paid to Agroasemex as stop-loss reinsurance premium (figure 6.4). 6.43. Since 1990, Agroasemex has provided advisory support and training to SIF members 6.46. In the event of a claim, the individual SIF to form and register SIFs. Agroasemex also as- is responsible for settling claims up to a loss sists the SIFs in accessing short- and medium- ratio of 52.5 percent plus any claims reserves term production and investment credit and pro- held over from previous years. Any claim in ex- Agricultural Insurance Feasibility Study  197 Table 6.3 Coverage by Self-Insurance Funds in Mexico, 2007 Insured area Total Premium (thousands of Total sum Premium premium Average subsidy Item hectares) or number insured (Mex$ subsidy (Mex$ rate (%) (Mex$ of insured animals millions) (%) millions) millions) (thousands) Crop 1,034.3 8,927.8 588.2 6.6 194.9 33 Livestock 4,106.6 15,154.9 59.0 0.4 19.0 32 Total n.a. 24,082.7 647.2 2.7 213.9 33 Source: Agroasemex 2008. Note: n.a. = Not applicable. Figure 6.4 Stop-Loss Reinsurance of Self-Insured Funds in Mexico Allocation of original gross 100% Agroasemex stop-reinsurance (22.5 % of gross) 77.50% Allocation of underwriting surplus SIF retained Special claims reserve, catastrophic losses 30% (52.5% of gross) Surplus for the Social Fund 70% 25% SIF administrative (25% of gross) Source: Authors. cess of this level is reinsured by Agroasemex. In to invest in income-generating activities or to turn, Agroasemex purchases stop-loss retroces- contribute toward crop and livestock insurance sion protection on the SIF program from inter- premiums. national reinsurers. In any underwriting year, if the SIF generates an underwriting surplus (prof- 6.47. Many of the features of the Mexican SIF it), 30 percent of the surplus is added to a special program and the lessons learned over the past claims reserve for catastrophic events and the re- 20 years are relevant to the future expansion of maining 70 percent is allocated to a special fund, mutual agricultural insurance in Kazakhstan. which may be divided among the SIF members One of the key differences is that Kazakhstan 198  Kazakhstan does not have a public sector organization like tural production loans or inputs, many farmers Agroasemex to provide a combination of techni- will lack both the capital to pay the insurance cal assistance and, most important, reinsurance premium and sufficient incentive to use scarce protection to the mutual insurers. This is a major resources on risk management. Placing these issue that will need to be addressed in the pro- products within complementary systems that posal to introduce a more market-based agricul- have broader linkages can also facilitate simpler tural insurance system based on a public-private contract design, as other mechanisms can deal partnership. One option for the GRK to consider more efficiently with the uninsurable risks. is whether to use the FFSA as its implementing agency to provide technical support and, most 6.50. Experience shows that bundling agri- important, formal excess of loss compensation cultural insurance with the provision of rural protection to the mutuals along lines similar to credit and input supplies could offer major ad- those of the Agroasemex model. As a starting vantages. The bundling of crop insurance with point to introducing such a program in Kazakh- credit and input supplies has been shown to pro- stan, the mutuals would need to be effectively vide a win-win situation for farmers, lending in- regulated and would need to adopt actuarially stitutions, and insurers alike. The farmer gains determined rating and standardized loss assess- access to seasonal crop credit, lending institu- ment systems and procedures. tions are more willing to lend to small farmers because their loans are protected by crop insur- ance, and the insurer benefits from (a) reduced Identification of Operational antiselection, which in turn reduces the need for Linkages to Bundle Programs preinspections; (b) lower costs of marketing crop insurance; and (c) better insurance uptake and 6.48. Agricultural insurance on its own is spread of risk than would normally be achieved not a solution. Agricultural insurance can help under a purely voluntary program. Malawi has a to stabilize agricultural production and farm program of bundled WII, credit and input sup- incomes in times of major production loss and ply, and output marketing that is showing early also to modernize agriculture through its abil- promise. ity to leverage access to credit, enabling farmers to purchase production-enhancing technology. 6.51. In many parts of the world, public or However, agricultural insurance cannot be ef- private sector provision of credit to agriculture fective if it is provided in isolation. It should be is protected by a compulsory crop insurance promoted only when other essential agricultural policy (crop credit insurance). The 2009 World services, including training and extension, the Bank survey of agricultural insurance provision timely availability of inputs (seeds, fertilizers, identified that, in 11 percent of the countries sur- and pesticides), and efficient marketing chan- veyed, public or private sector credit to agricul- nels for agricultural outputs, are in place (Mahul ture is protected by compulsory insurance cover and Stutley 2010). (Mahul and Stutley 2010). Examples of com- pulsory crop credit or livestock credit insurance 6.49. Agricultural insurance is only one tool schemes in Asia are found in India, the Philip- for mitigating the risks of agricultural produc- pines, Nepal, and Bangladesh. From an insurer’s tion, finance, and supply chain relationships. viewpoint, there are major advantages to auto- Therefore, other measures and complementary matic or compulsory crop credit insurance: (a) investments are needed to ensure that risk is antiselection is reduced, (b) there is less need for comprehensively managed and the value of in- preinspections, (c) the costs of promoting and surance is realized. In addition, without linking marketing the agricultural insurance program these insurance programs explicitly to finance, are reduced, and (d) the insurance uptake and such as bundling the insurance with agricul- spread of risk and premium volume are gener- Agricultural Insurance Feasibility Study  199 ally much higher than under a purely voluntary nonrepayment is protected. program. 6.55. In Kazakhstan, one of the key objectives 6.52. There are advantages to making a of the Law on Compulsory Crop Insurance was scheme involving small farmers compulsory to link crop insurance with the supply of credit, rather than voluntary, unless other circum- subsidized seeds, fertilizers, and other govern- stances allow the insurer to avoid adverse ment support programs. The obligatory ap- selection and high administrative costs. Even proach to crop insurance is, however, somewhat with a compulsory scheme, worthwhile incen- different from the voluntary bundled approach. tives must be built in to counter moral hazard. It is not possible to report on whether the obliga- Operating an insurance scheme together with a tory scheme in Kazakhstan has been successful credit program can offer the level of control that in improving farmers’ access to seasonal crop insurers require, reflecting the common interests production credit and to the subsidized govern- of banks and insurers: a farmer who does not ment support programs. take out insurance will not be eligible for a loan. Loan applicants also normally go through an Conclusions initial appraisal procedure that assists the lend- er in evaluating the management potential of 6.56. No single crop insurance product is best the farmer. suited to the needs of individual small and mar- ginal farmers in southern Kazakhstan. If crop 6.53. Where agricultural credit and insurance insurance is to be developed for this category are linked, there is a potential for the bank or of farmers, it will be necessary to study the risk MFI to lower its interest rates to the extent that exposures on a crop-by-crop and a location-by- climatic or natural risk exposures have been location basis and to select the most suitable transferred to the insurance policy. In Ma- product for that situation. For example, crop hail lawi’s weather-based crop insurance program insurance may be suitable for small cotton pro- and Mongolia’s livestock index-based insurance program, lending banks have reduced their in- ducers, while AYII might work best for rain-fed terest rates to producers who agree to purchase wheat producers. drought index insurance. 6.57. For subsistence farmers in southern 6.54. Wherever possible, agricultural insur- Kazakhstan, there may be potential to design ance should be led by demand, and the ideal macro-level (per rayon or oblast) catastrophe situation is for farmers and service providers index products along lines similar to those of to reach a voluntary agreement to bundle in- the Mexican FAPRAC program. Such pro- put supply, credit, and agricultural (crop) in- grams could be developed both for crops and for surance. The bundled approach is much more livestock. acceptable to farmers than compulsory linkage of credit provision and insurance and offers a 6.58. A more detailed feasibility study is need- potential win-win situation for all parties. The ed of the smallholder farming sector in southern farmer has timely and easy access to inputs of Kazakhstan. Such a study would provide guide- seeds, fertilizers, and credit, while the input sup- lines and recommendations for the future devel- plier’s and credit provider’s financial exposure opment of appropriate crop and livestock risk to climatic-induced crop failure and potential transfer solutions. 200  Kazakhstan Chapter 7: Fiscal Implication of Various Insurance Products on the GKR Budget 7.1. This Chapter briefly looks at the fiscal im- (numbers of farmers and insured crop area per plication of various crop insurance products Rayon and Oblast) as actually achieved under on the GKR budget, in particular, (a) the cur- the obligatory crop insurance scheme from rent LIC system (As if analysis of replacing the 2005 to 2010. The financial analyses in Tables current government participation on 50% of the 7.5 and 7.6 assume a voluntary scheme with dif- total liabilities with 50% premium subsidies and ferent levels of uptake varying from 10% to 50% providing an excess of loss protection in excess of the actual obligatory scheme levels. 100% of gross net premium income); (b) Spring Wheat MPCI program (Estimated Total Sum 7.3. With the assumption that under a fully Insured, Indicative Commercial Premium and market-based scheme, farmers will be permit- Fiscal Implications for different mechanisms ted to purchase crop insurance on a voluntary of Government Support, by Level of Coverage; basis, some estimates have been made in this and (c) Spring Wheat AYII Program (Estimat- report of the possible voluntary uptake rates ed Total Sum Insured, Indicative Commercial for spring wheat over a five year period rising Premium and Fiscal Implications for different from 10% of the current obligatory insurance mechanisms of Government Support, by Level scheme uptake levels in year 1 to 50% uptake of Coverage)62. level by year 5. The results of this analysis are presented for the new voluntary individual 7.2. The figures presented in Tables 7.1. to 7.4 grower spring-wheat MPCI program with 40% assume the same insurance penetration rates insured yield coverage level (Table 7.5) and for the new spring wheat AYII program with 50% insured yield coverage level(Table 5.6.), along 62  Hail and WII were not included in this analysis, with the estimated fiscal costs to government of as they were developed on prototype base only. different levels of premium subsidy. Agricultural Insurance Feasibility Study  201 Table 7.1. Adjustments in the current Loss of Investment average Premium Rates to Achieve an average 60 Percent Target Loss Ratio, by Oblast Adjusted gross Adjusted net Six-year premium rate premium rate Six-year Implied % actual average for 60% target for 60% target Oblast average loss change in rate premium rate loss ratio loss ratio cost (2005–10) required (2005–10) (before FFSA (after FFSA intervention ) intervention) Akmola 1.9 2.0 3.38 1.69 −13 Aktobe 5.8 22.2 36.97 18.48 217 Almaty 3.5 4.6 7.64 3.82 10 EKO 3.4 6.0 9.96 4.98 47 Karaganda 4.7 4.7 7.86 3.93 −16 Kostanay 1.7 1.3 2.10 1.05 −40 Kysylorda 5.2 0.1 0.15 0.08 −99 NKO 2.0 0.5 0.79 0.40 −80 Pavlodar 3.6 7.1 11.81 5.90 66 South Kazakhstan 2.7 0.8 1.34 0.67 −75 (SKO) WKO 7.8 39.6 65.94 32.97 323 Zhambyl 3.5 14.6 24.27 12.14 243 Total 2.4 3.4 5.65 2.82 17 Source: Authors based on crop insurance results for 2005–10. 202  Kazakhstan Table 7.2. “As If” Analysis of Costs to Government in order to replace the current participation on 50 percent of the agricultural insurance portfolio liabilities with an Aggregate Insurance Industry Excess of Loss Compensation Scheme and 50 Percent Premium Subsidies, 2005–10 KZT (thousands) “As if” cost Total cost to Total cost to “As if” 100% “As if” cost to to govern- government government gross premium government Government ment for for losses in for losses in Scenario (without for losses 50% losses in excess of 100% excess of 70% and year government in excess of premium excess of premium + premium + 50% claims 70% gross subsidies 100% gross 50% premium 50% premium reinsurance) premium premium subsidies subsidies With Aktobe and WKO 2005 1,797,214 0 0 898,607 898,607 898,607 2006 1,369,444 0 0 684,722 684,722 684,722 2007 1,994,785 0 0 997,392 997,392 997,392 2008 2,186,463 0 179,099 1,093,232 1,093,232 1,272,331 2009 2,228,732 0 0 1,114,366 1,114,366 1,114,366 2010 2,147,278 657,668 1,301,851 1,073,639 1,731,307 2,375,490 Total 11,723,915 657,668 1,480,950 5,861,958 6,519,625 7,342,908 Average 1,953,986 109,611 246,825 976,993 1,086,604 1,223,818 Without Aktobe and WKO 2005 1,603,315 0 0 801,657 801,657 801,657 2006 1,347,999 0 0 673,999 673,999 673,999 2007 1,805,232 0 0 902,616 902,616 902,616 2008 1,833,355 0 365,814 916,678 916,678 1,282,491 2009 1,795,005 0 0 897,503 897,503 897,503 2010 1,820,435 0 0 910,218 910,218 910,218 Total 10,205,340 0 365,814 5,102,670 5,102,670 5,468,484 Average 1,700,890 0 60,969 850,445 850,445 911,414 Source: Authors based on FFSA insurance data. Agricultural Insurance Feasibility Study  203 Table 7.3. Estimated Total Sum Insured, Indicative Commercial Premium and Fiscal Implications for different mechanisms of Government Support for Spring Wheat Multiple Peril Crop Insurance (MPCI) Program, by Level of Coverage Insured yield coverage level Oblast 10% 20% 30% 40% 50% Total sum insured (KZT millions) Akmola 10,985 21,970 32,956 43,941 54,926 EKO 1,181 2,361 3,542 4,723 5,904 Karaganda 1,346 2,691 4,037 5,383 6,729 Kostanay 14,786 29,573 44,359 59,145 73,931 NKO 13,216 26,432 39,648 52,864 66,080 Pavlodar 910 1,821 2,731 3,641 4,552 Total 42,424 84,849 127,273 169,697 212,121 Indicative commercial average premium rate at Oblast level for 60% loss ratio (%) Akmola 1.5 4.8 8.9 13.6 18.4 EKO 2.5 6.9 12.0 17.3 22.7 Karaganda 1.8 5.6 10.2 15.2 20.4 Kostanay 1.4 4.6 8.7 13.2 17.9 NKO 1.1 3.6 6.7 10.3 14.1 Pavlodar 2.1 6.2 11.1 16.3 21.5 Total 1.4 4.5 8.3 12.6 17.1 Indicative commercial premium for a 60% loss ratio (KZT millions) Akmola 165 1,056 2,949 5,956 10,088 EKO 30 163 424 819 1,341 Karaganda 24 149 411 819 1,370 Kostanay 210 1,366 3,844 7,803 13,267 NKO 142 943 2,659 5,420 9,290 Pavlodar 19 113 302 593 978 Total 589 3,790 10,590 21,410 36,334 GoK Fiscal Implications for providing premium Subsidies to AYII Insurance Premiums (KZT million/year) 30% Subsidies 177 1,137 3,177 6,423 10,900 40% Subsidies 236 1,516 4,236 8,564 14,534 204  Kazakhstan 50% Subsidies 295 1,895 5,295 10,705 18,167 60% Subsidies 353 2,274 6,354 12,846 21,800 GoK Fiscal Implications for providing a ceiling on AYII Insurance Losses in excess of different Priorities (KZT millions/year) Priority 70% GNPI 242 1,115 2,675 4,948 7,930 Priority 100% GNPI 227 991 2,243 3,895 5,845 Priority 150% GNPI 208 829 1,711 2,664 3,515 Source: Authors based on FFSA insurance data. Agricultural Insurance Feasibility Study  205 Table 7.4. Estimated Total Sum Insured, Indicative Commercial Premium and Fiscal Implications for different mechanisms of Government Support for Spring Wheat Area yield Index Insurance (AYII) Program, by Level of Coverage Insured yield coverage level (% of rayon average expected yield) Oblast 10% 20% 30% 40% 50% 60% 70% 80% Total sum insured (KZT millions) Akmola 10,985 21,970 32,956 43,941 54,926 65,911 76,897 87,882 EKO 1,181 2,361 3,542 4,723 5,904 7,084 8,265 9,446 Karaganda 1,346 2,691 4,037 5,383 6,729 8,074 9,420 10,766 Kostanay 14,786 29,573 44,359 59,145 73,931 88,718 103,504 118,290 NKO 13,216 26,432 39,648 52,864 66,080 79,296 92,512 105,728 Pavlodar 910 1,821 2,731 3,641 4,552 5,462 6,372 7,283 Total 42,424 84,849 127,273 169,697 212,121 254,546 296,970 339,394 Indicative commercial premium for 70% loss ratio (%) Akmola 0.24 1.03 2.43 4.52 7.07 10.34 14.26 18.69 EKO 0.74 2.25 4.50 7.41 10.72 14.64 19.03 23.67 Karaganda 0.33 1.31 3.05 5.58 8.62 12.28 16.48 21.09 Kostanay 0.24 1.01 2.38 4.42 6.93 10.17 14.04 18.44 NKO 0.17 0.73 1.70 3.14 4.85 7.17 10.15 13.80 Pavlodar 0.38 1.50 3.36 5.96 8.97 12.62 16.75 21.14 Total 0.24 0.98 2.28 4.20 6.52 9.52 13.16 17.35 Indicative commercial premium for a 70% loss ratio (KZT millions) Akmola 26 226 799 1,987 3,885 6,813 10,968 16,429 EKO 9 53 159 350 633 1,037 1,573 2,236 Karaganda 4 35 123 301 580 991 1,552 2,271 Kostanay 36 299 1,055 2,615 5,124 9,021 14,537 21,817 NKO 23 192 675 1,662 3,206 5,682 9,390 14,594 Pavlodar 3 27 92 217 408 689 1,068 1,539 Total 101 833 2,903 7,131 13,837 24,234 39,087 58,885 206  Kazakhstan GoK Fiscal Implications for providing premium Subsidies to AYII Insurance Premiums (KZT million/ year) 30% Subsidies 30 250 871 2,139 4,151 7,270 11,726 17,666 40% Subsidies 40 333 1,161 2,852 5,535 9,694 15,635 23,554 50% Subsidies 51 417 1,452 3,566 6,919 12,117 19,544 29,443 60% Subsidies 61 500 1,742 4,279 8,302 14,540 23,452 35,331 GoK Fiscal Implications for providing a ceiling on AYII Insurance Losses in excess of different Priorities63 (KZT millions/year) Priority 70% 54 349 1,046 2,295 4,297 7,112 10,733 15,157 GNPI64 Priority 100% GNPI 52 326 948 2,005 3,609 5,693 8,084 10,606 Priority 150% GNPI 49 295 820 1,634 2,764 4,008 5,078 5,731 Table 7.5. Five-Year Estimates of Voluntary MPCI Uptake, Total Sum Insured, Premium Income, and Costs of Premium Subsidies (40% Coverage Level) for spring wheat KZT (thousands) MPCI 40 % MPCI uptake Item coverage and Year 1 Year 2 Year 3 Year 4 Year 5 100% uptake 10% 20% 30% 40% 50% Total sum insured 169,697 16,970 33,939 50,909 67,879 84,849 Total commercial 21,410 2,141 4,282 6,423 8,564 10,705 premium Cost of GRK premium subsidies 25% of premium 5,353 535 1,071 1,606 2,141 2,676 50% of premium 10,705 1,071 2,141 3,212 4,282 5,353 65% of premium 13,917 1,392 2,783 4,175 5,567 6,958 Probable maximum loss 1 in 100 years 66,470 6,647 13,294 19,941 26,588 33,235 1 in 250 years 79,688 7,969 15,938 23,906 31,875 39,844 Source: Authors (Figures taken from Table 4.14 page 92). 63 Priority: Is the threshold, defined as a percentage of the Gross net premium income, above which the Government assume full liability arising from any risk or any one event for losses. In practice the Government puts a cap on the losses. 64 GNPI: Gross Net Premium Income Agricultural Insurance Feasibility Study  207 Table 7.6. Five-Year Estimates of Voluntary AYII Uptake, Total Sum Insured, Premium Income, and Costs of Premium Subsidies (50 Percent Coverage Level) AYII 50% AYII crop insurance uptake scenarios coverage level at 100% uptake Year 1 Year 2 Year 3 Year 4 Year 5 Item 5% 10% 15% 20% 25% Total sum insured (KZT millions) 212,121 10,606 21,212 31,818 42,424 53,030 Commercial premium (KZT 13,837 692 1,384 2,076 2,767 3,459 millions) Cost of GRK premium subsidies (% of premium) 25% 3,459 173 346 519 692 865 50% 6,918 346 692 1,038 1,384 1,730 65% 8,994 450 899 1,349 1,799 2,248 Probable maximum loss 1 in 100 years 51,516 2,576 5,152 7,727 10,303 12,879 1 in 250 years 70,634 3,532 7,063 10,595 14,127 17,659 Source: Authors. (Figures tken from Table 5.6) 208  Kazakhstan Bibliography Agroasemex. 2008. Sistema Nacional de Aseguramiento al Medio Rural: Informe final de opera- ciones al cierre del ejercicio fiscal 2007. Querétaro: Agroasemex. ———. 2009. Programa del seguro para contingencias climatológicas: Informe de avance al cierre del ejercicio fiscal 2009. Querétaro: Agroasemex. Agroseguro. 2011. Annual Report 2010. Madrid, 2011 Ahsan, Syed M (2009), “Microinsurance, Poverty and Vulnerability”, in Lazar, D., and M. Deo, Eds. Micro Finance: Performance Evaluation and Enterprise Development, 504-15, Allied Publishers, Chennai, January 2009. ARKS (National Statistics Agency of the Republic of Kazakhstan). 2004. Guideline on Crop Yield In- spection Arrangement. Almaty: Department for Industrial Statistics, Kazakhstan Statistics Agency. Baisholanov, Saken. 2011. “An Agro-Meterological Risk Assessment for Cereal Crop Production in Kazakhstan.” Prepared by Saken Baisholanov, former head of Agro Meteorology Department, Kazhydromet. Bora, Bulent. 2010. “Subsidised Agricultural Insurance in Turkey (TARSIM).” Trébol 54 (article 2). Available from http://www.mapfre.com/mapfrere/docs/html/revistas/trebol/n54/docs/Articulo2en. pdf. Burgaz Moreno F.. 2007. Nuevas perspectivas en la definición de políticas de ayuda a seguros agrar- ios y gestión de crisis Vida Rural, 2007, 261:27-29 Burt, C. C. 2007. Extreme Weather: A Guide and Record Book. New York: Norton. Celaya del Toro, V. 2010. “El uso de los seguros catastróficos para atender daños por contingencias climatológicas en el sector agropecuario de México.” In XI Congreso de ALASA: El seguro agrícola en América; Logros y desafíos. Cartagena de Indias, Colombia: ALASA. CJSC (Kazakh Actuarial Center). 2003. “Tariff Calculation Methodology for Compulsory Crop In- surance in the Republic of Kazakhstan.” Report prepared for the Ministry of Agriculture, CJSC, November 18. Monique Cohen, Michael J. McCord, and Jenefer Sebstad. Reducing Vulnerability: Demand for and Supply of Microinsurance in East Africa – A Synthesis Report. MicroSave, December 2003 Dick, W. 2009. Agricultural Insurance: Scope and Limitations for Rural Risk ManagementPeresenta- tion at the World Bank-Agricultural Rural Development Week. Washington DC, March 2009 Agricultural Insurance Feasibility Study  209 Economist Intelligence Unit. 2011a. Country Data: Annual Time Series. London: The Economist. ———. 2011b. “Current Policy: The 2010 Budget Deficit Is Below the Government’s Target.” Econ- omist Intelligence Unit, London, March 2. http://country.eiu.com/article.aspx?articleid=987853083. FAO (Food and Agriculture Organization) and World Bank. 2001. Farming Systems and Poverty: Improving Farmers’ Livelihoods in a Changing World. Rome: FAO. GFDRR (Global Fund for Disaster Risk Management). 2010a. “Brazilian Experience in the Field of Agricultural Insurance in Developing Countries: Challenges and Opportunities.” In GFDRR: 9th Meeting of the Consultative Group. Washington, DC: GFDRR. ———. 2010b. National Agricultural Insurance Scheme in India: Making Insurance Markets Work for Farmers. Washington, DC: World Bank, GFDRR. Hartell, J., H. Ibarra, J. R. Skees, and J. Syroka. 2006. Risk Management in Agriculture for Natural Hazards. Rome: ISMEA, May. Hazell P. B. R. 1992. “The Appropriate Role of Agricultural Insurance in Developing Countries.” Journal of International Development 4 (6): 567–81. Hazell, P., C. Pomareda, and A. Valdez. 1986. Crop Insurance for Agricultural Development. Balti- more, MD: Johns Hopkins University Press. Hellmuth, M. E., et al., 2009, ‘Index Insurance and Climate Risk: Prospects for Development and Disaster Management’, International Research Institute for Climate and Society, Columbia Univer- sity, New York IFAD (International Fund for Agricultural Development). Forthcoming. Weather Index-Based Insur- ance in Agricultural Development: A Technical Guide. Rome: IFAD. Iturrioz, R. 2009. Agricultural Insurance. Primer Series on Insurance 12. Washington, DC: World Bank. ———. 2010. “Seguro agropecuario en América Latina: Situación actual, desafíos y oportunidades.” In XI Congreso de ALASA: El seguro agrícola en América; Logros y desafíos. Cartagena de Indias, Colombia: ALASA. Latchininsky, Alexandre, and Ramesh Sivanpillai. “Locust in Kazakhstan.” Presentation at the Uni- versity of Wyoming. Mahul, O., and C. Stutley. 2010. Government Support to Agricultural Insurance: Challenges and Op- tions for Developing Countries. Washington, DC: World Bank. Morgounov, A., L. Rosseeva, and M. Koyshibayev. 2007. “Leaf Rust of Spring Wheat in Northern Kazakhstan and Siberia: Incidence, Virulence, and Breeding for Resistance.” Australian Journal of Agricultural Research 58 (9): 847–53. http://dx.doi.org/10.1071/AR07086 Munich Re Group. 2009. World Map of Natural Hazards. Munich: Münchener Rückversicherungs- Gesellschaft. Nomura. 2008 210  Kazakhstan ProVention. 2006. Rosema, A., M. de Weirdt, S. Foppes, and C. Wilczok. 2010. FESA Micro-Insurance: Methodology, Validation, Contract Design. Delft, the Netherlands: EARS. Skees, Jerry R., J. Roy Black, and Barry J. Barnet. 1997. “Designing and Rating an Area-Yield Crop Insurance Contract.” American Journal of Agricultural Economics 79 (May): 430–38. Skees, Jerry, and J. Hartell. 2008. “Developing Index-Based Agricultural Insurance to Enhance Fi- nancial Markets for Poverty Reduction in Vietnam.” Vietnam: Agricultural Insurance Product Brief- ing Note prepared for VIBARD Agricultural Insurance Working Group, Hanoi, Vietnam, November. Ford Foundation Project. Skees, Jerry R., J. Hartell, A. G. Murphy, and B. Collier. 2009. “Handbook IV: A Policy Vision for Developing Agricultural Insurance in Vietnam.” In Developing Agricultural Insurance in Vietnam: Four Educational Handbooks. Hanoi, Vietnam: AgroInfo. Skees, Jerry R., and A. G. Murphy. 2009. “ENSO Business Interruption Index Insurance for Cata- strophic Flooding in Piura, Peru.” GlobalAgrisk Inc, Lexington, KY, February. Stutley, C. 2010. “Crop Insurance Feasibility Study 2010.” Report prepared for the Innovative Insur- ance Products for the Adaptation to Climate Change Project (IIPACC) Project, Ghana. Uçak, Harun, and A. Berk. 2009. “Structural Change in Turkish Agricultural Insurance Policy and Recent Developments.” Wiadomości Ubezpieczeniowe 2/2009, Poland. http://www.piu.org.pl/pub- lic/upload/ibrowser/Wiadomosci%20Ubezpieczeniowe/numer2/WU2_09ucakberk.pdf. USDA (U.S. Department of Agriculture), FAS (Foreign Agricultural Service). 2010. Kazakhstan: Conmodity Inteligence Report. Washington, DC: USDA. http://www.pecad.fas.usda.gov/high- lights/2010/01/kaz_19jan2010/. Linnerooth-Bayer, J., K. Warner, et al. (2009). “Insurance, Developing Countries, and Climate Change.” The Geneva Papers on Risk and Insurance – Issues and Practice, Special issue on Climate Change and its Economic Impact on Insurance 34: 381-400 WFP (World Food Programme) and IFAD (International Fund for Agricultural Development). 2010. “The Potential for Scale and Sustainability in Weather Index Insurance for Agriculture and Rural Livelihoods.” International Fund for Agricultural Development and the World Food Programme, Rome, March. World Bank. 2005a. “Managing Agricultural Production Risk: Innovations in Developing Countries.” Agriculture and Rural Development Department (ARD), World Bank, Washington, DC. ———. 2005b. Natural Disaster Hotspots: A Global Risk Analysis. Disaster Risk Management Se- ries 5. Washington, DC: World Bank. ———. 2010a. “Kazakhstan: Public Expenditure and Institutional Review for the Agricultural Sec- tor.” World Bank, Washington, DC. ———. 2010b. World Development Indicators. Washington, DC: World Bank. Agricultural Insurance Feasibility Study  211 Agriculture and Rural Development Unit Sustainable Development Department Europe and Central Asia Region