AGRICULTURE GLOBAL PRACTICE TECHNICAL ASSISTANCE PAPER MALAWI AGRICULTURAL SECTOR RISK ASSESSMENT Åsa Giertz, Jorge Caballero, Diana Galperin, Donald Makoka, Jonathan Olson, and George German WORLD BANK GROUP REPORT NUMBER 99941-MW DECEMBER 2015 AGRICULTURE GLOBAL PRACTICE TECHNICAL ASSISTANCE PAPER MALAWI Agricultural Sector Risk Assessment Åsa Giertz, Jorge Caballero, Diana Galperin, Donald Makoka, Jonathan Olson, and George German © 2015 World Bank Group 1818 H Street NW Washington, DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org Email: feedback@worldbank.org All rights reserved. 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 necessarily reflect the views of the Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. 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. Rights and Permissions The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. World Bank Group encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clear- ance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone: 978-750-8400, fax: 978-750-4470, http://www.copyright .com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, World Bank Group, 1818 H Street NW, Washington, DC 20433, USA, fax: 202-522-2422, e-mail: pubrights@worldbank.org. Cover images: All images courtesy of Åsa Giertz/World Bank. CONTENTS Acronyms and Abbreviations ix Acknowledgments xi Executive Summary xiii Chapter One: Introduction and Context 1 Chapter Two: Malawi’s Agricultural System 5 Agriculture Sector Overview and Performance 5 Agroclimatic Conditions 7 Production and Market Trends 9 Chapter Three: Agriculture Sector Risks 11 Food Crops—Production Risks 11 Food Crops—Market Risks 16 Food Crops—Enabling Environment Risks 17 Export Crops Overview 19 Export Crops—Production Risks 21 Export Crops—Market Risks 23 Export Crops—Enabling Environment Risks 25 Weather-Yield Analysis 27 Chapter Four: Adverse Impact of Agricultural Risk 29 Overall Agricultural Losses 29 Production Volatility by Region 31 The Impacts of Agricultural Risks on Different Stakeholders 33 Vulnerable Groups 33 Gender Structures Add an Additional Layer of Vulnerability 35 Impact on Household Food Security 36 Chapter Five: Risk Prioritization And Management 37 Risk Prioritization 37 Priority Risk-Management Measures 37 References 45 Appendix A: Weather-Yield Analysis 49 Malawi’s Political Districts 49 Weather Information in Malawi 49 Distribution of Monthly Rainfall in Malawi 49 Drought and Excess Rainfall Analysis 49 Rainfall—Yield Regressions 50 Maize 54 Cassava 61 Malawi: Agricultural Sector Risk Assessment iii Appendix B: Climate and Climate Change 67 Current Climate and Weather Patterns in Malawi 67 Changes in Weather Patterns 68 Global Climate Change and Malawi 68 Impacts on Crops 69 Regional Variation of Climate Change Impacts 72 Appendix C: Vulnerability Analysis 75 Context 75 Common Shocks Faced by Malawian Households 79 Key Groups Vulnerable to Various Shocks 79 Factors Increasing Vulnerability to Shocks 81 Risk-Management Strategies to Main Shocks 83 BOXES Box 2.1: Malawi’s Farm Input Subsidy Program 8 Box 3.1: Glossary of Drought Events 14 Box 3.2: Aflatoxins 17 Box 3.3: National Food Reserve Agency (NFRA) 18 Box 3.4: Interviews with Small-Scale Maize Traders 20 Box 3.5: Case Study: Mary Mwase, Maize, and Tobacco Farmer Madisi 22 Box 3.6: Sugar Prices 25 Box 3.7: Recent Macroeconomic Reforms 26 Box C.1: Gender Vulnerability in Malawi 82 FIGURES Figure ES.1: GDP and Agricultural Value Added (% Growth) in Malawi, 1968–2011 xiv Figure ES.2: Value of Production Losses per Year as a Share of Total Agricultural Production Value xv Figure ES.3: Value and Frequency of Losses per Crop in Malawi, 1980–2012 xvi Figure ES.4: Costs and Government Budgetary Expenses for Activities Associated with Risk Mitigation and Risk Coping versus Losses from Risks in Malawi, 2008–12 xvii Figure 1.1: GDP and Agricultural Value Added (% Growth) in Malawi, 1968–2011 2 Figure 1.2: Agriculture Sector Risk-Management Process Flow 3 Figure 2.1: Gross Cereal Production Index (2004–06 = 100) in Malawi, 1968–2012 6 Figure 2.2: Commodities That Make Up the Top 80 Percent of Gross Agricultural Production Value (2009–11 Average) 6 Figure 2.3: Share of Area Harvested for Commodities That Make Up the Top 80 Percent of Gross Agricultural Production Value (2009–11 Average) 6 Figure 2.4: Agricultural Exports and Constant GDP (US$ ’000), 2001–12 7 Figure 2.5: Agricultural Exports from Malawi, 2012 7 iv Agriculture Global Practice Technical Assistance Paper Figure 2.6: Yield of Selected Crops in Malawi, 1961–2011 9 Figure 3.1: Maize Yields (MT/ha), 1980–2012 13 Figure 3.2: Groundnut Yields (MT/ha), 1980–2012 13 Figure 3.3: Potato Yields (MT/ha), 1980–2012 14 Figure 3.4: Banana Yields (MT/ha), 1980–2012 14 Figure 3.5: Bean Yields (MT/ha), 1980–2012 14 Figure 3.6: Monthly Maize Prices in Lilongwe, Blantyre, Mzuzu, and Zomba (Tambala/kg), 2005–12 18 Figure 3.7: Maize Interventions in Malawi since 2012 19 Figure 3.8: Market Interventions and Price Distortions in Malawi’s Maize Market 19 Figure 3.9: Cottonseed and Cotton Lint Yields and Major Drought Events in Malawi, 1980–2012 21 Figure 3.10: Sugarcane Yields (kg/ha) and Area Harvested (ha) in Malawi, 1980–2012 22 Figure 3.11: Average Auction Price (U.S. Cents/kg) and Volume (kg) of Tobacco Sold, Lagged One Year, 1995–2012 23 Figure 3.12: Annual Cotton Price Change (%) in Malawi, 1988–2012 24 Figure B3.6.1: Annual Prices Sugar and Sucrose (2006–14) 25 Figure B3.7.1: Exchange Rate (MK/US$) by Month, 1985–2014 26 Figure B3.7.2: Changes in Consumer Prices Compared with Same Time Previous Year, 1994–2013 26 Figure 3.13: Crop Calendar for Malawi 27 Figure 4.1: Example of How Indicative Losses Are Calculated 30 Figure 4.2: Value of Production Losses per Year as a Share of Total Agricultural Production Value 31 Figure 4.3: Value and Frequency of Losses per Crop, 1980–2012 31 Figure 5.1: Strategic Risk Instruments According to Risk Layers 39 Figure 5.2: Government Budgetary Expenses for Risk-Mitigating and Risk-Coping Interventions versus Losses from Risks 42 Figure A.1: Monthly Rainfall Pattern for Several Weather Stations 51 Figure A.2: Malawi’s Crop Calendar 54 Figure A.3: Maize Yield by Region, 1994–2013 55 Figure A.4: Maize Yield in Blantyre, 1994–2013 57 Figure A.5: Regression Results for Cumulative Rainfall and Maize Yield in Blantyre 57 Figure A.6: Maize Yield in Karonga, 1994–2013 57 Figure A.7: Regression Results for Cumulative Rainfall and Maize Yield in Karonga 58 Figure A.8: Maize Yield in Kasungu, 1994–2013 58 Figure A.9: Regression Results for Cumulative Rainfall and Maize Yield in Kasungu 58 Figure A.10: Maize Yield in Lilongwe, 1994–2013 58 Figure A.11: Regression Results for Cumulative Rainfall and Maize Yield in Lilongwe 59 Figure A.12: Maize Yield in Machinga, 1994–2013 59 Figure A.13: Regression Results for Rainy Events and Maize Yield in Machinga 59 Figure A.14: Maize Yield in Mzuzu, 1994–2013 60 Malawi: Agricultural Sector Risk Assessment v Figure A.15: Regression Results for Rainy Events and Maize Yield in Mzuzu 60 Figure A.16: Maize Yield in Salima, 1994–2013 60 Figure A.17: Regression Results for Rainy Events and Maize Yield in Salima 61 Figure A.18: Maize Yield in Shire Valley, 1994–2013 61 Figure A.19: Regression Results for Cumulative Rainfall and Maize Yield in Shire Valley 61 Figure A.20: Cassava Yield in Blantyre, 1994–2013 63 Figure A.21: Regression Results for Cumulative Rainfall and Cassava Yield in Blantyre 64 Figure A.22: Cassava Yield in Karonga, 1994–2013 64 Figure A.23: Regression Results for Cumulative Rainfall and Cassava Yield in Karonga 64 Figure A.24: Cassava Yield in Kasungu, 1994–2013 65 Figure A.25: Cassava Yield in Lilongwe, 1994–2013 65 Figure A.26: Cassava Yield in Machinga, 1994–2013 65 Figure A.27: Cassava Yield in Mzuzu, 1994–2013 65 Figure A.28: Regression Results for Rainy Events and Cassava Yield in Mzuzu 66 Figure A.29: Cassava Yield in Salima, 1994–2013 66 Figure A.30: Cassava Yield in Shire Valley, 1994–2013 66 Figure B.1: Average Monthly Temperature and Rainfall in Malawi 68 Figure B.2: Average Monthly Temperature and Rainfall for Malawi, 1900–1930 69 Figure B.3: Average Monthly Temperature and Rainfall for Malawi, 1990–2009 69 Figure B.4: Number of Hot Days over a Year in Malawi, 1960–2000 and 2046–65 69 Figure B.5: Projected Mean Temperature in Malawi according to Nine Climate Change Models, 2020–39 70 Figure B.6: Projected Mean Rainfall in Malawi according to Nine Climate Change Models, 2020–39 71 Figure B.7: Number of Days without Rain by Month, 1961–2000 and 2046–65 71 Figure B.8: Number of Days with Extreme Rain by Month, 1961–2000 and 2046–65 72 Figure B.9: Current Minimum and Maximum Temperatures in Malawi 73 Figure B.10: Results of Nine Climate Change Models for the Northern, Central, and Southern Parts of Malawi 74 MAPS Map 2.1: Average Annual Precipitation (mm) in Malawi 8 Map 2.2: Evolution of Land Cover in Malawi, 1973–2010 9 Map 4.1: Malawi’s Eight Agricultural Development Divisions 32 Map A.1: Political Districts in Malawi 50 TABLES Table 2.1: Land Use in Malawi (km2) 8 Table 3.1: Major Drought Incidents in Malawi, 1980–2012 13 Table 3.2: Pests and Diseases in Malawi for Analyzed Food Crops, in Field and Postharvest 15 Table 3.3: Trade Bans and Lifts in Malawi since 2008 20 vi Agriculture Global Practice Technical Assistance Paper Table B3.5.1: Risk Events in Order of Importance According to Mary 22 Table 4.1: Losses from Agricultural Production Risks, 1980–2012 30 Table 4.2: Maize Production by Agricultural Development Division in Malawi, 1983–2013 32 Table 4.3: Cassava Production by Agricultural Development Division in Malawi, 1983–2013 33 Table 4.4: Stakeholder Risk Profiles for Food and Export Crop Supply Chains 34 Table 4.5: Distribution of Poverty in Malawi 35 Table 4.6: Proportion of Plots by Type of Crop Cultivated and as a Share of Total Crops, 2011 36 Table 5.1: Risk Prioritization 38 Table 5.2: Long List of Agricultural Risk-management Solutions Identified for Malawi 40 Table 5.3: Value of Donor-Financed Agricultural Projects by Type of Activity, 2004–13 42 Table A.1: Rainfall Anomalies for Malawi’s 23 Weather Stations 52 Table A.2: Weather Stations Used in Each MAFS Region 54 Table A.3: Simple Linear Models’ Determination Coefficients for Maize Yield 55 Table A.4: Multiple Linear Models’ Determination Coefficients for Maize Yield 55 Table A.5: Average Maize Yield Before and After 2005 by Region 56 Table A.6: Simple Linear Models’ Determination Coefficients for Maize Yield Transformed 56 Table A.7: Multiple Linear Models’ Determination Coefficients for Maize Yield Transformed 56 Table A.8: Simple Linear Models’ Determination Coefficients for Cassava Yield 62 Table A.9: Multiple Linear Models’ Determination Coefficients for Cassava Yield 62 Table A.10: Average Cassava Yield Before and After 2005 by Region 62 Table A.11: Single Linear Models’ Determination Coefficients for Cassava Yield Transformed 63 Table A.12: Multiple Linear Models’ Determination Coefficients for Cassava Yield Transformed 63 Table C.1: Livelihood Options, Key Hazards, and Response Strategies in Malawi’s 11 Livelihood Zones 76 Table C.2: Proportion (%) of Households Severely Affected by Shocks during the Past 12 Months by Location, Sex, and Region in Malawi, 2011 80 Malawi: Agricultural Sector Risk Assessment vii ACRONYMS AND ABBREVIATIONS AfDB African Development Bank IHS Integrated Household Survey ADD Agricultural Development Division IHS2 Integrated Household Survey 2 ADMARC Agricultural Development and Marketing IHS3 Integrated Household Survey 3 Cooperation ISOs Intra-Seasonal Oscillations ARMT Agricultural Risk Management Team ITCZ Inter-Tropical Convergence Zone ASWAp Agricultural Sector Wide Approach kg Kilogram BBTV Banana Bunchy Top Virus M&E Monitoring and Evaluation CAADP Comprehensive Africa Agriculture MAFS Ministry of Agriculture and Food Security Development Programme MDTF Multi Donor Trust Fund CAB Congo Air Boundary MGDS Malawi Growth and Development Strategy CABS Common Approach to Budgetary Approach MK Malawi kwacha CRED Centre for Research on the Epidemiology of MT Metric Ton Disasters MVAC Malawi Vulnerability Assessment Committee CV Coefficient of variation NFRA National Food Reserve Agency DODMA Department of Disaster Management NGO Nongovernmental Organization EM-DAT International Disaster Database NSO National Statistical Office ENSO El Niño-Southern Oscillation OPV Open pollinated variety EU European Union QBO Quasi-Biennial Oscillation FAO Food and Agriculture Organization SECO Swiss Secretariat of Economic Affairs (of the UN) SGR Strategic Grain Reserve FAOSTAT FAO Corporate Statistical Database SST Sea surface temperature FISP Farm Input Subsidy Program TCC Tobacco Control Commission G-8 Group of Eight USAID U.S. Agency for International Development GDP Gross Domestic Product VSL Village Savings and Loan Group GRV Groundnut rosette virus WDI World Development Indicators ha Hectare WFP World Food Programme Malawi: Agricultural Sector Risk Assessment ix ACKNOWLEDGMENTS This report was developed by a team led by Åsa Giertz, to the stakeholders from major agricultural supply chains Agricultural Specialist from the Agricultural Risk Man- who participated at various moments during the field agement Team at the World Bank. The activities were work and during the workshops to discuss the findings. supported by the following agricultural specialists: Jorge Their active participation obliged the team to be realistic Caballero, Diana Galperin, Jonathan Olsson, Donald and practical. Makoka, George German, Traci Johnson, and Srilatha Shankar. The report was edited by Amy Gautam. This activity would not have been possible without the generous contributions from USAID, Ministry of Foreign The team is grateful for the leadership and coordination Affairs of the Government of the Netherlands, and State received from Vikas Choudhary and Olivier Durand. Secretariat for Economic Affairs (SECO) of the Govern- ment of Switzerland. The team would like to extend its appreciation to the Malawi Ministry of Agriculture and Food Security and Malawi: Agricultural Sector Risk Assessment xi EXECUTIVE SUMMARY Malawi is among the poorest countries in the world, with limited resources and an economy that relies heavily on agriculture. Per capita gross domestic product (GDP) is US$362 per year1 (World Development Indicators 2014) and 62 percent of the popu- lation lives on less than US$1.25 per day. Malawi is relatively small in size, is densely populated, and has high population growth, all of which put pressure on available land for smallholder farming and on the environment and the natural resource base, nota- bly land and forests. Officially, the population amounts to 15.9 million people, about 80 percent of whom live in rural areas (World Development Indicators, 2010–12 year figures, accessed March 2014). Agriculture is the backbone of Malawi’s economy, contributing 30 percent of total GDP (2011) and 76 percent of total national exports (2012). With 78 percent employed in the sector in 2013 Food and Agriculture Organization of the United Nations (FAO Country Profile, accessed May 2014), agriculture is also a main source of employ- ment and income. Increasing food security is one of the main objectives of Malawi’s Agricultural Sector Wide Approach (ASWAp 2010) and a strong focus on increasing maize production since the mid-2000 has resulted in rapidly increasing production. However, production risks continue to result in high losses to the sector, including for maize. Further, price interventions in the sector over the past year have implied greater price risks for producers and traders. As evident in Malawi, risks can have potentially significant implications on stakehold- ers, investments, and development in the agriculture sector. Adverse movements in agricultural commodity and input prices together with production-related shocks (for example, from weather, pests, diseases) not only affect farmers and firms active in particular supply chains, but may also put severe strains on a government’s finances. Rapid or significant declines in production and/or trade may reduce government tax revenues, affect balance of payments, necessitate compensatory (or recovery) expendi- tures, and/or otherwise adversely affect a government’s fiscal position. The prevalence 1 Current US$, 2010–11 average. Malawi: Agricultural Sector Risk Assessment xiii FIGURE ES.1. GDP AND AGRICULTURAL VALUE ADDED (% GROWTH) IN MALAWI, 1968–2011 60 GDP growth (annual %) 50 Agriculture, value added (annual % growth) Depression maize 40 prices and currency (by 40%) 30 20 Uneven rainfall 10 0 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 –10 –20 Drought Drought Political –30 instability, –40 Drought donors exit Source: World Development Indicators (WDI) 2014. of “shock-recovery-shock” cycles vastly reduces the abil- est commodities that jointly account for 80 percent of ity of many countries to plan for and concentrate on real Malawi’s agricultural production value (maize, cassava, development issues. potatoes, peas and beans, rice, groundnuts, bananas, tobacco, and sugar) plus tea and cotton because of their Over the past decades, Malawi has been struck by several export potential. Maize is by far the most important staple severe droughts that have resulted in spikes in food insecu- crop, accounting for more than 50 percent of the daily rity and prompted the need for humanitarian aid. During calorie intake in Malawi. Tobacco, tea, cotton, and sugar the last major drought in 2005, 40 percent of the popula- accounted for 67 percent of the total value of national tion was in immediate need of food aid as a result of a poor exports of goods in 2012, with tobacco alone accounting harvest. Because of the size of the sector in the economy for more than 54 percent. and the importance of agricultural products for export, agricultural growth correlates closely with GDP growth. The report takes a quantitative and qualitative approach This means that drops in agricultural growth affect the to assess risks that have occurred in the agriculture sector entire economy, as depicted in figure ES.1—agricultural since 1980. Productions risks are quantified in terms of GDP growth was negative in five years between 1992 and losses and mapped by different perils. Market and ena- 2010. Further, any drop in agricultural growth in a given bling environment risks are analyzed qualitatively. For the year will affect the ASWAp annual growth target of 6 per- purpose of this assessment, “risk” is defined as the pos- cent that Malawi has committed to under Comprehensive sibility that an event will occur and will potentially have Africa Agriculture Development Programme (CAADP. a negative impact on the achievement of a farm or firm’s For individual actors in the sector, these risks reinforce performance objectives and/or on successful functioning poverty traps by cycles of shock-recovery-shock and result of the overall supply chain. A broad spectrum of stake- in lower returns on investments in productive assets. holders was consulted throughout this work, including the Malawi government, farmers, traders, processors, The purpose of this report is to assess existing agricul- agricultural institutions, and academia. A consultative tural risks to the sector, prioritize them according to their stakeholder meeting was also held in Lilongwe to obtain frequency and impacts on the sector, and identify areas feedback on findings and to discuss areas for risk solution of risk-management solutions that need deeper special- interventions for deeper analysis. ized attention. Three levels of risks were assessed: produc- tion risks, market risks, and enabling environment risks Droughts and pests and diseases are cited by stakehold- to selected supply chains. To give a sectorwide overview ers as the most damaging production risks, especially for of the impacts of risks, the assessment looks at the larg- food crops. Drought is probably the most visible risk to xiv Agriculture Global Practice Technical Assistance Paper FIGURE ES.2. VALUE OF PRODUCTION LOSSES PER YEAR AS A SHARE OF TOTAL AGRICULTURAL PRODUCTION VALUE $1,000 2005 $800 Losses per year (US$ million) $600 1992 $400 1995 1999 1998 1994 2004 2002 $200 1989 1996 2003 1993 2001 1980 1983 19851986 1991 1997 2006 1987 1990 2000 $– 1980 1985 1990 1995 2000 2005 2010 $(200) Source: FAO Corporate Statistical Database (FAOSTAT); authors’ calculations. Note: These costs constitute only losses, not response costs, which would add to the cost of risks these years. the sector. Malawi has suffered very bad droughts in the the time they occur. Shocks affect household and national past that had strong fiscal impact and required help from food security, have important fiscal repercussions, reduce the international community. The damaging impact of the availability of foreign exchange, and generally have pests and diseases is significant but depends on agricul- an overall destabilizing effect on the macroeconomy. For tural practices and mitigation activities. The impacts of instance, during the 2001 drought, losses amounted to pests and diseases are at times also exacerbated by adverse US$161 million, or 4.3 percent of total agricultural pro- weather events. Erratic rainfall and hailstorms are fre- duction value; in 2005, losses were nearly US$900 mil- quent but of moderate or low impact. lion, 24 percent of total agricultural production (2006–08 average). Figure ES.2 shows the magnitude of losses for Price volatility is an important market risk in Malawi, par- individual years compared with the general yield trend ticularly in key crops such as maize, tobacco, and cotton. for assessed crops, where the size of the circle depicts the The causes of volatility depend on the crop: cotton prices losses as a share of total agricultural production value. fluctuate according to world prices, whereas tobacco and maize prices are mainly determined by the domestic mar- The losses in normal production value can be extreme for ket. Maize price volatility is largely a result of enabling important smallholder crops such as maize and tobacco environment risks because of unpredictable domestic (for example, 50 percent of maize value was lost in 2005), market interventions and export policies. Regardless of leading to disastrous impacts on household incomes, food the reason, sudden fluctuations in prices negatively affect security, and well-being. The magnitude of the losses when farmers, the segment of the supply chain with the least shocks occur is much greater for some crops than for oth- risk-management capacity. Exchange rate volatility and ers: maize, cassava, potatoes, and tobacco have the high- unreliable input markets add to these uncertainties for est average annual losses (figure ES.3). However, tobacco actors in the export crop sector. and tea incur losses more frequently, meaning that farm- ers involved in these crops are highly exposed to shocks. The impacts of individual shocks are at times devastating. Average figures are useful to understand the aggregate Understanding how risks affect different parts of the coun- costs of production risk yet tend to conceal the cata- try is important for risk-management purposes in an envi- strophic impact that some shocks have on individuals at ronment with limited resources. Maize yield volatilities Malawi: Agricultural Sector Risk Assessment xv FIGURE ES.3. VALUE AND FREQUENCY OF LOSSES PER CROP IN MALAWI, 1980–2012 1,600 Cassava Maize 1,400 Losses per crop (US$ million) 1,200 1,000 800 Potatoes Tobacco 600 400 Sugarcane Groundnuts 200 Rice Beans Bananas Tea – Cotton Pigeonpeas 0 0.1 0.2 0.3 0.4 0.5 0.6 (200) Frequency of losses are fairly even across Malawi’s eight Agricultural Devel- tyre (3 percent of total production) and Mzuzu (1.9 per- opment Divisions (ADDs), with Blantyre experiencing cent). If Salima and Machinga are included, these four the highest volatility and Kasungu the lowest. The ADDs regions jointly account for over 80 percent of total annual of Lilongwe and Kasungu, which have the largest exten- cassava losses in Malawi. sions of land cultivated to maize (almost 50 percent of the country’s total), exhibit relatively similar yield volatilities, Risks are costly for Malawi, not just for the private sector significantly lower than that of Blantyre. Cassava shows but also for the government. Malawi is one of the few similar differences in losses between ADDs, although its countries in Sub-Saharan Africa that adhere to CAADP’s coefficient of variation (CV) of yield is high in all ADDs goal of allocating 10 percent of the national budget to (likely due to the discrete jump in cassava yield in early agriculture, and the country spends about US$250 mil- 2000).2 lion on agriculture annually. Although this seems to have mitigated the impacts of risks since the mid-2000s, any Because of the different level of outputs between ADDs, losses in subsectors supported by the government imply these variations in yield have different impacts on total lost investments. And although the losses are smaller, the production. The eight ADDs produce a total of 2 mil- government and donors spend large amounts on emer- lion metric tons (MT) of maize annually but 70 percent gency aid and other coping mechanisms in response to of Malawi’s maize production is grown in three ADDs shocks, diverting funds that would otherwise be allocated (Blantyre, Lilongwe, and Kasungu), and 90 percent in five to long-term development investments. ADDs if Machinga and Muzuzu are included. Losses as a share of national production are largest in Kasungu, Figure ES.4 gives an overview of the cost of risks and Lilongwe, Blantyre, and Michnga, which together account risk management in Malawi. On the mitigation side are for over 9 percent of total production losses annually, and expenditures on activities that could potentially reduce 80 percent of total maize losses in Malawi. Similarly, two the impacts of identified risks, even though at the moment regions account for half of Malawi’s cassava losses: Blan- research and extension are not particularly geared toward risk mitigation but more toward general productivity- 2 The observed jump in cassava yields and subsequent discussions with Ministry enhancing practices. Nevertheless, the figure clearly of Agriculture and Food Security (MAFS) officials suggest that there are quality shows that risk-management expenditures are skewed concerns with the cassava yield data. The cassava loss estimates are based on national yield data, and should be adjusted if the national cassava yield data are toward coping mechanisms for ex post risks rather than revised. Total losses excluding cassava amount to US$103.5 million per year. ex ante risk-mitigating interventions that would decrease xvi Agriculture Global Practice Technical Assistance Paper FIGURE ES.4. COSTS AND GOVERNMENT to identify gaps in broad-based risk-management systems BUDGETARY EXPENSES FOR and to advise how these gaps can be bridged to minimize losses and strengthen Malawians’ resilience against future ACTIVITIES ASSOCIATED shocks. The short list of proposed solutions focuses on WITH RISK MITIGATION areas in which the intervention gaps are currently deemed AND RISK COPING VERSUS greatest. It comprises the following four broad areas: LOSSES FROM RISKS IN 1. Strengthen agricultural information sys- MALAWI, 2008–12 tems for effective policy development, NFRA Irrigation monitoring, and evaluation. Successful WFP cash transfer Extension services implementation of any risk-management instru- WFP food aid Research 160 ment depends on the ability to monitor the impacts 140 of risks and to evaluate the effectiveness of poli- 120 cies and investments. A solutions assessment in this area would (i) map out measures to strengthen USD (millions) 100 Malawi’s agricultural information systems so that 80 they contain reliable data for the development, 60 monitoring, and evaluation of agricultural poli- 40 cies; and (ii) propose measures to strengthen the 20 policy analysis and monitoring and evaluation capacity in the Ministry of Agriculture and Food 0 Mitigation Losses Coping Security (MAFS). An assessment could comprise Source: World Bank Ag. Public Expenditure Review 2014; National Food Reserve the following: Agency; authors’ calculations. » Identification of gaps in the current agricul- Note: Losses reflect average annual production losses from 1980–2012 according tural information system in terms of collection to the above calculations. Total losses would amount to US$103.5 million if cassava were excluded from the analysis. methods and management of data. » Assessment of existing equipment and infor- mation technology and a proposal for potential the losses from risks. Reallocating funds to risk-mitigating investments in agricultural information systems activities thus represents potentially large savings in terms to strengthen agricultural policy development of losses and coping activities. and evaluation. » Discussion of the technical skills needed to During the risk-assessment mission, a consultative stake- monitor and evaluate agricultural policies, and holder meeting was organized to solicit feedback on the areas for strengthening these skills within rel- long list of solutions from private and public sector stake- evant departments of MAFS. holders. Participating stakeholders were asked to grade 2. Implement measures to improve water the proposed solutions according to their alignment with management for crop production to miti- policy or business objectives; feasibility of implementa- gate current and projected future weather- tion in Malawi; affordability for the implementing party related risks. Given the farm structure in (whether public or private); potential for scaling up; and Malawi, with its large number of small-scale sustainability. The feedback from stakeholders and a gap farmers, water management will in part have to be analysis of already ongoing interventions were then used implemented through small-scale infrastructural to narrow the proposed solutions to a short list for a solu- investments and improved on-farm practices using tions assessment. a systems approach. Any analysis would have to be conducted with existing land use/ownership Myriad ongoing projects are already studying agricultural structure in mind. An assessment could comprise risks in Malawi. The goal of the solutions assessment is one or several of the following areas: Malawi: Agricultural Sector Risk Assessment xvii » The potential for expanding the use of small- inputs and technology to inadequate investments scale irrigation in Malawi and possible mod- in postharvest infrastructure, price uncertainty, els under which small-scale irrigation could be and contractual risks) could potentially be over- promoted. come through better organization of farmers. » The scope for improving relevant on-farm This intervention area is proposed to include the practices, including conservation agriculture following: and minimum tillage methods. » An assessment of existing farmers’ organiza- » The application of models for investing in on- tions (formal and informal) in Malawi. farm water harvesting infrastructure in the » A compilation of lessons learned from past context of Malaw’s agriculture sector. and ongoing initiatives to organize farmers in 3. Map existing functions and identify measures to Malawi, successful and unsuccessful, and con- improve coordination between the Stra- clusions about what determines their success. tegic Grain Reserve (SGR), Agricultural » Guidance on how farmers’ organizations can Development and Marketing Coopera- implement risk-management mechanisms in tion (ADMARC), and Malawi Vulnerability practice, focusing on a few specific areas (such Assessment Committee (MVAC) to better as adoption of new technology, price risks, con- target existing coping mechanisms toward their tractual risks, and so on). intended beneficiaries, to improve predictability of interventions, and to minimize market distor- Which of these areas will be included in a solutions assess- tions. Such work could include the following: ment will be determined together with the government » An outline of the roles and responsibilities (for- of Malawi. Ideally, the assessments will be conducted in mal and de facto) of SGR, ADMARC, and teams including relevant technical staff from the MAFS MVAC and proposed measures to strengthen and other technical bodies to ensure that the analyses and their coordination. proposed solutions are in line with the priorities and needs » An assessment of food security policies, includ- of the Ministry and/or relevant institution, and that the ing those related to trade, market interventions, knowledge acquired through the assessment remains with and grain subsidies. relevant staff. Preferably, any work will include gender- » An analysis of the financial costs and economic disaggregated assessments and proposals. impacts of these policies and if relevant, a proposal of alternative policies that can more This activity was requested by the Group of Eight (G-8) efficiently achieve the same objectives without and principally financed by the U.S. Agency for Interna- market distortions. tional Development and Feed the Futures programs. Con- 4. Provide opportunities to strengthen farm- tributions were also received by the Multi Donor Trust ers’ organizations for effective agricultural Fund on risk management, financed by the Dutch Min- risk management. Many of the challenges istry of Foreign Affairs and the Swiss Secretariat of Eco- in the sector that relate to risks (from uptake of nomic Affairs. xviii Agriculture Global Practice Technical Assistance Paper CHAPTER ONE INTRODUCTION AND CONTEXT With more than three-quarters of its workforce employed in agriculture, Malawi is highly vulnerable to any adverse events affecting the agriculture sector, and agricul- tural risks are ever present in the country. Over the past decades, Malawi has been struck by several severe droughts that have resulted in spikes in food insecurity and prompted the need for humanitarian aid. During the last major drought in 2005, 40 percent of the population was in immediate need of food aid as a result of poor harvest. Increasing food security is indeed one of the main objectives of Malawi’s Agricultural Sector Wide Approach (ASWAp 2010), and its strong focus on increasing maize pro- duction since the mid-2000s has resulted in rapidly increasing production. However, production risks continue to result in high losses to the sector, including for maize. Further, price interventions in the sector over the past year have induced greater price risks for producers and traders. Agricultural risks can obstruct development and enforce poverty traps, particu- larly for a country as reliant on agriculture as Malawi. Because of the size of the sector in the economy and the importance of agricultural products for export, agricultural growth correlates closely with GDP growth. This means that drops in agricultural growth affect the entire economy, as shown in figure 1.1—agricultural value added growth was negative in five years between 1992 and 2010, and the correlation coefficient between agricultural value added and GDP is 78 percent. Further, any drop in agricultural growth in a given year will affect the ASWAp annual growth target of 6 percent that Malawi committed to under Comprehen- sive Africa Agriculture Development Programme (CAADP). For individual actors in the sector, these risks reinforce poverty traps by cycles of loss-recovery-loss and result in lower returns on investments in productive assets. Malawi’s effort to manage risks and to provide relief in response to adverse events diverts significant resources from longer-term development investments. In recent years, the government and donors have spent US$80–US$100 million annually on coping mechanisms alone (such as food aid). This was in addition to the approximately Malawi: Agricultural Sector Risk Assessment 1 FIGURE 1.1. GDP AND AGRICULTURAL VALUE ADDED (% GROWTH) IN MALAWI, 1968–2011 60 GDP growth (annual %) Depreciation 50 Agriculture, value added (annual % growth) currency (by 40%) 40 Uneven rainfall and 30 depreciation of maize prices from market 20 interventions 10 0 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 –10 –20 Drought Drought Political –30 instability, donors exit –40 Drought Source: WDI 2014. US$250 million spent on average annually between 2008 of the impacts of risks, the assessment looks at the larg- and 2012 on regular government agricultural develop- est commodities that jointly account for 80 percent of ment programs, including the government’s Farm Input Malawi’s agricultural production value: Subsidy Program (FISP), which annually distributes Food crops: maize, cassava, potatoes, peas and beans, rice, inputs worth US$165 million to farmers. Any losses in the groundnuts, and bananas sector because of adverse events mean that these invest- Export/cash crops: tobacco, sugar, tea, and cotton ments were wasted. Maize is by far the most important staple crop, account- Improved agricultural risk management is one of the core ing for more than 50 percent of the daily calorie intake enabling actions of the G-8’s New Alliance for Food Secu- in Malawi. Tobacco, tea, cotton, and sugar accounted for rity and Nutrition. To better understand the dynamics 67 percent of the total value of national exports of goods of agricultural risks and identify appropriate responses, in 2012, with tobacco alone accounting for more than incorporate an agricultural risk perspective into decision 54 percent. It can be noted that tea and cotton do not making, and build the capacity of local stakeholders in risk belong to the largest crops that fall within the 80 percent assessment and management, the Agricultural Risk Man- threshold, but tea was included in the list of crops prior agement Team (ARMT) of the Agriculture and Environ- to the mission because of its contribution to total agri- ment Services Department of the World Bank conducted cultural export, and cotton because of its potential as an an agriculture sector risk assessment. This activity was export crop.3 requested by the G-8 and principally financed by USAID and Feed the Futures programs. Contributions were also The report takes a quantitative and qualitative approach received by the Multi Donor Trust Fund on risk manage- to assessing risk. Productions risks are quantified in terms ment, financed by the Dutch Ministry of Foreign Affairs of value of losses and then mapped by different perils. and SECO. Market and enabling environment risks are analyzed qualitatively through deskwork and stakeholder consulta- The purpose of this report is therefore to assess existing tions. For the purpose of this assessment, risk is defined as agricultural risks, prioritize them according to their fre- the possibility that an event will occur and will potentially quency and impacts on the sector, and identify areas of risk-management solutions that need deeper specialized 3 Although the sector in total makes up about 10 percent of total agricultural attention. Three levels of risks were assessed: produc- production value, livestock were not included in the assessment because no tion risks, market risks, and enabling environment risks single livestock product falls within the top 80 percent production value. Fishing to selected supply chains. To give a sectorwide overview and forestry were not included in the assessment. 2 Agriculture Global Practice Technical Assistance Paper FIGURE 1.2. AGRICULTURE SECTOR RISK-MANAGEMENT PROCESS FLOW PHASE I PHASE 2 PHASE 3 PHASE 4 Client demand Risk Solution Development of risk Implementation and assessment assessment management plan risk monitoring RM plan development Desk review Desk review Implementation by stakeholders Stakeholder In-country Monitoring risks consultations assessment mission Incorporation into existing govt. programs and Stakeholder development plans Finalize analysis Refining RM strategy workshop have a negative impact on the achievement of a farm or areas will be assessed and gaps mapped to determine firm’s performance objectives and/or on successful func- activities needed to minimize the impacts of risks on tioning of the overall supply chain. A broad spectrum of the sector. stakeholders was consulted throughout this work, includ- ing the Malawi government, farmers, traders, processors, This report is structured as follows: Chapter 2 provides an agricultural institutions, and academia. A consultative overview of the agriculture sector and the selected crops. stakeholder meeting was also held in Lilongwe to obtain Chapter 3 maps the production, market, and enabling feedback on findings and to discuss areas for risk solution environment risks to food crops and export crops. Chap- interventions for deeper analysis. ter 4 looks at the adverse impacts of agricultural risks in terms of losses, both at the national level and for different Figure 1.2 provides an overview of the full process the regions. It also discusses the impacts of risks on different World Bank’s ARMT has applied in the past. The Agri- stakeholders and identifies particularly vulnerable groups. cultural Sector Risk Assessment constitutes the first Finally, chapter 5 prioritizes the risks in terms of their fre- phase. Based on its results, a solutions assessment will quency and the severity of their impacts, and discusses be conducted under which a few potential risk-man- solutions based on this prioritization, ongoing risk-man- agement instruments will be further assessed. Under agement activities, and the feedback from the consultative this second phase, ongoing activities in the selected workshop. Malawi: Agricultural Sector Risk Assessment 3 CHAPTER TWO MALAWI’S AGRICULTURAL SYSTEM AGRICULTURE SECTOR OVERVIEW AND PERFORMANCE Malawi is among the poorest countries in the world, with limited resources and an economy that relies heavily on agriculture. Per capita GDP is US$362 per year4 (WDI 2014) and 62 percent of the population lives on less than US$1.25 per day (purchasing power parity [PPP]). Malawi is relatively small in size, is densely popu- lated, and has high population growth, all of which put pressure on land available for smallholder farming and on the environment and the natural resource base, notably land and forests. Officially, the population is 15.9 million people, about 80 percent of whom live in rural areas (World Development Indicators, 2010–12 year figures, accessed March 2014). Agriculture is the backbone of Malawi’s economy, contributing 30 percent of total GDP (2011) and 76 percent of total national exports (2012). With 78 percent of the population employed in the sector in 2013 (FAO Country Profile, accessed May 2014), agriculture is a main source of employment and income. The variability of agriculture has been a determinant of the overall economy’s volatility (recall figure 1.1). For exam- ple, in years when agriculture suffered significant setbacks because of weather-related or other risk events, GDP growth also experienced an inflexion. In figure 1.1, this can be seen in 1992 and 1994, when severe drought caused significant drops in agricultural production that translated into in negative GDP growth rates. More recently, such cor- relations occurred in 2001, 2005, and 2010. In terms of production indexes, the gross cereal production index shows a lot more volatility than do food production and total agricultural production indexes. The cereal production index in figure 2.1 also corresponds with the fluctuations in agricul- tural value added growth (figure 1.1). This is in line with findings from other studies that Malawi’s GDP is strongly correlated with maize production. 4 Current US$, 2010–11 average. Malawi: Agricultural Sector Risk Assessment 5 FIGURE 2.1. GROSS CEREAL PRODUCTION FIGURE 2.3. SHARE OF AREA HARVESTED INDEX (2004–06 = 100) IN FOR COMMODITIES THAT MALAWI, 1968–2012 MAKE UP THE TOP 80 250 PERCENT OF GROSS 200 AGRICULTURAL PRODUCTION 150 VALUE (2009–11 AVERAGE) 100 Tea Rice 1% Cotton 2% 2% Cassava 50 Bananas and plantains Pigeon peas 6% 0 0% 6% 19 8 19 0 72 19 4 76 19 8 19 0 82 19 4 19 6 88 19 0 19 2 19 4 19 6 20 8 20 0 02 20 4 20 6 20 8 10 12 6 7 7 7 8 8 8 9 9 9 9 9 0 0 0 0 19 19 19 19 19 20 20 Beans Source: FAOSTAT 2014. 9% Groundnuts 11% Maize FIGURE 2.2. COMMODITIES THAT MAKE 51% Sugar cane UP THE TOP 80 PERCENT 1% Tobacco OF GROSS AGRICULTURAL 5% Potato PRODUCTION VALUE (2009–11 6% Source: FAOSTAT; cotton area data are from MAFS. AVERAGE) Cotton (lint+seed) 0.5% In terms of land area, maize takes up the largest area Rice (paddy) Tea (more than 1.6 million hectares in 2012), followed by Other 2% 2% 17% Cassava groundnuts and beans (more than 300,000 hectares). Bananas and 24% Tobacco is the most important export crop in terms of plantains 3% area planted, with about 160,000 hectares (figure 2.3). Pigeon peas 3% Maize 17% Malawi has a dual structure of production whereby Beans, dry 3% the smallholder subsector is the major producer of Groundnuts food crops, especially maize, cassava, potatoes, beans, (with shell) Sugar cane 4% 4% and peas, whereas large estates specialize in export Tobacco crops such as tea and sugarcane. Tobacco was for- (unmanufactured) 5% Potato merly in the hands of estates but following the policy 16% reforms during the 1990s, it became a mostly small- Source: Calculation of production value based on FAOSTAT data. holder activity. Other export crops, such as cotton and groundnuts, have traditionally been produced in smallholder farming. Food crops account for the largest proportion of agri- culture sector production, and three crops—maize, cas- The marketing channels for food crops differ from those sava, and potatoes—contribute over half of the total of export crops. Food crop markets are for the most value. Figures 2.1 and 2.2 show the relative importance part informal and farmers often depend on traders or of specific agricultural products in terms of production transporters who come to villages and buy their pro- value and harvested area. Maize is the main staple food duce. Farmers are also restricted by limited means of for most people in rural and urban areas and is cultivated transportation, even though they are aware of better almost everywhere. Most of the other food crops and cer- prices at bigger markets around the country. Maize dif- tainly cash crops have a relatively well-defined geographic fers from other food crops in that the government-owned production location. ADMARC participates in the market, buying and selling 6 Agriculture Global Practice Technical Assistance Paper FIGURE 2.4. AGRICULTURAL EXPORTS FIGURE 2.5. AGRICULTURAL EXPORTS AND CONSTANT GDP (US$ FROM MALAWI, 2012 ’000), 2001–12 Cotton, carded or combed GDP (constant 2005 US$ 000) 1% 4,000,000 Other Agricultural exports (US$ 000) 3,500,000 Nuts nes 24% Live poultry 1% 3,000,000 2% 2,500,000 Tobacco 2,000,000 unmanufactured; Dried vegetables, tobacco refuse 1,500,000 shelled 54% 1,000,000 2% 500,000 Cotton, not carded or combed Ground-nuts, not 0 3% Tea roasted 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 6% 3% Cane or beet sugar and Source: World Bank and International Trade Centre. chemically pure sucrose, in solid form 4% Source: International Trade Centre. large quantities of maize throughout the season. Cash crops, on the other hand, have more formal supply chains and limited actors after farm gate. The extreme case is and unprocessed. India is the only export market, where the sugar sector, in which only one processor operates in Malawi has captured a high-price window because of Malawi (Illovo Sugar Malawi Ltd., owned by Associated seasonal advantage. However, pigeon peas are still not an British Foods, the biggest sugar producer in Africa). The important export crop. number of purchasers is also relatively small at the tea and tobacco auctions. The maize and tobacco subsectors face the highest lev- els of government policy intervention. Malawi’s main Although food crops (mainly maize, cassava, and potatoes) producer support program is the FISP, which subsi- account for the largest proportion of total agricultural dizes seeds, fertilizers, and certain chemicals for maize, production value and cultivated area, export crops have legume, and cotton (Makoka 2013a; see box 2.1). been the main drivers of economic growth. In 2012, agri- Additional policy interventions include maize export cultural export accounted for 76 percent of total export licensing and maize export bans. The tobacco market from Malawi. Figure 2.4 shows GDP and agricultural is extensively regulated but government intervention export trends over the past decade. is transparent and more predictable than in the maize sector. However, agricultural export is strongly dominated by a few products, mainly tobacco and tea, followed by AGROCLIMATIC sugar, groundnuts, and cotton. Exports of tobacco and tea accounted for 60 percent of total exports in 2012 CONDITIONS (figure 2.5). Until May 2012, when the Malawi kwacha Five main landform areas exist in Malawi: the highlands, (MK) was left to float against the U.S. dollar, exchange the escarpments, the plateaus, the lakeshore and Upper rate policy affected the country’s export competitiveness Shire Valley, and the Lower Shire Valley. The climate because of overvaluation of the local currency. changes from semi-arid in the Lower Shire Valley to semi- arid and subhumid on the plateaus to subhumid in the Malawi is the one of the largest producer of pigeon peas in highlands. Most of the country receives between 763 and eastern and southern Africa. Production is concentrated 1,143 mm of precipitation per year. Three main areas have in the southern region where they account for approxi- precipitation of more than 1,524 mm: Mulanje, Nkhata mately 20 percent of household income. About 35 per- Bay, and the northern end of Lake Malawi (map 2.1). cent of production is sold on the market, both processed Almost 90 percent of rainfall occurs between November Malawi: Agricultural Sector Risk Assessment 7 BOX 2.1. MALAWI’S FARM INPUT SUBSIDY MAP 2.1. AVERAGE PROGRAM ANNUAL In an attempt to boost production and increase food secu- PRECIPITATION rity, Malawi introduced an input subsidy program in 2005. (mm) IN MALAWI The purpose of FISP is to increase smallholder farmers’ access to improved agricultural farm inputs with the objec- tive of achieving food self-sufficiency and increased income for resource-poor households through increased maize and legume production. FISP also subsidizes certain posthar- vest infrastructure to decrease postharvest losses. FISP has since accounted for more than 50 percent of the Ministry of Agriculture and Food Security’s (MAFS) budget. The program subsidizes fertilizers, maize and legume seeds, and, in certain years, cottonseed and chemicals. Under the program, farmers receive vouchers that cover a share of the input cost. The number of vouchers went from 166,000 in 2005/06 to 216,000 in 2008/09 and to 140,000 in 2012. Maize seed subsidized under the program went from 4,524 MT in the 2006/07 season to 8,245 MT in 2011/12. For fertilizers, the biggest component of FISP, farmers’ contribution declined from MK 950 per bag in 2005 to MK 500 per bag in 2012, whereas the value of the voucher increased from MK 1,750 per bag to MK 6,536 per bag in the same period. The actual results of the program are mixed. According to a recent World Bank evaluation of the program, FISP has Source: Moriniere and Chimwaza 1996. had only a moderate impact on yields, prices, and agricul- tural wages. One possible reason is that maize in Malawi has low response rates to fertilizer is relatively low. Other reasons are that fertilizers are shared and therefore not TABLE 2.1. LAND USE IN MALAWI (km2) optimally applied, and vouchers are resold and therefore Total Land Area (km2) 94,281 % of Total do not have the intended effect on targeted farm house- Agricultural land (km2) 55,720 59 holds. Consequently, impacts among higher-income farm- Arable 21,174 38 ers can be linked to FISP. Nevertheless, many farmers have the perception that FISP contributes positively to the well- Forested 18,945 34 being of their households. Permanent crops 557 1 Permanent pastures 11,144 20 Sources: Makoka 2013a; World Bank 2013. Other 3,900 7 Source: FAO/WDI. and March, with no rain at all between May and Octo- ber over most of the country. Mean annual temperatures Map 2.2 shows the land cover in Malawi and its evolution vary with altitude, ranging from 25°C in the Lower Shire over the past 30 years. Forestland has reduced extensively Valley to 13°C on the Nyika Plateau. Frost occasionally whereas the area dedicated to agricultural crops has occurs in lower lying land on the plateaus. increased. This change is certainly connected to Malawi’s high population growth and density, and is a main con- Forty percent of the total land area in Malawi is suitable tributor to increased production risks and reduced human for agriculture, as shown in table 2.1 (based on data for resilience. As such, it is a key long-term issue for public 2000). policy. 8 Agriculture Global Practice Technical Assistance Paper 2005). The long-term increase is attributed to government PRODUCTION AND MARKET interventions through programs such as the Agricultural TRENDS Productivity Investment Program (APIP), the Starter The yields of Malawi’s main crops have followed very Pack Scheme, the Targeted Inputs Program (TIP), and different trends, depending on public policies and mar- the FISP. In any case, actual maize yields remain below ket developments over the past 30 years. Maize yields potential yield. Tobacco yields have also increased over have increased, though at a very modest rate and with the years, peaking at almost 1,400 kilograms (kg)/hectare great drops due to droughts (for example, in 2001 and (ha) in 1997. Yields then declined and leveled off, with year-to-year variations affected by weather and farmers’ access to fertilizer. Other main food crops cropped by MAP 2.2. EVOLUTION OF LAND COVER IN smallholders, such as groundnuts and beans, which have MALAWI, 1973–2010 the largest cultivated area after maize, experienced a slow decline in yield, most likely associated with the low avail- ability of fertilizer and other inputs (figure 2.6). In terms of market trends, some crops, such as cotton and maize, are marked by intense price volatility. Maize has a relatively thin and poorly functioning market, a major cause of high seasonal variation. Interannual price varia- tion is mostly connected to uncertain public policies and irregular access to modern production inputs, which in turn have led to limited productivity growth. Cotton price volatility is connected to international market volatility. A crop such as tobacco, which has a relatively well-devel- oped internal market and a relatively efficient technology transmission mechanism, is less exposed to production risks but is very sensitive to domestic supply and demand variation. These issues are discussed in chapter 3 within Source: LTS International 2013. each supply chain’s market risk assessment. FIGURE 2.6. YIELD OF SELECTED CROPS IN MALAWI, 1961–2011 Maize Groundnuts, with shell 25,000 Beans, dry Lineal (Maize) Lineal (Tobacco, unmanufactured) Lineal (Beans, dry) Tobacco, unmanufactured 20,000 Lineal (Groundnuts, with shell) Hg/Ha 15,000 10,000 5,000 0 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 Source: FAOSTAT 2014. Malawi: Agricultural Sector Risk Assessment 9 CHAPTER THREE AGRICULTURE SECTOR RISKS Droughts and pests and diseases are cited by Malawian stakeholders as the most dam- aging production risks, especially for food crops. Droughts are probably the most vis- ible risk to the sector; very bad droughts in the past have had a strong fiscal impact on Malawi, necessitating help from the international community. The damaging impact of pests and diseases is significant but the extent of damage depends on agricultural practices and mitigation activities. The effects of pests and diseases are at times exac- erbated by adverse weather events. Erratic rainfall and hailstorms are frequent but of moderate or low impact. Price volatility is an important market risk in Malawi, particularly in key crops such as maize, tobacco, and cotton. Causes for these volatilities depend on the crop: cot- ton prices fluctuate according to world prices, whereas tobacco and maize prices are mainly determined by the domestic market. Maize price volatilities are largely a result of enabling environment risks due to unpredictable domestic market interventions and export bans. Regardless of the reason, sudden fluctuations in prices negatively affect farmers, the segment of the supply chain with the least risk-management capacity. This chapter presents findings regarding the production, market, and enabling envi- ronment risks for selected food and export crops. The impact of adverse events on different stakeholders is discussed in chapter 4. FOOD CROPS—PRODUCTION RISKS WEATHER-RELATED RISKS Weather-related risks such as droughts, dry spells, and erratic rains constitute some of the most important risks to the sector, although they are more predictable than they might seem at first glance. Drought in Malawi happens in a number of different ways, notably in shortened rainy seasons (because of late starts, early cessation, or both) and/ or dry spells during the rainy season. Although these weather events often come as a shock to producers, there are certain patterns in their occurrence. The short cycles or waves of weather patterns are affected by so-called teleconnections, especially El Niño and La Niña. Teleconnections are linkages between weather variations or anomalies Malawi: Agricultural Sector Risk Assessment 11 in widely separate locations of the world that bring about harvested then. Predictability of the rainy season is thus temporary changes over a one- to two-year time frame. important for cassava producers. El Niño events are strongly connected with drought in Malawi, whereas La Niña is associated with unusually wet The impact of shorter rainy seasons and extended dry years. If there is an El Niño event, the following growing spells on maize depends on the maize variety. Broadly, season in Malawi is highly likely to experience a signifi- three different maize varieties are currently used in cant drought.5 Other teleconnections also affect Malawi’s Malawi: traditional, hybrid, and composite. Hybrids and weather patterns. composites are being promoted by the government and donors. Traditional varieties are particularly susceptible Further, in the medium term, analyses of rainfall data to shortened rainy seasons because they require a long have shown that Malawi goes through different multiyear growing season. Hybrids are considered drought toler- cycles of wet and dry periods. The climate in Malawi ant because they have shorter growing seasons, and thus alters between 11.1-year cycles with precipitation above can still produce normal yields even if the rainy season is average and precipitation below average (Mwafulirwa shorter than normal. However, because they are less able 1999). One theory is that this longer oscillation is related to absorb soil moisture, these types of drought-tolerant to regular changes in sunspot activity, but this has yet to varieties are typically sensitive to prolonged dry spells. be confirmed. Some stakeholders were of the opinion They also don’t cope well with high temperatures, as they that weather has become more unpredictable over the have been developed for other climates. Another challenge past two decades, with drought events more frequent and with hybrids is that they require fertilizer, which farmers intense and with more frequent floods with more severe often find prohibitively expensive. The drought resilience impacts in certain parts of the country. (More informa- of composites is not clear, although some studies report tion on weather cycles and climate change can be found that they are the most drought tolerant of the three varie- in appendix A.) ties. However, although composites possess some of the traits of hybrids, the seeds can be recycled and are there- Despite this, most food crops grown in Malawi are not fore popular among farmers in Malawi. The traditional particularly drought tolerant and are therefore sensitive varieties cope better during extended dry spells and with to dry spells and erratic rains. Irish potatoes, groundnuts, higher temperatures because they have adapted to local beans, and bananas are all susceptible to dry spells. For conditions over time. However, traditional maize varieties groundnuts, farmers reported losing more than half their require a full growing season and are therefore not con- harvest in a dry season in 2012. Drought-tolerant varieties sidered drought tolerant. Hence, the effect of drought on exist for groundnuts, but are not widely adopted by farm- maize depends on the type of drought (that is, extended ers, in part because of limited access and in part because dry spell, less rainfall, or shorter than normal growing sea- of the timing of harvest, which overlaps the harvest of son), and the variety planted. Chapter 4 shows that the other crops, making sufficient labor unavailable. Few impacts of production risks are incurred significantly dif- drought-resilient varieties exist for potatoes. Although ferently across regions in Malawi. banana plants are also sensitive to drought, the banana- growing zones are located in areas with higher annual Table 3.1 shows the main droughts experienced in Malawi rainfall and with more rainy days than elsewhere in the during the past 30 years. Figures 3.1–3.5 show how these country; further, banana farming is often conducted close droughts affected different food crops. As can be seen, to rivers and streams, so bananas tend to be fairly drought maize and groundnuts show great fluctuations in yield as resistant. Cassava is relatively drought tolerant, but inter- a result of drought. viewed farmers reported losses of more than 50 percent of cassava in dry years. In addition, cassava loses qual- Bananas show a more stable yield trend, which is partly in ity during rainy periods, and has a lower market price if line with the above discussion. However, this stable yield can also be questioned on the grounds of data quality, 5 According to Mwafulirwa (1999), the likelihood is 80–90 percent. particularly given the sharp jump in yield between 1998 12 Agriculture Global Practice Technical Assistance Paper TABLE 3.1. MAJOR DROUGHT INCIDENTS IN MALAWI, 1980–2012 November– Crops Affected Total # of March According to Year Start Month People Affected Rainfall (mm) Region(s) Affected Yield Trends 1991 Information not No data 696 8 total: 2 in north, 3 in central, Maize available 3 in south Potatoes 1992a April 490 21 total: 5 in north, 6 in central, Groundnuts and 10 in south Maize 1994 Information not 583 17 total: 5 in north, 3 in central, Groundnuts available 7,000,000 and 9 in south Maize Potatoes 1995 Information not 585 17 total: 4 in north, 3 in central, Groundnuts available and 10 in south Potatoes 2002a February 2,829,435 No data Information not available Beans Maize 2005–06a October through 5,100,000 754 (for 2005 11 total: 4 in north, 2 in central, Beans March event) and 5 in south Groundnuts Maize Potatoes 2012 August 1,900,000 No data Information not available Information not available Sources: EM-DAT, The International Disaster Database, Centre for Research on the Epidemiology of Disasters-CRED (http://www.emdat.be/search-details-disaster- list), RMSI, World Bank 2009; and appendix A of this report. FIGURE 3.1. MAIZE YIELDS (MT/ha), FIGURE 3.2. GROUNDNUT YIELDS (MT/ha), 1980–2012 1980–2012 3.0 1.2 Drought years: 1992,1994, 1995, 2005 Drought years in 1990, 2.5 1992, 1994, 2002, 2005 1.0 2.0 0.8 1.5 0.6 1.0 0.4 0.5 0.2 0 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Source: FAOSTAT 2013. Source: FAOSTAT 2013. and 1999. Because improved productivity cannot explain prior to and after 1998/99. As the regional analysis will this jump, it is reasonable to assume that there was a cor- show, national yield data do not capture the full picture of rection in the data. For the purpose of this report, banana banana production in Malawi. yields are given two trend lines because a single line would give the illusion of losses over seven years. In reality, and Beans show essentially no variation in yield up to the late as the dual trend lines show, yield remained flat both 1990s, when volatility increased. Based on the remarkably Malawi: Agricultural Sector Risk Assessment 13 FIGURE 3.3. POTATO YIELDS (MT/ha), BOX 3.1. GLOSSARY OF DROUGHT EVENTS 1980–2012 The word “drought” is commonly used when referring to a 20 Drought years: 1991, 1994, 1995, 2005 deficiency in precipitation in a certain period, but the way 18 16 in which this event occurs determines the impact. Three 14 such ways include the following: 12 10 Dry spell: A cessation in rainfall in a normally rainy sea- 8 son. Dry spells can be short or long, and their length will 6 determine the impact on the crops. They are especially 4 problematic for crops with poor ability to absorb moisture 2 0 in the soil. Similarly, the time at which they occur in the growing cycle of the crop will also determine the dam- 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 age because crops are differentially vulnerable in different Source: FAOSTAT 2013. stages of maturity. Dry spells are also sometimes referred to as erratic rains. FIGURE 3.4. BANANA YIELDS (MT/ha), Late onset or early cessation of the rainy season: 1980–2012 The rainy season starts later than normal or ends earlier 30 than normal, which affects the overall length of the rainy 25 season. Traditional crops are normally adapted to the nor- 20 mal local rainy season and therefore do not have time to 15 mature in this event. 10 High temperatures: Normal rainfall but temperature is 5 higher than normal. Global climate change models proj- 0 ect that temperature will increase in Malawi, and thus 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 incidents of high temperature will be more frequent in the future. High temperature is problematic for crops with low Source: FAOSTAT 2013. stress tolerance. FIGURE 3.5. BEAN YIELDS (MT/ha), 1980–2012 Table 3.1 provides an overview of the main droughts in 1.2 Malawi in terms of the number of affected people and the 1.0 impacts on various crops’ production. However, droughts 0.8 Drought years: 2002, 2004, 2005 can be measured according to different variables and their impacts depend on when in the season they occur, as this 0.6 can affect agriculture differentially. Table 3.1 is there- 0.4 fore not an exhaustive list of all droughts but reflects the 0.2 reported droughts’ impacts on the agriculture sector. (For a discussion of the various concepts associated with the 0 term “drought” in Malawi, see box 3.1.) 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Source: FAOSTAT 2013. PESTS AND DISEASES Pests and diseases are an important problem in Malawi stable yield from 1980 to 1998, it is reasonable to assume and although the exact figure is not known, a significant that yield data from this period were based on assump- share of food crops is lost annually as a result. The Min- tions rather than on actual yields. Nevertheless, three of istry of Agriculture and Food Security maps outbreaks to the drops in yield in the 2000s coincide with three main a certain extent, but it does not capture the full impact of drought years in Malawi. pests and diseases on the sector. Also, pests and diseases 14 Agriculture Global Practice Technical Assistance Paper TABLE 3.2. PESTS AND DISEASES IN MALAWI FOR ANALYZED FOOD CROPS, IN FIELD AND POSTHARVEST In the Field Pests Diseases Postharvest Maize • Stalk borers • Maize streak virus • Maize weevils • Gray leaf spot • Larger grain borers • Rust • Striga • Southern leaf blight • Whitegrubs • Wireworms • Termites Cassava • Cassava green mite • Cassava mosaic virus disease • Cassava mealy bug • Cassava bacterial blight • Termites • Cassava brown streak virus disease Potatoes • Weevils • Aphids • Nematodes • Bacterial wilt • Late blight Bananas • Banana weevils • Banana bunchy top virus • Nematodes • Fusarium wilt (Panama disease) • Black Sigatoka • Yellow Sigatoka Groundnuts • Whitegrubs • Rosette • Bruchids • Groundnut hoppers • Early leaf spot disease • Pod-sucking bugs • Termites • Late leaf spot disease Beans • Aphids • Bacterial blight • Beanflies • Angular leaf spot • Leaf beetles • Bacterial brown spot • Halo blight • Anthracnose, rust • Bean common mosaic virus Pigeon peas • Nematodes • Fusarium wilt (Panama disease) • Pod borer Helicoverpa armigera • Pod sucker Nezara viridula • Termites Sources: Monyo et al. 2012; Mih and Atirib 2003; Ngwira and Khonje 2005; and Authors’ interviews with stakeholders. are often closely tied to adverse weather events that exac- Pests and diseases are a problem for essentially all food erbate the impacts, which can make it difficult to attribute crops. Table 3.2 provides an overview of the plant pests losses to the different risks. Nevertheless, research and and diseases in existence in Malawi. As can be seen, vir- interviews with farmers show that pest and disease out- tually all food crops are subject to a variety of pests and breaks are regular occurrences in Malawi and that farm- diseases although some are more common than others. ers lose about 20–30 percent in the event of an outbreak. For example, a combination of bacteria and pests has In the worst cases, farmers can lose an entire harvest, spread between potato producers across Malawi and as from rosette diseases for groundnuts and the banana it is estimated that in some areas, up to 60 percent of bunchy top virus (BBTV). potatoes are contaminated. Similarly, it is estimated Malawi: Agricultural Sector Risk Assessment 15 that about 60 percent of Cavendish banana plants are because they often involve damaged infrastructure and infected by BBTV. Certain pests and diseases are also buildings, loss of livestock, and sometimes even loss of more localized than others. For maize, for example, human life. However, floods in Malawi are usually lim- southern leaf blight, rust, and stalk borers tend to be ited to a narrow geographic area and tend to not have problematic in middle elevation areas, whereas maize any visible impacts even on a regional level. Since 2007, streak virus is more of a problem at low elevations. Ter- only three flood events in Malawi have affected more mites pose a problem to maize and pigeon peas as they than 1,000 households. Farmers and other stakehold- will feed on plant residue. ers interviewed about major risks to agricultural pro- duction did not mention floods. The main exception is Agricultural practices significantly affect the occurrence around the Shire River and in other areas around rivers. of pests and diseases in Malawi. For example, plant dis- Land scarcity and the greater fertility of land on river eases are commonly transferred from harvest to harvest banks have encouraged farmers to cultivate areas close in vegetatively propagated crops such as cassava and to rivers, where flooding frequently occurs. However, potatoes, as farmers cannot afford to buy new and/or because this flooding can be expected (indeed, authori- certified seeds. Instead of culling diseased potato plants ties discourage farmers from taking this land under from their fields to mitigate the effect of potato diseases, cultivation), it should not be seen as a risk. farmers respond by harvesting potatoes earlier to avoid the rot spots that develop with the diseases and present toward the end of the growing season. However, this ANIMALS Elephants and hippos frequently damage harvests in the results in lower yields because there are smaller pota- field. This is especially problematic for farmers close to toes, which are priced less in the market. And the main national parks. Animals either cross fields and trample the reason for the spread of BBT is poor farmer practices, crops in their paths or they enter fields to eat the crops. For as farmers do not immediately remove infected banana example, in 1999, 369 households in the T.A. Chimwala plants when identified. Although diseases are gener- area had their crop damaged by elephants and became ally not a problem for pigeon pea producers, a high malnourished as a result. In 2005, elephants destroyed the incidence of Fusarium wilt occurs when farmers grow crops of 142 families in Machinga. Farmers in Mchinji pigeon peas in the same plot year after year. Finally, the estimate that about 10 percent of their crops are lost as improved bean varieties were all bred to be resistant a result of animals; they have limited options for protect- to one or more bean diseases, but farmers opt not to ing their crops, although farmers did report some coop- replace their seeds with new varieties. eration with the park services. However, although this is a Pests and diseases also pose a risk postharvest. Inadequate problem for individual farmers, it cannot be considered a infrastructural capacity along with lack of knowledge of structural risk. preventative storage methods result in stored grain being subject to pest infestations. For maize, for example, post- harvest losses from large grain borers and maize weevils FOOD CROPS—MARKET can be as high as 30 percent. Hybrid varieties are more RISKS pest prone during storage than other maize varieties, THE PRODUCTION OF OTHER CROPS complicating farmers’ possibilities to mitigate the effects Because of substitution effects, cassava prices tend to be of drought. affected by maize prices. Some cassava is milled into flour, which is a cheaper substitute for wheat and maize flour, FLOODS and sometimes it is mixed in with maize flour. As such, Floods are relatively frequent and problematic on a local when maize production is high and its price declines, so level, but do not constitute a structural risk to agricultural does the market price for cassava. When there is a short- production. Floods are frequently mentioned in moni- fall in maize production and its price rises, so too does the toring documents related to risk and disasters mainly price for cassava rise as its demand also increases. 16 Agriculture Global Practice Technical Assistance Paper Malawi is overdependent on a single export market— BOX 3.2. AFLATOXINS India—for pigeon peas. Experience has shown that this Aflatoxins are chemicals produced by fungi (in the case of is risky. Malawi’s pigeon peas have on occasion been Malawi, Aspergillus flavus) that live in the soil and flourish rejected due to poor quality and Malawi frequently fails particularly well under humid conditions. Aflatoxin con- to meet the Indian demand even though in theory its pro- tamination can take place before and during harvest and/ duction is sufficient. As India’s annual demand depends or during storage. Although aflatoxins flourish in humid on its own domestic production, it is difficult to predict for conditions, crops are also vulnerable to contamination dur- Malawian producers—a risk that the existence of alterna- ing droughts. Host crops include maize, groundnuts, and tive markets would somewhat mitigate. sorghum. Aflatoxins are carcinogenic, mutagenic, teratogenic, and immunosuppressive, and have other serious health implica- UNPREDICTABLE TRANSPORTATION tions, reasons that strict trade regulations related to aflatox- COSTS ins are in place. Transportation costs are not predictable, which is prob- In a recent survey, 30 percent of groundnuts sampled lematic especially for potato and banana producers. in Malawi were contaminated with unsafe levels of afla- Transportation costs are negotiated on an ad hoc basis, toxin. In general, improved varieties of groundnuts tend to and total transportation cost is dependent on the ultimate have lower rates of aflatoxin contamination because they path from point of sale by farmers to final market destina- are resistant to drought, pests, or diseases that can make tion. Prices often depend on truckers’ ability to backhaul groundnuts more susceptible to aflatoxin contamination. and cobble together multiple segments to reach their final But because most of groundnut production comes from smallholders, it is difficult to institute improved handling destination. For these reasons, truckers sometimes also and management practices and/or increase the use of change the price during transportation. improved varieties that would decrease the risk of con- tamination. A number of nongovernmental organizations (NGOs), donor, and government schemes intend to intro- AFLATOXINS duce improved varieties, but the high recycling rate self- Aflatoxins are a serious problem in Malawi (see box 3.2), limits farmers’ access to them. especially for groundnuts and maize, in some cases pos- Sources: Monyo et al. 2012; Ngwira and Khonje 2005. ing a risk to the entire sector as well as to consumers. For groundnuts, aflatoxin poses the biggest marketing risk, as experienced by Malawi in the 1990s when the United Kingdom banned all groundnuts imported from Malawi after detecting aflatoxins above permissible levels in ship- FOOD CROPS—ENABLING ments. Malawi is currently exporting to neighboring ENVIRONMENT RISKS countries and although they currently do not test for afla- The most serious enabling environment risk in Malawi toxins in shipments, they are likely to close their markets is unpredictable and opaque government interference. if and when new testing routines are introduced. Serious Currently, this is of particular concern in the maize sec- public health concerns are associated with the consump- tor, which over the past years has been subject to market tion of contaminated groundnuts, both domestically and interventions, price interference, and erratic policy bans. in export markets. It has also been estimated that as much Figure 3.6 shows maize prices in Malawi from 2005 to as 30 percent of all maize is contaminated with aflatoxin. 2012. In a normal year, prices should be lowest in May– Although contamination levels are not as severe as those June, immediately after harvest, and should show a marked for groundnuts, maize’s pronounced role in the national increase toward November–December, only to remain diet makes aflatoxin contamination a major health haz- high or increase further during the lean season until the ard for Malawi. Insufficient awareness and the lack of next harvest. In years of poor harvest, prices should be preventive measures, along with the prohibitive cost of higher than in other years, although imports and food aid testing, result in unacceptably high levels of aflatoxin may put downward pressure on prices. However, these contamination. cycles also depend on government interventions. Malawi: Agricultural Sector Risk Assessment 17 FIGURE 3.6. MONTHLY MAIZE PRICES IN LILONGWE, BLANTYRE, MZUZU, AND ZOMBA (TAMBALA/kg), 2005–2012 18000 Lilongwe Blantyre Mzuzu Zomba 16000 14000 12000 10000 8000 6000 4000 2000 0 Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct 2005 2006 2007 2008 2009 2010 2011 2012 Source: National Statistics Office (NSO) 2014. BOX 3.3. NATIONAL FOOD RESERVE sector that, with the help from the NFRA (see box 3.3), AGENCY (NFRA) has significantly decreased maize prices in the lean sea- The National Food Reserve Agency has been in existence son, when prices are normally higher than postharvest. since the early 1980s and is subordinated in the MAFS. Based on information from NFRA, traders, and farmers, The NFRA consists of seven branches around the country the chain of events has been as shown in figure 3.7. and employs 120 persons. Its mandate is to keep maize for the government to support vulnerable groups: These market interventions have distorted prices, lead- » In a normal year ing to abnormal patterns in maize price development. » For a humanitarian crisis Figure 3.8 describes the how the policy carried out in 2013 » For vulnerable groups in different parts of the country altered the normal price pattern over the season. Because of the timing of the intervention, farmers were not par- The annual budget allocation is in part based on the previ- ous year’s consumption and on production estimates made ticularly affected this year. Rather, traders bought at nor- by MAFS. Additional allocation is sometimes made over mal levels from farmers, expecting to make a profit later the year depending on domestic food security. in the season. In part because of the government budget cycle, ADMARC did not initiate procurement of maize To the extent possible, maize is procured domestically but imports of maize do take place. NFRA does not buy, sell, until after the new budget year had started in July. At that or distribute maize directly. ADMARC procures in part time, ADMARC procured maize at MK 125–145/kg, through public tenders and in part through direct procure- above the normal market price. The plan for 2013–14 was ment from traders, and sells maize on the market. Incomes to buy 120,000 MT and by early March, about 90,000 MT from sales are used as revolving funds for maize procure- had been procured through two official tenders. Purchases ment. Other outlets for the maize stored in the NFRA later in the season were done without public tenders, but are the World Food Programme (WFP) (both emergency relief and school feeding programs) and the Department instead through a network of traders in the interest of of Disaster Management (DODMA). time. In early 2014, the price offered was MK 130–160/ kg. By November 2013, ADMARC started selling maize Source: Interview with NFRA in March 2014. from the NFRA at MK 80/kg, much lower than the initial purchasing price and the price normally seen at that time. As a result of government policy, maize prices over past The price intervention by NFRA and ADMARC and the seasons have proved to be a significant risk to maize sec- accompanying export ban has meant that traders have had tor participants. Especially since 2012, the government to sell their stock at a loss, which the interviewed traders has developed a heavy interventionist policy in the maize signaled will be transferred to farmers next year. 18 Agriculture Global Practice Technical Assistance Paper FIGURE 3.7. MAIZE INTERVENTIONS IN MALAWI SINCE 2012 SGR releases Traders push maize on the prices down for Private traders SGR/ADMARC Prices down market via farmers next Harvest buy from tender and buy and traders sell ADMARC at season to farmers from traders at loss lower prices recoperate than procured losses FIGURE 3.8. MARKET INTERVENTIONS theft limit maize farmers’ interest in storing maize in sim- AND PRICE DISTORTIONS IN ple, raised silos. Instead, farmers store the maize inside MALAWI’S MAIZE MARKET their homes. Onions and potatoes are also stolen, some- Price times directly from fields. For potato traders, theft is a problem at “break of bulk points” (where goods are off- Normal price curve loaded and reloaded) during transport of potatoes to final market destinations. 2013–14 price curve EXPORT CROPS Time OVERVIEW Traders buy ADMARC buys ADMARC sells at Export crops differ from food crops in that most risks to market price at higher than market price lower than market price, forcing down the commodities are further down the supply chain. In prices in the sector. general, the supply chains for the four export crops ana- lyzed (tobacco, cotton, sugar, and tea) are better organized than for food crops. This means better access to inputs for farmers, and more incentives among processors to sup- The main enabling environment risk is changes in export port producers to minimize risks and losses at farm level regulations without any prior announcements, which (and indeed throughout the chain). However, export interviews with private and public sector actors indicated commodities are by nature more exposed to exogenous are a problem for both producers and traders. Although risk, such as foreign trade regulations and exchange rate trade policies de facto have been relatively stable over fluctuations. With a wider range of actors involved and the past years (table 3.3), frequent announcements about more value added throughout the chain, marketing and changes in trade policies are reportedly being made, gen- enabling environment risks tend to be more pronounced. erating uncertainty for participants in the sector. The interviews with small-scale maize traders in box 3.4 give insights to some of the concerns facing actors in the sector. For other food crops, enabling environment risks are less of a problem, mainly due to the lack of government inter- Risks are much more crop-specific for export crops than vention. The main exception is for beans, where export for food crops in Malawi. When analyzing the risks bans make trade uncertain and limit the potential for among export commodities, it is notable that the risks dif- returns for producers. Much of the Malawian bean belt fer between the supply chains and that the four subsectors lies near the border with Mozambique. analyzed face different challenges. Whereas for all seven food crops analyzed the main risks were drought, erratic THEFT rainfall, and pests and diseases, risks to individual export Theft is a problem for certain food crop producers and commodities differ to a greater extent because of more contributes to other issues for the sector. Concerns over complicated supply chains and markets. Malawi: Agricultural Sector Risk Assessment 19 TABLE 3.3. TRADE BANS AND LIFTS IN MALAWI SINCE 2008 Date Commodity Trade Ban Description 2008 25 assorted Restricted imports and exports Scarcity of products within Malawi prompted the commodities restriction. Export restriction included maize. July 23, Roundwood timber Total trade ban on roundwood timber Roundwood timber exports generated less revenue 2008 exports than processed timber, resulting in high foreign exchange losses. May 28, Cottonseed and Restricted ban on export of cottonseed Scarcity of cottonseed during the planting season 2012 seed cotton and seed cotton prompted the restriction. June 21, Soybeans Total trade ban on export of soybeans High levels of malnutrition among children under 2012 five in Malawi led to a total ban on soybean export to boost domestic consumption of protein. June 20, Export Export ban lifted on 15 commodities. Government lifted both total and restricted bans 2013 commodities Scrapped some commodities, reducing on some commodities, reducing the list from 25 to the list from 25 to 10 10. Maize remains on the list of restricted export products. Source: Authors, based on GOV press releases. BOX 3.4. INTERVIEWS WITH SMALL-SCALE MAIZE TRADERS This box summarizes interviews with a single-owner, family- farmer). Everyone in the outfit has a second job. The owner is run, maize-trading operation and with a single maize trader. a gas station attendant and the younger brothers all sell used shoes. Maize-Trading Operation. The two men interviewed work for the maize-trading operation owned by their elder During the interview, the brothers did relate that random brother. At least three other brothers with bicycles were out bans on trading maize were somewhat of a problem, particu- collecting maize from farmers and bringing it to this site for larly when a ban is announced when they have a lot of capital sale. The fact that they go to the farm gate, or at least to the tied up in inventory. However, they did not see bans as a par- farmers’ village center, is a critical success factor in their esti- ticularly large risk, stating, “We know we can always sell [the mation. The range of their operation is roughly 10 km from maize] eventually.” their trading post. Their peak period of activity is April–May, Single-Person Trader. The gentleman interviewed is cur- but they are open for business year-round. They typically rently a full-time driver. However, several years ago, he aug- hold back 60 to 75 50 kg bags of maize for sale later in the mented his driving income by trading. He raised the capital to year. Their method of minimizing storage costs is ingenious: start trading by selling the output from three acres of hybrid they own no warehouse or silo, instead renting two to three maize plus some savings. His modus operandi was to rent empty rooms per year in occupied houses to store their maize a truck, drive to farmers outside of Lilongwe, and then sell reserve. They lay mats down in the room, and put what they the maize he bought from them in downtown Lilongwe. He refer to as chemicals on the maize that act as a preservative. In stopped doing this when he encountered a problem fatal to addition to paying rent, they pay a small stipend to the home- owner to guard the maize from theft. his operation. When the current president came to office, she instituted a combination of policies and programs that had At the time of the interview, the brothers paid an effective the net effect of dropping the market price of maize, which rate of MK 5,500 per 50 kg bag and sold at an effective rate effectively wiped out his profit margin relative to his transpor- of MK 6,000 per 50 kg bag. They don’t see ADMARC as a tation costs. The price of maize has now risen sufficiently that particularly compelling competitor, because when ADMARC he is considering reentering the trading game next year. If he comes to town it only opens depots to which farmers have does, he will not give up driving. to bring their produce. They feel this gives them a compara- tive advantage, as they are collectors (that is, they go to the Source: Authors, based on interviews with stakeholders, March 2014. 20 Agriculture Global Practice Technical Assistance Paper mid-1990s, with a big drop in 1993. After 1997, fluctua- EXPORT CROPS— tions in yields essentially disappear as irrigation for sugar PRODUCTION RISKS production was largely developed by Illovo when the WEATHER-RELATED RISKS company established in Malawi in 1997 (figure 3.10). But Droughts, dry spells, erratic rains, and unpredict- excess rainfall during harvest is a challenge, not because it able weather patterns are problematic for export crops, affects yield, but because it complicates the extraction of although they are affected differently. Tea yields are cane from the field (trucks get stuck) and as the cane gets affected by irregular precipitation because the quantity of wet, additional power is needed at the plant to dry it, thus rainfall determines the frequency at which the plants can increasing processing cost. be plucked (twice a month during the rainy season and just once off season). One estate reported that this caused PESTS AND DISEASES yields to drop 25 percent in 2013. For tobacco producers, Pests and diseases are less of a problem for export crops dry spells are generally a problem, but so is late onset of than they are for food crops; the main pests are ballworms rain because the seedlings’ root-balls become too large for and aphids. Although pests and diseases exist for the four the tobacco to transplant well to the field, which means export crops, they are controlled with chemicals and at replanting. However, late onsets of rain do not affect pro- times through handpicking. For example, even small-scale duction on a large scale in Malawi. sugar outgrowers who have a pest problem do not experi- ence losses higher than 10 percent. Exceptions exist, of For cotton production, dry spells are common and may be course: Satemwa Tea Estate claims that in 2013 a pest damaging if they occur between the first and fourteenth attack during the dry season caused the complete cessa- week. The losses in the affected areas can be on the order tion of plucking for three months in the affected areas. of 10–50 percent. Figure 3.9 shows drops in cotton yields This risk is considered frequent but of low impact because since 1980 and their causes. pests are mitigated for with small amounts of pesticides. As with food crops, the occurrence and losses of export Rainfall is not a risk for sugarcane because sugar produc- crops also depend on agricultural practices, including the tion in Malawi is irrigated for both estates and outgrow- removal of infected plants, the handling of chemicals, and ers. Fluctuations in sugar yields can be seen prior to the the use of different varieties. FIGURE 3.9. COTTONSEED AND COTTON LINT YIELDS AND MAJOR DROUGHT EVENTS IN MALAWI, 1980–2012 1.2 y = 0.0077x + 0.5825 R² = 0.1711 1.0 0.8 0.6 Drought c Dry spellb 0.4 Droughta prices 0.2 Floodd Droughta Droughta 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 a National-level droughts according to RMSI 2009; b Reserve Bank of Malawi. Owing to low prices in preceding years, farmers were hesitant to plant cotton; c Kachule 2011; d National- level floods, according to RMSI 2009. These floods particularly affected the southern part of the country. Malawi: Agricultural Sector Risk Assessment 21 FIGURE 3.10. SUGARCANE YIELDS (KG/ha) AND AREA HARVESTED (HA) IN MALAWI, 1980–2012 Area harvested (Ha) Yield (tonnes/ha) Trend Linear (Yield (tonnes/ha)) 140 40,000 Drought 120 35,000 30,000 100 25,000 80 Kg/ha Ha 20,000 60 15,000 40 10,000 20 5,000 0 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 Source: FAOSTAT; authors’ calculations. BOX 3.5. CASE STUDY: MARY MWASE, MAIZE, AND TOBACCO FARMER MADISI Mary Mwase has 3 acres of land on which, together with her TABLE B3.5.1. RISK EVENTS IN ORDER OF husband, she cultivates 1 acre of tobacco, 1.5 acres of maize, and 0.5 acres of soybeans. Like many other women farmers IMPORTANCE ACCORDING in Madisi, Mary is engaged in other income-generating activi- TO MARY ties, such as baking doughnuts and selling cooked cassava and Affected sweet potatoes at Madisi Main Market. When sales are good, Crop Frequency Impact Mary earns a gross of MK 2,250 (US$5) in 3–4 selling days. For their tobacco production, Mary and her husband received Late onset Maize Frequent 20%–30% crop an in-kind loan of fertilizer, seed, and pesticides from Limbe of rains Tobacco loss Leaf Tobacco Company (LLTC—a tobacco leaf–buying Farm Maize Frequent 40%–70% yield company) through contract farming. For maize, they highly inputs’ price Tobacco loss depend on farm inputs provided by the government under volatility FISP. Their gross income comes mainly from tobacco sales Early ending Maize Frequent 20%–30% crop (MK 500,000) and to a lesser extent from maize and soybean of rains Tobacco loss sales (MK 10,500 and MK 9,600, respectively). Soybeans Pests and Maize Occasionally 5%–10 % crop At the production stage, the major risk Mary and her husband diseases Tobacco loss face is erratic rains, which take different forms such as early end- Soybeans ing of rains, late onset of rains, and dry spells in the midst of the cropping season (see the accompanying table). Mary believes changes in rainfall are due to poor management of the local ficient amounts. She also wondered why fertilizer and other vegetation, which has resulted in careless cutting down of trees inputs for tobacco production are not subsidized. Mary said she in the area. She recalls that when she was a girl, the area had is happy with the way the maize industry is regulated, especially a lot of trees and vegetation, and rainfall was not a problem, by ADMARC, because whenever she runs short of food (maize) unlike now. Because her maize, tobacco, and soybeans are rain during the lean months (December to March), she is able to buy fed, the changing rainfall patterns have negatively affected her maize from ADMARC at a lower price than from vendors. crops’ output. Mary also mentioned pests, and especially army- The major risk-management strategies adopted by Mary and worms, as serious production risks for many crops in Madisi. her husband include mitigation through contract farming Mary lamented the insufficient provision of farm inputs from with LLTC, use of early maturing varieties especially in maize FISP; the prices for farm inputs, especially fertilizer that she production, and chemical applications to control pests and uses, have varied widely over the past 10 years. For example, diseases. Mary had no strategies to cope with erratic rains. fertilizer prices oscillated from MK 4,000 per 50 kg bag in 2008 to MK 15,500 in 2013, making it difficult for her to use suf- Source: Author interview with Mary Mwase. 22 Agriculture Global Practice Technical Assistance Paper HAILSTORMS year’s auction price—the correlation coefficient is 84 per- Hailstorms are a problem for tobacco producers, but cent. In addition, due to the great importance of tobacco although they tend to be devastating where they strike, they production to so many smallholder farmers, these deci- are highly localized. They damage a limited number of sions affect the performance of other crops’ markets. hectares of crop rather than a region or even a village. The interview with farmer Mary Mwase in box 3.5 exemplifies A reserve (minimum) price is fixed every year just before some of the production risks facing small-scale producers in the start of the buying season, posing a risk to farmers Malawi and practices used to mitigate for these risks. who have already made their production decisions. In theory, the reserve prices are fixed at a level that covers EXPORT CROPS—MARKET the farming cost and allows for a 50 percent profit mar- gin. However, farmers complain that this minimum price RISKS is announced too late, when they have already made pro- PRICE VOLATILITY duction decisions and incurred most production costs. In Price volatility is an important market risk throughout sum, late reserve price announcements do not contribute the tobacco supply chain and although prices to a certain to stabilizing prices and supply. extent follow international price trends, price volatility is largely a result of changes in domestic supply. Farm- Price volatility is also very high in cotton, affecting both ers tend to take production decisions on the basis of the ginners and farmers. Figure 3.12 shows international cot- prices obtained at auction in the previous year, allocating ton price changes. A minimum procurement price is fixed more or less land to tobacco at the expense or advantage before harvest in an agreement between ginners, farmers’ of other cash and food crops. High prices the year before associations, and the government to facilitate transactions drive an increase in production and supply at the auc- between farmers and ginners and to assure returns that tion, whereas prices perceived as low the previous year recover farmers’ production costs. Drops in international cause farmers to restrain tobacco production. Because prices (export prices for ginners) may result in significant demand, in turn, tends to be far more stable from one disturbance of the domestic market, reduce demand year to the next, the result is continuing price and pro- from ginners at the minimum price, incentivize farmers duction volatility. Figure 3.11 shows the clear relationship to side-sell, and even cause some ginners to stop buying. between tobacco sales in the auction and the previous According to the Reserve Bank of Malawi, in addition to FIGURE 3.11. AVERAGE AUCTION PRICE (U.S. CENTS/kg) AND VOLUME (kg) OF TOBACCO SOLD, LAGGED ONE YEAR, 1995–2012 500,000,000 300 Volume (kg) Av. Price (U.S C/kg) 450,000,000 250 400,000,000 350,000,000 200 US$ cents/kg 300,000,000 Kg 250,000,000 150 200,000,000 100 150,000,000 100,000,000 50 50,000,000 0 0 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 Source: Tobacco Control Commission (TCC). Malawi: Agricultural Sector Risk Assessment 23 FIGURE 3.12. ANNUAL COTTON PRICE CHANGE (%) IN MALAWI, 1988–2012 30% 20% 10% 0% –10% –20% –30% Mar-88 Mar-89 Mar-90 Mar-91 Mar-92 Mar-93 Mar-94 Mar-95 Mar-96 Mar-97 Mar-98 Mar-99 Mar-00 Mar-01 Mar-02 Mar-03 Mar-04 Mar-05 Mar-06 Mar-07 Mar-08 Mar-09 Mar-10 Mar-11 Mar-12 Source: http://www.cotlook.com. Note: Cotton, Cotlook “A Index,” Middling 1-3 32 inch staple, CFR Far Eastern ports, U.S. cents/lb. the dry spell, the decline [of cotton production in 2010] is are as few as two buyers, resulting in limited competition attributed to lower prices offered in the preceding season and high price volatilities at the auction. Because of this, that resulted in the pulling out of a lot of farmers from estates prefer to make direct contracts with buyers in the growing the crop. Farmers would like to have a guaran- consuming countries, which in general means not only teed price at planting but this is not consistent with the more stable but also higher prices. The proportion of high volatility of international prices. direct sales versus auction sales varies between estates, and thus their exposure to price risks does too. For example, An effect of this is increased side-selling to ginners that Eastern Produce sells most of its produce through con- have lower operating costs and are willing to pay higher tracts all over the world, whereas Satemwa, a small estate, prices. Traditionally, a few ginners with large processing sells 80 percent of its produce at the Blantyre auction. facilities have dominated Malawi’s cotton market. These ginners both supplied inputs and provided extension ser- Uncertainty about selling prices is the most important risk vices to cotton producers. Extension services were offered for smallholder sugar farmers associated in trusts. The for “free,” the cost recovered through the price offered to price at which the trusts start selling to Illovo is decided at farmers. Recently, new ginners have entered the market the beginning of the marketing season but is continuously that do not provide services to producers and therefore adjusted according to variation in the exchange rate, infla- can offer higher prices for cotton. With the government tion rate, and interest rate (box 3.6). The function used to providing cotton inputs, fewer ties exist between produc- make the price adjustments on the basis of those variables ers and the three traditional ginners. This presents a risk is seemingly unknown to stakeholders, including the farm- for them because they no longer have reliable sources of ers’ organization (trust) interviewed for this analysis (Kas- cotton and therefore risk working at lower than optimal inthula). The leaders of the farmers’ organization would capacity. It is noted that since 2005, the government has like to see greater transparency and participation in the supplied inputs to the cotton sector in 2007/08, 2008/09, pricing process. 2011/12, 2012/13, and 2013/14, an unpredictable event for stakeholders in the past. For tea outgrowers, the price volatility risk is especially related to exchange rate volatility (see box 3.7). In fact, Limited buyers at the Blantyre tea auction make tea prices the outgrowers’ selling price is agreed with the estates on unpredictable and volatile. The auction in Blantyre is run the basis of negotiations between the tea growers’ asso- by two brokers and usually there are not more than five ciation and the estates’ association, but very often the active buyers (multinational companies). At times, there “agreed” price does not meet farmers’ expectations. At 24 Agriculture Global Practice Technical Assistance Paper BOX 3.6. SUGAR PRICES rising inflation, and the depletion of international gross reserves in the context of an overvalued exchange rate. Farmers are paid every month 85 percent of the amount The government that came to power in April 2012 has due resulting from the average of the current and previous month’s prices. This average price at which the trust sub- instituted macroeconomic policy adjustments to address mits its production to Illovo varies greatly from month to the imbalances (see box 3.7), including devaluation of the month and, at the end of the marketing year, is calculated Malawi kwacha and a move toward a flexible exchange as the average over all months (the marketing year aver- rate regime. The immediate result was a period of great age). This is the price used for finally settling the bill for the exchange rate volatility and a sense of financial instability sales of the entire marketing year. The trust, in turn, pays a among farmers and other actors in the tobacco and other fixed monthly amount to the member farmers (MK 34,000 supply chains. currently or a different amount depending on the arrange- ments regarding the repayment of each farm’s develop- ment loans) and a final bonus at the end of the year. Although there are signs that these reforms have started yielding results, economic recovery is fragile and the The Kasinthula Cane Growers Ltd., the trust interviewed exchange rate may take time to stabilize given the excess by the mission, has distributed benefits in only one year demand for foreign exchange. As long as instability con- since it was created in 1997. This bonus (15 percent) could even be negative if the final price is lower than the accumu- tinues to fuel shocks (input-product price imbalances), lated monthly payments. All costs (transportation, inputs, financial risk will exist along the agricultural supply and so on) are deducted from farmers’ final payment, so chains, caused by the variable exchange rate. In effect, the the exchange rate is a relevant risk. exchange rate has been very volatile since mid-2012 when the kwacha was devaluated. FIGURE B3.6.1. ANNUAL PRICES SUGAR AND SUCROSE (2006–14) Average local market price GOVERNMENT INPUT DISTRIBUTION 1000 800 Average export market price Delays in the supply and insufficient quantities of gov- US$/tonne Sucrose price 600 ernment-provided inputs are an important enabling envi- 400 ronment risk for cotton producers. Cotton inputs (seeds 200 0 and chemicals) have been supplied by the government for 2006 2007 2008 2009 2010 2011 2012 2013 2014 a few years now.6 Cotton inputs used to be provided by ginneries on a loan basis—a system that provided incen- tives for both ginners and farmers to be efficient in dis- present, the base price is fixed at US$0.13/kg of green tributing and using inputs, as ginners were interested in leaf, paid to farmers in Malawi kwachas. Then the estates obtaining as much raw material as possible and farmers pay a bonus twice a year calculated on the basis of the had to attain levels of production compatible with the final selling price and the industrial costs. At the time of loan repayment needs. Moreover, that system stimulated the interview (March 2014), the most recent price was loyalty between buyers and farmers and therefore prom- MK 17/kg. The exchange rate variation is a risk for ised to establish longer-term agreements. The success farmers as it affects both their returns and the cost of of the government input supply program depends upon inputs supplied by estates, which are also denominated in the government’s logistics for assuring the timely arrival foreign currency. of inputs (seeds in particular), the ginners’ distributing capacity, and sufficient quantities of chemicals for each farmer. Farmers interviewed for this report declared that EXPORT CROPS—ENABLING they preferred the previous system (input sales by gin- ENVIRONMENT RISKS ners) as distribution was more effective and timely. When The macroeconomic environment constitutes a inputs, particularly seeds, do not arrive on time, farmers constant risk to the export sector. Malawi faced serious macroeconomic challenges in 2011 and 2012 6 Under the FISP, the government provided cottonseed and chemicals in the as a result of policies that led to a growing fiscal deficit, 2007–08 and 2008–09 seasons, and since 2011. Malawi: Agricultural Sector Risk Assessment 25 BOX 3.7. RECENT MACROECONOMIC REFORMS During the period 2006–10, Malawi experienced strong eco- imposed. Other reform measures included: removal of subsi- nomic growth averaging 7.1 percent. Since 2009, however, the dies on fuel; cancellation of requirements for prior approval economic situation has worsened as a result of inappropriate and pre-vetting of all imports in excess of US$50,000; and macroeconomic policies, including rising budget deficits in an the reversal of surrender requirements on tobacco dollars, environment where the exchange rate was overvalued. These according to African Development Bank (AfDB 2013). The policy distortions contributed to a severe shortage of foreign high cost of finance remains a major obstacle to doing busi- exchange, which affected the availability of basic goods and ness in Malawi as the Reserve Bank’s key bank rate is very production inputs, including fuel, and higher inflation (see high. At the time of the study, the macroeconomic conditions accompanying figures). seemed to be worsening as the revelation of massive looting of public funds in 2013 is making continuation of general bud- Since April 2012, the new government has undertaken signifi- getary support by donors difficult. About 40 percent of the cant economic and governance reforms to address Malawi’s national budget is financed by donors, under the Common macroeconomic imbalances and resumption of donor sup- Approach to Budgetary Support (CABS), which includes the port. In May 2012, for example, the kwacha was devalued by AfDB, European Union (EU), Germany, Norway, the United 49 percent (from MK 167 to MK 250 to the U.S. dollar) and Kingdom, and the World Bank. subsequently floated. A tight monetary and fiscal policy was FIGURE B3.7.1. EXCHANGE RATE (MK/US$) BY MONTH, 1985–2014 500 450 400 350 300 250 200 150 100 50 0 Jan-85 Jan-86 Jan-87 Jan-88 Jan-89 Jan-90 Jan-91 Jan-92 Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Source: Reserve Bank of Malawi. FIGURE B3.7.2. CHANGES IN CONSUMER PRICES COMPARED WITH SAME TIME PREVIOUS YEAR, 1994–2013 100 80 60 40 20 0 1994 Jan 1994 Oct 1995 Jul 1996 Apr 1997 Jan 1997 Oct 1998 Jul 1999 Apr 2000 Jan 2000 Oct 2001 Jul 2002 Apr 2003 Jan 2003 Oct 2004 Jul 2005 Apr 2006 Jan 2006 Oct 2007 Jul 2008 Apr 2009 Jan 2009 Oct 2010 Jul 2011 Apr 2012 Jan 2012 Oct 2013 Jul Source: IMF, International Financial Statistics, accessed June 13, 2014. Sources: AfDB 2013; Reserve Bank of Malawi 2014. 26 Agriculture Global Practice Technical Assistance Paper FIGURE 3.13. CROP CALENDAR FOR MALAWI Malawi. October OCT OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP Planting Winter planting Rainy season Winter harvest Green harvest Main harvest Winter harvrat Lean season Peak Peak labor demand Tobacco sales and auction labor demand OCT OCT NOV DEC JAN EEB MAR APR MAY JUN JUL AUG SEP Source: USAID 2013. tend to shift to other crops. In addition, if farmers don’t containers was stolen on its way to the Port of Beira in get the quantities of chemicals required for the actual Mozambique in 2013. The cost is not only that of the lost planted area, they may lose part of the harvest to pests. products but also the increased costs for security arrange- Under the current government distribution program, ments. quantities distributed should be determined according to plot size but this is not usually the case and farmers often REJECTED SHIPMENTS get less than they need. This is a risk for both farmers and Although rare, tobacco shipments are sometimes rejected ginners. because they are contaminated with banned substances. To ensure that no pesticide residue is found in the tobacco POWER OUTAGES bought, a sampling analysis of tobacco sold in the auc- Especially for tea and sugar production, power outages tion is conducted. However, at times the samples do not are a problem. For sugar outgrowers, the problem is inter- reflect the quality of the full batch of tobacco and resi- rupted irrigation, which eventually leads to yield losses. due is found upon the tobacco’s arrival at the importer, For Illovo, this is less of a problem as the company has so the shipment is returned to Malawi. According to the its own power generating plant. Power outages also affect Tobacco Control Commission, a regulatory body, this is the functioning of tea-processing plants and can seriously not common, but when it happens it has devastating con- damage sensitive tea. It was reported that power outages sequences for the exporters. occur two or three times a week and sometimes last up to 12 hours. The tea-processing plants have their own power WEATHER-YIELD ANALYSIS generators but their use represents a considerable increase To determine whether and the extent to which yield is in the cost of tea processing. affected by climatic events, a study was conducted on the relationship that several climatic events have on different THEFT crops’ yield for Malawi. Daily weather data from 1961 Just as for food crops, theft is a problem for tea and sugar to 2011 from 23 weather stations across Malawi were producers. This happens both on the processing sites and used, correlated with disaggregate yield data for cassava during transportation. Illovo reported that as much as and maize for 1984–2012. To understand the impact of 3,000 tons were lost last year and managers from Eastern weather volatilities in different periods on yield, the crop Produce Malawi Ltd. reported that a tea shipment of 13 calendars for maize and cassava were used (figure 3.13). Malawi: Agricultural Sector Risk Assessment 27 Precipitation in Malawi follows a very clear pattern, with Indeed, these results show that rain explains a certain one rainy season from November to March and one dry amount of variability in yield. Particularly worth noting season from May to September. All regions within the is that both cumulative rainfall and rainy events during country follow this pattern, with few variations. Over the the harvesting phase help explain more yield variance in year, each crop goes through sowing, growing, and har- almost all regions. The exceptions are Blantyre, Karonga, vesting phases. and Kasungu, where the proportion of variance explained is less than 15 percent. Two different rainfall parameters were estimated for each of the three phases: Similar to maize, cassava experienced a jump in yield, as » Cumulative rainfall: the sum of daily precipita- the mean yield went from 3.8 MT/ha to 16.7 MT/ha at tion in mm for each phase; and the turn of the century. Thus, yields were also standard- » Number of rainy events: the number of days ized for cassava, but the result is less clear than for maize; in the phase in which rainfall is greater than 5 mm. rainfall explains very little of the variability in cassava yield. The cumulative rainfall in the three phases helps Maize: The relationship between cumulative rainfall and explain about 30 percent of yield variability in Blantyre, number of rainy events and maize yield is not significant, and indicates that the more rain, the better the yield in this except in the Salima region, where the number of rainy region. It is clear that the best production years were the events for the harvesting phase explains 25 percent of most humid ones, whereas the worst years had the least yield variability. However, because most of the determi- rainfall. The three driest years, 2004–05, 1994–95, and nation coefficients are very small, a multiple linear model 1991–92 (when about 600 mm fell throughout the whole was also run. According to the results, cumulative rain- seven month period), match some of the lowest produc- fall significantly explained variability (R2 > 20%) in maize tion years, so drought can be considered the main threat yield only for Salima and Shire Valley. In Salima, rainy in this region. But for the other regions, the relationship events indexes help explain yield, whereas in the Shire between rainfall and cassava yield is not as strong. Valley, cumulative rainfall through the three phases best explains yield. In summary, drought during the harvesting phase (March– April) helps explain most of the variability in maize yield, Except in the Salima and Shire Valley regions, there is particularly during the shock years of 1991–92, 1993–94, no linear yield trend for maize but rather two different and 2004–05. This applies for most of the country. How- yield levels with a cutoff point after year 2005. The fact ever, applying rainfall data and rainy event information that yields are almost twice as high from 2006 and onward on the three phases in cassava production does not signifi- (yield went from 1.1 MT/ha before 2005 to 1.95 MT/ha cantly explain cassava yield variability over the period stud- the year after) may explain why rainfall is not correlated ied (that is, 1984–2011). Nevertheless, it was found that the much with the variability of maize yield in the period droughts during 1991–92, 1993–94, and 2004–05 were the examined. To test for this, yields were kept constant. cause of low cassava yield, particularly in the Shire Valley. 28 Agriculture Global Practice Technical Assistance Paper CHAPTER FOUR ADVERSE IMPACT OF AGRICULTURAL RISK OVERALL AGRICULTURAL LOSSES Although the previous chapter provides a good overview of the types of risks pres- ent in Malawi, for policy purposes it is important to understand their impacts in terms of the magnitude of losses, geographic occurrence, and stakeholders affected. Without knowing how much the impacts of risks cost, where they occur, or whom they affect, it is difficult to target often limited resources in a manner that effectively minimizes the impacts. This chapter attempts to quantify losses in the sector that are larger than what could be considered as normal, to compare production and losses between regions, and to map how this affects different stakeholders in the agriculture sector. The quantification of losses captures production risks such as drought and pest and disease outbreaks. The indicative value of agricultural output lost for particular years (when yields are below one-third of the standard deviation of the long-term trend) is calculated as the deviation of the actual yield from a historic yield trend multiplied by the actual area that year. The production value is then multiplied with current pro- ducer prices and converted into US dollars at the prevalent exchange rate. Indicative loss values are also compared with agricultural GDP to provide a relative measure of the loss. Figure 4.1 shows the basis for estimating indicative losses. The dark red curve is the yield, the lighter red dotted line is the long-term trend, and the pink line with triangular shapes marks one-third of the standard deviation. Losses are measured in years where they fall below this point (denoted by the arrows in the figure). Table 4.1 shows the average annual losses from production risks for selected crops. The annual risk-related losses amount to US$149 million on average, or 3.98 per- cent of the total annual agricultural production value in Malawi. Almost 30 percent of losses is from maize, which suggests the great impact of agricultural produc- tion risks on smallholder households’ food security. Similarly, cassava and potatoes Malawi: Agricultural Sector Risk Assessment 29 account for 26 percent and 13 percent, respectively, of Although the average annual losses are high, the impacts total annual losses. Tobacco also forms an important part of the individual shocks are even more devastating. Aver- of the agricultural economy and many smallholders have age figures are useful to understand the aggregate costs of a significant part of their cash income compromised as production risk, yet they tend to conceal the catastrophic a result of tobacco production losses, which account for impact that some shocks have on participants in the sector 10 percent of total agricultural losses. These figures do at the time they occur. Shocks have an impact on house- not take into account losses caused postharvest, by price hold and national food security, have important fiscal volatility, for example. repercussions, reduce the availability of foreign exchange, and have an overall macroeconomic destabilizing effect. Chapter 2 showed the high correlation between agricul- tural and national GDP. For instance, during the 2001 FIGURE 4.1. EXAMPLE OF HOW INDICATIVE drought, losses amounted to US$161 million, or 4.3 per- LOSSES ARE CALCULATED cent of total agricultural production value, and in 2005 20 Yield (tonnes/ha) 0.3 Trend to nearly US$900 million, 24 percent of total agricultural 18 Trend Linear (yield (tonnes/ha)) production (2006–08 average). Figure 4.2 shows the mag- 16 14 nitude of losses for individual years, where the size of the 12 circle depicts the losses as a share of total agricultural pro- 10 duction value. 8 6 The losses in terms of the normal production value y = 0.3033x + 5.5697 4 R2 = 0.7238 were extreme for important smallholder crops such as 2 maize and tobacco (50 percent in maize in 2005), which 0 means disastrous impacts on household incomes, food 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 security, and well-being. Because of the contribution to TABLE 4.1. LOSSES FROM AGRICULTURAL PRODUCTION RISKS, 1980–2012 Annual % Loss of Average Annual Average Annual Ag. Production Total Losses Total Losses Crop Losses (MT) Losses (US$) (2006–08 Prices) (MT) (US$) Cassava 147,719 45,010,044 1.21 4,874,734 1,485,331,478 Maize 183,711 40,545,037 1.09 6,062,465 1,337,986,237 Potatoes 52,047 19,062,320 0.51 1,717,541 629,056,577 Sugarcane 27,956 3,628,714 0.10 922,548 119,747,574 Beans, dry 5,941 5,739,877 0.15 196,069 189,415,972 Rice, paddy 3,038 2,522,431 0.07 100,257 83,240,235 Tobacco 8,431 14,672,960 0.39 269,779 469,534,740 Pigeon peas 4,771 2,225,235 0.06 157,459 73,432,773 Groundnuts, with shells 7,612 7,295,940 0.20 251,203 240,766,028.02 Bananas 5,456 1,842,499 0.05 174,579 58,959,956 Tea 1,760 2,074,786 0.06 56,327 66,393,164 Cotton 2,851 4,090,601 0.11 91,241 130,899,241 TOTAL 451,294 148,710,449 3.98 14,874,205 4,884,763,980 Source: FAOSTAT; authors’ calculation. 30 Agriculture Global Practice Technical Assistance Paper FIGURE 4.2. VALUE OF PRODUCTION LOSSES PER YEAR AS A SHARE OF TOTAL AGRICULTURAL PRODUCTION VALUE $1,000 2005 $800 Losses per year (US$ million) $600 1992 $400 1995 1999 1998 1994 2004 2002 $200 1989 1996 2003 1993 2001 1980 1983 19851986 1991 1997 2006 1987 1990 2000 $0 1980 1985 1990 1995 2000 2005 2010 $(200) Source: FAOSTAT; authors’ calculations. FIGURE 4.3. VALUE AND FREQUENCY OF LOSSES PER CROP, 1980–2012 1,600 Cassava Maize 1,400 Losses per crop (million USD) 1,200 1,000 800 Potatoes Tobacco 600 400 Groundnuts Sugarcane 200 Rice Beans Cotton Bananas Tea 0 Pigeonpeas 0.00 0.10 0.20 0.30 0.40 0.50 0.60 –200 Frequency of Losses agricultural production value, the magnitude of the losses when shocks occur is much greater for some crops than PRODUCTION VOLATILITY BY others. Thus, maize, cassava, potatoes, and tobacco have REGION the highest average annual losses (figure 4.3).7 However, The relative production volatility among different tobacco and tea incur losses the most frequently, meaning regions is measured using the coefficient of variation8 that farmers involved in these sectors are highly exposed (CV) of yield. Because of the limited available data to shocks. for other crops, only maize and cassava were analyzed. 8 Calculated as the standard deviation divided by the series arithmetic media. It 7 The cassava loss estimates are based on national yield data and should be shows the extent of variability relative to the population mean: the higher the adjusted if the national cassava yield data from 1980 to 2012 are revised. CV, the higher the variability. Malawi: Agricultural Sector Risk Assessment 31 MAP 4.1. MALAWI’S EIGHT AGRICULTURAL Malawi’s eight ADDs were used to analyze the differ- DEVELOPMENT DIVISIONS ences in production volatility. The demarcation is shown in map 4.1. Maize production volatilities are fairly even across ADDs. Blantyre showed the highest varia- tion, with a CV of 48 percent, and Kasungu the lowest, with 31 percent (table 4.2). The ADDs of Lilongwe and Kasungu, which have the largest extensions of land culti- vated to maize (almost 50 percent of the country’s total), exhibit relatively similar production volatility (CVs of 36 percent and 31 percent, respectively). However, owing to the different output levels produced in each region, these variations have different impacts on total production. The ADDs produce a total of 2,016,170 MT of maize annually, but yield and area harvested vary significantly between ADDs (table 4.2). Seventy percent of Malawi’s maize produc- tion is grown in three ADDs, Blantyre, Lilongwe, and Kasungu, and 90 percent in five ADDs if Machinga and Muzuzu are included. Losses as a share of national pro- duction are largest in Kasungu, Lilongwe, Blantyre, and Michnga, which together account for over 9 percent of total production losses annually, and 80 percent of total maize losses in Malawi. Source: USAID. TABLE 4.2. MAIZE PRODUCTION BY AGRICULTURAL DEVELOPMENT DIVISION IN MALAWI, 1983–2013 Share of Average Production Area Harvested Yield 2009–12 Annual Average Annual CV of (% of Total, (% of Total, Average Losses Losses as % of Total Yield ADD 2009–12) 2009–12) (MT/ha) (MT) National Production (%) Blantyre 14.5 16 1.9 36,955 1.8 48 Karonga 4.2 3 2.7 5,798 0.3 44 Salima 4.5 4 2.5 13,905 0.7 41 Lilongwe 25.5 24 2.3 53,734 2.7 36 Mzuzu 9.5 9 2.2 18,303 0.9 34 Machinga 10 16 1.3 26,469 1.3 34 Shire Valley 1.9 4 1.1 6,476 0.3 34 Kasungu 29.9 24 2.7 67,967 3.4 31 Total 100 100 – 229,607 11.4 – Source: Data from MAFS 2013 Annual Statistical Bulletin. 32 Agriculture Global Practice Technical Assistance Paper TABLE 4.3. CASSAVA PRODUCTION BY AGRICULTURAL DEVELOPMENT DIVISION IN MALAWI, 1983–2013 Share of Average Production Area Harvested Yield 2009–12 Annual Average Annual CV of (% of Total, (% of Total, Average Losses Losses as % of Total Yield ADD 2009–12) 2009–12) (MT/ha) (MT) National Production (%) Karonga 25.3 18.7 9.923 13,136 0.9 80.6 Mzuzu 19.2 20.9 12.609 26,965 1.9 82.6 Kasungu 17.8 16.8 8.750 6,068 0.4 80.6 Lilongwe 8.9 13 7.427 5,846 0.4 77.7 Salima 9.2 9.6 10.788 20,138 1.4 78.6 Machinga 7.5 8.5 7.032 14,997 1.1 76.2 Blantyre 11. 7 11.7 7.397 42,874 3.0 89.7 Shire Valley 0.5 0.8 7.195 1,554 0.1 69.4 Total 100 100 – 131,577 9.2 – Cassava shows similar differences in losses between levels. Primary farmers’ organizations (clubs, trusts, coop- ADDs although the CVs are high in all of them (table eratives) are also a very weak segment in the supply chain. 4.3). The high CVs are likely due to the great jump Great product price variations (cotton), multipayment in cassava yields in early 2000. Total annual cassava systems (sugarcane), and variable input costs expose them production is 1,421,327 MT and four ADDs account to recurrent financial losses. They tend to have fragile for about three-quarters of the production: Karonga, financial structures (sugarcane trusts, for instance) and Mzuzu, Kasungu, and Blantyre. However, two regions sometimes rely on credit to finance their operations, alone account for half of national losses: Blantyre (3 thereby increasing their long-term commitments and percent of total production) and Mzuzu (1.9 percent). If risks. Estates, exporters, millers, and large trading com- Salima and Machinga are included, these four regions panies are in general less directly exposed to production jointly account for over 80 percent of total annual cas- risks and are able to hedge prices globally. Table 4.4 sum- sava losses in Malawi. This despite the fact that neither marizes stakeholders’ risk profile and their current man- Salima nor Machinga belong to the top cassava-produc- agement patterns. ing ADDs in Malawi. VULNERABLE GROUPS Vulnerability to agricultural risk is high across Malawi.9 THE IMPACTS OF Throughout the country, the main source of food is own AGRICULTURAL RISKS ON crop production. Both cash and food crops are impor- DIFFERENT STAKEHOLDERS tant sources of cash for households, which is important How the losses are distributed among stakeholders along because home food production is often supplemented by the supply chains is, to a great extent, a function of supply food purchased from local markets. Poor households also chain governance and stakeholders’ capability and oppor- tunities for risk management. Smallholder farmers are the most vulnerable segment in the supply chains. Their 9 The term “vulnerability” is used here to describe exposure to hazards and production and price risk-management strategies, usually shocks. Literature highlights the fact that vulnerability is a product of two com- based on low-risk and low-yield strategies, tend to result ponents: exposure to a hazard (a shock) and resilience (the ability to manage the in poor capital buildup and below-potential production hazard) (Devereux et al. 2007). Malawi: Agricultural Sector Risk Assessment 33 TABLE 4.4. STAKEHOLDER RISK PROFILES FOR FOOD AND EXPORT CROP SUPPLY CHAINS Most Stakeholder Common Risks Significance of Risk Current Risk Management Smallholder Weather risks Weather is a major risk for smallholders. It mostly Appropriate drought-tolerant varieties farmers (erratic rains, affects food production. Cash/export crops are and crop diversification to drought- drought, and so generally cultivated in appropriate agroecological tolerant crops (such as sorghum) are the on) zones and sometimes protected by irrigation most common, though not widespread, (sugarcane) contract farming arrangements that risk-management strategies. assure drought-tolerant varieties and other best practices (tea, partially). Smallholder Pests and diseases Pests and diseases can be controlled, and therefore Mitigation capacity is higher among farmers risks are limited, if technological knowledge and export crop producers under the resources are available. This is especially problematic umbrella of contract farming. for food crop producers, as cash crop producers have better access to mitigating instruments. Smallholder Same risks as Production, price, and exchange rate risks have Farmers’ organizations provide farmers’ smallholder financial repercussions on the organizations’ production and marketing services organizations farmers (above) finances and may jeopardize their existence. (technical assistance, input financing, and (clubs, trusts, plus a financial so on) that tend to reduce farmers’ risks, cooperatives) risk mostly production risks, but increase their exposure to financial risk. Grain traders Price risks Artificial price stabilization mechanisms derail Traders try to recuperate losses by Trade bans prices from their normal pattern, which results in pushing down farm gate prices the an unpredictable investment climate and potential following season; hence farmers bear the losses on investments. cost of the risk in the long run. Unforeseen and/or erratic trade policies add to this because they close any alternative markets when prices are low as a result of price interventions. Ginners Side-selling Ginners that are well established and have long- Ginners need to establish long- (cotton) Price volatility term contract farming arrangements (the few), term commitments with farmers. including provision of production support services This is becoming difficult under the to farmers and price negotiation, are more government’s free and untargeted input exposed to side-selling risks at moments of great distribution. price drops. More speculative ginners manage to profit from price fluctuations. Estates (tea, Market-related, By diversifying sales into export and domestic markets High management capacity. sugar) and particularly export (sugar), auctions and direct sales (tea), hedging, and so exporters price risks on, estates minimize the incidence of risks. often sell their household labor (locally known as “ganyu”) support rain-fed food production, makes households vul- in exchange for food.10 nerable to shock.11 Further, households’ reliance on ganyu is conditional on rainfall because the ganyu is usually pro- This heavy reliance on agriculture and the fact that the vision of farm labor. In 2007, 95 percent of the sampled majority of the population is dependent on rain-fed agri- households reported experiencing at least one shock in culture, whereas precipitation is frequently insufficient to 11 Shocks are defined as adverse events that lead to a loss of household wel- 10 The household economy approach distinguishes the source of food mainly fare, such as a reduction in consumption, income, and/or a loss of productive into “purchase,” “own crops,” and “ganyu.” assets (Dercon 2005). 34 Agriculture Global Practice Technical Assistance Paper TABLE 4.5. DISTRIBUTION OF POVERTY IN GENDER STRUCTURES ADD MALAWI AN ADDITIONAL LAYER OF Poverty Rate (% Below National Poverty Line) VULNERABILITY National average 50.7 Women make up 70 percent of the agricultural labor Group force but earn less for salaried work and own fewer assets Urban 17.3 than do men. The value of assets owned by male-headed Rural 56.6 households is more than double that of female-headed Male-headed household 49 households and male-headed households are more likely Female-headed household 57 to own agricultural assets. Women’s rate of pay for ganyu Geographic location is likely to be only two-thirds the rate paid to men. In North 54.3 2005, female-headed households had 14 percent less Central 44.5 consumption per capita than male-headed households, South 55.5 making them more vulnerable to adverse impacts on Highest: Chikwawa 81.6 production and incomes (Hay and Phiri 2008). Because Lowest: Nkhotakota 32.1 of their limited possession of assets, and hence collat- eral, women face more difficulties than men in access- Source: Malawi Integrated Household Survey (IHS) 3 (IHS3) of 2011. ing credit, an additional obstacle to mitigating risks and recuperating from shocks. A good example is cotton the past five years. More than 75 percent of rural house- production, which is heavily reliant on chemicals that holds reported encountering four or more shocks in the in general are too expensive for women to acquire. But past five years World Bank 2007). even in subsectors where female-headed households par- ticipate to a greater extent, gender biases against risk- Poor households that experience shocks are more likely mitigating investments pose a problem. This also applies to experience a decline in well-being than nonpoor to the FISP—women are reportedly subject to long households who experience the same number of shocks queues, sometimes lasting as long as two days, before (Devereux et al. 2006). Studies have shown that house- they can buy fertilizer because priority is usually given to holds are vulnerable to food insecurity because of their men (Mvula 2011). poverty situation (Makoka and Kumwenda 2013). In particular, poverty makes them susceptible to any food- Table 4.6 shows structural differences in the type of agri- related shock because they do not have the capacity to cultural activities performed by men and women. Ten per- prevent the food-insecurity shock or to manage its effects cent of the total plots managed by men are allocated to when it occurs. Table 4.5 shows the distribution of pov- tobacco, compared with 3 percent of plots managed by erty in Malawi. (More details on vulnerable groups can be women. Reportedly, women are less likely to engage in found in appendix C.) cash crop production due to labor and time constraints; thus female-headed households allocate larger portions Productive assets, including livestock, are an important of their plots to local maize and pigeon peas than their source of livelihood, especially in the face of shocks. In some male counterparts. This means that shocks to the tobacco livelihood zones, such as Western Rumphi and Mzimba, sector disproportionately affect male-headed households, Mzimba Self-Sufficient, and Lower Shire Valley, house- whereas shocks to maize and pigeon pea production and holds depend on livestock as a source of food and cash. markets disproportionately affect female-headed house- They are able to respond to shocks by increasing the sale of holds. their livestock, and thus cushion themselves against a range of shocks. Nevertheless, most farmers lack this option and Equally, women often rely on vendors who come to their therefore remain vulnerable after a shock occurs (Christi- doorsteps because they have difficulty transporting their aensen and Subbarao 2004; Dercon 2001; Makoka 2008). produce to more favorable markets. In many cases, better Malawi: Agricultural Sector Risk Assessment 35 TABLE 4.6. PROPORTION OF PLOTS BY TYPE OF CROP CULTIVATED AND AS A SHARE OF TOTAL CROPS, 2011 Share of Local OPV/ Pigeon Composite Total by Maize Hybrid Maize Peas Ground-nuts Tobacco Beans Sorghum Maize Rice Crop (%) (%) (%) (%) (%) (%) (%) (%) (%) Male 31.8 32.2 14.7 15.1 10.4 5.5 4.3 4.0 2.7 Female 45.3 29.7 21.3 17.0 3.3 6.2 6.4 3.9 3.1 Malawi 35.3 31.6 16.4 15.6 8.5 5.7 4.9 4.0 2.8 Share of crop Male 41 52 41 47 76 47 40 51 47 Female 59 48 59 53 24 53 60 49 53 Malawi 100 100 100 100 100 100 100 100 100 Source: Malawi IHS3 Report 2012; authors’ calculations. Note: OPV = Open pollinated variety prices are offered at district centers, which are often far of 4,908 households in 2009, WFP (2010) reported that away, but women do not have the time or resources to the most common coping strategy for various shocks transport their produce there. Reportedly, women also was a reduction in food portion size (reported by 57 per- often fall prey to unreliable weighing scales in transactions cent) followed by a reduction in the number of meals with vendors. (55 percent). The Malawi government and the World Bank (2007) reported that consuming less food was the IMPACT ON HOUSEHOLD first coping strategy for about 14 percent of all households that reported experiencing a shock. FOOD SECURITY For many households in Malawi, coping with shocks More details on vulnerability in Malawi as it relates to means changing household dietary patterns. In its study agricultural risks can be found in appendix C. 36 Agriculture Global Practice Technical Assistance Paper CHAPTER FIVE RISK PRIORITIZATION AND MANAGEMENT To better utilize scarce resources, it is important to understand which risks cause major shocks to the sector in terms of losses and to observe the frequency at which they occur. This chapter summarizes the risks faced by the agriculture sector and the pos- sible solutions, as identified by the mission and validated and prioritized with stake- holders at different levels and at a workshop with MAFS in Lilongwe. RISK PRIORITIZATION Table 5.1 summarizes stakeholders’ opinions regarding agricultural risk prioritization, defined on the basis of the probability of the event and its expected impact, for food and export crops. The darkest area of the table lists the most significant risks based on their potential to cause the greatest losses to the agriculture sector as a whole and the frequency of their occurrence for any of the crops affected. From the risk prioritization exercise, and based on the frequency of realized risk events, their capacity to cause losses, and stakeholders’ ability to manage the risks, the follow- ing emerged as the most important risks to Malawi’s agriculture sector: » Drought events » Price volatilities and government market interventions » Pests and diseases PRIORITY RISK-MANAGEMENT MEASURES The drought and diseases/pests hazards challenging Malawi confront it on a contin- uum of scales (for example, small, medium, and large). The vulnerability of Malawi’s food crop sector to these risks is largely a result of the often-poor performance by its stakeholders, institutions, and infrastructure. Low yields, inadequate infrastructure, underfunded and low-performing supporting institutions, and unstructured markets all leave those engaged in the agriculture sector with minimal incomes and the barest capacity to cope with hazards. The fact that Malawi’s climate patterns are somewhat predictable (particularly episodes of drought at intra-seasonal, seasonal, and decadal time scales) opens up pathways Malawi: Agricultural Sector Risk Assessment 37 TABLE 5.1. RISK PRIORITIZATION Impact Likelihood Low Moderate High Highly probable • Hailstorms • Pests and diseases (food and • Drought events, including: (1 in 3) • Untimely distribution of inputs (cotton) export crops) – False start of, or shorter • Theft (sugarcane, tea, food crops) • Price volatility and/or than normal, rainy season • Damage from wild animals uncertainty (tobacco, tea, – Extended dry spells • Power outage (sugarcane, tea) cotton, sugar) – Higher than average • Exchange rate (risk mainly for • Unpredictable regulatory temperatures smallholders) environment for traders Probable • Side-selling (cotton) • Unpredictable maize market (1 in 5) • Excess rainfall, increasing harvesting interventions causing price and processing cost (tea, sugar) volatilities in the maize • Floods (food crops) market (recent) Occasional • Export shipments rejected (1 in 10) (tobacco) for reducing the impacts of some hazards through bet- turing). How instruments are applied for a given risk will ter use of forecasting information. Being able to predict likely depend on the probability of the risk and the sever- late onset of the rainy season has allowed Malawi to insti- ity of its impacts (figure 5.1). tute a response farming program. The ability to reliably forecast major droughts six to seven months in advance ONGOING INTERVENTIONS allows the government, NGOs, and donors to reallocate Juxtaposed with high production losses, costly mar- resources and make plans and preparations for a response. ket uncertainties, and ever-present enabling environ- Understanding that annual variations from climate norms ment risks in the agriculture sector, significant financial and their parameters are expected can better inform risk- resources are allocated to the agriculture sector under mitigation strategies and the design of associated invest- regular government programs. Malawi is one of the ments in the future (for example, strategies around plant few countries that adhere to the commitment under the breeding programs, or assessment of the value proposi- CAADP to allocate at least 10 percent of its budget to tion behind infrastructure investments such as irrigation). the agriculture sector. Most of this (67 percent) goes to FISP in direct input support (seeds and fertilizers). The Long List of Solutions. The potential solutions identi- Between 2008 and 2012, the government spent on aver- fied during field interviews as well as suggested in various age US$12.6 million on agricultural research, US$9.1 government and nongovernmental documents are pre- million on extension services, and US$37.5 million on sented in this section (table 5.2). Risk strategies are usu- the crop production program for enhanced food security ally a combination of risk-mitigation, risk-transfer, and under Pillar 1 of ASWAp and US$21.5 million on irriga- risk-coping instruments. Risk mitigation refers to actions tion under Pillar 3. Although all of these investments do taken to eliminate or reduce events from occurring or to not automatically mitigate risks, research, extension, and reduce the severity of losses (for example, water-draining irrigation especially have the potential to do so if designed infrastructure, crop diversification, extension); risk trans- through a risk lens. because risk-mitigating practices are fer refers to mechanisms to transfer the risk to a willing often win-win investments (that is, while mitigating risks third party at a cost (for example, insurance, re-insurance, they increase productivity), much of the support under financial hedging tools); and, risk-coping refers to actions FISP could also have risk-mitigating impacts depending that help cope with the losses caused by a risk event (for on how the support is designed, what varieties are grown, example, government assistance to farmers, debt restruc- and how fertilizers are applied. 38 Agriculture Global Practice Technical Assistance Paper FIGURE 5.1. STRATEGIC RISK INSTRUMENTS ACCORDING TO RISK LAYERS PROBABILITY Layer 3 Very Low Frequency, Very High Losses Layer 2 Risk Mitigation Low Frequency, + Risk Transfer Medium Losses + Risk Coping Layer 1 Risk Mitigation + Risk Transfer High Frequency, Low Losses Risk Mitigation SEVERITY Source: World Bank ARMT. Despite all these efforts, as shown in the previous chapter, could potentially have risk-reducing effects in Malawi and the impacts of risks remain high and significant resources cover the focus crops in this report. In total, these projects are therefore allocated to coping mechanisms. In the case amounted to some US$765 million, or about US$85 mil- of disasters, such as the droughts in 2005 (when about lion per year. Of these, 15 projects totaling US$165 mil- 5 million Malawians received food aid) and in 2012 (when lion were oriented toward coping; that is, they supported more than 1.63 million people received aid), coping mech- postrisk interventions. In addition, hundreds of similar anisms financed by both government and donors provide projects worth less than US$1 million exist in Malawi that a vital safety net for people. But because of the impacts cover crops other than those focused on in this report. of risks on a regional and district level, the allocation of resources to ex post risk-coping mechanisms is also high in To better understand how resources to agriculture are normal years when agricultural losses are reportedly low allocated under these projects, the categories listed in at the national level. Over the past six years, DODMA table 5.3 were investigated. and WFP have distributed grain worth US$22.1 million, procured by the NFRA for the SGR. And since 2012, The amounts spent on mitigation and coping together ADMARC has released subsidized grain worth US$16.5 with the losses from risks are not just costly, but also rep- million via the reserve. This does not include other ongo- resent lost development opportunities: losses due to risks ing donor projects that support coping mechanisms, which are lost returns on investments and productive assets for have amounted to over US$16 million annually on aver- actors in the sector—money that could have reduced age over the past 10 years. Nor does it include food sup- poverty and generated multiplier effects in the economy; port provided by WFP that is not procured through the instead, resources spent by the government and donors on SGR. (In 2013/14, support provided by WFP reportedly coping mechanisms are diverted from longer-term devel- amounted to about US$92 million). Minimizing the losses opment investments in and outside the sector. from agricultural risks would, thus, free up resources for other longer-term development objectives. Figure 5.2 shows the amount spent annually from the regu- lar government budget on risk-mitigating and risk-coping In addition to the government’s efforts in the sector, a large mechanisms versus the amount lost due to risks on aver- number of donor-financed agricultural projects relate to age between 2008 and 2012. Risk-management expendi- risk reduction. Between 2004 and early 2013, more than tures are clearly skewed toward coping mechanisms for 60 projects above US$1 million supported activities that ex post risks rather than toward ex ante, risk-mitigating Malawi: Agricultural Sector Risk Assessment 39 TABLE 5.2. LONG LIST OF AGRICULTURAL RISK-MANAGEMENT SOLUTIONS IDENTIFIED FOR MALAWI Risk Mitigation Transfer Coping Food crops Production Improve resilience of production to Insurance: • Strengthen available information mitigate for droughts, pests, and diseases, • Macrolevel and analytical capacity in relevant including the following: insurance institutions (for example, MVAC, • Strengthen ag. research for more • Farmer-level DODMA) to improve timeliness and resilient varieties crop insurance effectiveness of responses through • Strengthen the capacity of ag. extension through banks better targeting services to more effectively promote and • Use improved information to plan improve adoption of risk-mitigation responses more effectively and thereby practices among farmers improve possibilities for price hedging, • Improve water management through budget planning, and so on for increased investment in infrastructure response programs (for example, irrigation, water harvesting) • Improve management of the • Improve access/timeliness/reliability of Strategic Grain Reserve (SGR) and inputs ADMARC to improve predictability of • Introduce conservation agriculture (that interventions is, minimum tillage) • Design rural food-for-work programs so • Strengthen weather early warning systems that they improve or build agricultural • Introduce response farming infrastructure • Use FISP to better mitigate risks • Support crop diversification / geographic optimization • Increase storage Mitigate the effects of floods: • Support reforestation and grass reclamation • Invest in water control (for example, dams, drainage infrastructure) Strengthen farmers’ organizations, traders’ associations, and trucking sector to bring more structure to markets and capture economies of scale Implement rural roads program to improve the quality of rural roads and reduce transportation costs Market Minimize market interventions to eliminate market and price distortions through improved management of the SGR, including adopting transparent and predictable purchases and releases of grain Strengthen market incentives for farmers to invest in risk-mitigation practices that result in increased production and productivity Introduce warehouse receipt system 40 Agriculture Global Practice Technical Assistance Paper TABLE 5.2. continued Risk Mitigation Transfer Coping Food crops Enabling Stabilize export policies (and related environment communication) so that market actors can operate in a predictable business environment Export crops High volatility Pursue production and price stability Insurance system of tobacco prices strategy, including enforcement of development could and production crop production planning, anticipated be an option to announcement of reserve prices, production volatility, intensification of crop diversification mostly if contract program, and strengthening of contract farming is further farming arrangements (for example, strengthened. reinforcement of integrated production system). Cotton price Strengthen cotton contract farming volatility and arrangements between ginners and untimely farmers, including inputs, technical availability of assistance, marketing, and so on. Follow a inputs loan-based system. Outgrowers’ Support greater pricing transparency price uncertainty and crop and income diversification, along the strengthening the contract farming extensive period arrangements between Illovo and the of sugarcane production trusts. harvesting and selling (April– November) Exchange rate Continue policy reforms that address volatility macroeconomic imbalances. interventions that would decrease the losses from risks. then used with the above gap analysis to narrow the pro- Thus potentially large savings exist in terms of avoided posed solutions to a short list for a solutions assessment. losses and expenditures saved on coping activities by real- Because the emphasis is placed on the more vulnerable seg- locating funds to risk-mitigating activities. ments of the supply chains, the proposed solutions would have a direct positive impact on reducing poverty. The Short List of Solutions. During the risk-assessment mis- sion, a consultative stakeholder meeting was organized to As discussed above, myriad ongoing projects relate to risk solicit feedback on the long list of solutions from private in Malawi. In brief, the short list of proposed solution and public sector stakeholders. Participating stakeholders areas comprises the following: were asked to grade each proposed solution according to its 1. Map out measures to strengthen agricultural infor- alignment with policy or business objectives; feasibility for mation systems so that they contain reliable data implementation in Malawi; affordability for the implement- useful for the development, monitoring, and eval- ing party (whether public or private); potential to be scaled uation of policies, and strengthen policy analysis up; and sustainability. The feedback from stakeholders was and monitoring and evaluation (M&E) capacity Malawi: Agricultural Sector Risk Assessment 41 TABLE 5.3. VALUE OF DONOR-FINANCED AGRICULTURAL PROJECTS BY TYPE OF ACTIVITY, 2004–13 Type of Mitigation Value of Project Type of Coping Value of Project Project (US$ Million) Project (US$, Millions) Irrigation 207.3 Drought and flood response 30.0 FISP 164.2 Emergency relief 78.1 Research 33.4 Grain storage 44.4 Natural resource 25.5 Emergency preparation 4.2 management Extension services 15.7 Coping and strategy support 12.0 Reforestation 4.1 Inputs 64.4 Production 74.9 Adaptation 10.7 Total 600.1 Total 165.7 Source: GOM Donor Database. Note: This table captures relevant donor-funded activities (grants and loans); some of these activities are incorporated under the government’s regular programs and are thus accounted for under MAFS’s budget. FIGURE 5.2. GOVERNMENT BUDGETARY in MAFS. Successful implementation of any risk- EXPENSES FOR RISK- management instruments will depend on the abil- ity to monitor the impacts of risks and to evaluate MITIGATING AND RISK-COPING the effectiveness of policies and investments. An INTERVENTIONS VERSUS assessment could comprise the following: LOSSES FROM RISKS » Identification of gaps in the current agricul- 160 NFRA Irrigation tural information system in terms of collection 140 WFP Cash Transfer Extension Services methods and the management of data. WFP Food Aid Research » Assessment of existing equipment and infor- 120 mation technology and a proposal for potential 100 investments in agricultural information systems US$ millions to strengthen agricultural policy development 80 and evaluation. 60 » Discussion of the technical skills needed for M&E 40 of policy and a proposal for areas for strengthen- ing these skills within relevant departments. 20 2. Assess measures to improve water management for 0 crop production to mitigate current and projected Mitigation Losses Coping future weather-related risks. Any analysis would Source: World Bank Ag. PER 2014. have to be conducted with existing land use/own- Note: Mitigation is calculated using an annual average of government expendi- ership structures in mind. An assessment could tures from 2008–12. Losses are an annual average from 1980–2012. Coping is an annual average of NFRA expenses to the WFP, DODMA, and ADMARC, comprise one or several of the following areas: as well as the amount spent in 2014 (the only year for which information was » The potential for expanding the use of small- available) for WFP food aid and cash transfer expenses. scale irrigation in Malawi and possible models 42 Agriculture Global Practice Technical Assistance Paper under which small-scale irrigation could be management. Many of the challenges in the sec- promoted. tor that relate to risks, from uptake of inputs and » The scope for improving on-farm practices, technology to inadequate investments in posthar- including conservation agriculture and mini- vest infrastructure, price uncertainty, and contrac- mum tillage methods. tual risks, could potentially be overcome through » Models for investing in on-farm water harvest- better organization of farmers. This assessment is ing infrastructure that would be applicable in proposed to include the following: the context of Malawi’s agriculture sector. » An assessment of existing farmers’ organiza- 3. Map existing functions and identify measures tions (formal and informal) in Malawi. to improve the coordination between the SRG, » A compilation of lessons from initiatives to ADMARC, and MVAC to better target existing organize farmers in Malawi, successful and coping mechanisms toward their intended benefi- unsuccessful, and conclusions regarding the ciaries, to improve predictability of interventions, determinants of their success. and to minimize market distortions. Such work » Proposals on how farmers’ organizations can could include the following: implement risk-management mechanisms in » An outline of the roles and responsibilities (for- practice, focusing on a few specific areas such mal and de facto) of SGR, ADMARC, and as adoption of new technology, price risks, con- MVAC, and proposed measures to strengthen tractual risks, and so on. their coordination. » An assessment of related food security policies, Which of these areas will be included in a solutions assess- including those of trade, market interventions, ment will be determined together with the government and grain subsidies. of Malawi. Ideally, the assessments will be conducted in » An analysis of the financial costs and economic teams including relevant technical staff from MAFS and impacts of these policies and if relevant, pro- other technical bodies to ensure that the analyses and pro- posed alternative policies that can more effi- posed solutions are in line with the priorities and needs ciently achieve the same objectives without of MAFS and/or other relevant institutions, and that the market distortions. knowledge acquired through the assessment remains with 4. Assess opportunities for strengthening farm- relevant staff. Preferably, any work will include gender- ers’ organizations for effective agricultural risk disaggregated assessments and proposals. Malawi: Agricultural Sector Risk Assessment 43 REFERENCES AfDB (African Development Bank). 2013. “Malawi Country Strategy Paper, 2013– 2017.” African Development Bank Group. http://www.afdb.org/fileadmin /uploads/afdb/Documents/Project-and-Operations/2013-2017%20-%20 Malawi%20-%20Country%20Strategy%20Paper.pdf. ASWAp. 2010. Agriculture Sector Wide Approach, Ministry of Agriculture and Food Security, Lilongwe. September 2010. ftp://ftp.fao.org/tc/tca/CAADP%20TT /CAADP%20Implementation/CAADP%20Post-Compact/Investment%20 Plans/National%20Agricultural%20Investment%20Plans/Malawi%20Post%20 Compact%20Investment%20Plan.pdf. Bardhan, P., and C. Udry. 1999. Development Microeconomics. Oxford: Oxford University Press. Christiaensen, L., and A. Sarris, eds. 2007. Rural Household Vulnerability and Insurance against Commodity Risk: Evidence from the United Republic of Tanzania. FAO CTTP10, Rome. Christiaensen, L., and K. Subbarao. 2004. “Towards an Understanding of Household Vulnerability in Rural Kenya.” Journal of African Economies 14 (4): 520–58. Dercon, S. 2000. “Income Risk, Coping Strategies and Safety Nets: Background Paper to World Development Report 2000/1.” WPS/2000.26. Dercon, S. 2001. “Assessing Vulnerability to Poverty.” Paper prepared for DfID (Department for International Development). Dercon, S. 2004. “Insurance against Poverty.” WIDER Studies in Development Eco- nomics, Oxford University Press, Oxford. Dercon, S. 2005. “Risk, Poverty and Vulnerability in Africa.” Journal of African Econo- mies 14 (4): 483–88. Devereux, S., B. Baulch, I. Maccauslan, A. Phiri, and R. Sabates-Wheeler. 2007. “Vulnerability and Social Protection in Malawi.” IDS Paper No. 387, Institute of Development Studies, Brighton. Devereux, S., B. Baulch, A. Phiri, and R. Sabates-Wheeler. 2006. “Vulnerability to Chronic Poverty and Malnutrition in Malawi.” A Report for DfID in Malawi. Ellis, F. 2003. “Human Vulnerability and Food Insecurity: Policy Implications, Forum for Food Security in Southern Africa.” Theme Paper. https://www.uea.ac.uk /polopoly_fs/1.53418!2003%20ffssa%20vulnerability.pdf. FAO Country Profile, accessed May 2014: http://faostat.fao.org/CountryProfiles /Country_Profile/Direct.aspx?lang=en&area=130. FAOSTAT (Food and Agriculture Organization of the United States). 2013. FAOSTAT database 2013. http://faostat3.fao.org/home/E. FAOSTAT (Food and Agriculture Organization of the United States). 2014. FAOSTAT database 2014. http://faostat3.fao.org/home/E. Grosh, M., C. Del Ninno, E. Tesliuc, and A. Ouerghi. 2008. For Protection and Promotion: The Design and Implementation of Effective Safety Nets. Washington, DC: World Bank. Malawi: Agricultural Sector Risk Assessment 45 Hay, E. R., and M. A. R. Phiri. 2008. “Situation Analysis of Disaster Risk Manage- ment Programmes and Practices, Final Draft.” World Bank Facility for Disaster Reduction and Recovery Track II: Malawi. Holzmann, R. 2001. “Risk and Vulnerability: The Forward Looking of Social Protec- tion in a Globalizing World.” Asian Development Bank, Manila. Kachule, R. N. 2011. “Performance of the Agricultural Sector in Malawi.” Agricul- tural Policy Research Unit Bunda College of Agriculture. http://community.eldis .org/.59ee3fb9/Perfomance%20of%20Agriculture%20Sector%20in%20Malawi.pdf. LTS International. 2013. “Mapping Land Cover and Future Land Cover Projec- tions: Scenario Analysis Technical Annex 1: Integrated Assessment of Land Use Options in Malawi,” 8. May. Makoka 2008. “Risk, Risk Management, and Vulnerability to Poverty in Rural Malawi.” Cuvillier Verlag, Gottingen. Makoka, D. 2011. “Identification of Key Vulnerable Groups in Malawi.” Report pre- pared for the World Bank under Effective and Inclusive Targeting Mechanism for Formal Social Support Programmes in Africa, July. Makoka, D. 2013a. “Smallholder Farmers’ Perception and Household Impact of the Malawi Farm Input Subsidy Program.” Center for Agricultural Research and Development (CARD), Lilongwe, University of Agriculture and Natural Resources. Makoka, D. 2013b. “Vulnerability and Resilience in Africa in the Context of People and Communities Living with HIV and AIDS: The Case of Malawi.” Report Prepared for the Resilient Africa Network (RAN). Makoka, D., and Kumwenda. 2013. “Study to Assess the Impact of Prolonged Dry Spells on the Livelihood of Vulnerable People in Balaka, Chikwawa and Nsanje District of Malawi.” Centre for Agricultural Research and Development (CARD). Report for UNDP Malawi, September. Mih, A. M., and G. I. Atiri 2003. Overview of Irish Potato Viruses and Virus Diseases chapter in “Plant Virology in Sub-Saharan Africa, Proceedings of a Conference Organized by the International Institute of Tropical Agriculture (IITA),” 4-8 June 2001, edited by Hughes J. d’A. and B. O. Odu, Nigeria. Monyo, E. S., et al. 2012. “Occurrence and Distribution of Aflatoxin Contamination in Groundnuts (Arachis hypogaea L) and Population Density of Aflatoxigenic Asper- gilli in Malawi.” Crop Protection, 42, 149–155. Moriniere, L., S. Chimwaza, and E. Weiss. 1996. “Malawi Vulnerability Assessment & Mapping (VAM) Baseline 1996: A Quest for Causality.” Lilongwe: USAID/ FEWS Project, and Arizona: University of Arizona, Office of Arid Land Studies. Mvula, P. et. al. 2011. Challenges of Access to Farm Input Subsidy by Vulnerable Groups in Malawi. London: Centre for Development, Environment and Policy, SOAS, Uni- versity of London, and Wadonda Consult. Mwafulirwa, N. D. 1999. “Climate Variability and Predictability in Tropical Southern Africa with Focus on Dry Spells over Malawi.” University of Zululand Institu- tional Repository. http://uzspace.uzulu.ac.za/handle/10530/898. National Food Reserve Agency (NFRA). March 2014. Interview with Chief Executive Officer Kelvin M’mangisa. 46 Agriculture Global Practice Technical Assistance Paper Ngwira, P., and P. T. Khonje. 2005. Managing Maize Diseases through Breeding under Malawi Field Conditions, African Crop Science Conference Proceedings, Vol. 6. 340-345: Uganda. http://www.acss.ws/Upload/XML/Research/41.pdf. NSO (National Statistical Office). 2012. Integrated Household Survey 2010–11: Household Socio-economic Characteristics Report. September. National Statistical Office, Zomba, Malawi. Pearce, J., A. Ngwira, and G. Chimseu. 1996. “Living on the Edge: A Study of the Rural Food Economy in the Mchinji and Salima Districts of Malawi.” London: Save the Children Fund (U.K.). Swift, J. 1989. “Why Are Rural Poor Vulnerable to Famine?” IDS Bulletin 20 (2): 8–15. USAID (U.S. Agency for International Development). 2013. Malawi Climate Change Vulnerability Assessment. Washington, DC: USAID. September. ———. 2013. Food Security Outlook. http://www.fews.net/southern-africa/malawi /food-security-outlook/october-2013. WDI (World Development Indicators). Accessed March 2014. http://data.worldbank .org/. WFP (World Food Programme). 2010. “Republic of Malawi, Comprehensive Food Security and Vulnerability Analysis (CFSVA).” WFP, Rome.http://documents.wfp .org/stellent/groups/public/documents/ena/wfp253658.pdf. World Bank. 2007. Malawi Poverty and Vulnerability Assessment: Investing in our Future. Washington, DC: World Bank. ———. 2009. Malawi: Economic Vulnerability and Disaster Risk Assessment. Vol. 1. Draft Report. World Bank, Washington, DC, August. ———. 2013. “Basic Agricultural Public Expenditure Diagnostic Review (2000–2013).” Malawi. Nov. 2013. ———. 2014. “Redefining the Goals and Objectives of the Farm Input Subsidy Program (FISP) in Malawi.” Policy brief. Malawi: Agricultural Sector Risk Assessment 47 APPENDIX A WEATHER-YIELD ANALYSIS MALAWI’S POLITICAL DISTRICTS To determine whether and the extent to which yield is affected by climatic events, a study was conducted on the relationship between several climatic events and differ- ent crops’ yield for Malawi. Malawi is divided into 28 political districts, as shown in map A.1. WEATHER INFORMATION IN MALAWI The Malawi Meteorological Services Department’s database of 23 weather stations has daily data from 1961–2011, or about 50 years of data from most of the stations. Map A.2 provides the geographic location of each weather station (red dots). DISTRIBUTION OF MONTHLY RAINFALL IN MALAWI Rain in Malawi follows a clear seasonal pattern throughout the year: most of the rain falls from November to March, with the months of May–September being generally dry. The period from January to March is usually the most humid across the entire country. All regions within the country follow this pattern with few varia- tions. Figure A.1 shows the average monthly distribution of rain for several weather stations: The pattern is evident in all stations: most of the rain falls from November to March, with approximately 200 mm falling per month; even though the period from May to September is generally dry, some stations receive almost no rain at all during these months. DROUGHT AND EXCESS RAINFALL ANALYSIS Once the yearly rainfall pattern has been established, it is useful to determine the annual variability in rainfall. To determine whether a year was dry or wet, the Malawi: Agricultural Sector Risk Assessment 49 MAP A.1. POLITICAL DISTRICTS IN MALAWI Using the standardized cumulative rainfall, drought and excess rainfall years are more clearly identified. Table A.1 shows the standardized cumulative rainfall by year and station; red blocks mean an extreme drought event (StdRain <–2); orange means a drought event (StdRain <–1); light blue means a light excess rainfall event (StdRain >1); and navy blue indicates an excess rainfall event (StdRain >2). Drought years were experienced in 1966, 1973, 1975, 1983, 1987, 1990, 1992, 1994, 1995, 1998, and 2005. During these 11 years, rain was more than one standard deviation below average at least five stations. The most extreme years were 1992, 1994, and 2005, when more than 10 stations were dry. The most extreme dry year was 1992, when 17 stations received less than 1 standard deviation less rain than average. The Mimosa, Mzuzu, Chitedze, and Makoka stations experienced the most severe droughts. The most recent dry year occurred in 2005, with most stations (18) having a negative anomaly of rainfall. Excess rainfall years were experienced in: 1961, 1962, 1963, 1974, 1976, 1978, 1979, 1980, 1984, 1985, 1986, 1989, 1996, 1997, 1999, and 2001. During these 16 Source: Wikimedia Commons. years, rainfall was more than one standard deviation Note: Districts: 1 = Dedza, 2 = Dowa, 3 = Kasungu, 4 = Lilongwe, 5 = Mchinji, 6 = Nkhotakota, 7 = Ntcheu, 8 = Ntchisi, 9 = Salima, 10 = Chitipa, 11 = above average for at least five stations, meaning that Karonga, 12 = Likoma, 13 = Mzimba, 14 = Nkhata Bay, 15 = Rumphi, 16 = rainfall was generally plenty during these years. The Balaka, 17 = Blantyre, 18 = Chikwawa, 19 = Chiradzulu, 20 = Machinga, 21 = Mangochi, 22 = Mulanje, 23 = Mwanza, 24 = Nsanje, 25 = Thyolo, 26 = most severe years were 1974, 1976, 1978, 1989, and Phalombe, 27 = Zomba, 28 = Neno. 1997. From 1974 to 1980, there was plenty of rainfall, because five of those seven years were extremely wet. standardized cumulative rainfall was calculated for each The most severe year in terms of excess rainfall was station, according to the following formula: 1989, when 13 stations had a positive anomaly, 6 of them extremely high. Mzimba, Chileka, Kasungu, KIA, ( Preci − mi ) Dedza, and Tembwe stations experienced the most rain StdRaini = in 1989. si Where StdRain, Standardized cumulative rainfall RAINFALL—YIELD Prec, yearly rainfall REGRESSIONS m, mean yearly rainfall A database of historical crop production information for s, standard deviation of yearly rainfall maize and cassava was provided by the MAFS statistical i, year bulletin. The database has 30 years of production and 50 Agriculture Global Practice Technical Assistance Paper FIGURE A.1. MONTHLY RAINFALL PATTERN FOR SEVERAL WEATHER STATIONS Karonga Weather Station Mzuzu Weather Station Salima Weather Station 300 300 300 292.8 323.8 250 250 250 287.6 229.2 252.2 200 200 213.1 200 190.2 192.9 180.6 184.6 184.6 177.8 178.7 150 164.3 150 150 100 100 100 81.3 78.1 50 53.3 50 50.0 50 31.2 30.3 36.3 25.8 25.2 4.3 1.8 0.5 0.3 1.5 12.5 13.2 9.7 1.6 0.4 0.5 0.2 6.3 0 0 0 br y ry ch ril ay ne A ly pt ust O er em r ec ber r br y ry ch ril ay ne Au ly pt ust O er em r ec ber r br y ry ch ril ay ne A ly pt ust O er em r ec ber r ov e be ov e be ov e be Fe uar Fe uar Fe uar Ju Ju Ju Ap Ap Ap ua ua ua b N tob b N tob b N tob ar M ar M ar M Ju Ju Ju Se ug Se g Se ug em em em em em em n M n M n M c c c Ja Ja Ja D D D Chitedze Weather Station Mangochi Weather station Thyolo weather station 300 300 300 271.0 250 250 250 244.5 229.7 216.4 200 200 201.5 200 188.6 190.5 181.6 172.5 150 150 150 143.9 135.8 135.0 100 100 100 96.9 74.5 77.1 50 50 50.6 50 39.5 35.8 32.3 23.0 24.3 10.4 1.6 0.8 0.2 1.7 9.8 5.0 3.1 2.8 4.0 1.7 14.3 18.0 11.2 8.0 0 0 0 Fe ary ry ch ril ay ne A ly pt ust O er em r ec ber r br y ry ch ril ay ne A ly pt ust O er em r ec ber r br y ry ch ril ay ne A ly pt ust O er em r ec ber r ov e be ov e be ov e be Fe uar Fe uar Ju Ju Ju Ap Ap Ap ua ua ua b N tob b N tob b N tob ar M ar M ar M Ju Ju Ju Se ug Se ug Se ug nu em em em em em em br M n M n M c c c Ja Ja Ja D D D Source: Authors, based on info from Malawi Meteorological Services Department. surface information for eight “regions”: Karonga, Mzuzu, Two different rainfall parameters were estimated for Kasungu, Lilongwe, Salima, Machinga, Blantyre, and each crop season (sowing, growing, and harvesting): Shire Valley. » Cumulative rainfall (CumRain)—the sum of daily pre- cipitation in millimeters (mm) for each of the sea- Because the geographic resolution of the crop data is sons described above; and not the same as the rainfall information, the following » Number of rainy events (Events)—the number of days convention was assumed: all weather stations close to in each season in which rainfall is greater than the region were used to determine a rainfall index for 5 mm. each region. Thus, the average of the available stations within a region were used as a proxy for each region’s To determine the relationship between yield and rain, lin- rainfall. Table A.2 shows which stations were used for ear regression models were run using both rain parame- each region. ters during each stage of the crop cycle as the explanatory variable for yield. The main objective of the regression Figure A.2 shows Malawi’s sowing calendar, which corre- analysis is to calculate the determination coefficient (R2) sponds to the November–March rains. Thus, three stages for each variable. The determination coefficient is a meas- were used to determine the relationship between rainfall ure of the proportion of the variability in yield explained and yield: a first stage from late October to December for by each rainfall variable. Therefore, a high R2 is a good the sowing season; a second stage from January to Febru- indication that the particular rain parameter and yield ary for the growing season; and a third stage from March are related. The results of the regression analysis for each to April for the harvesting season. crop and region follow. Malawi: Agricultural Sector Risk Assessment 51 TABLE A.1. RAINFALL ANOMALIES FOR MALAWI’S 23 WEATHER STATIONS Year Chitipa Mzimba NkhataB Salima Chileka Mimosa Ngabu Karonga Mzuzu Kasungu KIA Chitedze Nkhota Dedza 1957 –0.84 0.02 1.34 0.40 –0.67 1958 –1.42 –0.88 –1.33 –0.59 –0.82 –0.73 1959 0.40 –0.33 –1.22 0.18 –1.09 –0.37 1960 0.07 –0.32 0.27 –0.95 –0.28 –0.22 –0.45 1961 1.25 0.58 1.38 1.55 0.02 0.98 –1.52 1.86 2.71 0.22 0.22 1.59 1.53 1.40 1962 1.18 –1.38 1.24 –0.04 0.79 0.76 0.94 1.75 0.72 0.78 1.36 0.73 0.46 0.49 1963 0.46 0.10 –0.12 0.21 0.00 1.67 –0.11 2.31 1.20 0.24 –1.49 –0.55 1.10 0.96 1964 0.75 –0.54 –0.10 –0.63 –0.40 –1.12 –0.46 –0.53 –0.47 –0.44 –1.13 –0.59 –0.81 –0.71 1965 0.94 1.24 –0.60 –0.07 –0.92 0.04 –0.57 1.43 –0.07 –0.17 0.66 –0.45 0.90 0.36 1966 –0.81 –0.69 –1.01 –1.38 0.49 –0.80 –1.03 0.48 0.40 –0.35 –1.23 –1.56 –0.69 –1.52 1967 1.29 –0.56 –0.85 –0.12 0.35 –0.61 2.15 1.57 –0.10 –1.09 –0.80 0.68 –0.59 1968 1.50 –0.37 0.58 –0.40 –1.07 –0.66 –0.27 0.11 –0.03 –0.18 0.41 0.06 0.02 1969 –0.03 –1.44 –0.06 0.61 0.87 1.55 0.22 –0.33 –0.34 –1.11 –0.01 –0.21 0.53 –0.72 1970 1.58 –0.89 –0.53 0.26 –0.83 –0.75 –0.89 –0.29 –0.74 –0.69 0.12 0.53 0.23 –0.46 1971 0.67 2.32 –0.49 0.51 –0.26 –0.70 –0.70 –0.51 0.08 0.36 0.18 0.48 0.10 0.07 1972 –0.78 –0.91 0.09 –0.43 0.19 –0.96 –0.32 –0.35 –0.65 –1.73 –0.35 0.01 1.58 –0.78 1973 –1.31 –0.62 –0.66 –0.95 –0.87 –0.37 0.26 0.05 –0.93 –0.74 –1.86 –1.29 0.31 –1.20 1974 –0.74 1.69 0.36 1.31 2.11 1.59 0.45 1.83 0.18 0.78 1.96 1.92 0.96 0.65 1975 0.14 0.03 –0.30 –1.07 –1.90 –1.34 –0.24 0.16 1.21 –1.07 –0.44 0.51 –0.31 –0.21 1976 –0.56 1.56 1.95 1.62 1.21 1.10 0.75 –0.39 0.73 –0.04 0.43 0.42 1.72 0.42 1977 0.12 –0.61 –0.66 0.54 –0.07 –0.92 –0.62 0.56 –0.68 –1.02 1.17 1.25 0.59 0.24 1978 1.77 1.77 1.64 2.66 1.67 1.22 1.61 0.49 0.88 1.52 0.89 0.46 1.82 1.51 1979 1.30 0.13 1.16 –0.16 0.91 –0.34 –0.50 2.28 0.30 –0.50 1.02 –0.51 1.17 0.62 1980 –0.10 1.17 –0.08 0.57 –0.05 0.20 –0.79 –0.50 0.02 2.63 –0.25 –0.15 2.12 0.40 1981 –2.06 –0.33 0.31 –0.69 –0.81 0.40 –0.67 –0.61 –0.90 0.13 0.38 –0.08 –1.16 –0.37 1982 0.19 0.95 –0.22 –1.17 –0.30 0.11 0.66 –1.17 0.39 –0.23 –0.55 0.52 1.11 1.07 1983 –0.52 –0.19 –1.74 –0.52 –1.29 –1.14 –0.11 0.51 –0.59 –0.27 –0.76 –1.12 –1.12 –0.32 1984 2.24 0.55 –1.29 –0.95 0.95 0.78 1.32 0.45 1.37 –0.38 0.29 –0.06 –1.23 0.22 1985 0.58 0.10 –0.20 –0.21 2.42 0.71 1.09 –0.80 0.49 0.72 0.92 0.76 0.17 –0.02 1986 2.13 0.69 –0.69 0.04 1.04 0.73 0.40 –0.02 –0.01 1.11 –0.55 1.00 1.27 –0.17 1987 –0.78 –0.85 –1.81 –0.77 –0.56 –1.68 –1.76 –0.74 –1.24 –0.52 –1.43 0.33 –1.04 0.19 1988 –0.42 –0.65 0.14 2.15 0.28 0.83 1.39 –0.61 0.64 –0.07 –0.67 –0.16 –0.55 –0.06 1989 –0.01 3.13 0.65 0.85 2.13 1.88 1.49 –0.96 0.47 2.95 2.47 1.31 1.09 2.98 1990 –0.73 –0.33 0.29 –0.92 –1.45 –0.64 –1.17 –1.22 –0.99 –1.30 –0.97 –0.24 –0.40 –1.14 1991 –0.23 –0.35 0.44 –0.30 0.14 –0.41 –0.01 –0.55 0.78 0.22 0.36 –0.54 –1.15 1.92 1992 –0.16 –0.55 –1.52 –0.09 –1.22 –3.12 –1.55 –0.19 –2.18 –1.78 –0.30 –2.49 –1.19 –1.26 1993 –0.85 –1.75 –0.88 –0.57 –0.34 0.77 0.35 –1.32 –1.58 –0.43 –0.32 0.89 –0.64 0.81 1994 –1.39 –0.37 –0.89 –1.05 –1.36 –0.90 –1.62 –1.33 –0.92 –0.32 0.28 –0.78 –1.44 –0.98 1995 –0.46 0.39 0.38 –2.34 –0.26 –1.14 –0.04 –0.32 –0.84 –0.95 –1.16 –3.09 –1.17 –1.56 1996 –0.60 –0.19 1.52 0.04 0.54 0.83 0.08 –1.21 0.23 1.34 2.59 0.34 –0.65 0.47 1997 –0.25 –0.74 0.91 1.57 1.70 0.62 1.88 –0.33 –0.27 0.88 0.14 1.08 –0.22 0.99 1998 0.72 –1.54 –0.04 –0.45 –0.55 0.33 –0.21 0.13 –1.16 0.12 –0.69 0.34 –1.41 –1.92 1999 –0.39 0.72 2.44 –0.52 0.48 0.71 0.31 –0.78 2.01 –0.71 0.06 1.63 –0.46 –0.78 2000 –0.87 0.61 –0.96 –0.43 0.85 0.51 1.33 –0.01 –0.53 –0.88 –0.66 –1.19 –0.69 0.21 2001 –1.40 0.54 –0.31 1.64 0.52 1.24 2.29 –0.37 –0.85 –0.13 0.41 0.60 –0.74 –0.74 2002 0.02 0.12 1.63 0.41 –0.43 0.24 –1.19 0.77 1.71 –0.25 0.60 –0.44 0.16 –0.09 2003 –0.58 0.49 0.08 0.45 –1.02 –0.56 –0.59 –0.87 0.15 1.59 1.51 0.04 –1.07 1.36 2004 0.60 1.01 0.40 0.41 –0.03 0.24 1.18 0.09 1.60 0.24 0.83 –0.18 –0.94 2005 –1.18 –1.75 –1.77 –1.83 –1.52 –0.74 –0.78 –1.53 –0.81 –1.65 –1.42 –1.09 –1.56 2006 1.21 –0.48 0.54 1.34 0.99 0.00 –0.75 0.39 –0.21 –0.20 –0.33 –0.32 –0.15 0.96 2007 –0.34 –0.32 –0.59 0.71 0.28 –0.88 2.23 –1.26 0.57 0.91 –0.35 0.55 –0.22 2008 –1.32 0.00 0.35 0.31 –0.09 –0.39 0.18 –0.54 –1.17 –0.72 –0.44 –0.13 –0.06 ExtDro 1 0 0 1 0 1 0 0 1 0 0 2 0 0 Drought 7 5 8 6 9 6 7 6 6 6 8 7 11 7 Normal 35 39 34 38 36 36 33 34 35 35 33 34 25 34 Excess 10 8 9 8 7 7 9 8 7 7 7 7 10 6 Ext Exc 2 2 1 2 3 0 2 3 2 2 2 0 1 1 Prob Dry 13% 10% 16% 12% 17% 12% 14% 13% 13% 13% 17% 15% 24% 15% Prob Normal 67% 75% 67% 73% 69% 73% 67% 71% 73% 73% 69% 71% 54% 72% Prob Exc 19% 15% 18% 15% 13% 14% 18% 17% 15% 15% 15% 15% 22% 13% 52 Agriculture Global Practice Technical Assistance Paper Mangochi Bvumbwe Bolero Thyolo Makoka Chichri Tembwe Balaka Monkey ExtDry Drought Normal Excess ExtExc Conclusion 0 0 4 1 0 Normal 0 2 4 0 0 Normal 0 2 4 0 0 Normal 0 0 7 0 0 Normal 0.48 0.16 0 1 6 8 1 Excess 0.47 0.39 1.49 1.16 0 1 11 6 0 Excess 1.15 1.23 1.17 0.54 0 1 9 7 1 Excess –0.51 0.05 –1.05 –0.24 0 3 15 0 0 Normal –0.37 –0.35 3.85 0.20 –0.31 0 0 15 3 1 Normal 0.21 –1.57 –0.72 –0.79 –0.46 –0.82 –1.01 0 8 13 0 0 Drought 1.01 0.24 –0.41 –0.25 –0.06 –0.09 0.43 0 1 14 4 1 Normal –0.18 –0.76 0.09 –0.17 –0.73 –1.20 –0.64 0 2 17 1 0 Normal –0.77 0.84 –0.42 1.24 –0.65 1.82 –0.24 0 2 16 3 0 Normal –0.10 –0.52 0.08 –0.64 1.01 –0.63 –0.29 0 0 19 2 0 Normal –0.45 –0.94 0.69 –0.47 1.29 –0.68 0.27 0 0 18 2 1 Normal –0.74 –0.19 –0.64 –0.59 0.47 –0.19 –0.28 0 1 19 1 0 Normal –0.64 –0.36 –0.69 –0.74 –1.08 –0.52 –1.37 0 6 15 0 0 Drought 0.73 1.84 0.13 1.82 0.44 1.19 0.57 0 0 10 10 1 Ext Exc 0.64 –0.90 0.77 –1.25 –1.05 –0.91 –1.17 0 7 13 1 0 Drought 1.79 1.54 0.71 1.65 –0.99 1.57 2.51 0.00 0 0 10 11 1 Ext Exc –0.14 –1.47 0.11 –0.81 –0.01 –0.08 0.48 0.10 0 2 18 2 0 Normal 2.36 0.75 –0.25 0.92 0.48 0.94 2.68 1.29 0 0 6 13 3 Ext Exc 0.86 –1.32 0.34 –1.57 0.15 –0.30 0.01 –0.64 0 2 14 5 1 Excess –0.20 –0.47 1.12 –0.79 –1.69 –0.47 –0.54 –1.21 1.62 0 2 14 5 2 Excess –0.50 –0.62 –0.99 –0.56 0.07 –0.92 –0.53 –1.17 –0.56 1 3 19 0 0 Normal 0.04 0.26 –0.05 –0.06 –0.63 0.54 0.44 0.14 0.20 0 2 19 2 0 Normal –1.27 –0.65 –0.09 –1.32 –0.09 –1.44 –0.14 –0.08 –0.67 0 8 15 0 0 Drought 0.24 0.32 1.33 0.36 0.96 1.33 –0.09 –0.10 0.86 0 2 15 5 1 Excess –0.92 1.76 0.53 1.83 1.55 0.88 1.17 0.95 0.28 0 0 16 6 1 Excess –0.36 0.18 –0.02 0.30 1.64 1.14 0.92 1.24 –1.20 0 1 13 8 1 Excess –1.14 –0.90 –0.85 –1.06 –0.85 –0.77 –1.02 –0.26 –0.90 0 9 14 0 0 Drought –1.04 –0.40 –0.51 –0.29 0.84 0.16 0.11 –0.90 0.23 0 1 19 2 1 Normal 0.19 1.18 0.46 0.89 1.44 0.71 2.16 0.12 1.52 0 0 4 13 6 Ext Exc –0.76 –0.68 –1.76 –1.37 0.04 –1.60 –0.06 –0.88 0.03 0 8 15 0 0 Drought 0.66 –0.28 –0.24 0.19 0.54 –0.51 –0.58 0.38 –0.04 0 1 21 1 0 Normal –1.42 –1.50 –0.36 –1.59 –2.22 –1.46 –1.34 –1.90 –1.76 4 17 2 0 0 ExtDry –0.15 –0.05 –1.12 –0.23 –0.58 0.24 0.25 –0.42 –0.95 0 4 19 0 0 Normal –2.31 –1.50 –0.04 –1.31 –2.47 –1.26 –1.41 –0.43 –1.36 2 13 8 0 0 ExtDry –1.35 –0.64 –0.39 –0.49 –0.98 –0.19 –1.07 0.24 –1.78 2 9 12 0 0 Drought 0.13 1.06 0.08 0.12 0.58 –0.57 1.08 –0.08 0.94 0 1 16 5 1 Excess 1.67 2.90 0.16 1.88 2.34 1.68 –0.87 3.47 0.61 0 0 10 10 3 Ext Exc –1.42 –0.08 –1.61 –0.28 0.10 –0.54 –0.90 –1.00 –0.68 0 7 16 0 0 Drought –0.69 1.10 –0.95 1.33 0.10 1.48 1.16 1.00 –0.78 0 0 13 8 2 Excess 1.02 0.99 –0.62 0.28 –0.09 0.12 0.40 0.26 0.30 0 1 20 2 0 Normal 1.07 1.46 –0.80 2.03 0.48 0.61 0.73 0.20 1.95 0 1 13 7 2 Excess 0.19 –0.64 1.59 0.28 –0.96 2.10 –0.52 –0.90 0.83 0 1 17 4 1 Normal 0.74 –0.81 –0.38 –0.13 –0.47 –0.78 0.61 0.53 0.35 0 2 18 3 0 Normal 0.10 0.06 1.51 0.57 0.16 –0.27 –0.42 0.96 0.36 0 0 18 4 0 Normal –1.35 –1.23 –1.68 –1.49 –0.42 –1.46 –1.54 –1.13 –1.31 0 18 4 0 0 ExtDry 2.12 0.28 –0.07 0.84 0.72 0.94 0.67 0.89 0.47 0 0 19 3 1 Normal 1.19 0.53 0.61 0.84 1.36 0.64 –0.07 0.05 0.66 0 1 17 3 1 Normal –0.31 –0.27 –0.11 –0.78 0.01 –0.42 –0.57 –0.70 0.78 0 2 20 0 0 Normal 1 0 0 0 2 0 0 0 0 8 6 5 8 5 6 8 5 5 ExtDry 3 31 33 35 31 32 29 29 24 21 Drought 8 9 9 7 8 7 8 6 4 3 Normal 25 2 1 1 1 1 1 3 1 0 Excess 11 17% 13% 11% 17% 11% 14% 19% 15% 17% Ext Exc 5 65% 69% 74% 66% 73% 67% 67% 73% 72% 19% 19% 15% 17% 16% 19% 14% 12% 10% Malawi: Agricultural Sector Risk Assessment 53 TABLE A.2. WEATHER STATIONS USED IN EACH MAFS REGION Region Number Station 1 Station 2 Station 3 Station 4 Blantyre 1 Chileka Chichiri Bvumbwe Karonga 2 Chitipa Karonga Kasungu 3 Kasungu Nkhota Kota Lilongwe 4 KIA Chitedze Tembwe Machinga 5 Makoka Balaka Mzuzu 6 Bolero Mzimba Mzuzu Nkhata Bay Salima 7 Salima Dedza Mangochi Monkey Bay Shire Valley 8 Mimosa Thyolo Ngabu FIGURE A.2. MALAWI’S CROP CALENDAR Malawi. October OCT OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP Planting Winter planting Rainy season Winter harvest Green harvest Main harvest Winter harvrat Lean season Peak Peak labor demand Tobacco sales and auction labor demand OCT OCT NOV DEC JAN EEB MAR APR MAY JUN JUL AUG SEP Source: USAID 2013. variability in maize yield (R2 > 20%). In Salima, the com- MAIZE bined rainy events indexes help explain yield, whereas in Table A.3 summarizes the regression determination coef- the Shire Valley, the combined cumulative rainfall indexes ficient (R2) for each rain parameter by region. It shows that help explain yield best (24%). the relationship between cumulative rainfall and number of rainy events and maize yield is not significant, except in Figure A.3 illustrates the yield of maize for all regions the Salima region, where the number of rainy events for over time. It shows that except in the Salima and Shire the harvesting season explains 25 percent of yield variabil- Valley regions, where rainfall better explains the vari- ity. Because most of the determination coefficients are very ability in yield, there seem to be two different levels small, a multiple linear model was also run using each set of yield. There is not a linear upward trend, but rather of the three variables combined as regressors. Table A.4 two different levels of production, with a break point illustrates the results of these models. after 2005, when the level of production is clearly higher. Table A.4 shows that only for the Salima and Shire Valley Table A.5 illustrates the mean yield from 1984–2005 ver- regions do the two rain indexes barely significantly explain sus 2006–13 for each region. It shows that yield has been 54 Agriculture Global Practice Technical Assistance Paper TABLE A.3. SIMPLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR MAIZE YIELD No. Region CumRain1 CumRain2 CumRain3 Events1 Events2 Events3 1 Blantyre 5% 6% 0% 3% 5% 8% 2 Karonga 0% 1% 2% 0% 1% 0% 3 Kasungu 0% 1% 5% 0% 4% 6% 4 Lilongwe 3% 1% 3% 0% 0% 2% 5 Machinga 1% 3% 10% 0% 11% 10% 6 Mzuzu 3% 0% 1% 0% 1% 4% 7 Salima 2% 0% 14% 1% 1% 25% 8 Shire Valley 5% 17% 11% 4% 12% 6% TABLE A.4. MULTIPLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR MAIZE YIELD No. Region CumRain1+CumRain2+CumRain3 Events1+Events2+Events3 1 Blantyre 10% 11% 2 Karonga 4% 1% 3 Kasungu 7% 9% 4 Lilongwe 4% 3% 5 Machinga 14% 15% 6 Mzuzu 4% 5% 7 Salima 16% 26% 8 Shire Valley 24% 16% FIGURE A.3. MAIZE YIELD BY REGION, 1994–2013 3.5 3 Blantyre Karonga Kasungu Lilongwe Machinga Mzuzu Salima Shire valley 3 2.5 Yield (tons per hectare) Yield (tons per hectare) 2.5 2 2 1.5 1.5 1 1 0.5 0.5 0 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Season Season Source: Based on data sets provided to the World Bank by the Malawi Meteorological Services Department. almost two times higher from 2006 onward than in A.6 and A.7 show the determination coefficients of earlier years, perhaps explaining why rainfall does not these regressions. explain much of the variability in maize yield. To try to solve this problem, the standardized yield for each With the transformation of yield, rain explains more period of time was used instead of actual yield. Tables variability in yield. Particularly worth noting is that Malawi: Agricultural Sector Risk Assessment 55 TABLE A.5. AVERAGE MAIZE YIELD BEFORE AND AFTER 2005 BY REGION 1984–2005 2006–13 No. Region # Years Mean Yield (MT/ha) # Years Mean Yield (MT/ha) Ratio 1 Blantyre 22 1.090 8 1.962 1.8 2 Karonga 22 1.157 8 2.398 2.1 3 Kasungu 22 1.515 8 2.317 1.5 4 Lilongwe 22 1.123 8 2.033 1.8 5 Machinga 22 0.940 8 1.456 1.5 6 Mzuzu 22 1.221 8 2.201 1.8 7 Salima 22 1.224 8 2.112 1.7 8 Shire Valley 22 0.865 8 1.154 1.3 Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. TABLE A.6. SIMPLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR MAIZE YIELD TRANSFORMED No. Region CumRain1 CumRain2 CumRain3 Events1 Events2 Events3 1 Blantyre 1% 10% 0% 0% 7% 4% 2 Karonga 0% 1% 8% 0% 5% 2% 3 Kasungu 1% 1% 3% 0% 8% 4% 4 Lilongwe 7% 15% 27% 5% 18% 25% 5 Machinga 6% 6% 20% 6% 18% 24% 6 Mzuzu 3% 8% 28% 5% 2% 39% 7 Salima 15% 0% 9% 17% 1% 25% 8 Shire Valley 4% 21% 10% 4% 17% 4% TABLE A.7. MULTIPLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR MAIZE YIELD TRANSFORMED No. Region CumRain1+CumRain2+CumRain3 Events1+Events2+Events3 1 Blantyre 10% 9% 2 Karonga 10% 9% 3 Kasungu 5% 11% 4 Lilongwe 33% 32% 5 Machinga 34% 36% 6 Mzuzu 32% 39% 7 Salima 22% 37% 8 Shire Valley 26% 18% both cumulative rainfall and rainy events during the Kasungu, where the proportion of variance explained harvesting season help explain more yield variance in is less than 15 percent. A more detailed regional analy- almost all regions except for Blantyre, Karonga, and sis follows. 56 Agriculture Global Practice Technical Assistance Paper FIGURE A.4. MAIZE YIELD IN BLANTYRE, FIGURE A.5. REGRESSION RESULTS FOR 1994–2013 CUMULATIVE RAINFALL AND 3.5 MAIZE YIELD IN BLANTYRE 4 3 Yield (tons per hectare) 2.5 3 Standardized yield 2 2 y = 0.0011x-1.1912 R2 = 0.0665 1.5 1 1 0 0.5 –1 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 –2 0 200 400 600 800 1000 1200 1400 1600 Season Cumulative rainfall for the three stages (mm) Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin.. FIGURE A.6. MAIZE YIELD IN KARONGA, 1994–2013 BLANTYRE REGION 3 Figure A.4 shows maize yield in the Blantyre region over 2.5 time. The mean yield in the Blantyre region was 1 MT/ha Yield (tons per hectare) until 2005 (though in 1988 production was almost three 2 times more), but rose to 1.9 MT/ha after 2005. The three 1.5 worst seasons were in: 1991–92, when yield was 234 kg/ ha; 2004–05, when yield was 560 kg/ha; and 2009–10, 1 when yield was 1.4 MT/ha. 0.5 Figure A.5 shows that the relationship between cumula- 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 tive rainfall and yield for the Blantyre region is not strong. But some of the worst yield years can be explained by Season Source: Authors’ calculations, based on data from MAFS 2013 Annual the low cumulative rainfall during those seasons, as in Statistics Bulletin. 1991–92, 1994–95, and 2004–05 when total rainfall was about 600 mm and yield was relatively low. The number ha. Three dips on the chart indicate very low yields in the of rainy events during these three seasons was also small 1991–92, 1993–94, and 1996–97 seasons. (about 30 days throughout the whole seven months) com- pared with an average of 45 days in the whole 28 years As with the previous region, the relationship between of data. Hence, even though the relationship is not very rain and yield is not strong (5 percent), although the strong, drought explains why yield was low during those positive slope indicates that the more rain, the better the seasons. yield (figure A.7). For instance, rain during the 1993–94 and 1996–97 seasons was scarce (626 mm and 617 mm, respectively), explaining the low yield during those years; KARONGA REGION but the 1991–92 season, the lowest production year (444 In Karonga region, the difference in yield levels is more kg/ha), saw 933 mm of rainfall evenly scattered through evident (figure A.6). The mean yield before 2005 was the three stages, so drought does not explain such low 1.15 MT/ha, but since 2006 yield has been 2.39 MT/ yield during this season. Malawi: Agricultural Sector Risk Assessment 57 FIGURE A.7. REGRESSION RESULTS FOR FIGURE A.9. REGRESSION RESULTS FOR CUMULATIVE RAINFALL AND CUMULATIVE RAINFALL AND MAIZE YIELD IN KARONGA MAIZE YIELD IN KASUNGU 2.5 3 2 y = 0.0013x - 1.2424 2 1.5 R2 = 0.0486 y = 0.0005x - 0.5637 R2 = 0.0099 Standardized yield 1 Standardized yield 1 0.5 0 0 –0.5 –1 –1 –1.5 –2 –2 –2.5 –3 0 200 400 600 800 1000 1200 1400 1600 0 200 400 600 800 1000 1200 1400 1600 1800 Cumulative rainfall for the three stages (mm) Cumulative rainfall for the three stages (mm) FIGURE A.10. MAIZE YIELD IN LILONGWE, FIGURE A.8. MAIZE YIELD IN KASUNGU, 1994–2013 3 1994–2013 3 2.5 Yield (tons per hectare) 2.5 2 Yield (tons per hectare) 2 1.5 1.5 1 1 0.5 0.5 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Season Season Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. 1991–92 season (0.93 MT/ha) can be explained by low cumulative rainfall (689 mm) and few rainy events (32), KASUNGU REGION but a similarly “dry” season as in 1999–2000 (with 685 Yield in the earlier years in Kasungu region was the mm and 39 rainy events) had a much better yield of 2.07 highest with 1.5 MT/ha, but it also rose to 2.3 MT/ha MT/ha. During the 1996–97 season when yield was also after 2005 (figure A.8). The lowest yield years were the low (0.87 MT/ha), rain was normal with 912 mm and 1991–92 and 1996–97 seasons, when yield was less than 38 rainy events. 1 MT/ha. The determination coefficient (R2) is practically zero, LILONGWE REGION meaning that rain does not explain yield variability Figure A.10 illustrates that yield experienced a discrete in this region (figure A.9). The low yield during the jump after the 2005 season in Lilongwe region. Before then, 58 Agriculture Global Practice Technical Assistance Paper FIGURE A.11. REGRESSION RESULTS FOR FIGURE A.12. MAIZE YIELD IN MACHINGA, CUMULATIVE RAINFALL AND 1994–2013 2 MAIZE YIELD IN LILONGWE 1.8 2 1.6 Yield (tons per hectare) 1.5 1.4 1 1.2 Standardized yield 0.5 1 0 0.8 0.6 –0.5 0.4 y = 0.0057x - 1.0822 –1 R2 = 0.2672 0.2 –1.5 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 –2 –2.5 Season 0 50 100 150 200 250 300 350 400 Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Cumulative rainfall for the harvesting season (mm) Bulletin. mean yield was 1.12 MT/ha; afterward it was 2.03 MT/ FIGURE A.13. REGRESSION RESULTS FOR ha. Consistent with some of the conclusions in the regions discussed previously, the 1991–92, 1993–94, and 1996–97 RAINY EVENTS AND MAIZE seasons had the lowest yields over the entire time period. YIELD IN MACHINGA 3 Figure A.11 shows that rain during the harvesting season 2 explains 26 percent of the variability in yield. The positive slope indicates that drought during this period affected the Standardized yield 1 yield. There were three very dry years over this period: the 1993–94, 1994–95, and 2004–05 seasons, when rain 0 was less than 50 mm, consistent with relatively low yield –1 y = 0.1228x- 0.99 during those years. But the low yield during the 1991–92 R2 = 0.2386 season was not due to drought, because 186 mm of rain –2 fell during this season. –3 0 2 4 6 8 10 12 14 16 18 Rainy events for the harvesting season (days) MACHINGA REGION In Machinga region, the increasing trend in maize yield seems more gradual than that seen in the regions already discussed. Mean yield was 940 kg/ha before 2006 and rainy event occurred during the harvesting season, but 1.45 MT/ha after. As in other regions, the worst yields the sowing and growing seasons had an average number occurred in the 1991–92, 1993–94, and 2004–05 seasons of rainy events (16 and 14, respectively) and yield was (figure A.12). barely 662 kg/ha. It can be concluded that in this region, drought has mostly affected maize production during the In Machinga region, the number of rainy events dur- harvesting season. ing the harvesting season best help explain variability in maize yield (24 percent). The 1993–94 season was very dry, with only 4 rainy events, resulting in mean yield MZUZU REGION of 444 kg/ha (figure A.13). The 2004–05 drought dur- Mzuzu region in another region in which the discrete ing the harvesting season also affected yield: only one jump in maize yield is evident. Mean maize yield jumped Malawi: Agricultural Sector Risk Assessment 59 FIGURE A.14. MAIZE YIELD IN MZUZU, FIGURE A.16. MAIZE YIELD IN SALIMA, 1994–2013 1994–2013 3 3 2.5 2.5 Yield (tons per hectare) Yield (tons per hectare) 2 2 1.5 1.5 1 1 0.5 0.5 0 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Season Season Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. Bulletin. FIGURE A.15. REGRESSION RESULTS FOR is highly influenced by the 1998–99 season, in which yield RAINY EVENTS AND MAIZE was extremely good (2 MT/ha), corresponding to the YIELD IN MZUZU most humid season (28 rainy events and 808 mm of rain). 4 Most of the low yield years can be explained by the occur- rence of fewer rainy events (8). 3 y = 0.1368x- 1.896 Standardized yield R2 = 0.3881 2 SALIMA REGION As stated before, the rise in maize yield is not as clear 1 in the Salima region. The mean yield before 2005 was 0 1.22 MT/ha versus 2.1 MT/ha after. 2005 itself was generally a low yield year; mean yield increased in –1 2006–07 and again after 2009. As seen in other regions, the 1991–92, 1993–94, and 2004–05 seasons had the –2 0 5 10 15 20 25 30 lowest yields (figure A.16). Rainy events for the harvesting season (days) Similarly, the number of rainy events during the harvest- ing season best helps explain maize yield variability in from 1.22 MT/ha to 2.2 MT/ha after 2005, although the Salima region; even though the relationship is not very 1998–99 season yield was also high (1.99 MT/ha). Yield strong, the positive slope indicates that the higher the seems steadier in this region, but as in the other regions, rain, the better the yield—thus drought can be consid- the 1991–92, 1993–94, 1996–97, and 2004–05 seasons ered the main threat to production here. The 1993–94, had the lowest yields (figure A.14). 1994–95, and 2004–05 seasons each had approximately two rainy events, explaining the critically low yields in In Mzuzu region, the number of rainy events during the those seasons. But 1991–92 was not a dry season, so harvesting season helps explain maize yield variability (39 another reason may explain this year’s low yield (282 kg/ percent) more than in any other region (figure A.15). This ha) (figure A.17). 60 Agriculture Global Practice Technical Assistance Paper FIGURE A.17. REGRESSION RESULTS FOR FIGURE A.19. REGRESSION RESULTS FOR RAINY EVENTS AND MAIZE CUMULATIVE RAINFALL AND YIELD IN SALIMA MAIZE YIELD IN SHIRE VALLEY 2.5 2.5 2 2 1.5 1.5 y = 0.1484x- 1.3319 y = 0.0028x - 1.4478 1 Standardized yield 2 Standardized yield R = 0.2529 1 R2 = 0.2093 0.5 0.5 0 0 –0.5 –0.5 –1 –1 –1.5 –1.5 –2 –2 –2.5 –2.5 0 2 4 6 8 10 12 14 16 0 100 200 300 400 500 600 700 800 900 Rainy events for the harvesting season (days) Cumlative rainfall for the growing season (mm) FIGURE A.18. MAIZE YIELD IN SHIRE season was the driest, receiving 150 mm of rain during VALLEY, 1994–2013 the growing season, corresponding to the lowest yield 1.8 (245 kg/ha). The 2004–05 season was also dry (237 mm 1.6 of rain), explaining that year’s low yield of 415 kg/ha 1.4 (figure A.19). Yield (tons per hectare) 1.2 1 CASSAVA 0.8 Tables A.8 and A.9 summarize the regression determi- 0.6 nation coefficients for both the simple and multiple lin- 0.4 ear regression models, again using the three stages of 0.2 cumulative rainfall and rainy events variables by region. 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Tables A.8 and A.9 both show very small determina- Season tion coefficients, meaning that none of the different rain Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics indexes, even in the multiple linear regression models, Bulletin. help explain variability in cassava yield. Upon further review, cassava yield also had a discrete jump after 2000. SHIRE VALLEY REGION Table A.10 shows the mean cassava yield over 1984–2000 The rise in maize yield in Shire Valley region is not sig- versus 2001–13 for each region: nificant. The mean yield before 2005 was 865 kg/ha and 1.15 MT/ha after. As seen in previous regions, the Mean cassava yield increased from 3 MT/ha to about 1991–92, 1994–95, and 2004–05 seasons had the lowest 18 MT/ha; for some regions, yield was 4 times higher yields (figure A.18). after 2000. This difference in yield might explain why the determination coefficient is so low. The same transforma- In Shire Valley region, cumulative rainfall in the grow- tion applied to maize was hence also used for cassava. ing season has the highest impact on yield, explaining 21 Tables A.11 and A.12 show the determination coefficients percent of its variability. The positive slope indicates that for the simple and multiple linear models using the trans- the higher the rain, the better the yield. The 1991–92 formed cassava yield variable: Malawi: Agricultural Sector Risk Assessment 61 TABLE A.8. SIMPLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR CASSAVA YIELD No. Region CumRain1 CumRain2 CumRain3 Events1 Events2 Events3 1 Blantyre 6% 1% 3% 3% 0% 4% 2 Karonga 0% 1% 1% 3% 1% 0% 3 Kasungu 2% 0% 0% 0% 0% 2% 4 Lilongwe 0% 0% 4% 0% 2% 2% 5 Machinga 0% 0% 0% 1% 0% 0% 6 Mzuzu 0% 3% 1% 4% 6% 1% 7 Salima 10% 0% 2% 13% 0% 3% 8 Shire Valley 0% 1% 0% 0% 0% 0% TABLE A.9. MULTIPLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR CASSAVA YIELD No. Region CumRain1+CumRain2+CumRain3 Events1+Events2+Events3 1 Blantyre 9% 8% 2 Karonga 2% 5% 3 Kasungu 2% 2% 4 Lilongwe 6% 3% 5 Machinga 0% 1% 6 Mzuzu 4% 9% 7 Salima 13% 18% 8 Shire Valley 1% 0% TABLE A.10. AVERAGE CASSAVA YIELD BEFORE AND AFTER 2005 BY REGION 1984–2000 2001–13 No. Region # of Years Mean Yield (MT/ha) # of Years Mean Yield (MT/ha) Ratio 1 Blantyre 17 2.633 13 14.760 5.6 2 Karonga 17 4.320 13 18.582 4.3 3 Kasungu 17 3.639 13 16.650 4.6 4 Lilongwe 17 3.088 13 14.134 4.6 5 Machinga 17 3.320 13 12.768 3.8 6 Mzuzu 17 5.059 13 24.278 4.8 7 Salima 17 4.568 13 20.401 4.5 8 Shire Valley 17 3.968 13 12.184 3.1 Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. From Tables A.11 and A.12, it can be concluded variance explained, a more detailed regional analysis that even though the transformation of cassava yield follows. helped increase the determination coefficients, rainfall explains very little of cassava yield variability, except in BLANTYRE REGION the Blantyre region, where cumulative rainfall explains The mean yield in the Blantyre region increased about 40 percent. Because of the low proportion of from 2.6 MT/ha to 14.7 MT/ha, the highest relative 62 Agriculture Global Practice Technical Assistance Paper TABLE A.11. SINGLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR CASSAVA YIELD TRANSFORMED No. Region CumRain1 CumRain2 CumRain3 Events1 Events2 Events3 1 Blantyre 20% 23% 1% 12% 14% 8% 2 Karonga 5% 12% 2% 6% 8% 0% 3 Kasungu 5% 1% 0% 5% 0% 1% 4 Lilongwe 1% 0% 0% 4% 0% 1% 5 Machinga 0% 2% 3% 0% 3% 7% 6 Mzuzu 3% 4% 1% 0% 10% 0% 7 Salima 7% 1% 0% 4% 2% 2% 8 Shire Valley 0% 5% 1% 0% 4% 0% TABLE A.12. MULTIPLE LINEAR MODELS’ DETERMINATION COEFFICIENTS FOR CASSAVA YIELD TRANSFORMED No. Region CumRain1+CumRain2+CumRain3 Events1+Events2+Events3 1 Blantyre 40% 24% 2 Karonga 14% 12% 3 Kasungu 6% 6% 4 Lilongwe 1% 5% 5 Machinga 4% 8% 6 Mzuzu 9% 11% 7 Salima 9% 9% 8 Shire Valley 5% 5% increase of all regions. Yield was steady throughout FIGURE A.20. CASSAVA YIELD IN before 1997, oscillating about 2 MT/ha. The worst BLANTYRE, 1994–2013 season was 1991–92, when yield was barely 1.16 MT/ 25 ha (figure A.20). 20 Yield (tons per hectare) Cassava yield is most strongly correlated with rain in 15 the Blantyre region. The cumulative rainfall of the three stages helps explain about 30 percent of yield 10 variability, whereas the positive slope indicates that the more rain, the better the yield. It is clear that the 5 highest yield years were also the most humid ones, whereas the lowest yield years saw the least rain- 0 fall. The three driest years, 2004–05, 1994–95, and 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 1991–92 (when about 600 mm fell through the whole Season 7 month period), correspond to some of the lowest Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics yield years, so drought can be considered the main Bulletin. threat in this region (figure A.21). Malawi: Agricultural Sector Risk Assessment 63 FIGURE A.21. REGRESSION RESULTS FOR FIGURE A.23. REGRESSION RESULTS FOR CUMULATIVE RAINFALL AND CUMULATIVE RAINFALL AND CASSAVA YIELD IN BLANTYRE CASSAVA YIELD IN KARONGA 2.5 4 y = 0.0024x - 2.4885 R2 = 0.314 2 3 1.5 Standardized yield Standardized yield 2 1 y = -0.0041x +1.6104 R2 = 0.1183 0.5 1 0 0 –0.5 –1 –1 –1.5 –2 0 200 400 600 800 1000 1200 1400 1600 0 100 200 300 400 500 600 Cumulative rainfall for the three stages (mm) Cumulative rainfall for the growing season (mm) cates that the higher the rain, the worse the yield. This FIGURE A.22. CASSAVA YIELD IN KARONGA, regression is highly influenced by one outlying observa- 1994–2013 tion: the 2000 yield was very high despite the fact that 25 only about 200 mm of rain fell in that year. In general, however, it can be concluded that rain is of little impact 20 on cassava yield in this region (figure A.23). Yield (tons per hectare) 15 KASUNGU REGION In Kasungu region, yield follows an upward trend from 10 2000 onward; the mean yield before 2000 was 3.6 MT/ ha and has steadily increased to more than 20 MT/ha in 5 recent years. The 2001–02 season appears to have been inexplicably bad; yield decreased to 10 MT/ha despite 0 being higher before and after. However, rain was nor- 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 mal during this season (999 mm of cumulative rainfall Season and 45 rainy events), so rain does not explain this fall Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. in yield. No regression results are shown for this region because all determination coefficients were rather small (figure A.24). KARONGA REGION In Karonga region, cassava yield rose from an aver- LILONGWE REGION age of 4.32 MT/ha to 18.5 MT/ha, clearly showing a The discrete increase in cassava yield can be seen in completely different level after 2000. 2004–05 had the Lilongwe region: it increased from roughly 3 MT/ lowest yield in the post-2000 period, at only 15 MT/ha ha to 14 MT/ha. It is worth noting that the 2004–05 (figure A.22). season had relatively low yield, because the harvesting season was very dry (only 3 rainy days and 45 mm of Cumulative rainfall in the growing stage for cassava in cumulative rainfall). However, none of the rain indexes Karonga region has the highest determination coefficient were significant enough to explain the variability in cas- but it is barely 12 percent. Besides, the negative slope indi- sava yield (figure A.25). 64 Agriculture Global Practice Technical Assistance Paper FIGURE A.24. CASSAVA YIELD IN KASUNGU, FIGURE A.26. CASSAVA YIELD IN 1994–2013 MACHINGA, 1994–2013 25 16 14 20 Yield (tons per hectare) Yield (tons per hectare) 12 15 10 8 10 6 4 5 2 0 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Season Season Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. Bulletin. FIGURE A.25. CASSAVA YIELD IN FIGURE A.27. CASSAVA YIELD IN MZUZU, LILONGWE, 1994–2013 18 1994–2013 35 16 30 14 Yield (tons per hectare) Yield (tons per hectare) 12 25 10 20 8 15 6 4 10 2 5 0 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 Season Season Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. Bulletin. MACHINGA REGION In Machinga region, yield was fairly steady before 1996, MZUZU REGION oscillating about 2 MT/ha. In 1991–92, yield dipped to As in the other regions, the jump in cassava yield in the almost half that (1.14 MT/ha). Once the new level was Mzuzu region is evident after 2000. Again, the worst year reached, the 2001–02, 2002–03, and most importantly, was 2004–05, when yield was 19.8 MT/ha even though the 2004–05 seasons had low yields as well. As already the new mean was more than 24 MT/ha. Clearly some- stated, the 1991–92 and 2004–05 seasons were dry (for thing else affected yield during this season. The harvest example, in 2004–05, there was only 1 rainy event and season was the driest in this year (only 166 mm of rain in 43 mm of rainfall), explaining these seasons’ low yields. a region where 334 mm is normal), so the dry months of No regression results are shown for this region because March and April may explain the relatively lower cassava the determination coefficients were so small (figure A.26). yield in 2004–05 (figure A.27). Malawi: Agricultural Sector Risk Assessment 65 FIGURE A.28. REGRESSION RESULTS FIGURE A.29. CASSAVA YIELD IN SALIMA, FOR RAINY EVENTS AND 1994–2013 30 CASSAVA YIELD IN MZUZU 4 25 y = 0.0916x - 1.7756 Yield (tons per hectare) 3 R2 = 0.096 20 2 Standardized yield 15 1 10 0 –1 5 –2 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 –3 0 5 10 15 20 25 30 Season Rainy events for the growing season (days) Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. The number of rainy events during the growing sea- son in Mzuzu region had the highest determination coefficient but still only explained 10 percent of the FIGURE A.30. CASSAVA YIELD IN SHIRE variability in cassava yield, which is not significant VALLEY, 1994–2013 (figure A.28). 18 16 SALIMA REGION 14 Yield (tons per hectare) A similar pattern can be seen in the Salima region, which 12 had a steady yield of 4.5 MT/ha before 2000 and 20.4 10 MT/ha after. Similarly, 1991–92 had the lowest yield (1.4 8 MT/ha), explained by the low rain during the growing season (204 mm in a region where 485 mm are normal). 6 Since 2000, yield has increased steadily, with no shock 4 events, perhaps explaining why the relationship between 2 cassava yield and rain is not significant. No regression 0 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 results are shown for this region because all determination coefficients were insignificant (figure A.29). Season Source: Authors’ calculations, based on data from MAFS 2013 Annual Statistics Bulletin. SHIRE VALLEY REGION In Shire Valley region, the shocks of the 1991–92 and 2004–05 seasons are more extreme. Cassava yield during the 1991–92 season was only 269 kg/ha ver- Even though none of the rain variables were significant sus a mean yield of roughly 4 MT/ha. During the enough to explain cassava yield variability, a dry sowing 2004–05 season, yield decreased to 5 MT/ha, ver- season during 1991–92 (151 mm and 7 rainy events) and sus the post-2000 mean of 12 MT/ha. This indicates a dry harvesting season during 2004–05 (67 mm and 4 that something else affected cassava yield in this year rainy events) explain the relatively lower yields in these (figure A.30). years. 66 Agriculture Global Practice Technical Assistance Paper APPENDIX B CLIMATE AND CLIMATE CHANGE CURRENT CLIMATE AND WEATHER PATTERNS IN MALAWI Malawi has a subtropical climate (meaning hot, humid summers and mild win- ters) that is distinctly seasonal. The warm, wet season runs from November to March, during which most of the annual rainfall takes place. This is the main agricultural growing season. May to August is the cool, dry season; and Septem- ber and October constitute the hot, dry season. Although Malawi is a relatively small country, it has large variations in topography that create significant dif- ferences in temperature ranges and rainfall totals across the country, and thus a diverse range of agroecological zones. Higher elevations typically see cooler temperatures and more rainfall; for example, in the northern and southern high- lands. The hotter and drier zones are located at lower elevations, as is the case in the Shire River valley. The main drivers of rainfall are the Inter-Tropical Convergence Zone (ITCZ) and the Congo Air Boundary (CAB). The ITCZ is where the northern and southern hemi- spheres’ weather systems meet. The CAB is where Indian Ocean and southern Atlan- tic Ocean air masses meet. Flooding in Malawi is associated with both the ITCZ and CAB bringing rain at the same time. The rainy season in Malawi is demarcated by the passage of the ITCZ over the country. In normal years, the ITCZ begins to move across Malawi in late October, moving southward throughout November, and begins its return north in late March–April, marking the beginning of the dry season. Late arrival of the ITCZ means a late start to the rainy season, and an early departure means an early cessation. Intra-Seasonal Oscillations (ISOs), or dry spells of 10–60 days duration, can be caused by a number of atmospheric circulation patterns, includ- ing episodes of tropical cyclone disturbances east of Madagascar and high-pressure cells over South Africa. Figure B.1 shows annual temperature patterns, with warmer temperatures corre- sponding to the wet season and cooler periods to the dry season(s). Malawi: Agricultural Sector Risk Assessment 67 FIGURE B.1. AVERAGE MONTHLY a regular basis. Malawi’s climate also oscillates between TEMPERATURE AND RAINFALL decade-long wet and dry spells with a periodicity of 11.1 years. It is speculated that this longer oscillation is related IN MALAWI 400 mm 25.0°C to regular changes in sunspot activity. 300 mm 22.5°C GLOBAL CLIMATE CHANGE Temperature AND MALAWI Rainfall 200 mm 20.0°C The long waves of climate change are the permanent 100 mm 17.5°C shifts of average temperatures and precipitation caused by global increases in temperature brought on by increased 0 mm 15.0°C concentrations of greenhouse gases in the atmosphere Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec (that is, global climate change). Source: World Bank Group, Climate Change Knowledge Portal. See http://sdwebx .worldbank.org/climateportal/index.cfm?page=country_historical_climate& ThisRegion=Africa&ThisCCode=MWI#. Global climate change (hereafter referred to as climate change) is forecast to change temperatures and precipi- tation in Malawi over the next 50 years. The average CHANGES IN WEATHER annual temperature is forecast to increase 1°C–3.5°C, and the number of hot days12 is also forecast to increase. PATTERNS This level of increase is significant enough to raise Weather pattern deviations, meaning pronounced depar- evapotranspiration rates. The food crops and varieties tures from normal climate patterns, occur in Malawi over grown in Malawi that are heat intolerant will have trou- different time scales. The short waves of climate change ble absorbing sufficient moisture from the soil at those involve teleconnections, which are linkages between cli- temperatures. mate oscillations or anomalies that are widely separated across the globe. The changes they bring about are tem- Figures B.2 and B.3 show that monthly temperatures have porary, and generally happen within a one- to two-year already increased significantly over the past 100 years. time frame. In the medium term, analyses of rainfall data have shown that Malawi goes through several different The number of hot days is projected to increase signifi- wardyear cycles of wet and dry periods. cantly (figure B.4). Whereas from 1961–2000 the high- est scenario reported 18.9 hot days at its maximum, by Malawi’s climate is affected by several different telecon- 2046–65, the number jumps up to 28.3 days. The mean nections. Chief among them is El Niño/La Niña or El temperature is projected to change from 1 to 3 degrees Niño-Southern Oscillation (ENSO). El Niño events are every month from 2020 until 2039 (figure B.5). strongly connected with drought in Malawi, whereas La Niña is associated with unusually wet years. If there is an The distribution of rainfall is forecast to change in sig- El Niño event, the following growing season in Malawi is nificant ways. For example, more heavy rainfall days are 80 to 90 percent likely to experience a significant drought. anticipated. Figure B.6 shows the results of nine climate ENSO events change Malawi’s climate by causing changes change models for the 2020–39 time frame. According in the prevailing wind patterns. Other influential telecon- to the models, January and February will see markedly nections are the Quasi-Biennial Oscillation (QBO), which heavier rainfalls compared with current levels. involves oscillations of the wind patterns in the strato- sphere, and sea surface temperature (SST) anomalies in Extreme rain patterns are forecast to become more fre- the southern Atlantic and Indian Oceans. quent. Figures B.7 and B.8 show the projected number of days without rain and with extreme rain, respectively, in Some of these cycles appear to be to be associated with ENSO and QBO events, meaning they both happen on 12 A hot day is one that exceeds the hottest 10 percent of all days per year. 68 Agriculture Global Practice Technical Assistance Paper FIGURE B.2. AVERAGE MONTHLY FIGURE B.3. AVERAGE MONTHLY TEMPERATURE AND RAINFALL TEMPERATURE AND RAINFALL FOR MALAWI, 1900–1930 FOR MALAWI, 1990–2009 400 mm 25.0°C 400 mm 27.5°C 300 mm 22.5°C 300 mm 25.0°C Temperature Temperature Rainfall Rainfall 200 mm 20.0°C 200 mm 22.5°C 100 mm 17.5°C 100 mm 20.0°C 0 mm 17.5°C 0 mm 15.0°C Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Source: World Bank Group, Climate Change Portal. Source: World Bank Group, Climate Change Portal. FIGURE B.4. NUMBER OF HOT DAYS OVER A YEAR IN MALAWI, 1960–2000 AND 2046–65 20 Ensemble median (50%) 40 Ensemble median (10%) Ensemble low (10%) Ensemble low (50%) Ensemble high (90%) Ensemble high (90%) 10 20 Days Days 0 0 –10 –20 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Actual hot days for Malawi from 1961 to 2000 Projected hot days for Malawi from 2046 to 2065 Source: World Bank Group, Climate Change Portal. 2046–46 compared with 1961–2000. It can be seen that to cause outbreaks of pests and diseases in maize. There days without rain and days with heavy rain are expected is currently no ideal maize variety for the projected cli- to occur more often. mate change in Malawi. Improved and hybrid varieties, touted for their ability to grow in short seasons, still pos- IMPACTS ON CROPS sess numerous disadvantages over traditional varieties. For instance, they are more susceptible to prolonged dry One recent analysis13 of the projected effects of climate spells, are more easily introduced to pests in storage, and change on key crops in Malawi from 2020–60 reached the require fertilizer to attain yields similar to traditional vari- following conclusions: eties (USAID 2013, 42). Traditional varieties meanwhile Maize: There will be a high to very high likelihood are no panacea. Although they can produce favorable of decreased yield due to periods of extreme heat and yields in high temperatures, yield rates are very vulner- drought. On the other hand, increased rainfall is likely able to water stress and poor levels of micronutrients in soil (ibid., annex D, 2). 13 USAID 2013: “The Global Climate Models used to downscale climate change projections in the USAID report came from the 2012 Coupled Model Groundnuts: Increases in temperature and variable Inter-comparisons Project Phase 5 (CMIP5 [Taylor 2012]) archive. This archive precipitation decrease groundnut productivity. Heavy contains simulations of the historic and future climate yielded by multiple Global Climate Models (GCMs), assumes a range of emission scenarios, and is late rains promote aflatoxins, which limit export produced by the world’s leading climate modeling institutions.” potential. Additionally, pests and diseases become a Malawi: Agricultural Sector Risk Assessment 69 FIGURE B.5. PROJECTED MEAN TEMPERATURE IN MALAWI ACCORDING TO NINE CLIMATE CHANGE MODELS, 2020–39 Source: World Bank, Climate Change Portal. The World Bank graphs use the IPCC scenario A2. The A2 storyline and scenario family describe a very heterogeneous world. The underlying theme is self-reliance and preservation of local identities. Fertility patterns across regions converge very slowly, which results in continuously increasing population. Economic development is primarily regionally ori- ented and per capita economic growth and technological change are more fragmented and slower than in other storylines. For more information, see http://sdwebx.worldbank.org/climateportal/index.cfm?page=country_future_climate_down&ThisRegion=Africa& ThisCcode=MWI. greater risk for groundnuts under both decreased and consuming harvesting interferes heavily with other increased rainfall conditions. Of particular note is the crops. possibility of a groundnut rosette virus (GRV), which occurs in decreased rainfall conditions, and can cause Pigeon peas: Pigeon peas show favorable yields even in losses of up to 90 percent for the crop (ibid., annex D, areas with low moisture. However, earlier-maturing varie- 5). Uptake of early-maturing varieties of groundnuts, ties are more likely to show lower yields overall. Increased which perform better in low rainfall conditions, has rainfall raises the potential for greater bouts of diseases, been very low, perhaps due to the fact that their time- which thrive under such conditions. 70 Agriculture Global Practice Technical Assistance Paper FIGURE B.6. PROJECTED MEAN RAINFALL IN MALAWI ACCORDING TO NINE CLIMATE CHANGE MODELS, 2020–39 Source: World Bank, Climate Change Portal. FIGURE B.7. NUMBER OF DAYS WITHOUT RAIN BY MONTH, 1961–2000 AND 2046–65 Ensemble low (10%) Ensemble median (50%) Ensemble high (90%) 150 150 Ensemble high (90%) Ensemble median (50%) Ensemble low (10%) 100 100 Days Days 50 50 0 0 –50 –50 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Actual days without rain, 1961–2000 Projected days without rain, 2046–65 Source: World Bank Group, Climate Change Portal. Malawi: Agricultural Sector Risk Assessment 71 FIGURE B.8. NUMBER OF DAYS WITH EXTREME RAIN BY MONTH, 1961–2000 AND 2046–65 Ensemble median (50%) Ensemble low (10%) Ensemble high (90%) Ensemble high (90%) Ensemble low (10%) Ensemble median (50%) 150 10 100 5 Days Days 50 0 0 –50 –5 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Actual days with extreme rain, 1961–2000 Projected days with extreme rain, 2046–65 Source: World Bank, Climate Change Portal. Soybeans: Soybeans have very good drought toler- ance. However, they are also very sensitive during REGIONAL VARIATION particular portions of their growing cycle. Therefore OF CLIMATE CHANGE climate change may still be a source of stress for the IMPACTS plant, particularly at an early stage when it is drought intolerant. Although only slight decreases in productiv- RAINFALL In one study of the 2020–40 time period, rainfall ity are expected for soybeans, the potential is high for in the northern part of the country was predicted the increased prevalence of diseases under increased to remain at similar levels and frequencies, with the rainfall and warmer temperature conditions. This exception of a decrease in November rainfall levels. includes soybean rust, which affects all stages of the For 2040–60, however, the impact was much clearer— crop’s production. the dry season was predicted to extend to December and rainfall levels to increase in February and March Export crops are also likely to be affected by water and (USAID 2013). electricity shortages. Water availability, which is critical to crops such as sugar, is likely to be significantly affected in In the lakeshore area, most of the studied areas are pro- the country. On the whole, the country’s water balance is jected to have less rainfall in early and late summer from expected to drop by half by 2035.14 Adding to this pres- 2020–40. In the 2040–60 period, both November and sure will be the increased use of small-scale irrigation by December are expected to become drier, whereas January smallholder farmers, thereby reducing water sources for and February will be wetter (USAID 2013). large-scale (mostly export-heavy) irrigation systems. Elec- tricity, another key component for processing most export In the south, the 2020–40 and 2040–60 time periods crops, will also pose significant production challenges with show similar results. November and April will become the onset of climate change. Most of the country’s elec- drier, both in terms of days and rainfall levels, whereas tricity production is currently obtained through hydro- rain will increase in the same manner in February and power. Extended dry seasons, population growth, and March. The only difference between the two periods is increased demand will tax the already overburdened elec- that in 2040–60, total monthly rainfall is expected to tricity system. decline (USAID 2013). 14 Water balance here refers to availability-demand (USAID 2013, 4). 72 Agriculture Global Practice Technical Assistance Paper FIGURE B.9. CURRENT MINIMUM AND MAXIMUM TEMPERATURES IN MALAWI Source: Ministry of Natural Resources, Energy and Environment, Department of Climate Change and Meteo- rological Services. “Temperature Maps” found at http://www.metmalawi.com/climate/temperature.php. TEMPERATURE This trend is expected to continue to hold in the future, Current and projected future temperatures also vary by but will be exacerbated by hotter temperatures overall. region. Figure B.9 shows current minimum and maxi- Figure B.10 shows the results of nine climate change mum temperatures in Malawi. The south of the coun- models. try is significantly hotter than other parts of the country. Malawi: Agricultural Sector Risk Assessment 73 FIGURE B.10. RESULTS OF NINE CLIMATE CHANGE MODELS FOR THE NORTHERN, CENTRAL, AND SOUTHERN PARTS OF MALAWI Source: World Bank, Climate Change Portal. 74 Agriculture Global Practice Technical Assistance Paper APPENDIX C VULNERABILITY ANALYSIS15 CONTEXT The Malawi Vulnerability Assessment Committee16 divided the country into 11 live- lihood zones based on the livelihood options that households use to make a living (table C.1). In all livelihood zones, the main source of food is own crop production, often supplemented by food purchases from local markets. Poor households also often sell their household labor (locally known as “ganyu”) in exchange for food.17 In some livelihood zones, wild foods are also an important source of food, especially dur- ing lean periods. Crop sales remain an important source of cash for households in all the livelihood zones. In some zones where cash crops (such as tobacco and cot- ton) are widely grown, they provide an important source of cash for households. In all zones, food crop sales also contribute significantly to household incomes. Other important sources of income for the majority of households, especially the poor, include ganyu, self-employment, and sale of nonfarm products, such as firewood and charcoal. Table C.1 provides details of sources of food and cash in each of Malawi’s 11 livelihood zones. Vulnerability can be perceived as the existence and the extent of a threat of pov- erty and destitution (Dercon 2005). Regardless of how vulnerability is defined, its underlying factor is a sense of insecurity regarding the extent to which a shock or a hazard will result in a decline in household or community welfare (Makoka 2008). Although poverty is perceived as a static phenomenon, vulnerability is a forward- looking measure of household welfare. Poverty can therefore be defined as an ex ante measure, whereas vulnerability is an ex post measure of household well-being (Dercon 2001). 15 This appendix borrows heavily from Makoka 2011 and 2013b. 16 MVAC is a consortium of the Malawi government, NGOs, and UN agencies in Malawi and is chaired by the Minis- try of Economic Planning and Development. Its role is to provide accurate and timely information on food insecurity, thereby informing policy formulation, development programs, and emergency interventions to reduce food insecurity and vulnerability of the population. 17 The household economy approach distinguishes the sources of food mainly into “purchase,’” “own crops,” and “ganyu.” Malawi: Agricultural Sector Risk Assessment 75 TABLE C.1. LIVELIHOOD OPTIONS, KEY HAZARDS, AND RESPONSE STRATEGIES IN MALAWI’S 11 LIVELIHOOD ZONES Districts Response Under the Main Food Main Cash Strategies to Livelihood Zone Zone Sources Sources Key Hazards Hazards Central Karonga Karonga • Own crops (maize, • Sale of own crops • Dry spells affect • Increased livestock cassava, sweet (maize, cassava, rice, crop production sales potatoes, rice) sweet potatoes) • Flooding • Local sale of • Food purchase • Self-employment • Armyworms household labor • Food in exchange • Sale of livestock attack maize crop (ganyu) for labor (ganyu) • Sale of household • Sale of household • Own milk/meat labor (ganyu) assets • Reduced number of meals • Consumption of maize husks Western Rumphi Rumphi • Own crops • Sale of tobacco • Dry spells affect • Local and distant and Mzimba Mzimba (maize, pulses, • Sale of maize crop production ganyu sweet potatoes, • Sale of other • Newcastle disease • Increased groundnuts) crops (pulses, sweet affects chickens consumption of • Food purchase potatoes) • Highly volatile wild foods and • Food in exchange • Self-employment maize and roots for labor (ganyu) • Sale of livestock tobacco output • Sale of household • Sale of household prices assets labor (ganyu) • Extreme reduction in number of meals • Mzimba Self- Mzimba • Own crops • Sale of tobacco • Dry spells • Local and distant Sufficient (maize, cassava, • Sale of maize • Excessive rainfall ganyu sweet potatoes, • Sale of other crops and waterlogging • Increased pulses, millet) (cassava, sweet • Crop diseases consumption of • Food purchase potatoes, soybeans) • Cattle diseases less preferred food • Wild foods • Self-employment (for example, (cassava) • Own milk/meat • Sale of livestock foot-and-mouth) • Sale of household • Food in exchange • Sale of milk assets for labor (ganyu) • Sale of household • Excessive labor (ganyu) livestock sales • Extreme reduction in number of meals Nkhatabay Cassava Nkhatabay • Own crops • Sale of cassava • Flooding • Local and distant Karonga (cassava, maize, • Sale of bananas • Dry spells ganyu Rumphi sweet potatoes, • Sale of other crops • Crop pests (such • Sale of household Nkhotakota groundnuts, rice, (groundnuts, sweet as armyworms) assets pulses, bananas) potatoes, pulses) • Drought • Increased sale of • Food purchase • Small business nonfarm products • Food in exchange enterprises (firewood, fish) for labor (ganyu) • Sale of household • Expenditure labor (ganyu) switching 76 Agriculture Global Practice Technical Assistance Paper TABLE C.1. continued Districts Response Under the Main Food Main Cash Strategies to Livelihood Zone Zone Sources Sources Key Hazards Hazards • Kasungu- Mzimba • Own crops • Sale of tobacco • Waterlogging • Increased local Lilongwe Plain Kasungu (maize, sweet • Sale of maize • Dry spells and distant ganyu Lilongwe potatoes, • Sale of other crops • Livestock theft • Sale of household Dowa groundnuts, (groundnuts, sweet • Crop pests (such assets Ntchisi pulses) potatoes, soybeans, as armyworms) • Increased sale of Dedza • Food purchase pulses) • Wildfires nonfarm products Mchinji • Food in exchange • Small business • Drought (firewood, fish) for labor (ganyu) enterprises • Consumption of • Own milk/meat • Sale of household maize bran • Wild foods labor (ganyu) • Consumption of • Sale of livestock wild roots Southern Lakeshore Nkhotakota • Own crops • Fishing • Flooding • Increased local Salima (maize, rice, • Fishing ganyu • Dry spells and distant ganyu Mangochi cassava, sweet • Crop sales (rice, • Threat from • Migration potatoes, sweet potatoes, wild animals • Sale of household sorghum) maize, cassava) (elephants, assets • Food purchase • Small business hippos) • Eating less • Food in exchange enterprises • Drought preferred foods for labor (ganyu) • Self-employment • Reduction in • Own milk/meat (firewood sales, mat- number of meals • Wild foods making, and so on) • Sale of livestock Lake Chilwa and Machinga • Own crops • Sale of tobacco • Flooding (of • Increased local Phalombe Zomba (maize, rice, • Sale of maize Lake Chilwa) and distant ganyu Chiradzulu cassava, sweet • Sale of other crops • Dry spells • Sale of household Phalombe potatoes, pigeon (groundnuts, sweet • Drought assets Thyolo peas, sorghum, potatoes, soybeans, • Increased sale of Mulanje cowpeas) pulses) nonfarm products • Food purchase • Small business (firewood, fish) • Food in exchange enterprises • Consumption of for labor (ganyu) • Sale of household maize bran • Wild foods labor (ganyu) • Consumption of • Sale of livestock wild roots (goats) Southern Lakeshore Salima • Own crops • Sale of rice • Flooding • Increased local Dedza (maize, rice, • Sale of maize • Dry spells and distant ganyu Ntcheu cassava, sweet • Sale of other crops • Drought • Migration Mangochi potatoes, pigeon (groundnuts, sweet • Sale of household peas, sorghum, potatoes, soybeans, assets cowpeas) pulses) • Consumption • Food purchase • Livestock sales of less preferred • Food in exchange • Small business foods for labor (ganyu) enterprises • Reduction in • Wild foods • Sale of household number of meals • Own milk/meat labor (ganyu) (continued) Malawi: Agricultural Sector Risk Assessment 77 TABLE C.1. continued Districts Response Under the Main Food Main Cash Strategies to Livelihood Zone Zone Sources Sources Key Hazards Hazards Shire Highlands Machinga • Own crops • Fishing • Flooding • Increased local Mangochi (maize, rice, • Fishing ganyu • Dry spells and distant ganyu cassava, sweet • Crop sales (rice, • Threat from • Migration potatoes, pigeon sweet potatoes, wild animals • Sale of household peas, sorghum) maize, cassava) (elephants, assets • Food purchase • Small business hippos) • Consumption • Food in exchange enterprises • Drought of less preferred for labor (ganyu) • Self-employment foods (firewood sales, mat- • Reduction in making, and so on) number of meals Middle Shire Valley Blantyre • Own crops • Sale of cotton • Flooding • Increased local Mangochi (maize, rice, • Sale of pigeon peas • Dry spells and distant ganyu Balaka cassava, sweet • Sale of other crops • Drought • Sale of household Zomba potatoes, pigeon (rice, sweet potatoes, assets Mwanza peas, sorghum, soybeans, pulses) • Increased sale of Neno cowpeas) • Livestock sales nonfarm products • Food purchase • Fish sales (firewood, fish) • Food in exchange • Sale of charcoal/ • Consumption of for labor (ganyu) firewood maize bran • Own milk/meat • Sale of household • Consumption of labor (ganyu) wild roots • Thyolo-Mulanje Thyolo • Own crops • Sale of pigeon peas • Dry spells • Increased local Tea Estates Mulanje (maize, cassava, • Sale of other crops • Drought and distant ganyu sweet potatoes, (sweet potatoes, • Banana diseases • Migration pigeon peas, cowpeas, bananas) • Sale of household cowpeas, • Livestock sales assets bananas) • Fish sales • Consumption • Food purchase • Sale of charcoal/ of less preferred • Food in exchange firewood foods for labor (ganyu) • Sale of household • Reduction in • Own milk/meat labor (ganyu) number of meals Lower Shire Valley Chikwawa • Own crops • Sale of cotton • Flooding • Increased local Nsanje (maize, rice, • Sale of pigeon peas • Dry spells and distant ganyu millet, sweet • Sale of other crops • Drought • Sale of household potatoes, pigeon (rice, sweet potatoes, • Livestock diseases assets peas, sorghum, soybeans, pulses) • Increased cowpeas) • Livestock sales livestock sales • Food purchase • Fish sales • Increased sale of • Food in exchange • Sale of household nonfarm products for labor (ganyu) labor (ganyu) (firewood, fish) • Own milk/meat • Eating less preferred foods • Consumption of wild roots 78 Agriculture Global Practice Technical Assistance Paper In the context of this study, vulnerability is a term used 2010). Using IHS2 data, Devereux and others (2006) were to describe exposure to hazards and shocks. Literature able to show that poor households who experience shocks highlights the fact that vulnerability is a product of two are more likely to experience a decline in well-being than components: exposure to a hazard (a shock) and resilience (the nonpoor households who experience the same number of ability to manage the hazard) (Devereux et al. 2006). shocks. An assessment of the major types of shocks facing COMMON SHOCKS FACED BY Malawian households shows that climate and environ- MALAWIAN HOUSEHOLDS mental shocks (such as droughts and floods) and eco- nomic shocks (such as rising food prices, falling prices for Households in Malawi face a wide range of shocks, most cash crops, household business failure) are the underly- of which threaten their livelihoods and their survival. ing factors contributing to high vulnerability in Malawi. Shocks are defined as adverse events that lead to a loss For instance, using data on 12,288 households collected of household welfare via a reduction in consumption, during IHS3 from 27 districts of Malawi between 2010 income, and/or a loss of productive assets (Dercon 2005). and 2011, NSO (2012) shows the major type of shocks Shocks are classified into two groups: idiosyncratic shocks, reported by households (table C.2). Among the most com- which are household specific, such as death and illness; mon shocks are: drought (reported by 38.7 percent of the and covariate shocks, which are communitywide, affect- households); the high cost of agricultural inputs (reported ing all households. Examples include floods, drought, and by 26.2 percent); and unusually high prices of food agricultural pests and diseases, among others (Makoka (24.5 percent). As table C.2 shows, floods (reported by only 2008). These shocks may push an already poor household 3.5 percent of the population) and crop pests and diseases deeper into poverty or drive a nonpoor household below (5.2 percent) are less common shocks. The statistics also the poverty line (Grosh et al. 2008). show that the proportions of female-headed households Households in Malawi, especially those residing in the that face various shocks are similar to those of male- rural areas, live in environments where shocks are com- headed households (table C.2). mon. In particular, smallholder farmers who are depend- ent on rain-fed agriculture in Malawi often cope not only with severe poverty but also extremely variable incomes KEY GROUPS VULNERABLE because of the wide range of shocks they face (Bardhan TO VARIOUS SHOCKS and Udry 1999). Studies have shown that the majority Vulnerable groups are defined as individuals or house- of rural households in Malawi are exposed to a num- holds characterized by exceptionally low levels of income ber of shocks, most of which are livelihood threaten- or high levels of poverty (World Bank 2007). Grosh and ing. For example, using Integrated Household Survey 2 others (2008) identify vulnerable groups as individuals who (IHS2) data, the Malawi government and the World Bank face special difficulties in supporting themselves because (2007) report that 95 percent of the sampled households of some particular aspect of their situation. According reported experiencing at least one shock in the past five to the authors, these groups typically include the elderly, years. Further, literature suggests that urban households orphans, widows, people with disabilities, people with tend to experience fewer shocks than rural households. HIV/AIDS, refugees, and internally displaced persons, For example, in the IHS2 data, about 60 percent of urban among others. households reported experiencing three or fewer shocks, whereas over 75 percent of rural households reported The Malawi Growth and Development Strategy (MGDS) encountering four or more shocks in the last five years (2006–11) provides an excellent exposition of vulnerable (World Bank 2007). In the WFP study of 2009, whereas groups in Malawi. The MGDS defines the most vulner- 36 percent of rural households reported not experienc- able as including individuals or households affected by ing a shock, 29 percent reported experiencing one shock, disasters; households headed by orphaned children, the and 35 percent experienced more than one shock (WFP elderly, and single parents (especially female headed); Malawi: Agricultural Sector Risk Assessment 79 TABLE C.2. PROPORTION (%) OF HOUSEHOLDS SEVERELY AFFECTED BY SHOCKS DURING THE PAST 12 MONTHS BY LOCATION, SEX, AND REGION IN MALAWI, 2011 Place of Shock Residence Sex Region Total (%) Urban (%) Rural (%) Male (%) Female (%) North (%) Central (%) South (%) Drought/irregular rains 37.8 9.1 43.1 36.2 42.8 27.9 17.3 58.3 Unusually high costs of 26.2 8.5 29.5 26.1 26.4 26.0 36.5 17.3 agricultural inputs Unusually high prices for 24.5 17.7 25.7 23.8 26.5 24.8 26.2 22.9 food Unusually low prices for 12.2 2.0 14.1 12.9 10.0 10.1 20.4 5.6 agricultural output Serious illness or accident 11.5 6.2 12.5 11.6 11.1 10.0 12.7 10.8 of household member Unusually high level of 5.7 1.1 6.5 6.0 4.9 6.8 7.7 3.7 livestock disease Theft of money/ 5.6 5.6 5.6 5.6 5.8 3.2 6.0 5.9 valuables/assets/ agricultural output Unusually high level of 5.2 0.7 6.0 5.3 4.8 3.3 8.2 3.0 crop pests or disease Floods/landslides 3.5 1.1 4.0 3.6 3.5 5.3 4.7 2.1 Conflict/violence 3.2 3.3 3.2 3.1 3.8 1.9 3.7 3.2 Death of other household 3.1 2.6 3.2 2.8 4.1 2.1 3.0 3.5 member(s) Earthquakes 2.9 2.7 2.9 3.0 2.4 14.7 2.3 0.2 Break-up of household 2.4 1.2 2.6 1.2 6.1 1.7 2.0 2.9 Birth in the household 2.3 1.6 2.4 2.6 1.2 2.7 2.2 2.3 Reduction in earnings 1.7 2.9 1.5 1.8 1.6 1.4 1.4 2.1 from household End of regular assistance/ 1.6 0.6 1.7 1.2 2.6 1.0 1.6 1.7 aid/ remittances from outside Household 1.5 2.1 1.4 1.6 1.2 2.0 1.2 1.6 (nonagricultural) business failure Death of income earner(s) 1.2 0.6 1.3 0.5 3.4 1.0 1.0 1.5 Reduction in the earnings 0.9 2.1 0.7 1.0 0.5 0.3 1.0 1.0 of currently salaried household member Loss of employment 0.7 1.1 0.7 0.9 0.3 0.4 0.6 0.9 of previously salaried member Other 1.9 2.1 1.8 1.9 1.7 1.6 2.0 1.8 Source: Makoka 2013b. 80 Agriculture Global Practice Technical Assistance Paper persons with disabilities; children under five and lactating that larger households20 and households with more young and pregnant mothers; orphans in streets, orphanages, children are more likely to be ultrapoor. Box C.1 high- foster homes, and extended family member households; lights the major gender vulnerabilities to which widows, the unemployed and underemployed in urban areas; and divorced women, and female-headed households are sub- the land constrained in rural areas. However, the MGDS ject in Malawi. emphasizes that not all individuals in the above categories are classified as most vulnerable. The determining factor Ellis (2003) also notes that children under the age of is made based on their inability to meet their basic needs five are a key group vulnerable to undernutrition, mal- and on the basis of poverty characteristics. nutrition, and infectious diseases. Further, child-headed households are an important vulnerable group in Malawi. Grosh and others (2008) highlight that vulnerable groups A child-headed household may be defined as a household tend to have a low level of education, are poorly inte- characterized by a child under age 18 years acting as a guardian for grated in the labor market, and own few assets. Further, siblings, relatives, and other children. Child-headed households many vulnerable groups face discrimination, making it are vulnerable because the head is not old enough to take even more difficult to generate independent income to over the responsibility of looking after siblings and taking support themselves. It is important to note that different care of household affairs.21 vulnerable groups face problems specific to that group. Ellis (2003) describes vulnerable groups as those “living FACTORS INCREASING on the edge.18 VULNERABILITY TO SHOCKS Using ultrapoverty as a proxy for vulnerability because Limited Livelihood Options: In all the districts of of data limitations, the Malawi Poverty and Vulner- Malawi, the majority of the population is dependent on ability Assessment report of 2007 identifies a number of rain-fed agriculture. However, in many livelihood zones, ultrapoor households. Female-headed households were found the annual precipitation rates are usually not sufficient to to be significantly more likely to be ultrapoor, and are support rain-fed food production. As a result, dry spells therefore seen as one of the vulnerable groups in Malawi are a frequent hazard that affects food production (see (World Bank 2007).19 A number of other studies also clas- table C.1). For the households to be able to withstand food sify female-headed households as a vulnerable group, insecurity-related shocks, livelihood opportunities must including Grosh and others (2008) and Christiaensen and exist outside rain-fed agriculture. Households that have Subbarao (2004). In his study of vulnerability in southern access to land along rivers are able to grow maize and Africa, Ellis (2003) argues that female-headed households other crops along the wetlands, thereby widening their are vulnerable because of women’s lack of access rights sources of food and cash. However, the majority of house- to land and their lack of time to cultivate land, among holds that do not have such access. Further, households’ others. Along the same line, widows and divorced women are reliance on ganyu is also conditional on rainfall since the classified as vulnerable because of loss of a previous part- ganyu is usually provision of farm labor. Lack of adequate ner’s contribution to household livelihood (Ellis 2003). Malawi Government and World Bank (2007) also report 20 Although larger households are associated with increasing vulnerability, some studies have found larger family size associated with decreasing vulnerability to poverty, including Christiaensen and Subbarao (2004) in their study of vulner- ability in rural Kenya. The authors argue that larger household size may reduce 18 The phrase “living on the edge” provides a graphic image of the livelihood household vulnerability because of the larger supply of labor, which may be circumstances that vulnerability tries to convey (Ellis 2003). It was first used as a useful during periods of consumption shortfall. title of a Save the Children report (namely, Pearce, Ngwira, and Chimseu 1996). 21 Factors contributing to the rising phenomenon of child-headed households in 19 It is important to note that using the IHS2 data, female-headed households Malawi include frequent deaths due to HIV/AIDS; abject poverty; the weaken- were also found to be poorer than male-headed households in Malawi. Holding ing of the extended family support system; poor long-term planning for fami- all other factors constant, a female-headed household had 14 percent less con- lies; and the lack of adequate support to the existing community-based OVC sumption per capita than a male-headed household (World Bank 2007). structures (Makoka 2011). Malawi: Agricultural Sector Risk Assessment 81 BOX C.1. GENDER VULNERABILITY IN MALAWI rate was highest in Chikwawa (81.6 percent) and lowest in Nkhotakota (32.1 percent), with a national average rate » Women make up 70 percent of the agricultural labor of 50.7 percent. Poverty remains a more serious problem force but are less likely to engage in cash crop pro- duction because of labor and time constraints. in the rural areas, where 56.6 percent of the population » In 2005, a female-headed household had 14 percent is estimated to live below the national poverty line, ver- less consumption per capita than a male-headed sus 17.3 percent in urban areas (2011). Regionally, the household, according to the Malawi Poverty and poverty rate is highest in the southern region (55.5 per- Vulnerability Assessment Report. cent), followed by the northern region (54.3 percent); it is » The value of assets owned by male-headed house- lowest in the central region (44.5 percent). Studies have holds is more than double that of female-headed shown that households are vulnerable to food-insecurity households and male-headed households are more likely to own agricultural assets. shocks because of their poverty situation (Makoka and » Women’s rate of pay for ganyu is likely to be only two- Kumwenda 2013). In particular, poverty makes them thirds the rate paid to men. susceptible to any food-related shock as they do not have » Women face more difficulties in accessing credit, the capacity to prevent the shock or to manage its effects because many do not possess the assets required as when it occurs. This is a more serious problem for female- collateral. headed households, as 57 percent of people living in » According to the 2008 Malawi Population and Hous- ing Census, 59 percent of women were literate com- female-headed households are poor, versus 49 percent in pared with 69 percent of men. male-headed households (2011). » Unequal employment opportunities exist between men and women outside the agriculture sector in Limited Productive Assets: Another key factor that is Malawi. For example, according to the 2010 Malawi a major source of vulnerability to a range of idiosyncratic Millennium Development Goal Report, the share of and covariate shocks is households’ limited assets. There women in wage employment outside the agriculture sector was only 15 percent in 2006, and is projected is vast literature on the use of household assets to pro- to be 18.8 percent in 2015. tect households from shocks (see Dercon 2000; Makoka » As household assets are depleted, women are more 2008). Many households do not have assets to cushion likely to engage in sexual transactions and other risky themselves against a range of shocks, including drought. behaviors to meet household subsistence needs. Productive assets, including livestock, are an important » Women and girls typically take on the burden of car- source of livelihood, especially in the face of shocks. As ing for sick family members. » Young girls are more likely than young boys to be table C.1 shows, in some livelihood zones (such as West- withdrawn from school to care for younger siblings or ern Rumphi and Mzimba, Mzimba Self-Sufficient, and the sick and to assist with domestic and agricultural Lower Shire Valley), households depend on livestock as work following a livelihood shock to the household. a source of food and cash. They are able to respond to » Female-headed households are more dependent on shocks by increasing the sale of their livestock. Initiatives external support (gifts from relatives, food aid, pub- that build households’ asset base would therefore be effec- lic works programs) for subsistence than are male- tive in ensuring that households’ vulnerability to various headed households. » Women are rarely represented on councils of elders, livelihood shocks is minimized. and so are unable to influence decisions over access to land and inheritance rights, among others. Low Own-Food Production: As table C.1 shows, the Source: Adapted from Hay and Phiri 2008. main source of food across all livelihood zones is own pro- duction. However, in many households, own-food produc- tion is too low to last the whole food consumption year. livelihood options outside agriculture is therefore a key Therefore, they depend on the market to fill their food source of vulnerability in many livelihood zones. gap. Unfortunately, the majority of food-deficit house- holds do not have the financial capacity to get sufficient Poverty: Poverty is an important driver of vulnerability food from the market. This makes them more vulnerable in Malawi. According to the IHS3 of 2011, the poverty to any food-related shock. Prolonged dry spells, droughts, 82 Agriculture Global Practice Technical Assistance Paper unreliable rainfall, lack of inorganic fertilizer, and poor sources for rural households reported by the Malawi gov- soils are all factors responsible for low own-food produc- ernment and the World Bank (2007) include tobacco sales tion. For female-headed households, low landholdings (16 percent of households), nontobacco crop sales (53 and lack of household labor exacerbate the problem of percent), livestock sales (30 percent), and informal sale of low own-food production (see box C.1). household labor (ganyu) (52 percent), among others. Illnesses Due to HIV and AIDS: HIV/AIDS-related Other ex ante risk-management strategies include illnesses in communities are another important factor migration of household members as a way of diver- contributing to the high vulnerability of households to sifying income for the household (World Bank 2007). food-related shocks. Illnesses disrupt households from However, although migration can be seen as a means undertaking productive activities. At times, even healthy for individuals to seek new opportunities and to members of the household, especially women, withdraw diversify income sources for the household, it can household labor to nurse sick relatives. also arise due not to economic reasons but fam- ily issues. In the IHS2 data, most reported migration RISK-MANAGEMENT was related to family issues, such as marriage and divorce (World Bank 2007). It is important to note, STRATEGIES TO MAIN however, that migration can be undertaken ex ante or SHOCKS ex post. There is evidence in the literature that vulnerable groups undertake different risk-management strategies in the Further, informal insurance (via village savings and loan face of shocks. A distinction is made between the strate- groups [VSLs]) to protect households against future gies undertaken before a shock occurs—ex ante risk-man- shocks is known to exist in Malawi but has not been accu- agement strategies—and those taken after a shock has rately captured by nationally representative data. Never- already occurred—ex post coping strategies. The goal of theless, informal group-based insurance schemes, as well ex ante risk-management measures is to prevent the shock as formal group-based lending facilitated by microfinance from occurring, or if prevention is not possible, to miti- institutions, are an important source of household income gate the effects of the shock (Holzmann 2001; Makoka that reduces the impact of shocks when they occur. In 2008). Studies have shown that households’ level of eco- many communities, VSLs are usually used by women to nomic vulnerability is a function of not only the degree shield their households from livelihood shocks. to which they are exposed to negative shocks that have an effect on their welfare, but also the extent to which they Because drought or irregular rainfall is one of the most can cope with the shocks when they occur (Christiaensen severe shocks in Malawi, one of the most common ex ante and Subbarao 2004; Dercon 2001; Makoka 2008). strategies employed is to grow drought-resistant crops. In drought-prone areas of Balaka, Chikhwawa, and Nsanje, Ex Ante Risk-Prevention Strategies: The com- for example, planting crops such as millet and cassava mon ex ante risk-mitigating strategies in Malawi include is encouraged to enable farmers to manage the risk of income diversification, especially through crop diversifi- drought. Regardless of the form it takes, an ex ante risk- cation, and nonfarm income-generating activities. Using management strategy is largely about building up assets to data from IHS2, the Malawi government and the World provide households with buffers against uncertain events Bank (2007) report that large shares of both urban and (Swift 1989). It also entails diversifying activities on and rural households have nonfarm income sources, with off farm and this diversity needs to comprise activities that wealthier households in rural areas earning income from have risk profiles that differ from one another (Ellis 2003). nonagricultural household enterprises.22 Other income reported earning income from nonfarm enterprises than did poorer households. For instance, 27 percent of the poorest 20 percent of rural households had an 22 In the IHS2 data, about 34 percent of all rural households reported earning enterprise income compared with 38 percent of the richest 20 percent of rural an income from household enterprise. In particular, more wealthy households households. Malawi: Agricultural Sector Risk Assessment 83 Ex Post Coping Strategies: Most households have Malawi. In all livelihood zones the poor, who are often limited ex ante strategies to mitigate risks in Malawi. subject to food-related shocks, use ganyu as a major source As a result, when a particular shock occurs, households of exchange for food (see table C.1). undertake a number of strategies to relieve the impact of the shock. From a range of coping options, house- Households also get support from social networks, by bor- holds initially adopt a coping strategy that is “nonero- rowing from relatives and neighbors or sending children sive” to enable it to survive without disintegration or to live with their relatives elsewhere, as a means of coping significant cost (World Bank 2007). Examples of non- with shocks (Makoka 2008). Further, Makoka (2008) was erosive responses include reducing consumption of non- able to show that wealthier households use social networks food items, sending a family member to town to look as a coping strategy more often than poorer households. for work, and gathering wild food, among others (Ellis 2003). Other viable strategies include getting assistance Another important form of coping with shocks, especially from neighbors and family friends or using a modest those that affected households’ ability to access food, is amount of household savings. changing household dietary patterns. In its study of 4,908 house- holds in 2009, WFP (2010) reports that the most common One of the first responses to major shocks reported by coping strategy to cope with various shocks is reduction households is the use of cash savings. In his two-period of food portion size (reported by 57 percent) followed by study of 259 rural households in Malawi, Makoka (2008) a reduction in the number of meals (55 percent). Malawi indicates that 10 percent of the sampled households Government and World Bank (2007) report that consum- reported using cash savings to cope with shocks in 2004 ing less food was the first coping strategy for about 14 per- and 9 percent used the strategy in 2006. Christiaensen cent of all households that reported experiencing a shock. and Sarris (2007) argue that the use of liquid savings does Table C.1 shows that changing dietary patterns is an not disrupt households’ productive resource base. important coping strategy in many livelihood zones. Ellis (2003) also points outs that as a coping strategy, house- Households may also sell assets to cope with shocks. Table holds may substitute between foods, for instance eating C.1 shows that sale of household assets is an important cassava instead of maize. coping strategy in all 11 livelihood zones. Literature sug- gests that households that respond to shocks by selling Poor and vulnerable households cope with shocks through assets are those that had built up assets (such as livestock, support from social support programs. In IHS2, about 3 per- farmland) in “good” years to deplete in “bad” years, a cent of households that experienced shocks used assis- form of self-insurance (Christiaensen and Subbarao 2004; tance from different programs as a first coping strategy Dercon 2004). Makoka (2008) noted that the majority of (World Bank 2007). Makoka (2008) reports that about 25 household that employ this strategy may be vulnerable percent of the sampled households reported using sup- but are usually nonpoor. However, the sale of productive port from social safety net programs as the first response assets (such as land) can put households on a long-term to cope with shocks. lower earning path, as it undermines households’ future productive capacity (Christiaensen and Sarris 2007). Sale Other erosive coping strategies include withdrawing chil- of household assets is therefore an erosive response, caus- dren from school, engaging in commercial sex work, and ing a downward spiral in the asset status of the social unit overexploiting natural resources (World Bank 2007). It (Ellis 2003) and its future ability to manage shocks. is important to note that households employ nonerosive responses first; if they still cannot cope after using the Another important coping strategy is household supply of initial strategy, they move to erosive strategies that entail temporary labor, both on and off farm, commonly known as substantial permanent damage to their ability to engage ganyu. Ganyu is a major coping strategy employed in rural in productive activities. 84 Agriculture Global Practice Technical Assistance Paper A G R I C U LT U R E G L O B A L P R A C T I C E T E C H N I C A L A S S I S TA N C E P A P E R W O R L D B A N K G R O U P R E P O R T N U M B E R 99941-MW 1818 H Street, NW Washington, D.C. 20433 USA Telephone: 202-473-1000 Internet: www.worldbank.org/agriculture