Report No. 25923-UZ Uzbekistan Living Standards Assessment Policies To Improve Living Standards (In Two Volumes) Volume II: Full Report May 2003 Human Development Sector Unit Europe and Central Asia Region Document of the World Bank CURRENCY AND EQUIVALENTUNITS (ExchangeRates effective May 12,2003) Currency Unit = Soum US$l = Soum 968.42 (oficial exchange rate) US$1 = Soum 969.64 (OTC exchange rate) US$l = Soum 1,000.00-1,020.00(consumer gooddexchange bureau exchange rate) US$1 = Soum 1,120.00-1,140.00 (parallel market exchangerate) ACRONYMS AND ABBREVIATIONS AIDS Acquired Immune Deficiency Syndrome IMF InternationalMonetary Fund ALMP Active Labor Market Program LFS Labor Force Survey BEEPS Business Environment and Enterprise MAWR Ministry of Agriculture and Water Performance Survey Resources CBU Central Bank of Uzbekistan MICS Multiple Indicator Cluster Survey CEE Central and Eastern European MLSP Ministry of Labor and Social Protection CIS Commonwealth of Independent States MOF Ministry of Finance COM Cabinet of Ministers MOH Ministry of Health CPI Consumer Price Index MPE Ministry of Public Economics CTS Common Tariff Schedule NBU National Bank of Uzbekistan DHS Demographic and Health Survey NGO Non-Governmental Organization EBRD European Bank for Reconstruction and OTC Over-the-counter Development EPL Employment Protection Legislation PPI Producer Price Index EU European Union PPP Purchasing Power Parity FA0 Food and Agricultural Organization PTA Production and Trade Association FBS Family Budget Survey RESP Rural Enterprise Support Project FDI Foreign Direct Investment SME Small and Medium-SizeEnterprises GDP Gross Domestic Product SOE State-Owned Enterprise GKI State Committee on Privatization UNDP United Nations Development Program GOU Government of Uzbekistan UNICEF United Nations International Children's Emergency Fund ILO International Labor Organization VAT Value-Added Tax Vice President : Johannes Linn (ECAVP) Country Director : Dennis de Tray (ECCOS) Sector Director : Annette Dixon (ECSHD) SectorManager : Maureen Lewis (ECSHD) Task Team Leader : Reema Navar (ECSHD) TABLE OF CONTENTS ACKNOWLEDGEMENTS .............................................................................................................................. 1 CHAPTER 1:INTRODUCTION:UNDERSTANDINGLIVING STANDARDSINUZBEKISTAN .....3 What has happened to living standards?......................................................................................................... 3 Government Strategy ...................................................................................................................................... 4 Growth and Poverty ........................................................................................................................................ 5 The role of external shocks............................................................................................................................. 8 CHAPTER2: A PROFILEOFPOVERTYAND LIVING STANDARDS ................................................. 9 Measuring Income Poverty ............................................................................................................................. 9 A Profile ofIncome Poverty inUzbekistan.................................................................................................. 10 Characteristics ofthe Poor ............................................................................................................................ 18 CHAPTER3: POVERTYAND THE LABORMARKET: GENERATINGEMPLOYMENT OPPORTUNITIES .......................................................................................................................................... 23 Characteristics Of The Labor Market InUzbekistan.................................................................................... 23 The Labor MarketAnd Poverty .................................................................................................................... 32 Generating Employment Opportunities ........................................................................................................ 33 Implications For Policy ................................................................................................................................. 52 CHAPTER4: IMPROVINGRURALINCOMES: THE ROLE OFAGRICULTURALPOLICIES 56 ... Structure and Trends in Agricultural Output................................................................................................. 56 Key Agricultural Policies :Objectives and Directions ................................................................................. 57 The experience to date: Agricultural policies and farm productivity............................................................ 59 Employment, Incomes and Rural Welfare .................................................................................................... 64 Conclusions and Policy Recommendations .................................................................................................. 67 CHAPTER5: IMPROVING CAPABILITIESPROTECTINGHEALTHSTATUSAND PREVENTINGIMPOVERISHMENT ......................................................................................................... 70 Health Status and Epidemiology ................................................................................................................... 70 Patterns o f Health Service Utilization........................................................................................................... 73 Health Financingand Delivery: The inherited system, policy transition and the introduction o f comprehensive reforms................................................................................................................................. 76 Health Reforms to date: the ImplementationExperience ............................................................................. 79 Incidence ofpublic health care expenditures ................................................................................................ 86 Provider Accountability ................................................................................................................................ 87 Policy Implications and Recommendations: ................................................................................................. 87 CHAPTER6: IMPROVINGCAPABILITIES: PROTECTINGINVESTMENTSINEDUCATION 91 . Trends ineducational attainment and existing disparities ............................................................................ 91 3 The Education System: the Legacy. Policy Responseto Transition Challenges. andthe Introduction of a More Active Approach.................................................................................................................................. 99 The Experience to date: Costs, Quality and Relevance, and Attendance.................................................... 100 The incidence o f public expenditures on education .................................................................................... 105 Conclusions and Policy Recommendations ............................................................................................... 107 CHAPTER7: PROVIDINGSECURITY:THE ROLEOF SOCIALPROTECTIONPROGRAMS 110 . Overview of Social ProtectionPrograms and Financing Trends ................................................................ 110 The Performance of Social Protection Programs........................................................................................ 113 Economic Reform and Social Protection.................................................................................................... 128 Conclusions and Policy Options ................................................................................................................. 129 CHAPTER8: IMPROVINGPOVERTYMONITORING INUZBEKISTAN ...................................... 133 4 ACKNOWLEDGEMENTS This report is the outcome of substantial collaboration between a Government o f Uzbekistanteam and World Bank team. We are extremely grateful to all our counterparts in the Government o f Uzbekistan for their keen interest and valuable inputs throughout the process o f preparation of this report. The comments and inputs provided by the Government Working Group to both volumes of the report, have been reflectedto the extent possible. The final views expressed inboth reports, however, are those of the World Bank team. The Government Working Group consists of a Policy Level Group that was chaired by Mr. Furkat Baratov (Deputy Minister o f Economy), and includes Mr. Rustam Akhliddinov (First Deputy Minister of Public Education), Mr. Botir Alimukhamedov (Deputy Minister of Labor and Social Protection), Mr. Durbek Akhmedov (Director, Center of Effective Economic Policy), Mr. Damin Asadov (First Deputy Minister of Health), Mr. Abduvohid Juraev (Deputy Minister o f Agriculture and Water Resources), Ms. Reyhanat Makhmudova (Deputy Head o f the State Statistics Committee), Mr.Zavikhulla Nasretdinkhodjaev, (Head of Division, Information-Analytical Department, Cabinet of Ministers), Mr. Khurram Teshabaev (Director, Center o f FiscalPolicy), Mr.Shukrat Vafaev (Head of Division, Cabinet of Ministers). The Technical Level Group was chaired by Mr. Bekzod Musaev (Ministry of Finance), and includes Mr. Tursun Akhmedov (Center of Effective Economic Policy), Ms. Lyudmila Babadjanova (State Statistics Committee), Ms.Karimova (Center of Effective Economic Policy), Ms. Zulfia Khamidova (Center of Fiscal Policy), Mr. Bakhtiyor Khoshimov (Ministry o f Health), Mr. Valeriy Lapyngin (Minstry of Economy), Ms. Zulkhumor Mutalova (Ministry o f Health), Mr. Alexander Novotny (Ministry of Labor and Social Protection), Mr. Artur Tukhtarov (Information and Analytical Department, Cabinet o f Ministers), Mr. Erkin Yusupov (Ministry o f Agriculture and Water Resources), Mr. Anvar Zokirov (Ministry o f Public Education), Mr. Ulugbek Zakirov (State Statistics Committee). The Technical Group worked closely with the World Bank team inexamining the results o f the analysis of the Family Budget Survey, while also providing detailed comments on policies, facts and relevant data. The poverty analysis and the determination of the poverty line in particular, were conducted in close collaboration between the World Bank and Government of Uzbekistan teams. Much of the analysis reported inthis volume was finalized during a visit o f members o f the Technical Working Group to Washington for two weeks inMarch, 2002. The Government Working Group also guided and facilitated, together with the Bank team, the qualitative researchon health sector and governance issues. The World Bank Team benefited from substantial advise and support from Maureen Lewis (Sector Manager, ECSHD). We are also extremelygrateful for the guidance and support from Annette Dixon (Sector Director, ECSHD), Dennis De Tray (Country Director), Kadir T. Yurukoglu (Senior Advisor), and David Pearce (Country Manager). Peer Reviewers, Peter Lanjouw and Ana Revenga provided extensive comments and guidance and participated in many helpful discussions at various stages throughout the preparation of the report. The core World Bank team was led by Reema Nayar and includes Dilnara Isamiddinova, Brian Kropp, John C. Langenbrunner, Mark Lundell, Caralee McLiesh, Kalpana Mehra, Michael Mertaugh, Edmundo Murrugarra, Yoshimi Nishino, Anahit Poghosyan, Mansoora Rashid, Kinnon Scott , Bekzod Shamsiev, Andriy Storozhuk, William Martin Tracy. Additional contributions were received from Bekzhod Abdurazzakov, Mariam Claeson, Dina N.Abu-Ghaida, Arvo Kuddo, ManjuRani and Moukim Temorov. We are also grateful to several other colleagues in the World Bank Group for helpful discussions and for their presentations and discussions duringthe visit o f the Government o f Uzbekistanteam to Washington. We would liketo acknowledge helpful discussions and comments from staff o f the Asian Development Bank, UNDP, USAID (Zdrav-Plus project, Abt Associates). Finally, we thank the Multi-Sector Team Learning Program at the World Bank for providing financial and other support to further the collaboration between the World Bank and Government of UzbekistanTeams. We 1 would also like to acknowledge supplementary financial support for parts of the research from the World DevelopmentReport(ZOOZ), the Health Sector DecentralizationThematic Group, and aJapaneseTrust Fund. 2 Chapter 1: INTRODUCTION: UNDERSTANDINGLIVING STANDARDSINUZBEKISTAN Uzbekistan occupies a special place among the CIS countries in terms of its growth performance and economic strategy since the beginningof transition. As with other countries o f the Former Soviet Union, a decline in output followed independence in 1991. Output continued to decline until 1995, at which point it was 81 percent o f its 1990 level. While this constitutes a deep recession, all other CIS countries experienced far more severe declines, and many faced more protracted declines, in output (Table 1). Uzbekistan's relatively small output declines have been attributed in part to a successful reorientation of its main exports (cotton and gold) to non-FSU markets, development of energy self-sufficiency, and an initially low level of industrialization. Since 1996, however, Uzbekistan has reported steady positive economic growth averaging 4 percent for seven consecutive years now. Nevertheless, Table 1 shows that Uzbekistan's growth has not been high compared to the growth rates observed inother CIS countries after they returned to positive growth. With an average growth rate o f population of 1.5 percent duringthe period, Uzbekistan's cumulative real per capita GDP growth since the end o f the transition recession (18.2 percent) was in fact the second lowest among the CIS countries. IUzbekistan* 1 I ~~ * 81.1 73.0 31.2 4.0 18.2 11 According to official statistics. Altemative real GDP growth estimatedby internationalfinancial organizations have been lower. Source WorldBank staff calculations What hashappenedto livingstandards? Poverty was not unknown in Uzbekistan prior to the transition and indeed Uzbekistan was one o f the poorest republics of the Former Soviet Union. As a result o f its socialist legacy, however, Uzbekistan's health and education indicators compared favorably to countries with similar levels of income. Comparable data are not available over time to assess trends in income poverty. The decline inoutput and household incomes together with hyperinflation in the early years of the transition were likely associated with a deterioration in living standards and an increase in poverty, although the smaller falls inreal incomes may have been associatedwith smaller declines in living standards compared to many other CIS countries. 3 Some indicators o f well-being show improvements associated with Uzbekistan's growth since 1996. According to administrative data, indicators such as life expectancy, infant and child mortality rates, participation rates in basic education have recovered or improved substantially with the resumption of economic growth. Other gains that had started in earlier decades, seem to have continued since independence: illiteracy has now been completely eliminated, and gender gaps in educational achievement have been closing and remain primarily at higher levels. Reflecting the early privatization o f housing, 95% of all households own their dwellings, and the majority of rural households have access to household plots for cultivation. However, serious challenges remain. Growth has not generated sufficient employment opportunities or substantially improved incomes o f the population. In 2002, Uzbekistan's GDP per capita at the official exchange rate was $384, the third lowest in the CIS (after Kyrgyz Republic and Tajikistan)'. A little over a quarter (27.5%) of the population i s estimated to be living inabsolute income deprivation, and a third of them could be considered extremely and unsustainably poor. Regional variation inpoverty is high. Some non-income dimensions o f well being are o f concern. Contrary to administrative data, survey data show that access to basic services such as safe water and sanitation remains low, particularly in rural areas. Survey data show much higher infant mortality rates compared to administrative sources, some infectious diseases, such as tuberculosis and AIDS, are increasing, and despite recent increases enrollment rates particularly for pre-school, secondary and higher education, remain substantially below levels at independence. As inother countries, the poor, particularly inrural areas, face a higher incidence o f infectious disease and worse nutritional outcomes, but many non-poor but chronically illare also at risk of falling into poverty. There are large disparities in educational outcomes, particularly at teknikum and university levels, between the poor and non-poor. As in the case o f income dimensions, there are large variations in health and education outcomes between regions. Many segments of the population, but especially the poor, are negatively impactedby informal payments, and corruption, including in health and education. Why has growth not been accompanied by broad-based improvements in living standards? We look now at the potential role played by policies and institutions. Government Strategy The strategic objectives o f economic policy announced soon after independence included diversifying the economy from the cotton mono-culture to a more industrialized economy, expanding employment opportunities, and achieving energy and food security to protect the economy and population from major external shocks. To achieve these objectives, the government has adopted a gradual, evolutionary approach to reforms, with an emphasis on first creating the legal and institutional foundations for economic reforms, gradually changing the ownership structure to lay the basis for a mixed economy, maintaining a strong role for the state as 'regulator' o f the economy, and ensuring a strong social policy. To develop industrial and manufacturing capacity in a predominantly agricultural economy, the government has relied on a capital intensive, import-substituting industrialization strategy with a high degree of government intervention, substantial external public and publicly-guaranteed borrowing, and implicit resource redistribution towards priority sectors, including through a restrictive foreign exchange and trade regime, and energy pricing below full cost recovery. Inline with the stated objectives ofthe government, the country achieved self-sufficiency in energy by 1995 and, more recently, in wheat. On the other hand, the overall implementation of market reforms has been unevenand a number o f structural distortions constrained economic growth and broad-based improvements inliving standards. 'At the commercial bankiconsumer goods, indicative, or black market exchange rates, GDP per capita was lower, ranging between$226 and $336 in 2002 dependingon the rate. 4 Growth and Poverty Why has steady economic growth not resulted in more substantial improvements in living standards in Uzbekistan? Due to problems with official statistics, especially in the area o f national accounts and consumer prices, actual growth might have been lower than reported. Alternative estimates indicate that GDP growth was below the officially reported levels, although these estimates still show positive growth2. Growth has not been high enough nor suflciently broad based. As discussed earlier, the growth rate in Uzbekistan lags behind most other CIS countries. Growth has not been broad based: according to official statistics, during the period from 1996-2002, real value added in agriculture grew on average at about 3.3% per annum, and in industry it grew at only 2.0% per annum. Growth was higher in the trade sector, which however, accounts for only a small share o f employment. Thus, incomes and wages remain low. In 2001, GDP per capita at purchasing power parity was $2,440, third lowest among CIS countries. The average monthly wage in2002 was $29 at the official exchangerate (or $17 at the parallel market rateQ. Thegrowth inproductivity has been limited. While international experience has demonstrated the importance of productivity growth for increases in incomes and wages o f workers, in Uzbekistan labor productivity4 in industry has grown at only 0.8% per annum in 1996-2002, in agriculture at about 5.5% per annum, and services at 1.6% per annum. The limited productivity growth reflects insufficient restructuring (particularly outside o f agriculture), and associated movement o f resources to higher productivity activities in the private sector and in services and trade. Although some restructuring has been initiated in the agricultural sector, Uzbekistan has not seen the dramatic sectoral shifts inoutput or employment witnessed in Central and Eastern European countries, and public services still dominate the services sector. Private sector share o f GDP i s amon the lowest in transition countries and over 60% o f investments came in recent years from the state sector .Foreign direct investmentflows have been falling steadily, and at an average o f $3 per capita in2000- 2002 are among the lowest in CIS countries. Growth has not generated suf$cient employment opportunities for the growing population. Although employment in absolute terms has increased according to official statistics, the recent economic growth has not translated into a sufficient number o f productive jobs for the growing working age population. The employment rate (ratio o f employment to working age population), based on official statistics, has declined steadily since independence (by 12 percentage points between 1991 and 2002), and the recent growth has not reversed this trend. Activity rates are particularly low among women (only 33%, comparable to Turkmenistan, Tajikistan, and Azerbaijan). Open unemployment (6%), high youth unemployment (13%), underemployment, the significance o f temporary work arrangements, and informal sector employment are all additional manifestations of inadequate utilization of labor resources6. The experience o f East Asian countries, where a growing workforce was accompanied by faster GDP growth per capita, demonstrates the importance o f tapping the potential o f a growing workforce. 2 See IMF, Recent Economic Developments 2000, 2001 and World Bank, Social and Structural Policy Review, 1999. Alternate IMFestimatesput the averagegrowth inreal GDP during 1996-2002at around2/3rds ofthe official rate. Dynamics of U S dollar-denominatedincome indicators varied between 1996 and 2001/2002. GDP per capita in PPP terms steadily increased(from $2,000 in 1996) while US$ GDP per capita at current exchangerates declined(from $601 in 1996to $384 in2002 at the official exchange rate, and from $409 to $226 at the black marketrate). Average wage duringthe same perioddeclined(from $54 at the official exchangerate and from $37 at the blackmarket rate). Labor productivity figures are World Bank staff calculations based on official national accounts and employment statistics. In2002, share ofstate investment fell sharply (to 41 percent). 6 Numbers quoted here are from Family Budget Survey which differ quantitatively, although not qualitatively, from official labor force monitoring. 5 Income inequality has increasedand has prevented a higher impact of growth onpoverty reduction. As inthe case o f poverty, comparable data on income inequality are not available over time for Uzbekistan. However, wage inequality has increased, particularly between workers in the agricultural sector and all other workers, suggestingthat rural-urban disparities have widened. While the average agricultural wage was comparable to that in industry in 1991, by 2002 it was only 21%7. Unlike the initial increases in inequality in Central European transition countries, this increase in inequality has not been a result o f alignment o f wages with productivity, as agricultural productivity growth has been higher than that in industry. Levels of wage and income inequality in Uzbekistantoday are comparable or higher than those observed in other CIS countries, and significantly higherthan those incountries of Central and Eastern Europe and the Baltics. The Role of Macroeconomic Policies'. Beginning in the mid-1990s - and especially since 2000 - the government has made substantial efforts to tighten macroeconomic policies and address key macroeconomicchallenges such as containing inflation and cutting state budget deficits. Nevertheless, macroeconomic stability has remained elusive, and together with foreign exchange and trade restrictions, which resulted in large implicit resource transfers between sectors, as well as excessive state intervention in the economy, this has constrained growth below its potential and prevented it from translating into more significant improvements inliving standards. There has been some streamlining o fthe foreign exchange regime since 2000 and a major reduction inthe gap between the official and black market exchange ratesg However, the Government still maintains restrictions on current account convertibility and foreign trade through a system of import controls and non-tariff barriers. These restrictions, inplace since late 1996, have created disincentives for exporters and compressed imports. Together with a fall in world market prices o f Uzbekistan's key export commodities, they led to a significant contraction in foreign trade since 1996-97. Openness to internationaltrade is essential for relative prices to stimulate the expansion of competitive sectors and for higher returns to labor in these sectors. Uzbekistan's administrative restrictions in the foreign exchange market are distorting relative prices and redistributing large amounts o f income from the competitive export-oriented sectors, such as agriculture, to less efficient import substituting sectors such as car manufacturing, electronics, etc. Moreover foreign exchange restrictions combined with an unattractive domestic business environment have created strong incentives for capital flight, which represents a drain on resources that would otherwise be available for investment and further hurt economic growth andpotential for improving living standards. Major monetary tightening begun in the mid-1990's helped to curb the hyperinflation by 1997. However, further tightening of monetary policies was postponed. Together with large devaluations o f the official exchange rate, this has prevented inflation from declining further. CPI inflation stayed, according to official statistics, in the 25-30 percent range between 1997 and 2001, declining to 22 percent in 2002. Alternative IMFstaff estimates put inflation at about twice that level although the difference was reduced substantially in 2002. The relatively highinflation not only constrains growth and improvementsin living standards, but also erodes real earnings and tends to disproportionately affect the poor who are more likely to have fixed- incomes, fewer real assets and lack access to foreign exchange. Inflation and an unstable exchange rate also create additional uncertainties and risks for business operation, which adversely affect the investment climate. Since workers in agriculturalcooperatives additionally may receivea share of cooperative's profits -if cooperative is profitable-as well as in-kind payments, wage figures alone may overstate the disparities. 8 For a more detailed discussion of Uzbekistan's macroeconomic and structural policies, see World Bank, Country EconomicMemorandum, 2003. The parallel market premium over the official exchange rate was reduced from the peak of over 300 percent in early 2000 to around 10-15 percent in April 2002, through a combination of accelerated official exchange rate devaluation, expanded access to some segments of the foreign exchange market, and administrative measures (such as import restrictionsand border closures) to depress foreign exchange demand in the black market. 6 Uzbekistan substantially liberalized prices during the period of accelerated market reform in 1994-96. However, the prices of a limited set o f essential consumer goods and services (as well as those o f strategic commodities such as cotton and energy) are subject to price controls intended to keep these prices below international levels". Despite the price controls, however, many consumer goods are resold to final consumers through market channels and priced at the black market exchange rate close to the level of neighboring countries where such prices have already been liberalized. Thus the impact o f price controls on the costs of essential commodities purchased by the poor might be more limited than anticipated. While low energy tariffs undoubtedly reduce the cost of this essential utility to households, a less costly and better targeted approach would do muchbetter at protectingthe poor. The role of Jiscal policy. Unlike most CIS economies, Uzbekistan managed to avoid the collapses of government revenues during early transition period, in part because of tax receipts from captive state enterprises and the cotton sector, but also due to stronger tax administration than insome other CIS countries. Nevertheless, tax revenues have declined since the mid-ninetiesfor various reasons, including a reduction in tax rates to stimulate domestic producers, the introduction o f simplified taxation for agricultural producers and small and individual enterprises, and a narrowing o f the tax base as a result o f the proliferation o f tax breaks. Despite this, consolidated budget revenues in Uzbekistan were high, at about 35 percent of GDP in 2001-2002, compared to its income level and the other CIS countries whose revenues averaged around 26 percent. These relatively strong budget revenues have allowed the government to maintain a relatively high level of public spending. The Government has consistently directed over 40 percent of state budget expenditures towards social expenditures such as education, health and social security. As government revenues have declined in recent years, however, expenditures have had to be compressed accordingly. While some rationalization o f facilities, for example, inthe health sector is underway, the implementation has been slower than envisaged. As inother countries inthe region, declining expenditures in both health and education, have been accompanied by the introduction o f formal payments and the increase in informal payments. Although there are policy provisions to protect the disadvantaged from formal payments, out of pocket expenditures constitute a greater burden for the poor. Given limited financial resources o f the poor ,this has resulted in uneven access to - and quality o f - services for the poor, which are affecting the ability o f the poor to maintaintheir human capital. The wide coverage o fthe pension system and its generosity has been an important source ofsocial protection, not only for the pensioners, but also their family members. Since most pensioners live in extended families, the poverty reduction impact o f the pension system has been significant and it has likely provided an important safety net during the transition. Social assistance expenditures have begun to be targeted in recent years through an innovative scheme relying on decentralized delivery, and have played a role especially in protecting the living standards of some poor families with young children. As in the case o f social expenditures, expenditures on irrigation infrastructure have declined, reducing agricultural productivity and increasing rural vulnerability. While the narrowly definedbudget deficit was well contained -and remainedunder 2.5 percent o f GDP since 1997, overall fiscal stance o f the public sector, including state enterprises, has been more expansionary, loThe Government either directly sets these prices or influencesthem in some other way (such as imposingprice caps or retail trade margins). These are intendedto either keep prices low to maintainliving standards, as in the case of energy or consumer goods and services, or to generate additional revenues for the Government (cotton). Price controls are supported by export bans or export restrictions. For goods subject to price controlsthat are importedat the overvalued exchange rate, such as vegetable oil or sugar, the price controls are designed to pass on the implicit foreign exchange subsidy to the population. 7 especially in 1997-1999, as the relatively high investments averaging 29-30 percent during 1996-2001l1 supporting economic growth, were to a large extent financed through publicly guaranteed external borrowing. This was reflected inexternal debt accumulation: public and publicly-guaranteedexternal debt stock reached $4.0 billion at the end o f 2002, or around 135 percent o f exports o f goods and services12. Moreover, given Uzbekistan's reliance on commercial borrowing, the degree o f concessionality of Uzbekistan's external debt is low and debt service high. The high levels of debt and debt service slow down economic growth and leave the economy vulnerableto shocks. The role of external shocks Uzbekistan's economy has also faced external and domestic shocks that have complicated economic management and adversely affected living standards. These include: The East Asian and Russian crises in 1997-1998, which hit Uzbekistan's exports. Since the foreign exchange policy was not adjusted, the official exchange rate grew increasingly overvalued and in foreign trade declined. Water shortagesduring2000-2001, caused by natural factors but also by the lack o f regional coordination in water resource distribution. These harmed agricultural output and rural incomes, particularly in the downstream areas. Terms of trade shocks due to sharp declines inthe prices o f cotton and gold, which comprise over 50% of Uzbekistan's exports. The price o f cotton fiber dropped by over 40 percent in early 2001 compared to its peak in 1995, before starting to recover somewhat in late 2002. Gold prices have also been lagging over the past five years with some recovery beginning in2002. While prices of Uzbekistan's imports such as manufactured products, especially machinery, also tended to decline, these declines were modest compared to the fall inexport prices. Political factors such as the bombings in Tashkent city in 1999, which were followed by increased restrictions, for example, on labor mobility, particularly into Tashkent. Uzbekistanhas pursued a gradual approach to market reform, inpart to avoid the social costs associated with a rapid transition. Has this approach alleviated the problems o f poverty inheritedby the country? This report provides the first national level picture of living standards in Uzbekistan. It examines the current status of living standards in Uzbekistan (next chapter), identifies key challenges and constraints to improving living standards, and suggests priority policy actions that are needed for broad based improvements in living standards (chapters 3-7). Chapter 8 provides suggestions for strengthening poverty monitoring. The report suggests that the present economic strategy may have limited effectiveness in improving living standards inthe future. Inother words, the future opportunity costs o f non-reform for living standards may be high. Inorder to reduce poverty and achieve broad-based improvements in living standards, Uzbekistanmust not delay market reforms. Such reforms, however, must be accompanied by the strengthening o f policies that protect the ability of the poor to build their human capital as well as strong social protection programs for those who are negatively affected by the reforms. Understanding and reducing regional disparities is also essential to ensure broad based improvements in living standards. Reforms must also be accompanied by institutionsthat protect the poor and provide them voice while increasing accountability of state institutions. ''Alternative estimates basedon availablesourcesof savings indicate lower level of investment, however. Accordingto the IMF's estimates, for example, investment was just 11-12 percent during 1998-2001. See IMF, Republic of Uzbekistan- Statistical Appendix, 2002, p. 10. l2Total extemal debt stock, including private non-guaranteedand short-term debt, is estimated at $4.7 bln., or around 160percentof exports of goods and services, at the end of 2002. 8 Chapter2 A PROFILEOFPOVERTYAND LIVING STANDARDS Poverty was not unknown in Uzbekistan prior to the transition. Indeed, available information shows that, at independence, Uzbekistan was one o f the poorest republics o f the Former Soviet Union. The distribution of poverty showed a strong rural bias and levels of inequality were similar to other Central Asian Republics, although per capita incomes were somewhat more unequal than other parts of the FSU. Although comparable data are not available over time, newly available household data for Uzbekistan confirms that poverty is still significant and nearly seven million people do not have enough resources to meet basic consumption needs. Two thirds of the poor are in rural areas, and poverty risks differ sharply between regions. Education, particularly higher education, i s also an important determinant o f poverty risks. Poverty i s also correlated with labor market status, particularly inactivity and unemployment, although half of the poor live in families where the head is working. Demographic characteristics are also correlated with poverty. The poor have fewer workers for each dependent household member. Access to land is widespread, and land i s an important source o f consumption for the rural population. Ownership of dwellings is also almost universal. The poor, however, have lower access to public services and utilities. Overall, however, access to safe water and sewerage is quite low. Some human capital indicators are declining, and the poor also have worse health and education indicators. MeasuringIncomePoverty To measure material well being, the standard approach is to first determine some measure o f welfare, preferably based on total household consumption (of food, non-food items as well as the flow o f services from durable goods and housing) to rank individ~als'~.This measure of welfare needs to be adjusted for household size and composition to take into account differing needs of different family members as well as possible economies of scale in consumption. Finally, a poverty line needs to be set. An absolute poverty line sets an absolute minimum standard o f living and is typically based on a fixed basket of food products (deemed to represent minimum nutritional intake necessary for good health) plus an allowance for other expenditures(such as housing and clothing). Inthis report we use data from the Uzbekistan Family Budget Survey from April 2000 to March 2001 to measure poverty. The data is, for the first time, nationally representative. Data from twelve months are also representative at the oblast level which permitsthe first comprehensive look at regional differences. Although it is a good quality income and expenditure survey, the data have some serious drawbacks for poverty analysis. In particular, they do not permit the construction o f a total consumption aggregate. Only a reliable food consumptionaggregate can be c~nstructed'~. l3Householdconsumption is believed to be more reliablethan income, particularly intransitionaleconomies . This is because in the context of wage arrears, income is erratic while consumption is smoothed. In the presence of a large informal economy, underreporting of income could be significant. Finally produce from the household plot is an importantsource of food consumption and this is not a standard component of monetary income. Constructionof a total income measure is hrther complicatedbecause of the short recall period for income in the UzbekistanFamily Budget Survey and the difficulties with disentangling gross and net income, particularly for small or informal business or agriculturalactivities. 14No information was collected on the value of housingor durable goods. Inthe case of other non-foodconsumption, a two week reference period was used to collect information. This reference period is appropriate for the consumer price index goals of the Family Budget Survey, but so short as to create enormous noise in the data when trying to measure household welfare, not national consumption. Non-food purchases tend to be very lumpy, thus, the day in which the household is interviewedwill affect the overall welfare level that is recorded for that household, regardless of its true welfare level 9 To deal with these difficulties, we adopt a somewhat unorthodox approach to measuring and analyzing poverty. We draw on the findings in Olsen Lanjouw and Lanjouw (2001) which shows that under certain theoretical assumptions, the poverty rate as measured by a head count measure will be the same under the case where a food consumption measure is compared to a food poverty line and one where a total consumption measure is compared to a total poverty line15. Olsen Lanjouw and Lanjouw examine the methodology empirically and find that it gives sensible results in many practical cases in Latin America and Asia. Scott and Mehra (2002) further examines the empirical performance of this approach in neighboring transition countries, Kazakhstan and the Kyrgyz Republic, and finds that this approach works well empirically. This strong empirical support for these results suggests that we can expect this approach to give reasonable results for Uzbekistan. A limitation o f this approach i s that there i s no reason to expect it to give reasonable results for other measures o f poverty such as inequality, depth, and severity. The approach does, however, permit the best possible use o f available data to understand the extent o f poverty and to obtain some insights into the characteristics o fthe poor. We thus, first construct a consumption aggregate based on food consumption only. We correct for price differences between different months o f the survey as well as across regions. We use a per capita measure of consumption, but test the sensitivity o f the results to an alternate assumption that adjusts for differences in household composition. Except where the results are sensitive to the measure used, this chapter presents only the results basedon a per capita measure.The sensitivity results are reported inScott and Mehra (2002). To set the food poverty we determine the cost o f a consumption basket that will provide 2100 calories per person per day. The Food and Agricultural Organization (FAO) has developed six basic caloric levels depending on the primary staples consumed ina country. The 2100 caloric level is from the FA0wheat based diet and should be appropriate in the Uzbekistan context. The consumption basket is based on actual consumption patterns o f the poor population and i s converted into soums usingthe prices faced by the poorer population. A second poverty line was computed, based on a caloric intake of 1500 calories per person per day. This could be considered an "extreme" poverty line. Thejustification for selecting this line i s that while such a low caloric intake is not life threatening in the short run, it is not sustainable in the medium or long term. Table 1 shows the value o f the food poverty line, the extreme poverty line, and their comparisons with the minimum and average wage in 2000. The value o f the food poverty line thus obtained, in October 2000 prices, i s 3601 soums per month, which is 50% higherthan the value o f the minimumwage in2000, while the extreme poverty line is quite close Gust 5 percent higher)thanthe minimumwage inOctober 2000. Source: FBS, 2000-01: Value o f food poverty line calculated by author; value o f minimumwage and average wage in 2000 from SDOS A Profile of IncomePoverty inUzbekistan Using the approach described above, an estimated 27.5 percent of the population, or 6.8 million people in Uzbekistan are unable to meet basic consumption needs. Approximately one-third of all poor households can be class@ed as extremely poor. Thus absolute poverty in Uzbekistan is significant. The use o f food consumption alone complicates international comparisons of poverty rates. Nevertheless, rough comparisons l5See Olsen Lanjouw and Lanjouw for a discussion of the assumptions. 10 suggest that the poverty rate in Uzbekistan is lower than in the Kyrgyz Republic and Vietnam, while it is higherthan inGeorgia, Braziland Indonesia. Examining the sensitivity o f the poverty rate to the poverty line provides a useful way o f understanding how many individuals are close to the poverty line and therefore might either fall into poverty as a result of small shocks (and could therefore be considered vulnerable), or alternatively how many might be able to escape poverty. We find that a large number o f individuals are clustered around the poverty line. A 10 percent increase in the poverty line leads to an increase inthe poverty rate of 26.5 percent. Similarly, decreasing the poverty line by 10 percent lowers the poverty rate by 26 percent (Table 3, Figurel). This suggests that while a large number of people could fall into poverty as a result o f economic risks, such as recession, equally a significant number of households could escape poverty more easily as a result o f growth or good policies. We look next at the poverty risks for different groups o f the population. This provides some usefulinsights into factors associated with poverty. We find that the oblast or region o f residence and higher levels of education of the household head are the strongest correlates of poverty. Labor market status is also correlated with poverty, and individuals living in households with unemployed heads are more likely to be poor. Demographic characteristics o f households, particularly the number o f children and household size is also important. These are examined below. Location andPoverty Rural populations have higher rates o f poverty and extreme poverty than their urban counterparts. They are 35 percent more likely to be poor, and 58 percent more likely to be extremely poor. Approximately 4.5 million people, or 70 percent o f Uzbekistan's poor, live in rural areas. This finding suggests that agricultural policies that can lead to higher incomes in rural areas will have a significant impact on poverty reduction. These policies and their linkages with rural welfare are discussed inthe next chapter. As will be seen later in this chapter, the widespread access to household plots in rural areas and their importance for meeting the consumption needs o f rural families, and especially the rural poor, i s probably an important reason why the risk ofpoverty inrural areas is not more elevated comparedto urbanareas. Region of residence or oblast are more closely associated with poverty than rural or urban residence16. The risk of poverty inthe Southern region, which consists of Kashkadarya and Surkhandarya oblasts is 3.5 times that in the Tashkent region (Tashkent city and oblast) and nearly twice the national average. The relative risks of extreme poverty inthis region are even higher, at 8 times that in the Tashkent region, and nearly three timesthe national average. Thus, althoughthe Southern region accounts for only 16percent of the population, it accounts for 28 percent ofthe poor, and 46 percent of the extreme poor. On the other hand, only 9 percent o f the poor and 7 percent o f the extreme poor live in Tashkent region. Figure 1 shows that Tashkent is unambiguously less poor than any other region, regardless o f where the poverty line i s set. At the opposite end o f the spectrum is the Southern Economic Region which has the highest rate o f poverty regardless of wherethe poverty line is set. l6 Indeed in the multivariate analysis, reported in Scott and Mehra (2002) 2002, in Volume 2, the inclusion of region of residence significantly reduces (pc method) or wipes out (pae method) the independentimpact of rural residence. 11 Characteristic Incidence of Incidence of extreme Share of Share of Share of IArea poverty Poverty population poor extreme poor Urban 22.5 7.1 37.4 30.6 27.4 Rural 30.5 11.2 62.6 69.4 72.3 /EconomicRegions /Tashkent II 13.3 1I 3.4 II 18.2 I1 8.8 I1 6.5 I1 Mirzachul 21.3 5.2 6.6 5.1 3.5 Ferghana 28.6 8.0 27.6 28.7 22.8 Northem 33.5 8.0 11.6 14.1 9.6 Central 21.4 6.0 19.9 15.5 12.3 Southem Region 47.4 27.4 16.1 27.8 45.5 Oblasts Karakalpakstan 36.4 7.7 6.2 8.2 4.9 Andijan 31.8 9.1 8.9 10.3 8.3 Bukhara 13.4 1.9 5.8 2.8 1.1 IKashkadarya 62.6 41.6 8.9 20.3 38.2 /Navoi 18.7 5.6 3.2 2.2 1.8 INamangan III 1 Samarkhand 39.7 12.2 7.8 11.3 9.8 26.4 8.4 10.9 10.5 9.4 1ISurkhandarya 28.4 9.7 7.2 7.4 7.2 Syrdarya II1III 8.4 II1I1I 2 IIIIII 2.6 IIIIII 0.8 III 0.5 IIIIII Tashkent Region 16.9 3.8 9.6 5.9 3.8 Ferghana 18.1 4 10.9 7.2 4.5 - Khorezm 30.1 8.3 5.4 5.9 4.6 Tashkent City 9.2 2.9 8.7 2.9 2.6 National 27.5 9.7 100 100 100 12 Figure 1: CumulativeDistribution of Food Consumption Cdf of Per Capita Annual Consumption, 2001 1 - - .a -- .6 - - .4 - - .2 - - n - u l 7I 7I5 8I 8I5 9I 9I5 10 I Log Per Capita Annual Consumption Source: FBS, 2000-01. Note: This is basedon aper capitamethodof assigning householdconsumptionto householdmembers. The middle vertical line is the food poverty line. The linesto either side are set at plus (or minus) ten percent From directly to the right of the highest vertical line, the curves represent the following regions (from bottom to top): Tashkent, Mirzachul ,Central, Ferghana, Northern, Southern. Looking at poverty rates by the more disaggregated oblasts, wefind that disparities are even more severe, with thepoverty risk in Kashkadarya being nearly 7 times as high as that in Tashkentcity. The risk o f being extremely poor is highest in Kashkadarya oblast, while it is lowest in Bukhara, Syrdarya, and Tashkent city. The more disaggregated picture also shows high poverty risks in Namangan (44 percent higher than the national average) and in Karakalpakastan (32 percent higher than the national average). Kashkadarya alone accounts for 38 percent of the extremely poor. Table 3 shows that although poverty rates are highest in the Southern economic region, the poor are less clustered around the poverty line in this region. Thus while they are less vulnerable to falling into poverty than other regions, the poor in this region are also less likely to escape overty easily. This suggests that targeted interventionsinKashkadaryamightbe neededto alleviate poverty77. On the other hand, residents in Tashkent region are more vulnerable to economic risks while at the same time many more residents than in other regions could escape poverty easily. Since the poor are spread out across all oblasts, however, policies that rely on geographical targeting alone are unlikelyto be sufficient. Why are there such strong disparities across regions, and particularly oblasts? Geographical differences could be due to a variety o f reasons, such as differences in educational attainment, environmental factors (affecting health or land quality), lack of infrastructure, all o f which are associated with lower endowments in disadvantaged regions. They could also be due to differences in opportunities (such as access to markets or labor market opportunities) and differences in institutions and policies. As we shall see later in the report, human capital endowments such as health or education do indeed differ across regions, as do labor market opportunities. The multivariate analysis of the determinants o f poverty (See Scott and Mehra, 2002), however, shows that controlling for differences in education, labor market participation and demographic 17Since these results are based on a measure o f food consumption, some caution must be exercised in interpreting these. Ideally these results should be verified with a measure o f total consumption once the data is available. 13 factors does reduce differences in poverty risks across regions. However, significant differences remain across regions, with poverty risks substantially higher in the Southern, Northern, and Ferghana regions relative to Tashkent. Further understanding o f the differences between regions, such as differences in other endowments, and in polices and institutions i s a priority for devising adequate policies to reduce poverty. HeadCount Ratio Change inPovertyRate EconomicRegion wlexisting wlpoverty line w/ poverty line wl poverty line wl poverty line poverty line + 10% -10% + 10% -10% Tashkent 13.3 18.5 8 39.1 -39.8 Mirzachul 21.3 27.6 16.2 29.6 -23.9 Ferghana 28.6 37.1 20.6 29.7 -28.0 Northern 33.5 42.5 23.6 26.9 -29.6 Central 21.4 28.7 14.4 34.1 -32.7 Southern 47.7 54.5 40.1 14.3 -15.9 UZBEKISTAN 27.5 34.8 20.3 26.5 -26.2 Source: The significant geographical differences are also consistent with the limited internal migration discussed in the next chapter. Limited economic opportunities, the importance o f household plots for consumption, strong family ties combined with affordability o f housing in prosperous areas, as well as institutional barriers to mobility (such as the propiska) to Tashkent city are some o f the key reasons why people do not move to other regions insearch of better living conditions. EducationandPoverq Table 4 shows the poverty risk o f individuals by the educational attainment o f household heads. Education of household head, particularly the attainment of some higher education, is strongly correlated with lower poverty o f individuals inthe household. Families where the head has higher education are halfas likely to be poor as those where the head has nine or fewer years of education. Those living in households where the head is illiterate or has four or fewer years o f education have a poverty risk that is 23 percent higher than average. However, education does notprotect completely+om poverty and there are poor inhouseholds from all categories o f education, including higher levels. Importantly families living in households where the head has completed secondary education have a higher than average risk o f poverty and their risk of poverty does not differ significantly from those living inhouseholds with less education. The multivariate analysis (see Scott and Mehra,2002), also confirms these findings: education o f the household head, particularly the attainment o f higher levels (completed technical, incomplete or completed highereducation), is a key determinant of household welfare. Compared to a household with no (or less than 5 years education), having a head with higher than secondary education increases welfare levels substantially. The multivariate analysis confirms that there is no significant difference in welfare between households with no (or less than 5 years education), and those with primary or secondary education. What is behind these results? The next chapter shows that more educated individuals are more likely to participate inthe labor market. However the labor market rewards teknikumand higher education much more than it does lower levels of education. Indeedthere is no premiumto lower levels inthe private sector. Thus the labor market might be demanding higher skills or different skills than those offered by graduates of secondary schools. The lower premium for secondary graduates might also be related to the large share of working age adults with secondary education relative to employment opportunities for individuals with this level of education. The importance o f connections (rather than education alone) in securing jobs and their 14 quality, could explain why education does not protect completely from poverty. Improving the quality and relevance of the general education system as well as ensuring access to higher education, especially for the poor, will be key priorities for reducing poverty and preventing the intergenerationaltransmission of poverty. These issues are discussed inchapter 6. Table 4: Educationand Labor Market Status ofHouseholdHead and Poverty /Education of Head IIncidence of Poverty Share of Population Share of Poor I I I Noneto four years 34.0 8.4 10.4 Fiveto nine years 32.4 13.3 15.7 Full secondary 31.3 47.4 53.9 Incomplete Technikum 24.6 2.8 2.5 ]CompleteTecWSomehigher I 17.6 13.4 IHiehereducation I 16.6 I 14.7 I 8.9 I Inactive 30.5 46.2 51.2 Ofwhich: Pensioners 29.7 63.9 62.2 Disabled 23.7 5.6 4.4 Other 33.5 30.5 33.5 Employed 24.6 51.6 46.2 Unemployed 35 2.2 2.8 National 27.5 100.0 100.0 Labor Market Status andPoverty Table 4 also shows the poverty risk of individuals accordingto the labor market status ofthe householdhead. Households with heads who are not employed have a 37 percent higher risk of poverty relative to those in households with employed heads. Households whose heads are inactive (but not drawing pensions or disability benefits) are also at a high risk of poverty. However, having ajob does notprovidefullprotection @om poverty and individuals living in households with employed heads constitute 50 percent of the poor. Individuals living inhouseholdswith unemployedheads constitutea small fraction ofthe population and thus of the poor, and inactivity is more prevalent. Table 4 also shows that households with heads who are pensioners are not significantly more likely to be poor. We see below as well, that the presence of additional pensionage individuals inthe householdis associated with lower poverty. This is due to the high coverage of the pensionsystem, and its generosity (chapter 7). Chapter 3 looks more closely at the labor market, and determinants of labor force participation, and at unemployment. It also provides useful insights into why employment does not protect against poverty in Uzbekistan. The poor are less likely to participate inthe labor market and are more likely to be unemployed and underemployed. However, even when they work, they are more likely to work in lower paying jobs, temporary or casualjobs, or work in the informal sector. Two thirds of the working poor live in households whose heads work in industry or agriculture (Scott and Mehra, 2002). The rural poor, like the non-poor, work predominantly in agriculture (nearly half) followed by industry, while just under a third of the urban poor work in industry. As seen in chapter 3, these are among the lower paying sectors, particularly agriculture, where wages are half of the national average, and have the highest incidence of wage arrears. Interestingly, the poverty risk for households whose heads work in industry are substantially higher than average householdswith employedheads, while for those whose heads work in agriculture have poverty risks 15 that compare to other households with employed heads (Table 5). This is most likely due to the importance of household plots for those working in agriculture (see below on access to land and Chapter 4). Thus, reducing unemployment and inactivity among the poor is important, but it will be essential to improve the access of the poor to better paying and regular employment. Industrv 40 Agriculture 24.3 All employed 24.6 Demographiccharacteristicsof households Poverty is related to demographic characteristics of the households. Poorer households are larger and have more children, but also more adults. Table 6 shows, however that the relative poverty risks by demographic characteristics are sensitive to the measure of consumption and specifically to adjustments made for different consumption needs of different household members. As there is no unassailable methodology for determining the adjustment, we use here a per equivalent measure based on weights developed by the MinistryofLabor and Social Protection18. HouseholdComposition HeadCount Ratio Share of Share of Poor PC Method PAEMethod Population PC Method PAE Method Number of elderly (pension age) Noneinhhld 27.0 21.9 68.2 67.0 71.1 1 One 28.7 I 20.2 19.4 20.2 1 18.7 I lTwo 28.2 17.6 12.4 12.7 10.4 INumber of childrenunder 14 /Noneinhhld IIII 7.5 III 10.8 IIII 6.4 IIII 1.7 III 3.3 III One 14.0 16.3 12.0 6.1 9.3 Two 20.7 17.2 22.7 17.1 18.6 Three 27.1 19.6 25.1 24.7 23.4 Four 34.3 23.4 18.7 23.3 20.8 /Fiveor more 1 48.9 I 34.0 I 15.2 I 27.0 1 24.6 I National 27.5 21.0 Individuals living in larger households are more likely to be poor, by both measures. This i s true even when other characteristics of such households are taken into account, inthe multivariate analysis, such as household composition, education and other characteristics of the household head and location (Scott and Mehra, 2002). Table 6 shows however, that while households with four or more children have elevated risksaccordingto the See Scott and Mehra (2002) for more information on this adjustment. The implied economies of scale in this adjustment are relativelysmall. The sensitivity of the results to this adjustment, therefore is significant. 16 per capita measure, the magnitude o frelative risks is reduced when the per adult equivalent measure is used19 However, individuals living in families with five or more children, comprising 15 percent of the population and a quarter o f the poor, have an elevated risk o f poverty according to both measures. When we take into account other characteristics o f these households (such as lower education o f the head, location, etc.), also take into account differences in consumption needs of different members o f the family, we find that households with many children actually are at lower risk o f poverty (Scott and Mehra, 2002). Thus if children are less expensive to support than adults, then programs that target resources to families based on the number o f children alone would not necessarily go to the poorest families. The move in recent years from universal child allowances to a targeted program, has therefore likely resulted in significant improvements in the targeting efficiency o f the program. At the same time the relative effectiveness o f this program has might be having a significant impact on the welfare of the individuals inthe household. This is examined in greater detail in Chapter 7 where we see that the child benefits program constitutes approximately a quarter of household food consumption for the poorest quintile. The results on the correlation of the presence of elderly members in the household with poverty is also sensitive to the measure of poverty used (Table 6). Specifically, using a per capita consumption measure, individuals living in households with elderly members have a poverty risk that is comparable to the national poverty risk as well as that of households without elderly members, but when adjustments are made for the lower consumption needs of these individuals their poverty risk falls below that o f all other households. In addition, ifwe take into account other characteristics o f such households inthe multivariate analysis, we find that households with elderly members present are substantially less likely to be poor: the predicted probability risk for individuals living with two elderly (pension age) members falls to 10 percent. This is a direct result of the wide coverage o f the pensions system and its generosity. Since most pensioners live in extended families, the poverty alleviation impact o fthe pension system is significant (Chapter 7). Other characteristics o f the household head such as gender, marital status, age, and ethnicity are also correlated with poverty. However, when other variables are taken into account, gender and ethnicity are no longer significant determinants of poverty. When an adult equivalent measure i s used in the multivariate analysis, married heads are more likely to be poor compared to other heads (single, divorced or widowed). Dependency andpoverty Poor households are larger than other households with both more children and more adults in each household2'. Interestingly,however, the number o f people who depend on each person of working age21is fairly similar across poverty groups: for every person o f working age, there i s approximately one person who is potentially a "dependent". However, as a result of the lower levels o f labor force participationamong the poor, combined with the significantly higher unemployment rates, the dependency ratio, or the number of dependents for each employed household member, is higher. The dependency ratio is 4.8 for the urban poor (that i s 4.8 dependents for each employed member) compared to 3.7 for the urban non-poor. Similarly, the dependency ratio among the poor inrural areas is 5.3 compared to 4.5 among the non-poor. 19 Care should be taken ininterpretingthese results because of large standard errors associatedwith small cell sizes. 20 On average, non-poorhouseholdshave six members, 2.8 o f these beingchildren. Poor householdshave on average4 adults and 3.6 children. 2 1Definedas 16-55 for women and 16-59 for men. 17 Groups Excludedfrom the Survey: Street Children in Uzbekistan With Uzbekistan's economic transition, there has been a decrease inthe provision o f services for at-risk children and an increase in the number o f street children. InTashkent city there are several thousand street children, and the number is increasing. These children tend to be older (about 75% are over 10 years) and have at least one living parent. The number o f orphans among street children is relatively small, as children without living parents are typically assigned to institutionalized care facilities or to relatives.' In Tashkent city about 90% o f street children have families. While some are abandoned, many leave home for domestic reasons, such as poverty, parents' alcoholism, poor opportunities to find employment, domestic violence ,etc. However, there are also a number ofchildrenwho runaway from children's homes or stay at the homes only to sleep at night.. The ethnic composition o f street children is diverse. More than 50% o f them are Russian-speaking, many o f whom come from mixed families with some with roots in other former Soviet Union republics. The number o f ethnically local street children is also growing; there are children coming to the country capital from Kashkadarya and Surkhandarya, as well as from industrial cities in the greater Tashkent area, such as Chirchik- city andAngren-city. Street children concentrate mostly inurban areas, such as Tashkent, because it is easier to find means for living. Of the street children interviewed for Mekhr Tayanchi's "Child labor in Uzbekistan, 2000" report, the main income sources for street children include: begging (82%), thieving (27%), and washing cars (26%). Other sources o f minor weight include trading, carrying loads and goods inthe market, assisting in cafes, and cleaning homes. Many o f these occupations are seasonalinnature. There is a severe deficiency in access to health services and education among street children. Approximately 11 percent o f street children have never been to school, 25 percent o f street children cannot read and write, and 20 percent o f street children do not know elementary mathematics. Street children mainly cope with health problems on their own, although a few receive assistance from NGOs, which help them seek hospital care for serious aliments. There is anecdotal evidence o f TB cases among street children, however there has not yet been enough research to support this claim. While the programs aimed at helping street and at-risk children are still inadequate, there has recently been an increase in aid for these children. Mekhr Tayanchi, (also known as the Tashkent Center for Development o f Community Based Social Services for Children in Difficult Circumstances) has recently initiated the development o f a number o f social services for street children in basic education, health, and professional education. Additionally, the Ministry o f Public Education, in collaboration with Mekhr Tayanchi, is working to create a number o f new schools in Tashkent, which will address the specific education problems o f at-risk children. Finally, there is now a mechanism established by the Tashkent city health department providing hospital care for sick street children. Source: Child Labor inUzbekistan, Save the Children and Tahlil, 2002. Characteristicsofthe Poor Inorder to understand better the risks faced bypoor households and the conditions under which they live, we look now at some the characteristics of the poor. This gives us some insights into the channels by which the poor suffer from income poverty, as well as how they cope with risks, but also into some non-income dimensions of poverty such as their living conditions, health and education, which are important indicators of living standards inand ofthemselves. Access to Land Access to land, and specifically household plots is widespread and household plots have provided an importantmeans of survival for the ruralpopulation. Nationally, 82 percent of all households have access to household plots, 97 percent o f all rural households have access to household plots. In urban areas, as would be expected, the size of the land is small, and more than half o f this is covered by buildings or housing. Rural 18 inhabitantson the other hand, access more land on average, and a larger fraction (about 60 percent) is useable for agricultural purposes. Interestingly, the rural poor have more available land: according to the FBS, the rural poor had 19.7 sotoks of useable land while the non-poor had on average 14.9sotoks of useable land. The data do not provide information, however onthe quality of land. Table 7 provides some insights into the importance of land for own food consumption. It shows that over a quarter of the food consumption of rural households comes from own production from household plots (comparedto 7 percent for urban households). Interestingly, among the rural population, the better off rely more on consumptionof food from their householdplots. This result, while surprising, is consistent with the reported inthe RESP study2 Nevertheless,the widespreadand relatively egalitarian distribution of land has hypothesis that the poor ma7 have access to poorer quality land or other agricultural inputs. This was also been an importantfeature of the Uzbek transition experience and has played an important role in protecting the living standards of the ruralpopulation. 1 2 3 4 5 Total Purchasedfood 92.1% 92.8% 93.1% 92.7% 92.6% 92.8% Nonpurchasedfood 7.9% 7.2% 6.9% 7.3% 7.4% 7.2% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 1 2 3 4 5 Total Purchasedfood 79.3% 76.1% 73.9% 72.4% 70.3% 74.0% Nonpurchasedfood 20.7% 23.9% 26.1% 27.6% 29.7% 26.0% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Ownership of dwellingsand living conditions As a result of early privatization of housing, ownership of dwellings is almost universal. Approximately 95 percent of households own their own dwelling. In the rural areas, poor and non-poor alike live in separate houses while in urban areas, approximately a quarter of the poor and nearly a half of the non-poor live in apartments. The poor do not seem to have less living space than better off households. However, there are differences in the quality of housing between the poor and non-poor. The poor live in sparsely furnished homes and have fewer amenities. More SignZiJicant,however is the lower access to public services and utilities. Figure 2 shows that poor people are underserved by public water system. They also tend to be further from the nearest source of drinking water. The poor are also underservedby centralizedservices such as hot water, sewerage andnatural gas (Figure 2). 22ThurmanandLundell,2001. 19 Figure 2: Access to Public Services and Utilities by Poverty Status I 100 1 z -gm 80 Q 3 ! a 60 ~ 0 + o 40 c, 5 20 n Running Central Central Central Natural Telephone I Water HotWater Heat Sewage Gas , Source: UzbekistanFBS 2000-01, calculations by author Figure 3: Access to Public Services and Utilities by Geographic Location Running Central Central Central Natural Telephone I Water Hot Water Heat Sewage Gas Source: UzbekistanFBS 2000-01 Figure 3, however, reveals that the geographic distribution o f all these services (between urban and rural areas) is an important reason for differences between the poor and non-poor. Ruralpopulations simply do not have adequate access to systemsfor running water, central heat, hot water and sewerage. It should also be noted that overall levels of access to running water and sewerage are quite The data here on access is also likely to overstate the actual availability o f these services as supply o f gas, electricity and water, is often irregular. The low levels o f access to water and sewerage in the country as a whole, and in rural areas in particular, combined with other factors such as agricultural practices24that affect the quality o f the drinking water supplies, is likely to have been an important reason for the high prevalence of childhood diarrheal diseases, gastrointestinal infections, viral hepatitis, and typhoid outbreaks. In 1997, for example, the number 23 Government sources report that access to safe water has improved, especially in rural areas, where access has increasedto 78% from 62% in 1996. 24 Inaddition, the Soviet era expansionofthe cotton crop, diversions of water flows, and use of fertilizers and pesticides have led to water pollution and soil degradationand have contributed to the poor and declining quality o f drinking water particularly in rural areas. The salinization of the land and extensive use of pesticides may also pose hazards. 20 of people diagnosed with hepatitis in Uzbekistan was three times higher than the CIS average and 26 times higherthan the EUaverage. These are likely to have disproportionatelyaffectedrural and poor populations. Education As we have seen, education is associated with improved material well being. In addition, educational attainment and the ability to access good quality education i s an important non-income dimension of welfare. Uzbekistan has made impressive gains ineducation over the past several decades. Illiteracy has virtually been eliminated. Over 90 percent o f the 25-44 year old cohort, for example, had had at least some secondary or higher education. Despite these gains, the transition, as in other countries has posed challenges, and there have been some reversals ineducation investments. Official data show that while enrollment rates at basic education stayed more or less constant in the early years o f the transition, they increased after 1995/96, and the net enrollment ratio reached 95 percent in 2000/01. For other levels the picture is different. Although absolute enrollments and enrollment ratios have begunto recover inrecent ears, absolute enrollments for the pre-school, secondary and higher levels remain below pre-transition levels . The declines in absolute enrollments are especially severe for pre-school and 1 5 higher levels. For higher education, the Family Budget Survey shows that the fraction of younger cohorts (those who are presently 35-44 years old or 25-34 year old) with higher levels of education (and particularly university education), have fallen relative to older cohorts, indicating that the declines in higher education investments startedbefore independence26. There are also disparities ineducational attainment between different groups. The poor appear to have started with less education than the non-poor, on average, and some o f that gap remains. Chapter 6 takes a closer look at current levels of investments ineducation and finds that the gaps are particularly sharp at higher levels o f education: at teknikum level, the non-poor are 3 or 4 times more likely to attend, and at university, the non- poor are 4 to 7 times more likely to participate. The available evidence from the Family Budget Survey suggests that these gaps between poor and non-poor may have increased since independen~e.~~ These gaps are o f concern since these are precisely the levels that are most correlated with labor market earnings and welfare. As discussed in chapter 6, gender gaps have been closing but remain at these higher levels o f education, and regional disparities in educational attendance are significant. Chapter 6 provides a more in- depthdiscussion ofthe barriers faced by the poor ininvestinginhuman capital. Health As inthe case o f education, good healthstatus and access to health facilities are important indicators o fwell being inand o fthemselves. Heretoo, the overall levels and general trends insome o fthe health indicators are o f concern. The collapse inincomes inthe early years o f the transition was accompanied, as inother countries, by a worsening of health outcomes. The recovery since 1995 has seen some improvements in health outcomes; for example, the Governmentreported infant mortality has dropped, and official data also show that life expectancy recovered to pre-transitionlevels by 2000. However, many health indicators remain below regional or EUaverages. Communicable diseases remain a major problem, althoughthere have been some significant achievements2' . The incidence of some infectious diseases, such as tuberculosis and AIDS are clearly increasing and the incidence of AIDS i s now higher on a per capita basis than inother neighboringCentral Asian republics.The 1996 Demographicand Health ~~ 25Dataprovidedby State Departmentof Statistics andMinistry of Public Education. 26Some ofthis couldbe explained by the possibleout-migrationofyounger andmoreeducatedindividuals. 21Scott and Mehra(2002) find that for the most recent cohort (25 to 34 year olds), the non-poorare fifty to 100 percent more likely to have completed technikum or universitythanthe poor. 28 There has not been a single case of poliomyelitis for several years, for example, and incidence o f communicable diseases is often lower than inneighboringcountries. 21 Surveys showed that childmalnutrition was highwith stunting (chronic malnutrition) at 3 1%, wasting (acute malnutrition) at 12%, and underweightchildren at almost 19percent. Iodine deficiency goiter is 18% and iron deficiency anemia is found in 61% o f children under 3 years and 60% o f women aged 15-49. Although not available for citation at the time this report was going to press, preliminary results from the 2002 DHS show significant improvements inchild nutritional status. As in other countries inthe region, however, the survey data from the DHS infant mortality rates are significantly higherfrom survey data, compared to administrative data. Available data from the 1996 DHS show that while differences between income groups in infant and child mortality rates are not as severe as in some other countries, there are severe disparities by income groups in the nutritional status of children and women. This is particularly true in the case o f acute malnutrition and underweight children. These disparities in nutritional outcomes reflect the difficulties faced by poor households inmeetingtheir basic needs2'. Despite this, self reported health status fi-om the FBS compares favorably to neighboring countries. Only 5.6 percent o f the population reported that they were sick in the last 30 days. Among those 16 and older, less than 7 percent report any long term or chronic ailment (almost half that found in the Kyrgyz republic). As in other countries, the poor are slightly less likely to report poor health, but thee differences are smaller than in other countries. The poor are also slightly less likely to seek care and more likely to rely on home visits than the non-poor. Patterns of utilization o f primary care or higher end care are closely related to urban or rural residence, with the highest quintiles more likely to seek higher end care (in urban hospitals), and the poorest quintile more likely to seek primary care. These differences and potential barriers to obtaining health care are examined further inchapter 5. LYIt should be notedhere, that the measure ofwealth inthe DHS is basedon ownership of assets andtherefore is not a very reliablemeasureof wealth, especially in atransitioncountry. Nevertheless, the findings suggestthe importance of monitoring nutritional and health status indicatorsamong different segments of the population. 22 Chapter 3: POVERTYAND THE LABORMARKET: GENERATINGEMPLOYMENTOPPORTUNITIES Unemployed individuals and their families are at a highrisk o f living in poverty. Nevertheless, employment does not protect against poverty and over 11 million poor individuals live in households where the head i s employed. The characteristics of the labor market inUzbekistan today reflect some o f the challenges that the country faces inutilizing its labor resourcesto its full potential and in increasing labor incomes. Labor force participation is low, particularly among women. Open unemployment has emerged and coexists with underemployment and a significant fraction o f the labor force are engaged in informal labor activity. Wages are low for the majority o f workers and wage arrears are predominant. The poor are disproportionately affected by adverse labor market outcomes. The labor market i s the main channel through which growth affects poverty. For growth to reduce poverty, it must be accompanied by substantial increases in the demand for labor. The private sector, including small and mediumenterprises, needs to emerge as the major provider o f employment opportunities. Ifgrowth i s to be accompanied by broad based improvements in living standards, it will be essential to improve the environment for private sector development, ensure a flexible labor market that permits labor to be employed in its most productive use and rewards productivity, ensure the relevance of skills of the labor force, and addressregional, gender and ethnic inequalities in labor market opportunities. CharacteristicsOfThe Labor MarketInUzbekistan Labor ForceParticipation According to the Family Budget Survey (FBS), only about 45 percent of the population between the ages o f 16 and 6430participate in the labor market in Uzbekistan. Using the official definition of working age31a slightly higher fraction, about 48 percent, of the working age population is actively participating in the labor market32. Although comparisons of labor force participation rates across countries is complicated by differences in definitions, and data sources, activity rates in Uzbekistan are lower than those observed internati~nally~~as well as to many countries in the region including Russia and Georgia, and are closer to those observed in neighboring countries such as Tajikistan and Turkmenistan. The low labor force participation rates reflect in largepart the significantly lower participation rates of women. According to the FBS, the labor force participation rate among men using the ILO definition is 60 percent, which compares favorably with other countries. However, for women, the labor force participation rate is only 31 percent for 30ILO definition ofworking age population 31 16to 54 for females and 16to 59 for males, taking into accountofficial retirementages 32This estimate of labor force participation is based on the Family Budget Survey. As such it is different from official statistics. Survey based data are generally considered more reliable than data from the "Balance of Labor Resources". However, the Family Budget Survey does not permit the inclusion of unpaid family workers in the estimate of the employed. Given the importance of this category of workers, especially inrural areas and among women, the estimates of employment, and hence of labor force participation may be underestimated. On the other hand, the official data, although basedon improved labor force monitoring, likely over-estimatethe activity rate of the population. Nevertheless the basic message on the importance of higher labor force activity and employment is valid; while further efforts to implement improved labor force surveys and specializedsurveys to capture aspects of rural employment are needed(see Box 1and Chapter 8). 33 According to the SIMA database (World Bank), Uzbekistan's labor force participation rate is lower than that observedinother regions includingEast Asia andPacific, USA and OECD countries, and for males it is lower thanthose observedfor the Latin America and Caribbeanregion. However, while for Uzbekistan,the SIMA data base gives a more or less comparable total labor force participation rate as the Family Budget Survey, there are major discrepancies betweenthe male and female labor force participation rates obtained from the two sources. According to the SIMA data base, the labor force participation rate for men in Uzbekistanwas 45 percent and for women 39 percent in 1999. 23 women, comparable to the low participation rates in Turkmenistan, Tajikistan and Azerbaijan (see Box 1 for discrepancies inlabor statistics). Low labor force participation rates reflect the lower supply of individuals who may be choosing to study longer, to stay at home for child care, or to retire earlier. However, the decision not to participate in the labor market reflects the opportunity costs o f not participating. Many potential workers may also be discouraged from entering into the labor market due to poor employment prospects. Examining the determinants of labor force participation provides some interesting insights into why these rates are low. As i s to be expected, for both men and women, participation increases with age and reaches a peak before declining at older ages reflecting retirementpatterns. Activity rates are also very different across regions, with some differences between men and women in different regions. For both men and women, however after controlling for other factors such as education, ethnicity, age, etc., activity rates are lowest in Samarkhand, Karakalpakstan, Surkhandarya, and Kashkadarya, likely reflecting the lack o f availability o f productive employment opportunities in these poor regions. Women o f Russian ethnicity are more likely to participate inthe labor market compared to all other women. For men, on the other hand, while there are no significant differences between Russian, Uzbek or the other major Central Asian ethnic groups (Tajiks, Karakalpak, and Kazak ethnicities), men from "other" ethnic groups (including Kyrgyz, Turkmen, Ukranian or Azeris among others) are less likely to participate inthe labor market. Menfrom the lower income groups are less likely to participate, likely reflecting higher levels o f discouragement from limited employment opportunities. Cultural factors are likely to be important reasons for lower female participationrates. Highfertility rates are also important. Women with very young children are much less likely to participate in the labor force. An additional child under five reduces the female participation rate by more than 7 percentage points. The likelihood o f female participation increases, however, with more members inthe household. This might be because of the childcare support available in larger families and suggests the potential impact of the availability o f alternate child care arrangements on increasing women's participation particularly in urban areas. Thestrongest impact on women s laborforce participation comes+om education. Although men with more than 5 years o f education are also significantly more likely to participate in the labor market, the gradients between different levels of education are flat. On the other hand, for women participation rates increase steeply with higher levels o f education (Table 1). Inparticular, participation rates increase by more than 24 percent for women who attended either Technical or Professional Schools (PTU, SPTU, Academic Licea), and by almost 40 percent for women going into highereducation. These aggregate results hide interesting differences between those that completed each education level and those that left the level without completion. For example, participation among females who completed Technikum is more than 52 percent higher than those with less than 5 years of education, compared to only 29 percent higher among those that left the Technikum incomplete. The recent declines in fertility rates could result in higher labor force participation rates for women in the future34. Chapter 6 shows that the gap in higher education between men and women is closing with successive generations. Nevertheless, the declining investment in higher education by men and women is o f concern and may limit the potential increases in labor force participation o f women. Reforms in the pension system that reduce incentives for early retirement could also result in increased labor force participation of men and particularly women (chapter 7). However, it will be important for the labor market to absorb any increases in participation. The following discussion on unemployment and underemployment suggests that j4Accordingto official statistics, birth rates decreased from 33.7 per thousand in 1990to 21.5 per thousand in2000. 24 without substantial increases inthe demandfor labor, Uzbekistan will not be able to exploit the full potential o f its abundant labor resources. Table 1: Effects of education on female laborforce participationrates Educationlevel LFP Gain (in%points) Standarderror - 5-9 12.2 (3.9) - 10-11 9.3 (3.2) - SPTU, PTU 24.0 (4.6) - Technikum 28.3 (4.2) - Higher 39.5 (4.3) - Overallfemale 31.3 Theproblems of labor absorption: unemployment and underemployment Official unemployment rates35are extremely low (about 0.4 percent), and likely reflect poor incentives to register as unemployed. Using survey data from the FBS, however, and the internationally accepted definition o f ~nemployment~~,the estimated unemployment rate is approximately 6 percent37, with male unemployment at 6.7 percent compared to female unemployment at 4.4 percent. The majority o f the unemployed are unregistered, with only 8 percent registered for unemployment benefas with the labor offices. Some of the factors behindthe poor coverage and relative ineffectiveness o f the unemployment programs are discussed inChapter 7. An unemployment rate of 6 percent is not especially high. However, in a country like Uzbekistan, where poverty i s significant, unemployment programs are not particularly effective and the informal sector i s significant, open unemployment rates do not tell the full story. Households cannot afford to be unemployed and are more likely to choose low productivity activities inthe informal sector or temporary or casual work to open unemployment. There i s no official definition of underemployment that is used, reflecting in part the absence of this concept in the Soviet system. However, estimates of the size of the informal sector (ranging from 20 to 50 percent depending on the definition) the significance of temporary work arrangements (approximately 19 percent of employed), and measures o f partial employment (approximately 10 percent o f employed work fewer than 80 hours per month?' suggest that Uzbekistan's labor absorption problem is more significant than suggestedby the unemployment rate. 35 Official unemployment rates are estimatedusing administrativeinformation on the number of registeredunemployed andthe administrativedefmition of the labor force. 36This is the definition usedby the ILO (1996-2002). It includes amongthe unemployedall those inthe labor force who are either employed or seeking ajob regardless of registration status. The FBS was not designed as a typical Labor Force Survey and thus has some significant differences. See Murrugarra(2002) for a more complete discussionon the constructionofthe unemployment variable from the FBS. 37 Official estimates of unemployment from the new Labor Force Monitoring Surveys regularly conducted by the Ministry of Laborand SocialProtectionstartingin2002, show a slightly lower estimate ofunemployment (at about 4%). As discussed earlier, boththe FBS andthe labor force monitoring surveys couldbe improved. While the Family Budget Survey is usedhere, the basic messagesonunemploymentare similar from both sources. 38 Part time work and temporary work arrangements might also be out of choice. Nevertheless, there is a strong correlationbetweenthese and poverty. 25 Box 1: Improving Labor Statistics in Uzbekistan Labor statistics are usually obtained from three broad sources: (i)population censuses and labor force surveys, (ii) establishment or firm levels surveys; and (iii) administrative data or register based statistics. In Uzbekistan, the last population census took place in 1989. Regular laborforce surveys (LFS) are an essentialtool to monitor and analyze information on employment, underemployment, and unemployment, including seasonal, occasional and informal employment, hours of work, sector and occupationof employment, alternatives to formal employment including work in the household or informal sector, the earnings associated with these types of employment and to relatethem to characteristics, such as age, education, etc. of the individual. Regular LFS have not been conducted in Uzbekistan, althoughthe Government has recentlytaken importantsteps to addressthis omission. A LFSwas piloted inNovember 2001 with a view to full and regular implementation. For this report, we use information from the employmentmoduleo f the FBS. Although it is not formally a LFS, it provides information from a nationally representative sample of the population, and permitsthe estimation of some key labor market indicators consistent with ILO methodology. There are some discrepancies between activity rates from the FBS andthose from the World Bank's SIMA data base which derives its labor force information from an ILO database. Howevertheratesfrom the FBSare likely to be more reliableas the methodologyis closer to ratesobtained from a standardLFS. Uzbekistandoes not have readily usable firm or establishment level information surveys at the national level, nor a recent industrial census. Available data from firms is collected by the GOU but is aggregated at each level of government, and only summary level data rather than unit firm data is available for analysis at the national level. Unit level data onfirms, including on new entrants, exit, job creation, destruction, growth, ownership, sector of operation, and financial performance can provide important information for policy design. While aggregatedata can providea valuablegeneralpicture, firm level data is essentialfor answeringcentralquestions for industrial development policy, such as: What is the relativeproductivity growthof firms of different sizes and indifferentsectors? What are the sources ofjob creationand destruction? What is the difference in sources of growth for rural versus urban firms? What impact does ownership structure have on firm performance? Analysis of such firm level data has been central to identifying reform options in other transition countries, for example improving labor market flexibility in Lithuania, and evaluating the impact of enterprise assistance programs in Moldova. Integrationof the GOU's statistical records into a unit firm level database, or conducting a comprehensiveindustrial census, could providethe basis for similar analysis to better inform policy inUzbekistan.Additional, more systematic assessments of the business environment would also be of great benefit in identifying institutional bottlenecks to firm performance. The World Bank has recently strengthened its measurement systems of the investmentclimate with new firm level surveys. A stratified nationally representative sample of firms are surveyed with a standardized questionnaire that covers the production and financing of the firm, as well as quantitative indicators of the obstacles to business development. The surveys also enable measurements of outcomes such as total factor productivity or growth rate of output. They can be used to obtain comparable country benchmarks, identify which of the obstacles have the most impact on firm performance, and thereby help to focus reform efforts. Official labor force data have so far primarily been derived from the administrative records of enterprises inthe formal sector of the economy, which are submitted to statistical bodies on a quarterly basis. Until very recently, these statistics were based on the old Soviet concept of labor resources(Balance of Labor Resources, BLR), and characterize the employment data only by sectors of the economy. However, since February 2002, the GOU adopted internationally accepted methodologies for labor statistics (ILO), a positive step which will improvethe official labor statistics. Becauseof the problems inthe availableadministrativedata at the time of this analysis, there are discrepancies between labor market information from administrative data and that obtainedfrom the FBS. For example, while employmentrate from the BLR i s nearly 70 percent in 2000, the labor force participation rate from the FBS is only 48 percent. Two major sources of discrepancy in implied labor force participation rates are the number of employed which is significantly greater in official statistics, and the estimated Unemployment which is lower in official statistics as it is based on those who register for programs. The FBS might be underestimatingto some degree, the extent of informality inthe economy. It also does not permit the inclusionof unpaidfamily work. The true employment may lie somewhere in between the two sources. A further discrepancy between oflcial data and the FBS is observed on the degree of employment in state ownedfirms. According to the FBSdata, approximatelyone half of total employment is instate ownedenterprises while official statisticsgive afigure closer to 40 percent. This is not surprising,giventhe differences in the two data sources on employment and other basic indicators. An additional source of complication could be in the definition of the "state sector", official data are based on the official definition while self reportedFBS data are based on individuals understanding of whether their firm is public or private. Inthe contextof limited privatization, many individuals might also not be aware of whether or not their firm has been privatized. Moreover, the average unemployment rate hides signijkant variations over the year, for individuals of different age groups and by region. The unemploymentrate among youth (defined as 16 to 25 year olds) is 12.7 percent, or more than twice as high as the national average. It is high for both men and women. Given the demographic composition of Uzbekistan's population and labor force, the high incidence of youth unemployment means that youth represent about half of all unemployed. Among females they represent nearly 60% of all unemployed.The high incidence of youth unemploymentis evidencethat the economy is not 26 suficiently absorbing recent cohorts in the labor market. Youth represent a slightly higher share of the rural unemployed (about 53 percent) than the urban unemployed (about 47 percent), suggesting that this is somewhat more problematic in rural areas. The high incidence o f youth unemployment may also point at potential mismatch of the skills o f new entrants into the labor market with demand, and suggests the importance o f evaluating the relevance of educational programs. There are also significant regional differences in unemployment rates (Table 2). Unemployment rates are higher in rural areas compared to urban areas, and range from 9.4 percent in Samarkhand to 2.5 percent in Syrdarya . The regions with higher rates of unemployment also have higherpoverty incidence. Although we cannot examine the trends in unemployment (as measured by survey data) over time, the trends inregistered unemployedsuggest that poorer regions or regions o f higher demographic density saw the highest increases in unemployment between 1998 and 2000. While the national number of registered unemployed increased by 4.5 percent, it increased much more in Samarkand (35.3%), Surkhandarya (18.5%) and Dzhizak (33.3%), Fergana (58%) and Khoresm (61%) oblasts. Regional disparities in unemployment rates suggest considerable diflerences in labor market opportunities across regions. They are also symptomatic o fpotential barriers to labor mobility between regions with fewer job opportunities to those with more opportunities and between rural and urban areas. These are discussed briefly later inthis chapter. Table 2. Locationand Unemployment(% of the labor force participantswho are1unemployed) I I Unemployment Rate Total 5.8% Area ofResidence Rural 6.2% ukhara 5.4% Dzhizak 8.4% Kashkadarya 5.1% Navoi 6.9% Namangan 7.2% Samarkand 9.4% Surkhandarva 7.4% Syrdarya 2.5% Tashkent oblast 6.2% Fergana 5.7% Khorezm 4.6% Tashkent city 3.8% Unemployment i s also higher during the Spring (Table 3). During the months of April to June, unemployment is 50 percent higher than the average for the year, both in urban and rural areas. Seasonality i s slightly more severe in rural areas, and is especially severe among the youth: about 20percent of youth are unemployed during the Spring. The high seasonality of unemployment reflects the heavy dependence o f the 27 economy on agriculture. Significant variation in labor market outcomes suggests that many households may face significant shocks inlabor earnings duringthe year. Quarter Rural Urban Total Ad-June 2000 9.1 8.1 8.6 otal Significance of the informal labor market Definitions and estimates o f the degree of informality o f labor markets are fiaught with problems. Uzbekistan is no exception. However the available evidence suggests that, like many o f its neighbors, Uzbekistan might have seen an increase in its informal economic activities. Informallabor arrangements could take a variety of forms including, for example, workers employed informally in officially registered enterprises, individual entrepreneurs or self employed workers and workers in small firms not officially registered, or informal income of workers in formally registered firms. Informality i s typically associated with greater insecurity in labor market status and lower transparency in labor markets. Participants in the informal sector are also typically excluded from the social protection benefits associated with formal labor markets and are therefore more vulnerable to risks. Informality is also associated with increased difficulties in tax collection and therefore reduced resources for funding re-distributive expenditures and basic social services. Informality in labor markets is another indication of the low demand for labor from the formal sector. It is caused by incentives set by institutions in the business environment, such as hightaxes and burdensome regulations, as well as the use o f centralized wage setting (CTS). The relevance ofthese institutions inthe case of Uzbekistan i s discussed later inthis chapter. A recent study that estimated the importance of the informal sector in a number of countries found that one thirdofthe Uzbek population is inthe informal economy (Schneider, 2002p9. This estimate is similarto that inKazakhstanbutmore thanthat inKyrgyz Republic (30 percent), and substantially less thanthat incountries from the Caucasus: Armenia (40 percent), Azerbaijan (50 percent) and Georgia (53 percent) . Using a measure of compliance to economic regulations such as taxes (MacIsaac and Rama, 1997), we find that the informal sector in Uzbekistan might be as high as 51 percent based on the FBS4'. Such a measure must be used with caution, since it relies on self reported information on tax compliance which are subject to misreportin or error and since it does not fully examine other institutional arrangements such as tax exemptionsfl . An alternate measure which attempts to capture the degree o f precariousness or job security puts the estimate of informality at about 20 percent42. The correlates of all the measures, however suggest that informality is higher in sectors like agriculture and tradeand services and inpoorer regions. Retailtrade and services are activities typical o f new entrants to informal markets, and anecdotal evidence corroborates this inUzbekistanparticularly amongthe youth (Save the Children, 2002). 39 This is an estimate o f the "shadow" economy which is not the same as, although it is closely linked to, the "shadow" labormarket. 40 Individualsare consideredto comply ifthey pay either payroll or pensiontaxes. 41 Individualentrepreneursare exempt from payroll taxes. Ifthey are excluded inthe analysis, the estimate goes downto about 47 percent. 42 This measures informality as incidence of temporary, seasonal or casual jobs. Strictly speaking not all these workers are "informal sector" as some may have contracts. See Murrugarra and Tracy for various measures and potential problems. 28 Labor Market Earnings: determination, returns to human capital and the role of wage arrears How are wages determined in Uzbekistan? Different regulations are applied to different sectors. Workers in budget organizations, state owned and state stock companies43 are required to use a Common Tariff Schedule44 and thus have the most regulated earnings. The schedule is recommended for other entities, although in the private sector, wages may be set according to the specific characteristics o f the negotiation, from collective bargaining to individual agreements. As discussed later, the state sector is quite large, accounting for approximately 40 percent of employment accordingto official definitions. Taking into account the fact that the state has ownership inmanyjoint stock companies, this means that a significant share of the work force may be required to use the CTS45.As discussed in box 1, approximately half o f all employed report working in exclusively state owned enterprises. Sectors with highest employment in the non-state sector inthe FBS (definedbroadly to include mixed ownership companies) include agriculture and trade and services. Apart from health, education and government and non-profit sectors, public sector employment is highest in transportation and communication, IT and Finance. This is also confirmed by official data and usingbothnarrow and broaddefinitions ofthe non-state sector (McLiesh and Tracy 2002). How does this translate into wages? According to the FBS overall, there is no significant difference between public and private46hourly earnings (Table 4). The only significant differences are observed inTransportation and Trade, and Retail sectors, where the private workers are earning almost twice the hourly wages paid inthe public sector, and in Communication, Technology and Finance, where private workers have a 30 percent premi~m.4~ Table 4 also shows that despite the restructuring o f the state and collective farms into shirkats and the emergence o f private farms, there are no significant wage premiums to workers in the private sector relative to the state sector in agriculture, and in fact, wages are slightly lower than in the government sector. Official data also confirm that agriculture is the lowest paying sector. This i s directly related to the high 43 State organizations are 100percent state owned. The state may also have a significant share injoint stock companies, while these are not officially classifiedas "state" but as non-stateeven ifstate share is significant. 44 Cabinet of Ministers Resolution ,March 18, 1996 replacedthe previous 28 tariff schedule with a 22 tariff schedule. The CTS is an administrativetool that determines the salary accordingto the level, training and otherjob specifications. The basic wage is specified in terms of multiples of the minimum wage, and the bonuses or other compensations are specifiedas a percent of the basic wage. The minimum wage is adjustedregularly, usually once or twice a year. Starting August 1, 2002 it will be 4535 soum. In addition to assuring a minimum taxable base, are a centerpiece to the Government's strategy to control wage and price inflation in enterprises inthe environment of lax budgetconstraints to enterprises. 45 The Resolutionstates that his scheme is compulsory for all organizations andestablishmentswhich are fmancedfrom the budget, as well as for state f m s and state stock companiesand organizations the activitiesof which are basedon so- called "self management". For firms, organizations and establishments of other types of ownership (without state participation), it is recommendedthat the Tariff Schedule be established according to this new 22-grid Unified Wage Tariff Schedule. However self managing enterprises (state and non-state) are allowedto establish, within the limits of a salary fund, their own tariff coefficients and zero (first) level wage grid which is for other establishments a minimum wage. Particular wage schemes (e.g., which occupationlprofession corresponds to particular wage grid) should be approved by higher level organizations (typically ministries), and confirmed by the Ministry of Labor and Social Protection, including for self managing state owned and state stock companies. Interviewswithjoint ventures andjoint stock companies, (even with minority share of state property) confirm that the Tariff Schedule applies to these enterprises. 46 The definition of private used here is a broad one that includes mixed-ownership enterprises. According to self reported data in the FBS, only a minority of employment was in mixed-property firms. Only those who reported working in a state organizationare includedinpublic. See Box 1for possible problems with this self reportedownership status in Uzbekistan. The Government sector accordingto the FBS is much larger than that reportedin official statistics, likely reflecting at least in part the inability of respondents to distinguish between state owned companies and "privatized" companies. 47 These findings are corroborated when median wages are examined instead o f average wages. In the median wage analysis, significant premiums are found also in Health and private sector losses in Agriculture. See Murmgarra and Tracy, 2002. . 29 degree o f government intervention in production levels, crop prices and access to inputs in the agricultural sector. (see also Chapter 4). Table 4: Hourlywages for Governmentand Private Employmentby Sector (average insoums per hour and standard ekors) Sector Private Government Difference Industry 87.1 129.8 -32.9% (18.2) (2.6) Gov't & Non-Profit 78.9 82.5 -4.4% (11.6) (3.9) Total 84.4 82.8 1.9% (3.6) (8.2) How are earnings linked to human capital investments? The table below reports the returns to education estimated controlling for other characteristic^.^^The findings show that among private workers, only higher education offers a significant premium in earnings. Investments in basic and general secondary education (up to grade 11) do not significantly improve wages compared to those with 4 years o f education or less. Inthe public sector, a very different picture is observed: wage premiums increase with education reflecting the effects of the Common Tariff Schedule in public sector wages - a system that administratively increases wage with higher educational diplomas. Three points should be noted here. First, to the extent that returns to education in the private sector are likely to be more closely related to underlying productivity differences and therefore reflect demand, this suggests that higher levels o f skills are in greater demand. Second, the experience o f other transition countries, where reforms have been more extensive, such as Poland or the Czech republic, suggests that these differences inthe 48See Murrugarra and Tracy (2002). 30 Table 5: Returnsto educationby education levelin the Government and Private sectors % gain compared to 4 years of education or less (standard errors in Darenthesis) Education level Private sector Public sector 5-9 years -0.3 42.7** (18.9) (13.7) 10-11 years -1.6 52.6** (18.1) (13.1) Professional College -0.9 63.7** (18.7) (14.7) Technikum 28.5 83.9** (17.5) (15.9) Pigher education 61.8 ** 122.3"" (18.4) (17.2) returns to higher levels of education may increase further if Uzbekistan were to accelerate its transition to a market economy. Third, the use o f the CTS likely contributes to rigidities in the labor market by preventing firms that have to comply with it from rewardingon the basis of performance and productivity. While the use o f the CTS i s associatedwith lower wage inequality (for example, the gini coefficient of median hourly wages inthe private sector is 0.59 compared to 0.52 inthe public sector and other measures of dispersion are also higher inthe private sector), the costs are high. We return to these sources o f rigidity later. The role of wage arrears: As in other transition countries, wage arrears are common in Uzbekistan. About 27 percent o f workers in Uzbekistan reported arrears on their wage payments. While arrears have been consistently linked to public sector workers in other countries, according to the FBS, wage arrears in Uzbekistan is mainly a problem among workers who report that they work in the private sector: 18 percent of workers in public and mixed property organizations reported arrears compared to 39 percent in the private sector. Wagearrears are signijicant among those working in agriculture (54 percent of workers report wage arrears), construction (34 percent) and industry (21percent). These three sectors together account for more than half of total employment. Wage arrears do not affect all groups of workers equally and are more significant among low paid workers: approximately 45 percent of workers earning less than the minimum wage reported wage arrears. Among these low paid workers, 62 percent o f those working in agriculture and 48 percent of the low paid workers in constructionreported wage arrears. The high incidence o f wage arrears can be related to the fiscal problems o f the enterprises in these sectors. The high incidence of arrears in agriculture demonstrates once again the significant costs o f the state intervention in the sector, despite privatization. 31 Box 2: Gender and Ethnicity and wage differentials As with other countries inthe Europe and Central Asia region, the overall gender gaps inearnings inUzbekistan are not excessive by international standards. According to the Family Budget Survey women in Uzbekistan eam about 70 percent compared to males. Once those gross earnings are adjusted for the number o f hours worked (Le. hourly earnings), however, the difference is only 7 percent. These gender gaps are comparable to those in other countries inthe region (Paci, 2002). However, the average gender differential marks substantial differences in pay between men and women in rural and urban areas. Gender differences in wages may be explained by differences in education and training, experience, sector o f employment and position. For instance, part o f the large hourly wage difference inrural areas may be explained by the type o f work performed by males (for example processing) and females (such as cotton picking). Of course, many o f these differences themselves are often a result o f past or present differences in treatment between males and females, within households, in educational institutions, or inthe labor market. Nevertheless, it is useful to examine the extent o f the wage differential that is explained by observable characteristics, such as human capital or sectoral choices. Murrugarra and Tracy (2002) estimate that after taking into account differences in observed characteristics such as education or sector o f employment, the gender wage difference is around 18 percent in the govemment sector and 22 percent inthe private sector. Again, these gaps are not high by international standards. However, as with other countries inthe region, the unexplainedfraction ofthat wage difference ranges between 66 and 78 percent, suggesting the possible concentration o f women inless desirable, low productivity and lower paying occupations within sectors. This calls for a fuller understanding o f the source o f these gaps and the possible design o f policies to promote gender equality in the labor market. Early studies on ethnicity and labor market outcomes (Lubin, 1984), suggested significant advantages to ethnic Russians in the labor market and disadvantages to other ethnic minorities in the seventies inUzbekistan. Preliminary evidence from the recent Family Budget Survey suggests that some unexplained differences between different nationalities or ethnic groups continue to persist in Uzbekistan today. Ethnic Russians continue to have significant wage premiums (23 percent and 17 percent respectively) in both the public and private sectors that cannot be explained by differential educational attainment, sector o f work or location of residence. This may explain some o f the higher labor force participation o f ethnic Russian females noted earlier. For individuals o f non-Russian and non-Uzbek nationalities, the outcomes are very different depending on whether they work in the public or private sector. For example, ethnically Tajik individuals have 20 percent lower wages in the public sector compared to ethnic Uzbeks (although those from Karakalpakstan have wages that are almost 70 percent lower). On the other hand, non Uzbek and non Russian individuals from Karakalpakstan, ethnic Kazakhs and from other ethnicities (except Tajiks) have between 12 and 16 percent wage gains compared to the Uzbeks. These results suggest potential difficulties faced by these ethnic minorities in obtaining better paid positions in the public sector, while they can attain higher productivity in the private sector. Giventhe ethnic diversity o f the population and increases inmigrationinCentral Asia, a fuller analysis o fthe source and reasons for some o f these differences and policies to promote ethnic equity in labor marketsmight be warranted. The Labor Market And Poverty The poor have larger families and more working age adults than non-poor families, and therefore have more abundant labor resources, although o f lower quality as measured by educational attainment. The poor, however, are less able to utilize their labor resources as evidenced by the lower labor force participation rates and the higher unemployment rates among the poor. The unemployment rate among the poorest quintile is 8.6 percent compared to only 3.8 percent among the highest income quintile. The poor are also exposed to greater seasonal variations in labor market conditions, and therefore face larger shocks intheir labor earnings. Table 6 shows that unemployment rates among the poorest two quintiles reach as high as 12% or higher inthe spring. 32 Among those who are working, there are considerable differences in the type of employment between the poor and non-poor. The poor are disproportionately found inthe agriculture and industry. The poor are also substantially more likely to work inthe informal sector and have less stable employment. Thus 23 percent of workers from the poorest quintile have temporary or seasonal employment compared to 15 percent from the highest income quintile. This difference is larger in urban areas where 22 percent o f the poorest quintiles have temporary work arrangements compared to only 12 percent of the highest income quintile. The poor are also more likely to be found in part time jobs and work fewer hours on average, once again with a larger difference in urban compared to rural areas. As discussed later, the poor are more likely to be individual entrepreneurs or work invery small f m s compared to larger firms. Not surprisingly, the poor earn less than the non-poor in the labor market: the difference in median wages between the richest and poorest quintiles is 78 percent inthe private sector and 52 percent inthe public sector. Given their predominance in sectors such as agriculture where wage arrears are significant, the poorest workers face a higher incidence o f wage arrears compared to the better off workers. The incidence of wage arrears was about 32 or 33 percent for workers from the poorest quintiles compared to 21 percent for the better off. GeneratingEmploymentOpportunities Labor Market Adjustment during Transition The foregoing discussion has demonstrated the challenges faced by the labor market today in utilizing its labor resources and in rewarding labor effort. How do the characteristics of the labor market today relate to the pattern of adjustment during transition? Table 7 shows the trends in GDP, employment and growth in working age population. Until 1995 while GDP fell, employment stayed stable or increased in almost every year. Thus the decrease in output did not see equivalent falls in employment, resulting in falling labor productivity (Table 8). Unlike other CIS countries which, for the most part, saw substantial and continuous falls in the real wage, the real wage in Uzbekistan appeared to have been relatively well maintained during this period4'. 49Wage data over time are difficult to interpret. It is our understanding that at least during this period, official wage data represented wages in public organizations only and did not include the collective farms in agriculture. The wage series began to include collective farms after 1999. 33 Table 7: Growth and Employment inUzbekistan, 1992-2001 RealGDPgrowth(percentagechange) -11.2 -2.3 -5.2 -0.9 1.7 5.2 4.3 4.3 3.8 4.2 Employment(thousands) 8,271 8,259 8,379 8,449 8,561 8;680 8,800 8,885 8,983 9,136 Public sector employment(thousands)* 4,743 4,647 4,338 4,109 3,897 3,835 3,693 Workingage population(thousands) 10,463 10,707 10,975 11,222 11,508 11,824 12,176 12,550 12,942 13,353 Employmentrate (percent) 790 77.1 76.3 15.3 74.4 73.4 72.3 70.8 69.4 68.4 Public sector share ofemployment(percent 57.3 56.3 51.8 48.6 45.5 44.2 42.0 After bottoming out in 1995,real GDP has grown, on average at 4 percent per annum according to official statistics5'. Employment has also grown, although at an average rate of just over 1 percent per annum, far lower than any estimate of GDP and lower than the growth of the working age population. As a result, the employment rate, an indicator o f absorption o f labor resources, has fallen substantially, from 75 percent in 1995 to 68 percent in2001, Inother words, although growth has beenpositive, it has not generated sufficient employment opportunities to absorb the growing workforce. Real wages according to official statistics have more than tripled between 1995 and 2002, although at alternate estimates of inflation or in U S dollar terms at different exchange rate they have not risen much or even may have fallen (Figure 1). Moreover, as we have seen, wage arrears have emerged as a significant problem. The larger prevalence of wage arrears among low wage workers, and in lower paid sectors, the increase inthe importance of the informal sector and the slight widening o fwage differentials between sectors (Murrugarra and Tracy, 2002), all suggest that inequality in labor earnings has increased, despite the use of centralizedwage setting. Figure 1:Real and US Dollar Wages, 1995-2002 (1995=100) ' 350 - GDPdeflator y x -US%(at official exchange rate) --CUSS (at parallel market exchange rate) +CPI (official) -JK-CPI (alterntive, IMF) 250 200 - I 150 100 t I 0 1 1995 1996 1997 1998 1999 2000 2001 2002 Source: Ministry of Macroeconomics and Statistics, IMF and World Bank estimates 50Alternate estimates from the IMFput the estimate o f average annual growth in real GDP at 2 percent per year for this period. 34 ThePatterns of Growth Why has growth since 1995 not been accompanied by sufficient increases in employment and earnings? An analysis of the patterns of growth provides useful insights. Table 8 shows sectoral changes in employment, value added and productivity duringtransition. Table 8: Output, Employmentand Productivity,1993-2002 (averageannualpercentagechangefor the period) 1993-1995 1996-2002 1993-2002 I GDP growth: -2.8 4.0 1.9 Industry -5.5 2.0 -0.3 Agriculture 0.0 3.3 2.3 Transport and communication -11.6 4.3 -0.5 Construction -11.5 3.3 -1.2 Trade -5.8 9.4 4.9 Services 2.9 3.6 3.4 Employment growth: 0.7 1.4 1.2 Industry -1.5 1.2 0.4 Agriculture -1.2 -1.9 -1.7 Transport and communication -1.8 2.5 1.2 Construction -3.3 4.5 2.1 Trade 16.5 2.0 6.3 Services -0.7 3.4 2.2 Implicit productivitygrowth: -3.5 2.5 0.7 Industry -3.9 0.8 -0.6 Agriculture 1.3 5.5 4.2 Transport and communication -10.0 1.8 -1.7 Construction -8.5 -1.1 -3.3 Trade -18.0 7.4 -0.2 Services 3.6 0.2 1.2 Source. Ministry of Macroeconomics and Statistics; Ministry ofLabor and Social Protection; and IMF Agricultural GDP has grown at an average rate of around 3 percent since 1995, and since 1997, even somewhat faster than the overall GDP. This was accompanied by a reduction in employment between 1998 and 2002, which coincided with the restructuring o f rural enterprises. While this suggests possible productivity gains, actual labor effort in agriculture might be understated in the official statistics, and has likely increased as a result of greater and more intensive use of personal land plots. In addition, the large implicit taxation of agricultureto support the Government's industrialization strategy, represents a substantial drain on resourcesS1.Thus real wages in agriculture have remained more or less flat, and agricultural wages declined to around a fifth of industrial wages by 200252,while wage arrears predominate. The magnitudeofthe implicit taxation has beenfalling inrecentyears. See chapter 4. 52Inadditionto wages, agriculturalemployeesmayreceive a share inprofits of agricultural enterprisesand inkind paymentswhich may partly offset the difference. 35 Figure2 :SectoralDistributionof Employment l Agriculture =Industry OCanstruction O T r a n s p a r t and cammunlcntlan l T r a d e U S e r " l C e S 1 9 9 2 1 9 9 6 2 0 0 2 According to official data, industrial real value added has grown slower, consistently below the overall GDP growth rate. Reflecting this relative stagnation in output, industrial employment grew very slowly (around 1 percent per year) and productivity growth has been low. Industry's share o f employment has remained at about 13 percent. Interestingly, real wages in industry grew continuously during this period, despite the relatively low growth inproductivity, although as noted before the wage series likely represents wages inthe official sector. Industry, particularly energy, chemical and petrochemicals and machine building, have receiveda large share o f investment including public investment (Table S), and were provided with implicit subsidies through favored access to foreign exchange53, publicly guaranteed foreign loans, tax exemptions, and directed domestic credits. The case o f UzDaewoo automobile factory in the Ferghana valley (Box 2) provides useful insights into the reasons for low employment generation in industry and the potential role played by the Government's industrialization strategy which relied on high cost and high capital intensive investments in industryinthe midto late 1990's. As discussed below, inaddition to resulting in a capital intensivepattern of growth, the soft budget constraints for protected enterprises have been preventing the shedding of less productive employment in these enterprises while contributing to an environment that discourages private sector entry. Value added in services has grown at a respectable rate o f 3.5 percent since 1995 and employment in the sector has grown steadily after fall inthe first half o f the nineties. However, public services (health, education and the majority o f housing services) account for almost 90% of employment inthis sector inrecent years. Since 1995, growth has been highest in the retail trade sector. Despite its relative dynamism in output growth,54 however, employment growth inthis sector has slowed down since 1995 and the retail trade sector accounts for approximately 8-9 percent oftotal employment (Figure 2). Reflecting these patterns, Figure 2 shows a relatively stagnant picture of changes in the sectoral distribution o f employment inthe first half of the nineties, with the exception of an increase in the share of retail trade. In the second half of the nineties, while agriculture's share intotal employment declined reflectingthe release of labor from the large enterprise sector, the weight of services and construction in total employment increased. 53 Rosenberg, C. and de Zeeuw, M., Welfare Effects of Uzbekistan's Foreign Exchange Regime (IMF Staff Papers, December2001) 54 Itshouldbe notedthat growth inthis sector might be overstateddue to problems with the retailtrade deflator(see IMF Recent EconomicDevelopments2000). 36 Box 3. Welfare Implications of a Capital-Intensive Investment: An Example of UzDaewoo UzDaewoo automobile plant inAsaka, Andijan Region, is considered by the Uzbek authorities as a flagship o f the Uzbek industry, a symbol o f the country's industrial development. However, it is also an example o f a high-cost, capital intensive investment which limited employment generation and income impact. The plant with the production capacity o f 200,000 vehicles per year was built in 1996 as a 50-50 joint venture between the Uzbek government and the Korean conglomerate Daewoo at the cost o f over $600 mn. The Uzbek and Korean partners in the joint venture partners contributed $100 mn.to its statutory capital. Inaddition, thejoint venture received two loans under public guarantees - $200 mn. from Uzbek commercial banks which had been repaid by 1998 and $222 mn. from the Daewoo mother company which is still outstanding. Publicly guaranteed loans were also taken to establish joint-ventures (such as U z Dong Won Co) to supply components to the Asaka plant, and in 2001, an additional $36 mn. loan was taken to start production o f a new model, Matiz. The joint venture was also given sweeping tax exemptions, including from VAT, customs duties, and profit tax. In2001, thejoint venture produced 41,000 vehicles (Tico sub-compacts, Nexia and Matiz compacts, and Damas minivans), o f which 15,000 were exported, mainly to Russia, and employed around 2,000 workers inthe Asaka plant itself. Ifsales, marketing and supplierskontractors' jobs are taken into account, the number ofjobs created would increase to 6,000. The average salary at the plant is around 70,000 sums a month (less than $100 at the official exchange rate). The joint venture depends crucially on imported high-tech components such as engines and transmissions which account for over 60 percent o f costs o f production. As a result, its imports ($178 mn. in 2001) is more than double its exports ($84 mn.in2001) and the venture required implicit subsidies through preferential access to foreign exchange at the overvalued exchange rate. This makes the plant highly vulnerable to exchange rate depreciation: it is estimated that incase of foreign exchange regime liberalization, the plant would face an annual loss of up to $100 mn., despite the fact that its domestic pricing is already effectively based on black market exchange rate. Raising domestic or export prices o f the vehicles produced hrther would result in a substantial drop o f the joint venture's sales, even though it is a monopolist in the domestic market with imports o f new and second-hand cars restricted by high import tariffs. By all standards, the cost o f the venture is very high- to create domestic jobs that generate less than $10 mn. inannual wages and social security taxes* at thejoint venture and its domestic suppliers, over $250 mn. inexternal debt has been incurred plus dozens o f millions o f dollars in implicit subsidies have been provided annually through the foreign exchange regime. It appears the government is beginning to realize the benefits o f investing in more labor intensive technologies to expand employment: the new Matiz line installed in 2001 is around 30 percent robotized compared to 60 percent for the earlier lines. *As note above, UzDaewoo does not pay other taxes. Restructuring,privatization and the climatefor private sector investment The aggregate picture suggests that productivity increases were not widespread, and were limited to trade and to some extent a g r i ~ u l t u r eAggregate trends, however, hide useful information on the reallocation ofjobs ~ ~ . between firms of different productivity, the extent o fjob creation and destruction, and the linkages between employment creation and the policy and institutional environment. M o r e disaggregated, firm level information could provide useful insights for policy development. Such data is not, however, available in Uzbekistan (Box 4). The experience of other transition countries has shown that for growth to generate sufficient employment, serious enterprise restructuring that leads to shedding of jobs (and other resources) in existing lower productivitystate enterprises, needs to be accompanied by quick implementation of reforms needed to allow a new, job-creating private sector to emerge, including through foreign investment and especially through development of Small and Medium Enterprises (SMEs). Privatization: Small privatization involving primarily retail trade and catering enterprises, and housing made significant progress in Uzbekistan during the first half in the nineties. Privatization of medium and large 55 World Bank staff calculations suggest that within agriculture these occurred largely outside the agricultural enterprises. 37 enterprises, however, has proceeded much more slowly. Particularly damaging to the privatization process were the passage o f decrees in late 1997 and in August 1998 which unilaterally increased state stakes in the statutory capital o f privatized enterprises, and gave disproportionate power to state representatives on the boards o f privatized enterprises. Followingthis, the planned privatization through an innovative Privatization InvestmentFund Scheme and case by case privatization o f large firms essentially stalled. Thus, in 1998, approximately 90 percent of small enterprises had been privatized, compared to 76 percent of medium enterprises and less than 25 percent of large enterprises. Private sector share of GDP in 1999 was among the lowest intransition countries, comparable to the shares of Moldova and Azerbaijan, and higher only than the shares o f Belarus, Turkmenistan and Tajikistanj6 . Official statistics suggest a decline in the share o f public employment between 1992 and 1998 from 56 percent to 42 percent (Table 7). However, this decline might be overstated becauseof the broad definition o f the non-state sector where a company could be classified as non-state even if majority ownership i s with the state. Usingofficial datas7,we get significant variations inthe indicated share of the non-state sector (ranging from about 72 percent using a broad definition to 40 percent using the official definition o f private, and 9 percent usinga narrow definition that i s closer to that used inOECD countries5*). As mentionedearlier, self reported data in the FBS suggest that public sector share of employment was about 54 percent. All data sources indicate a high share o f employment in publicly owned property, even in competitive sectors such as industryand services. Enterpriserestructuringand exit of non-viable enterprises has also been slower than desirable. The number o f bankruptcy proceedings has risen since 1999 but only around 10 percent of all the bankruptcies have been amongst state enterprises. In general, the Government prefers to prevent bankruptcy o f state enterprises by restructuring enterprise debts, extending new directed credits or providing other support. Enterprises themselvesmay not initiate bankruptcy, and enterprise management can be brought to criminalresponsibility for bankrupting their enterprises. Table 9 shows investment by ownership structure since 1995. While investment rates, averaging 28 percent during 1996-2002, have been high5' compared to other CIS countries, there has been a pronounced trend towards the decline o f investment - by 2002 they fell to under 20 percent of GDP - and the majority o f investment until 2002 came from the state sector, even when the narrow official definition o f the state sector is used. Thus investment from the more productive private sector has been low. Despite annual decrees in recent years, offering new enterprises for privatization, including in the strategic energy, cotton, telecom, and railroad sectors that were previously considered off bounds for privatization, and other supporting legislation to increaseprivate sector participation, investor interest remains low. 56 EBRD, 1999,Transitionthe first ten years, World Bank2002. Accordingto this data, privatesector share inGDP was 50 percent.Accordingto official statistics, athird of GDPwas producedby the privatesector in2001. Giventhe broader officialdefiition ofprivate sector inUzbekistan, however, this may be an overestimate. 57Total employment accordingto this database is lower than that indicatedin the BalanceofLabor Resources, 58 The "broad" defiition of private in the official database includes employment in all non public property including: citizen's private, farms, cooperatives, and consumer cooperatives, agriculturalenterprises, Kolkhoz (cooperative farms), unitary enterprises, household associations,joint stock companies, public associations, joint ventures and "other types of private property". The "official" definition of private is the sum of citizen's private property, cooperatives property, propertyof householdassociations, properly of public associations,property ofjoint ventures, and "other types of private property". The narrow definition of privateproperty includesonly citizen'sprivate properly. 59Alternate estimates from the IMF, based on availablesources of savings indicatea lower level of investment, however, atjust 11-12 percent during 1998-2001.See IMF Republic o f Uzbekistan, Statistical Appendix, 2002. 38 Table 9: Compositionof Investment by Sector and Ownership, 1995-2002 1995 1996 1997 1998 1999 2000 2001 20021 Total investment 100 100 100 100 100 100 100 100 Agriculture 7 7 7 6 8 6 6 6 Industry 49 34 32 29 33 30 39 32 Other sectors(construction, etc.) 44 59 61 65 59 64 55 62 Total investment 100 100 100 100 100 100 100 100 State 55 73 68 59 66 65 59 41 Private enterprises andpopulation 10 9 10 19 15 16 13 15 Joint-stock companies 8 9 8 8 11 11 13 29 Joint ventures, foreign citizens and intemationalorganizations 21 6 10 10 7 7 14 11 Other forms o fownership 6 3 4 4 1 1 1 4 Memo: Gross fixed capital investment (as apercento f GDP)* 33 37 34 30 27 24 25 19 Centralizedinvestment (as percent of GDP)** 6 7 7 7 7 6 5 5 ***According tooffcial statistics Financedfromthe state budget Source: Ministry ofMacroeconomicsand Statistics, Ministry ofFinance Foreign Direct Investment: Inaddition to low investments from the domestic private sector, the interest of foreign investors also remains low. Table10 shows thatforeign investmentflows per capita in Uzbekistanare among the lowest in the CIScountries, with only Tajikistan having similarly low inflows inrecent years. Table 10: ForeignDirectInvestmentin CIS, 1994-2002 (net, per BOP) 1994 1995 1996 1997 1998 1999 2000 2001 2002 Armenia 0.7 5.1 4.7 13.5 58.2 32.1 27.3 18.4 23.6 Azerbaijan 2.9 42.9 80.8 142.2 129.3 63.9 16.0 27.9 130.3 Belarus 1.o 1.4 10.3 34.5 20.0 44.2 11.9 9.6 22.9 Georgia 1.5 1.1 10.0 43.5 40.6 11.3 28.4 15.1 25.0 Kazakhstan 50.3 60.0 71.4 83.8 73.7 96.0 84.9 187.7 144.5 Kyrgyz Republic* 8.4 20.9 10.0 17.6 18.1 7.9 5.8 4.1 4.2 Moldova 4.9 19.9 6.3 21.3 20.8 11.0 39.2 44.0 29.8 Russian Federation* 4.7 13.9 17.5 33.0 18.8 22.6 18.6 17.1 16.9 Tajikistan 2.1 3.4 4.2 5.0 3.9 3.4 3.9 1.4 3.2 Turkmenistan 23.4 51.4 23.3 21.4 12.6 24.5 23.8 24.3 22.5 Ukraine 2.9 5.0 10.3 11.5 14.9 9.8 12.0 15.7 14.3 Uzbekistan 3.3 -1.1 3.9 7.1 5.8 5.0 3.0 3.3 2.6 CIS average: - - 21.1 36.2 8.8 18.7 34.7 27.6 22.9 30.7 36.7 * Gross inbound foreign direct investment. Source: ECSPE country teams FDI flows are generally believed to be beneficial to employment creation. Foreign ownership is typically associatedwith greater restructuring, and the transfer o f technology and modern management practices and is therefore associated with increased efficiency. Moreover, positive spillover employment effects of large foreign investors on local subcontractors and small and medium enterprises could be significant. Although foreign investors receive preferential tax treatment and selective exemption from foreign exchange restrictions, these provisions have clearly not been sufficient to attract foreign capital, in the absence of foreign exchange convertibility and a favorable overall investment climate. 39 Therole of smaII and medium enterprises The paths o f development across countries have shown that SMEs are an important source of growth and employment creation in many o f the successful transition countries in Central and Eastern Europe and also in other parts o f the world. A vibrant SME sector can also help provide the poor with opportunities to participate in markets and improve living standards, as well minimize the social costs of reforms by absorbing labor released from stateenterprises. In recognition of the importance of the small firm sector as a source of growth and poverty reduction, the Government of Uzbekistan has since 2000 stepped up the pace o f institutional reform to actively promote SME development6', introducingmany progressive reforms. These include streamlining businessregistration procedures, eliminating exchange surrender requirements, reducing business inspections, simplifying tax and reporting requirements, establishing a department for SME protection within the Ministry o f Justice, and introducing a new approach for access to micro-credit6'. Partly inresponse to these reforms, the SME sector has increased with the number o f officially recorded SMEs having doubled since 1998, which has been accompanied by an increase in SME share o f employment. Individualentrepreneurs and dehqan (household plot) farmers have been the most dynamic sector, accounting for most o f the growth in SMEs. According to official statistics, in 2001 individual entrepreneurs and dehqan farmer together contributed 19 percent to GDP and accounted for 41 percent of total employment62 while SMEs accounted for 15 percent of GDP and 9 percent o f total employment. The growing share of SMEs inGDP, accelerated b privatization and break-ups o f larger firms, has contributed a large part of overall GDP growth inUzbekistan . Ci; Although international comparisons are hampered by inconsistencies in definitions across countries and the non-availability of current data, a comparison of the share of employment fiom smallJirms and individual entrepreneurs in 1998 ranks Uzbekistan's small firm sector approximately in the middle of transition countries (Figure 3). While Uzbekistan's share of individual entrepreneurs and employment in small firms was significantly higher than in Russia, Ukraine, Belarus and Kazakhstan, it was lower than for Central Europeancountries and the Baltic states. 60Official definitions of firm size vary across countries. In Uzbekistan, small and medium enterprises in industry are definedas firms employing less than 100 workers. Official GOU categorizations of SMEs inthis sector are: individual entrepreneurs (who do not have the right to hire employees), micro firms (up to 10 employees), small f i s (up to 40 employees), and medium firms (40 to 100 employees). There are some variations in categorizations according to the sector of operations. For instance in the trade and services sector, micro firms are defined as those with less than 5 employees and small firms as those with less than 10 employees. 61 In the second half of 2002 and early 2003 some reversals took place, which raise concern. These include a clamp- down on individual traders, sharp tightening of import procedures, attempts to centralizewholesale trade, a threat to re- privatize some trading enterprisesandmore pervasive controlover financial transactionsof enterprises. 62Individual entrepreneurs proper (Le., excluding dehqan farmers) accounted for 9 percent of GDP and 28 percent of total employment.The official statistics may overstate the number o f individual entrepreneurs and their contributionto GDP, however. According to tax authorities, the number of individual entrepreneurs registered as taxpayers is much lower (see footnote 8 to Chapter 1). 63Although detailed firm level data are not available to examine the sources of growth more carefully- for example whether due to productivity improvements or increased use of capital or labor, a simple decompositionof Uzbekistan's growth suggests that SMEs accounted for 81 percent of GDP growth in 2000 and 96 percent o f the growth in 2001 (McLiesh and Tracy, 2002). 40 Figure3: SmallFirm Share ofEmployment,1998 Lithuania 55 1 Hungary 54 9 Czech Republic Poland 45 7 Latvia 45 5 Georgia 39 6 Uzbekistan - SMEs (400) 39.5 Uzbekistan small ( ~ 4 0 ) ~ Russia 16 6 Ukraine Belarus Kazakhstan Source World Bank, GOU. UNDP Despite these positive developments, two factors indicatethat Uzbekistan has yet to fully exploit the potential of its SME sector as an engine o f growth and job creation. First, state participation in SME ownership is significant. Much of the argument on the benefits of SME development in transition assumes that SME growth i s comprised of new, private and more productive firms64. InUzbekistan, in 2001, only 38.5 percent of SMEs were classified as private enterprises. A remaining 54.lpercent are classified as non-state owned SMEs (but not private enterprises e.g. they are cooperatives, associations, joint stock companies, individual entrepreneurs, etc). And a surprisingly high number o f registered SMEs are classified as fully state owned SMEs (accounting for 7.4 percent of all registered SMEs). Given that this figure includes only fully state owned enterprises and notjoint stock companies, this estimate of state participation in ownership of SMEs is conservative. It i s substantially higher than in other countries - for example inBulgaria the state ownership accounts for 1.6% of registered firms. Moreover, because state owned SMEs are on average much larger than non-state SMEs, they account for a significantly greater proportion of employment than indicated by the numbers of registered firms.65 Thus the relatively low proportion of private, commercial enterprise suggests some caution in concluding that Uzbekistan's growth in SMEs will automatically lead to substantial productivity improvementsover time. The second caution about the role played by the SME sector comes from the sectors and nature of employment provided by the SME sector. In other transition countries, particularly in Central and Eastem Europe, the sources of growth have been primarily industrial, non agriculture and services rather than more marginal trade and agricultural activities. In contrast, in Uzbekistan, agriculture and trade accountfor the majority (68percent) of active registered SMEs. While this proportion is larger for the micro firms and individual entrepreneuddehqan farmers, these sectors still account for 50 percent o f the firms in small enterprises and 35 percent of those in medium enterprises. Reflecting the dominance of micro firms and dehqan farmers in agriculture and the significant growth in their number, approximately 85 percent of the total net increase in registered SMEs and individual entrepreneuddehqanfarmers between 2000-2002 was 64See, for example, World Bank 2002. Transition: the first 10 years Accordingto the FBS, state owned enterprises account for 54 percent o femployment in SMEs, althoughcautionmust be exercised in interpretingself-reportedownership status (Box 1). 41 attributable to agriculture, with administration accounting for another II percent and trade for the remaining 4percent. There waspractically no growth in industrial SMEs between 2000 and 2002.66 This pattern is also reflected in the type of employment provided by SMEs which suggests not only a high share of microfirms and individual entrepreneurs, but also a large association, especially of self employment and employment in very small firms (micro firms) with informality and the associated insecurity of employment. Individual entrepreneurs and workers inmicro firms are far more likely to work on a seasonal, temporary or casual basis. Only 23 percent of individual entrepreneurs and 58 percent o f workers in micro firms report working on a regular, full time basis. For small, medium firms, however, the share o f regular employment is 83 percent and 87 percent respectively (compared to 84 percent for large firms). Similar results hold with the tax compliance measure o f informality with 66 percent o f workers in micro firms reporting not paying payroll taxes compared to 41 percent of workers in large firms with over 100 employees61. Despitethese caveats, the SMEs are likely to have a direct impact onpoverty reduction in Uzbekistan. SMEs provide an important source of employment for poor people. Workers from the two lowest income quintiles are more likely to work as individual entrepreneurs or inmicro or small firms than in medium or large firms. Multivariate analysis of employment in SMEs from the FBS shows that SMEs are more likely to employ workers from the lower income quintiles, as well as women and youth. And accordingto available qualitative research (Voices of the Poor, 2000, and Expert KKP qualitative study), people perceive starting small businesses not just as a coping strategy, but as the main way to overcome poverty and improve living standards. Respondents in the KKP study stated that entrepreneurs who established their own businesses became richer in comparison with others in the community. Consistent with this, individuals who reported trying to set up small non-farm businesses in the FBS were approximately equally distributed among all income quintiles, with a slightly higher fraction from the poorest quintiles. The higher share of the higher income quintiles in private SMEs compared to public SMEs suggests that the wealth enhancing opportunities inthe private SMEsector may be greater. Increasingly, evidence is accumulating from other countries on the direct linkages between SMEs and poverty. For example, it is widely accepted that productive non farm rural enterprises played an important role in poverty reduction in China, India and Vietnam, particularly in reducing rural poverty. There i s also evidence from several countries that shows that entrepreneurial activity i s important for upward income mobility o f the poor.68The available evidence from Uzbekistan suggests significant potential for a growing private sector, to which SMEs are central, in providing productive employment opportunities for the poor and lifting many o f them out of poverty. Inthe next section, we examine the main obstacles to entry and growth of a vibrant private sector, and especially SMEs. Obstaclesto Private Sector and SME Development The above discussion suggests that a vibrant private sector that includes new small and mediumenterprises, has yet to fully emerge as a significant provider o f high productivityjobs. The overall picture is one of high state ownership and investment, significant employment in enterprises with state participation as well as in the informal sector, and insufficient entry o f the private sector as a provider of employment. A somewhat more positive picture emerges from the SME sector, although there i s evidence that the full potential of SMEs has yet to be realized. 66 Consistent with this, and reflecting some o f the difficulties faced by SMEs, S M E contribution to total exports i s small and concentrated in a small number o f firms: only 2091 SMEs (1.2 percent) exported in2001, with the vast majority o f these being from Tashkent city. The number of SMEs exporting declined by 35 percent inthe year 2000. 67 The figure is approximately 90% for individualentrepreneurs, who are not requiredto pay payroll tax 68 For example Fields and Pfeffermann, forthcoming 42 An attractive environment for investment and growth is crucial for the development of the private sector. Insufficient macroeconomic stability, an essential element for private sector entry, contributes to an unfavorable investment climate in Uzbekistan. Private investors are also deterred by structural distortions induced by high levels of state intervention in the economy, including restrictions on current account convertibility and foreign trade, and significant state intervention in the commercial activities o f enterprises (including production, marketing, allocation of inputs) either directly or through Production and Trade Associations. High taxes and burdensome regulatory environment, and low levels of transparency also contribute, as does a banking system that does not provide effective intermediation. The protection and preferential treatment o f favored enterprises through soft budget constraints, in addition to preventing restructuring and the sheddingof resources from less productive enterprises, tilts the playing field against new private entrants and particularly small and medium enterprises. In 1999, the European Bank for Reconstruction and Development and the World Bank, conducted a major survey on the business environment and enterprise performance (BEEPs Survey) in transition countries. In Uzbekistan the sample consisted o f 125 firms, stratified by size, activity and ownership. The following discussion reports the findings of this survey on the major obstacles faced by businesses. These findings are complementedby results from surveys o f 4000 entrepreneurs and managers o f SMEs as well as private and dehqan farms, conducted by Ijtimoiy Fikr (Center for Public Opinion), for the Government o f Uzbekistan in 2000 and2001. Firms in Uzbekistan report an array o f impediments to their operation and growth. At the top o f the list of obstacles identified by private f m s , both SMEs as well as larger firms, inthe BEEPs Survey69,was inflation, followed by Jinancing, exchange rates, taxes and regulations, corruption, and anti-competitive practices. Although these were identified as top barriers by firms o f all sizes, smaller firms were more likely to rate them as significant constraints, suggesting that the business environment is particularly unfavorable for SMEs. This is consistent with cross country studies, which find that SMEs face highercorru tion and greater financial and legal obstacles, and also their growth i s most affected by these obstacle&. Although the questions were somewhat different, these results reported in the BEEPs are broadly consistent with the findings on major obstacles from the Ijtimoiy Fikr Survey of entrepreneurs of SMEs commissioned by the GOU(McLiesh and Tracy 2002). Although cross country comparisons o f absolute rankings and perceptions are generally less reliable than comparisons of rankings within a country, it is interestingto note that compared to other transition countries, Jirms in Uzbekistanrate inflation and exchange rates as greater obstacles.While firms in Uzbekistan tend to report other constraints as less binding than other transition countries taken as a whole, they consistently report all constraints as more binding than their counterparts in countries such as Poland, Estonia, Czech Republic and Hungary, where new private SMEs have proven to be the engine o f growth and employment creation. This suggests a still significant potential in Uzbekistan for policies and institutions that can reduce impedimentsto private sector and SMEdevelopment. Inflation and Exchange Rate: Inthe BEEPS survey, inflation and exchange rate were reported as the two most significant constraints by small firms and were among the top two constraints faced by firms in the private sector. Insufficiently tight macroeconomic policies and large devaluations o f the official exchange rate have contributed to a relatively high inflation environment. The overall fiscal stance of the public sector, including state enterprises has been expansionary. Exchange rates and access to foreign currency have been subject to extensive regulation in Uzbekistan. Recent reforms in this area may have reduced this constraint somewhat since the BEEPs survey, at least for SMEs. Since July 1, 2001, SMEs have been exempt from 69 Relative to state owned firms, private firms were more likely to report being constrained by financing, taxes and regulations, corruption, policy instability and infrastructure. 70Beck, Demirgiiq-Kunt and Maksimovic, 2002, Schiffer and Weder, 2001 43 obligatory surrender of hard currency receipts for exports of goods and services of their own production. However, all other exports are still subject to heavy foreign exchange surrender requirements, and access to foreign exchange remains limited by import contract registrationrequirements and strict limits on the amount of foreign exchange to be sold to consumer good importers. Despite recent steps towards unification of its multiple exchange rate regime, Uzbekistan still has two legal exchange rates with a gap o f about 10-15 percent between the official and the black market rate in April 2003. This problem i s consistently rated as a major barrier to business. Unless the issue of currency convertibility, unifzcation and related trade restrictions is resolved, therewill be only a minimal improvement in the overall investmentclimate. Financin~Constraints: Among financing constraints the most significant ones inthe BEEPs Survey relate to difficulties in accessingbank loans, and the high interest rates. Access to finance was also rated as the most serious obstacle by 35 percent o f the entrepreneurs of SMEs interviewed for the Ijtimoiy Fikr Survey. These constraints can be directly related to inadequate macroeconomic stability and structural distortions which raise the cost of funds to the private sector. They can also be attributed to the significant intervention in the banking system which undermines its credibility and diminishes its ability to channel resources to productive investments. Financing problems are associated with poor performance: analysis of the Uzbekistan BEEPs results show that the extent to which firms inthe BEEPs survey view financing as a constraint is negatively associatedwith firm growth. In accordance with the elevated priority accorded by the GOU to encouraging SMEs in recent years, institutions to support the development o f financing options for SMEs have been introducedover the previous two years71. The reforms permit commercial banks to make loans to individuals, micro and small firms at the central bank refinancing rate under certain condition^^^. It is not clear whether these measureswill increase access to credit for SMEs significantly. Because the interest rate is capped at the lower Central Bank refinancing rate, banks are unlikely to make a margin on most SME loans. Moreover, few micro and individual entrepreneurs have title to property or are able to obtain the other necessary guarantees to secure the loan. An additional resolution on the provision o f subsidized credit (interest rate at about one-sixth o f the central bank refinancing rate) for initial capital o f start-up entrepreneurs from non-budgetary funds was passed in September 2001. While this could have a positive impact on access to financing, cross country experience shows that extensive use o f credit subsidies to small firms could be counterproductive, by providingincentives against firm growth and channeling funds to unproductive use. Other legal institutions to support credit markets are also yet to be fully developed. Until May 2002, when the law on Credit Unions was approved, regulations assigned lending activities to commercial banks only. This prohibited the development o f non bank intermediaries, which in other countries are major sources o f finance for micro firms. The new law may have a significant effect -already the Credit Union of Farmers has been established in the Namangan region. Despite this, NGOs still cannot legally provide microfinance because of the restrictions enforced by the Central Bank regulations on micro lending procedures73. Uzbekistan does not currently have either public or private credit information registries, which have proven effective in expanding credit market depth and lower loan default rates across countries including several transition nations such as Lithuania, Slovak Republic, and Bulgaria74. Although Uzbekistan's 1998 Law on Pledges provides for a relatively wide scope o f property that can be used as collateral for loans, it contains ''IncludingCabinet of Ministers Resolutions 902 and 903 Feb29" 2000 72 There are several restrictions: loan amounts are limited to the equivalent of 5,OOOUSD for individual entrepreneurs and 10,000USD for micro and small firms, they must be secured, and only 50% of the credit may be disbursed in cash (which can be usedonly for procurementof certain inputs),providedit is repaidwithin 18 months. 73NGOs do provide this service on an informalbasis. 74 See for example, Japelli and Pagano, 1999. Credit registriesare databases that facilitate information sharing on loan payment histories and borrower characteristics, thus enabling better risk assessment of borrowers and more efficient allocationof credit. 44 several features that may reduce its effectiveness in facilitating access to credit - including bureaucratic requirements in creating security interests, lack of an operational and unified registry of moveable property, unclear priority rights amongst competing secured creditors, and relatively high court involvement in the procedures to realize collateral. The operation of firms continues to be constrained significantly by limitations on access to and usage of -- - cash. Despite formally allowing all economic entities to withdraw cash from their bank accounts to the amount that had been previously deposited in cash,75 the number of transactions through which economic entities can receive cash inthe first place has been reduced and there are significant restrictions on the use o f cash after it has been withdrawn. Under current regulations, firms may withdraw cash for a limited range o f purposes - mainly wages and employee allowances76- and cash may not be used for purposes other than those declared upon ~ i t h d r a w a l .Settlements with other legal entities and individual entrepreneurs must be ~ ~ made in non-cash form only78- often at higher prices, which reflect the premium for scarce cash payments. There are limits on the cash on hand assignedto legal entities, and any cash in excess of these limits must be surrendered to the bank. Although it is argued this system helps to control fraud and trace financial flows to enforce tax payments, it creates a high fixed cost o f business transactions through increased interaction with the bank for non-cash settlement, and undermines property rights, inthat it reduces flexibility o f entrepreneurs to choose how to use their own cash property. Constraints on cash usage are particularly costly for SMEs, because SMEs usually make payments for smaller and more frequent transactions. It also contributes to a greater informal sector, as firms attempt to avoid the implicit tax of channeling funds through the formal financial system.79 Removal o f constraints on the use of cash will significantly reduce the transactions costs for firms and encourage more productive use o f funds in the private sector. Taxes and Rewlations: The top three tax and regulatory concerns reported by firms in the BEEPS survey are foreign exchange regulations, high levels of taxes, and customshade regulation, followed by tax administration and business licensing. Customdtrade regulations were a particular constraint in the agriculture sector. Key regulations affecting the agricultural sector and their impact on farm incomes are discussed inthe next chapter. Taxes are high in Uzbekistan. Basic profits tax on legal entities have declined from 33 percent in 1999 to 20 percent in2003. Other taxes levied on businesses include a value added tax (VAT) of 20 percent, local taxes (including social infrastructure tax on net profits o f up to 8 percent), property tax o f 3 percent, road fund tax 1.5 percent, ecology tax 1 percent, and also taxes on water, land, and excise. Payroll taxes are also high, and as discussed later, raise the non-wage costs o f labor, particularly for smaller firms. Tax administration and compliance also constitute a significant burden for private sector development. Reporting requirements are onerous: VAT is generally paid three times each month, tax returns are submitted quarterly and firms must pay advance installments o f estimated tax 6 times in every quarter. Taxes are collectedthrough the banking system. The State Tax Committee can, without notice, withdraw funds directly from an entrepreneur's account. Taxes have priority over cash withdrawals for wages. Although withdrawal of funds i s automatic, compliance costs are further increased by the requirement that the taxpayer accountant have returns and payments stamped and confirmed by the relevant authority, which must then be presentedto statistical and other agencies. Cabinet o f Ministers Resolutionnumber 280 of August 5, 2002 and Cabinet of Ministers Resolutionnumber 1156 of May 25 2002. There are additionalallowances for productionindividual entrepreneursand agro-processing enterprises. 76Cabinet of MinistersResolutionnumber 1156of May 25 2002 77Cabinet o f Ministers Resolutionnumber 376 of January 24, 1998 78Cabinet of MinistersResolution376 of January 24 1998 79For example a reported commonpracticeis cash salary withdrawalfor nonexistent staff. 45 Uzbekistan's tax base has narrowed in recent years with the proliferation of tax breaks. Strategic sectors of the economy are granted tax incentives and major investments are subject to individual decrees based on negotiation on tax incentives with the government. Small firms also receive several exemptions and, as a result of recent efforts to encourage the SME sector, face a simplified tax regime and tax administration. For exam le, small firms (with up to 40 employees) may elect to pay a single tax of maximum 12 percent o f sales &. A hightax burden on firms and their collection through the banking system, contributes to higher informality settingoff a vicious cycle. There is scope for reducingthe tax burdenon firms inUzbekistan. Elimination o f tax privileges and imposition o f discipline on existingenterprises through hard budget constraints could help tighten the fiscal stance and therefore contribute to lower taxes. Simplified tax systems for both SMEs and larger firms are needed to avoid the unintended consequence of encouraging small firms to remain small and larger firms to break up. There is also scope for reduction on the payroll taxes through reform of the pension system. Comprehensive reform of the tax administration and collection system could reduce the administrative burdenon firms and banks, as well as reduce incentives for informality and improve the fiscal position. Onerous regulations on business registration discourage new firms from establishing operations and are associated with a greater share of the informal sector and higher levels of corruption. A recent survey o f regulations governing entry (McLiesh and Tracy, 2002) shows that while regulations o f business entry were an important barrier to businesses prior to October 20Ols1, following the progressive reforms in this area to streamline the process for business registration, they are no longer a significant constraint for new firms in Uzbekistan. Prior to the reforms it took 14 separate steps, 52 days and cost 17 percent o f per capita GDP in official fees to establish a standard industrial firm in Uzbekistan. A number o f additional permits and procedures (for example sanitary inspections) were required for other firms. Compared with 85 other countries, this rankedUzbekistan 67'h, 57' and 35th,respectively. O f transition countries, only Romania and Russia had greater official regulatory barriers to entry. Following the reforms, it requires 7steps, 18 days and 8 percent of per capita GDP, placing Uzbekistan lgth, 13thand 14'h, and rating it the most efficient transition country in terms o f official steps and time to register, trailing only Lithuania in terms o f costss2. Overall, Uzbekistan now places between Switzerland and Singapore inentry procedures. The new system is not, however, strictly speaking a 'one window process' as advertised, since the hokimiyats do not coordinate the requirement to open a bank account, prepare seals, or clear the trademark name. Moreover, it i s not yet clear whether the new registration procedures are being adhered to, especially in regions with lower capacity, are recent moves in late 2002 to impose greater restrictions on traders have reversed some o f the positive developments in this area. Nevertheless the reforms are an example of successful efforts to reduce administrative obstacles, which could be expanded to other areas o f government intervention infirm activity. Prior to 2001, levels of inspections o f business for compliance with regulations and regulatory instruments were extraordinarily high. Over 60 different government agencies inspected businesseson average between 20 and 30 times per year. The 2000 Ijtimoiy Fikr survey o f 4000 entrepreneurs rated bureaucratic interference resulting from these inspections as one o f the most significant barriers to operation. Extensive 80The rates may be lower or higher, depending upon the sector. For instance, the rate is only 6 percent in agriculture while trading enterprises the rate is 30 percent. Individual entrepreneursmay pay a monthly flat tax rate between 5 and 30 minimum wage. Further, SMEs only have to reportto statisticalandtax agencies quarterly. There are no exemptions from payroll tax for SMEs, although individual entrepreneursdo not pay. 81Cabinet of Ministers Resolution347, August 2001. 82 Itshould be noted, however, that the costs per capita have declined only because o f the change in official exchange rates, andalso that these are official costs, andthe reportedincidence of additionalpaymentsis significant. 46 reforms during the year 2000 to the inspections process have help to reduce this problem. The system i s reinforced by a new complaints procedure administered by the Ministry of Justice. Under the new system, no more than one inspection per year i s permitted, the inspection may not exceed 30 calendar days without permission, and all interested supervisory agencies participate. 83 As o f 2000, less than 40% o f registered enterprises are inspected during a given year.84 Although this is a substantial improvement, it is still extraordinarily high in an absolute sense and dramatically higher than the experience of firms in the vast majority o f countries. InOECD countries, most firms will never be inspected duringtheir operation. Licensing and related regulatory requirements increase the official costs o f firm operation, through official fees and transactions costs of compliance. They also provide officials with opportunities for direct contact with firms, greater discretion, and therefore more possibilities to extract informal payments from entrepreneurs. Despite the significant success of reforms in licensing and inspections, there remains scope to further reduce bureaucratic intervention - by reducing the number o f licenses, permits, standards, certifications, authorizations, orders, controls and inspections to only a minimum required for essential consumer protection and social purposes. Corruption and Governance: Firms in the BEEPS Survey also rate corruption as a major constraint to businesses. Follow up questions reveal that informal payments are widespread in Uzbekistan. Only 20 percent of firms in the BEEPs survey report never having had to make an informal payment to public officials, whereas over 43 percent o f firms report paying bribes frequently, mostly or always. SMEs report making bribes more frequently than large firms, and firms in commerce find bribes more necessary than in other sectors. Within transition counties, Uzbekistan ranks third in the frequency of informal payments, behind Azerbaijan and Romania (Figure 4). Bribes are most common for getting licenses and permits, dealing with tax authorities, and interacting with customs authorities. Although the results o f the Ijtimoiy Fikr survey indicate that the incidence of corruption is lower, a highproportion of entrepreneurs declined to reply (8.7 percent) or were uncertain (22.9 percent), suggesting that the issue is sensitive. Figure4: FrequencyofPaymentsto PublicOfficials:(% FirmsResponding"Frequently-Always") UzbekistanTotal Uzbekistan. small Uzbekistan. medium Uzbekistan large - CEEAverage FSUAverage Estonia Lnhuania Czech Republic Kazakhstan Russia Hungary Poland Georgia Armenia R0 mania A zerbaiia n 0% M% 20% 30% 40% 50% 60% 70% Source: BEEPs 83 Inspections are limited to once every two years for tax-payers with positive audit results. Other reforms include: streamliningproceduresfor obtaininglicenses, statutory limits on the time to issue licenses; abolishingannualrenewalof licenses; reducing the number of inspections agencies; requiring a coordination plan of inspections agencies, which is approved by a RepublicanCouncil on Coordination of Activity of SupervisingBodies; and introductionof a "book" on inspections, which is held by 90% of registered firms. Inspectors must sign the book before commencing inspections and provide evidencethat the inspectionwas authorized. 84 UNDP, 2000 47 International experience has established that governance - or the process by which authority is exercised - influences development outcomes, including firm performance. Corruption has a corrosive effect on investment and economic growth rates, and is associated with larger unofficial economies. By contrast, greater enforcement of property rightsand stronger rule of law promotes investment, firm growth and overall economic growth. A cross country study of governance issues in 173 counties carried out this year by the World Banks5 reveals that, although it has improved between 1997f98 and 2000101, Uzbekistanperforms poorly in governance (Figure 5). Despite the improvements, governance indicators for voice and accountability, and quality of regulation remain particularly low in 2000101. Uzbekistan still ranks in the bottom 13 percent of all countries on voice and accountability and in the bottom 11 percent in regulatory quality, performing worse than the average for FSU countries. For all governance indicators, Uzbekistan ranks below the average for countries of Central and Eastern Europe. Figure5: GovernanceIndicatorsinUzbekistanand Other TransitionCountries, - i Controlo f Cormption Ruleo fLaw 57 0 I Regulatory Quality Govemment Effectiveness Voice and Accountability 0 20 40 60 80 100 poor govemance good govemance Percentage of countries with lower score FSUA v. CEEAv. =Uzbekistan 1997/98 17Uzbekistan2000/01 Source: Kaufman, Kraay and Zoido-Lobaton 2002. The Governance indicators are compiled by the World Bank. Indicators for FSU average and CEE average are for 2000101, For each category o f governance indicator, countries are given a score between -3 and 3, based on the component indices o f each indicator. A lower score represents poorer governance performance. For each country or country grouping, the graph shows the percentage o f countries with a lower score. Governance and accountability i s severely affected by limited transparency and access to information. Firms inUzbekistanreport they are not informed inadvance when there are reformsthat affect their operations. In the Ijtimoiy Fikr survey, only 29.4 percent of respondents reported that there were sufficiently well aware of entrepreneurs rights - the remainder rate their knowledge as superficial or poor. Although most laws are publicly available, lower level regulations, ministerial and association norms and directives sometimes are not, and there is extensive reliance in the current legislative framework, on lower level laws such as decrees, directives, and norms which are subject to more discretion and are less publicly accessible. Macroeconomic data availability is limited mainly to quarterly aggregate accounts. Disclosure requirements for firms are relatively weak, and the proportion o f firms using International Accounting Standards i s significantly lower than in other transition countries. Institutions to support access to information are weak. The media i s almost entirely state owned and restrictedby tight licensing and defamation laws. Uzbekistan does have a freedom of information act, although the level o f generality of the provisions means it i s more symbolic than 85Kaufman, Kraay, and Zoido-Lobaton(2002) 48 operational. Provisions for what types o f information can be accessed as well as exemptions in the case where information will harmful are vague, andthe act makes no detailed provision for implementation. The predictability and relevance o f policy can also be improved with mechanisms that improve voice and participation and ensure that needs o f business are adequately incorporated into policy formulation, either directly or through consultations with representative business associations. Neither yet occurs at a significant level in Uzbekistan. Many firms are members o f the Chamber of Commodity Producers, or the Association o f Private and Dekhan Farms - although these associations are not entirely independent from government, having been established by a government decree. Of entrepreneurs polled in the Ijtimoiy Fikr survey, 27.2 percent reported to be members of the Chamber o f Commodity Producers, o f whom 31.7 percent were fully satisfied with the Chamber's performance. By contrast 79 percent of those surveyed reported to be members o f the Association of Dehqan and Private Farms, 43.9 percent of whom were fully satisfied with the Association's performance. Like many other transition countries, civil society groups and alternative business associations are still nascent and extensive policy dialogue between these groups and government i s rare. Relative to other factors, rule of law and contract enforcement in Uzbekistan is not a significant impediment to private sector development. A survey of dispute resolution procedures (McLiesh and Tracy, 2002), finds that judicial efficiency inUzbekistan compares favorably with other transition countries. Firms in the BEEPs survey also rate the judiciary amongst the top one or two transition countries in terms o f fairness, speed and enforcement of judgment. The honesty and corruptibility o f the court system was rated lower, however, at around the average of transition countries and well below international best practice. And enforcement i s hindered by the fragmentation o f laws that has accompanied reforms, which increases the likelihood o f inconsistent interpretation. Enforcement i s also constrained by inadequate transparency and dissemination, especially o f lower level law, as discussed earlier. Inadequate Commtition: Private firms in the BEEPs Survey reported anti-competitive practices as a significant obstacle to business development. Regional comparisons provide evidence that the environment for firms inUzbekistan is less competitive and more heavily dominated by state firms. Compared withfirms in other transition countriesJirms in Uzbekistan are more likely to report theyface no competitors at all, and greater competition fiom state owned enterprises and micro firms. Firms in other transition countries are significantly more likely to report SMEs, foreign firms, imports, and smuggled goods as their main source of competition. Also, firms in Uzbekistan report that domestic competitors provide only slight pressure to develop new products and markets. Of transition countries, only Belarus and Armenia report similarly low levels of competitive pressureto innovate. Competition and contestability in Uzbekistan are constrained by highlevels o f state ownership in key sectors (including manufacturing, banks, transport, construction, health and education, and media) and a poor climate for private and foreign investment. In addition to high state ownership, competition is also restricted by numerous exemptions and preferences for various groups in many business regulations. Whilst some of these preferences now benefit SMEs (such as taxes and credit for micro firms), the playing field is not level. Large, and in particular state owned enterprises receive implicit subsidies through preferential access to foreign exchange at below market rates as well as subsidized credit. The use ofthe banking sector as a major conduit to loans to state owned enterprises and farm collectives limits the availability o f finances for new and less well connected enterprises. At the end o f 2001. approximately 64 percent o f banking system credit was extended to state owned enterprises. The energy sector is a major source of soft budget constraints to enterprises through low energy prices, payment arrears, and budget offsets. This delays the restructuring o f industrial enterprises and thus the shedding of assets which could be usedby smaller, new enterprises. Inthe Ijtimoiy Fikr survey of approximately 4000 entrepreneurs, 37 percent of respondents reported that they had unequal opportunities to do business. Perceptions of unequal opportunities for business were higher in urban 49 areas (51.3 percent) as compared with rural areas (35.5 percent), and were also higher for respondents representing micro firms (47.3 percent) and small enterprises (48.3 percent). Extensive state intervention, which restricts the development of a competitive private sector, also takes the form of controls exercised through Production and Trade Associations (PTAs). There are currently 56 PTAs covering a range of sectors - from mining, to agricultural industries, to power, to services. Although officially non governmental, these associations are created by Cabinet of Ministers resolutions, and in some cases act as holding companies for Government stakes in PTA memberjoint stock companies. Managers o f PTAs hold the rank equivalent to a government minister. The PTAs exercise significant power over their members - allocating funding, material resources / production inputs, issuing production orders, controlling prices, industry standards, certifications, and even reprimanding, fining and removing the management o f their member firms. Without membership, it i s difficult for a firm to operate effectively in that industry. Controls are exercised through PTA orders, administrative norms and protocols, and are therefore not evident ingovernment ownership statistics, or laws andregulations. Inaddition, the governmentdistributesaround 60 key commodities such as energy and metals through a system o f quotas called "material balances." As a result, inability or difficulties in obtaining material resources or access to production inputs were the second most reported obstacles to business (18.2 percent), following access to finance (35.2 percent) in the Ijtimoiy Fikr survey. Major challenges of respondents included obtaining access to water (49 percent of respondents), gas, fuel and lubricants (3 1 percent) and spare parts for machinery (31.8 percent). Uzbekistan's competition policy "The Concept o f State Anti-Monopoly Policy" aims to develop competition and entrepreneurship; regulate monopolies, prevent abuse o f dominance and protect consumers. An Anti- Monopoly Committee implements the policy, supported by 4 major laws on competition. The AMC has taken steps to reforming many o f its approachesto regulating competition, including committing to transition away from price controls to monitoring anticompetitive behavior. Nevertheless, the AMC's effectiveness is limited by its definitions of dominance and economic market, lack o f autonomy from the government, and inadequate linkswith structural reforms, including eliminating soft budget constraints and encouraging exit of non viable firmss6. As such it has had little impact on the competitive environment. Continued efforts in privatization, elimination of soft budget constraints and subsidies, and encouraging exit o f non viable firms and reduced state intervention in the product and input markets and in firm activity, i s necessaryto improve levels of competition. High Taxationof Labor Firms in Uzbekistan face highpayrolltaxes, totaling 39.7 percent in 2003.87 Highpayroll taxes are common inmany countries of the region. Although payroll taxes in Uzbekistan are not as highas those in Slovakia, Poland and the Czech Republic, they are higher than in many other transition countries and significantly higher than the average for EUand OECD countries. High payroll taxes increase the non-wage cost of labor, and are likely to be a disproportionate burden for SMEs bothinterms ofthe cost, but also the administrative requirements". They also introduce a bias against unskilledand low wage labor as the proportion of this non-wage costs is higher for these workers, and it is more difficult to understate wages for these workers. This contributes to reduced demand or informal employment of such workers. Because payrolltaxes are inevitably shared with employees, high payroll taxes also reduce incentives to work and therefore labor supply. Thus high payroll taxes contribute to a vicious cycle o f high taxes and reduced formal employment. Reducing social security taxes could contribute to 86Broadman, 2000 "In2003 payroll taxes are 35.0 percent for the Pension Fund, 1.5 percent compulsory payment to the Employment Assistance Fund, 0.7 percent to the trade unions fund. A 2.5 percent mandatory insurance fee is levied on employees' wages. Inaddition, legal entitiespay a0.7 percentsales tax to the PensionFund. There are no exemptions for payroll taxes for SMEs althoughindividual entrepreneurs do not pay. 50 increased employment opportunities in the formal sector, but should be carried out in the context o f broader reform o f the social protection system (chapter 7). The role of labor marketpolicies Flexible labor market regulations enable firms to adapt resources to market conditions and provide mobility o f workers to more productive enterprises and regions. One important type o f labor market regulation is Employment Protection Legislation (EPL). Strict EPL may increase labor costs by placing constraints on firm flexibility to adjust its labor force according to market conditions, and can also encourage informal labor contracts. Across countries, strict EPL is associated with lowerjob creation ratios and higher unemployment, especially o f long-term natures9. I n Uzbekistan, EPL does not appear to be a signijkant barrier to labor market jlexibility. In an index o f the strength o f key aspects o f EPL - including grounds for dismissal, notification and approval requirements for dismissals, rules for reemployment, retraining requirements, notice period and severance pay - Uzbekistan scores o f 7 out o f 11, where 11 represents the most flexible EPL. This places Uzbekistan within the top half o f a sample o f 80 countries, and amongst the most flexible EPL for transition countries. Although employment protection legislation might not be a significant barrier, the use o f the Common Tariff Scheme among a significant fraction o f the workforce might be a significant source o f labor market rigidity, preventing firms from rewarding workers according to productivity and performance. It also contributes to segmentation and informality inthe labor market. Minimumwages inUzbekistanare not very highand have consistently remainedbelow 20 percent o f average wages, a level which does not typically constrain the demand for labor by firms. Moreover, the FBS indicates that approximately 10 percent o f workers in the private sector earn below the minimum wage suggesting that it is not always complied with. Complianceis also lower inthe rural areas. Only a small fraction inthe public sector earn below this wage, reflecting the use o f the minimum wage in the CTS. Although the minimum wage might not be a binding constraint overall, it may be a more significant constraint for youth or new entrants inthe labor market or low wage workers. The low levels o f participation in unemployment and active labor market programs suggests that these programs do not provide significant disincentives to work (Chapter 7). As discussed earlier, disparities in unemployment rates between urban and rural areas, and particularly between regions suggest potential barriers to labor mobility. Unfortunately available data to not shed adequate light on the extent o f migration between regions, although a recent UNHCR report suggests a reversal in the direction o f urban-rural migration in the second half o f the nineties, with the overall net migration being quite low9'. Are labor market policies a source o f barriers to labor mobility? Like other countries in the Former Soviet Union, Uzbekistan inherited some o f the mechanisms, and particularly the use o f the propiska , to control urban expansion 91. The propiska system is only enforced at present in Tashkent a9 See, for example, Blanchard (2000) According to this report, during the first half of the nineties net migration was from urban to rural areas, reflecting the collapse in incomes and employment opportunities, the higher prices for food, other consumption items, and housing and the distribution of land in rural areas. Between 1995 and 1999, the net migration was from rural to urban areas, reportedly due to better opportunities and low incomes and wage arrears on collective farms. The FBS provides possible evidence of moderate migration, with more than 6 percent on the working age population reportedly absent from their households,although these individuals mightbe inthe same region. 91 Mobility restrictions in the former Soviet Union has been extensively discussed and somewhat overstated in the western literature. These were created in the fifties but their effects on controlling net migration into cities was clearly observedduring the late eighties when the implementation was already inplace and different economic performancewas observedacross regions in the FSU. One of the control mechanismswas called "total" control and involved restrictions in obtaining a document indicating proper residence in a specific area (propisku).Other controls involved limitations on 51 for security purposes. While there are no significant formal restrictionsto obtaining apropiska document, the administrative processes and direct and opportunity costs may represent a barrier to entering the Tashkent labor market. Limits to formal internal migration lead into the illegality (informality) of workers without a propiska who are then subject to more precarious labor market conditions, probably involved in jobs of shorter duration and without any coverage o f social insurance. With the possible exception of thepropiska for Tashkent, barriers to mobility are more likely to come from outside the labor market, including for example, high housing costs in more prosperous regions, the importance o f household plots inrural areas, and the significance o f family networks. In sum, with the exception of the Common Tariff Schedule, and the possible impact of minimumwages on the demand for particular groups of workers, and o f thepropiska inTashkent, labor market regulations do not appear to be a major obstacle to labor reallocation and employment generation. Firms in the BEEPSsurvey rate labor regulations as one o fthe least constrainingtypes o f regulations. Wideningskills mismatches As discussed earlier, the analysis of returns to education inthe labor market suggeststhat the private sector is demanding higher levels o f education, particularly university education. This may represent a demand for higher levels of skills while indicating a possible mismatch in the training received by graduates of the secondary education system and o f the vocational training system. At the same time, the majority of Uzbekistan's labor force and two thirds o f its unemployedhave secondary or lower levels of education. More problematic, Uzbekistan is one of the few transition countries which has registered consistent disinvestments inhigher education for the past two decades. The reasons for this needto be better understood and the trend reversed. While the disinvestments in education may be a reflection o f inadequate expansion o f employment opportunities for better educated workers, they may also partly be a result o f the distortions created in the labor market by the centralized wage setting mechanism. Whatever the reason, this raises concerns about generating adequate employment opportunities in the future. To ensure that growth results in increasing employment opportunities and incomes in the future, it will be important to increase the pace o f reform, reduce the role of the Government in wage setting leaving these to more flexible negotiations between workers and employers, and to reverse the declines in higher education investments. ImplicationsFor Policy The characteristics o f the labor market in Uzbekistan today reflect some o f the challenges that Uzbekistan faces in utilizing its abundant labor resources to their full potential. Low activity rates, particularly for women, represent low labor supply but also reflects low opportunity costs of staying at home as a result of poor employment opportunities. Higher unemployment than suggested by official statistics, its significance among the youth and in some regions, the presence of underemployment and informal labor arrangements, are all indicative o f insufficient demand for labor in the formal sector. Wages are low in sectors such as agriculture, and wage arrears predominate. The poor are disproportionately affected: they are less able to utilize their labor resources, have higher unemployment rates, are more exposed to seasonal fluctuations in labor market opportunities, and are more likely to be inthe informal sector or sectors characterized by wage arrears. For growth to result inpoverty reduction, it must be accompanied by an increase in employment opportunities and higher incomes for all workers, including the poor. Although growth has been positive in recent years in Uzbekistan, it has not generated sufficient employment opportunities to absorb the growing work force. Employment and incomes remain constrained by significant state intervention in the economy and an the business activities that could encourage population growth. For a detailed discussion o f the effects o fthepropiska on city growth see Gang and Stuart (1999). 52 investment climate that is unfavorable to private sector development, including the development of SMEs that have been a major source o f growth and employment creation in many other countries. For growth to reduce poverty in the future, it will be necessary to improve the environment for private sector development , including further reducing obstacles for private firms and SMEs, and addressing regional, gender and ethnic disparities in employment opportunities. While the SME sector has benefited from a number o f exemptions for example, from foreign exchange regulations, subsidized credit, lower taxes and a simplified tax regime, these privileges provided in isolation o f an overall improvement inthe environment for the private sector can only have a limited impact, as well as have the unintended consequence o f providing small firms with incentives to remain small and large firms to break up. Improving the environmentfor private sector development Macroeconomic environment: Increasing macroeconomic stabiliv and proceeding with fundamental structural reforms, particularly the liberalization of foreign exchange and trade regime i s a pre-requisite for economic growth andjob creation. Despite recent progress with unification and the elimination o f surrender requirements for production SMEs, the foreign exchange and trade regime remain heavily regulated. Elimination o f restrictions on access to foreign exchange, unification of exchange rates and further reduction or elimination of mandatory surrender requirements for exports are essential to improving the overall investment climate and reducing the bias against exports. As discussed inthe next chapter, this will reduce the scale o f implicit taxation on agriculture and improve agricultural incomes. Some groups will be negatively impacted by such a liberalization and it would be necessary to strengthen social protection for these groups. This is discussed briefly in Chapter 7 and the Government is currently analyzing the potential impacts o f liberalization. Improving availabilitv of finance: Addressing financing constraints calls for financial sector reforms to improve the credibility o f the banking system. This will require significant reduction in state intervention in the banking system to ease constraints on credit and cash for private firms and SMEs. Eliminating preferential access to credit for state owned enterprises and effective privatization o f banks will level the playing field for the private sector, and improve financial sector efficiency. To improve access of SMEs to loans, the Government has recently introduced credit subsidies for these firms. While limited subsidies to smaller firms may be justified on the basis o f disproportionately large informational asymmetries in credit markets, extendedcredit subsidies to small firms may prove counterproductive: by discouragingexpansion of small firms, encouraging fragmentation of large firms, and possibly allocating credit to unproductive enterprises. Further efforts to develop institutions that help the financial sector to allocate credit to its more productive use - such as streamlined collateral laws, encouraging the development o f credit information registries, and a legal framework for microcredit (including through provision by NGOs) - are likely to expand access to credit for the new private, productive sector, while not introducing distortions that discourage firm growth and reduce financial sector efficiency. Eliminating restrictions on the use of case will reduce transactions costs and decreaseincentives for informal activity. Reducing the burden of the tax svstem: The high tax burden on firms could be reduced through a broadeningo f the tax base and elimination of tax privileges. This will also contribute to reduced discretion in the tax regime, increasing transparency. Reduction of payroll taxes in the context o f overall reform o f the social protection system could also contribute to increased employment and lower informality. Comprehensivereform of the tax administration and collection system could reduce the administrative burden on firms and banks, as well as reduce incentives for informality and improve the fiscal position. Reducing licensing and regulatory requirements: It will also be important tofurther reduce licensing and regulatory requirements (including the number of licenses,permits, standards, cert$cations, authorizations, orders, controls and inspections) to only a minimum requiredfor essential consumer protection and social purposes. A comprehensive review o f current licensing and related regulatory requirements of private 53 enterprises, and identification of non essential regulatory instruments is a necessary first step. Further steps include additional simplification o f administrative procedures to comply with regulatory requirements, standardization o f procedures across different agencies, reducing the number of agencies involved in supervision and issuanceo f licenses, making the regulatory requirements fully transparent (including through use o f information technology), locating regulatory information and decision-making close to businesses. Lessons from other countries that have reformedtheir licensing systems, such as Sweden, Portugal, Australia, Italy, Lithuania, Poland, and Latvia, provide instructive examples o f successful reforms (McLiesh and Tracy, 2002). Uzbekistanhas had successful experience reforms of entry regulations, expanding this type o f reform to regulations o f operations will significantly reduce impedimentsto firm performance. Reducing corruption and imtwoving governance: Lower taxes, elimination o f discretion in the tax regime, and reduced o f the burden o f regulations will contribute to increased transparency and therefore reduced corruption and informality and improved governance. Governance can also be improved by greater consistency and clarity in regulations, improved access to and dissemination o f information, increasing disclosure requirements on firms, and ensuring freedom o f the media to investigate and report on governance failures. It will also be important to examine and reduce incentives for informal payments (including options for civil service and procurement reform), enforce penalties for corruption, and monitor performance of officials. Voice and participation can be improved by introducing mechanisms to incorporate the needs of business into policy formulation, encouraging the development of independent business associations and civil society groups and increasing consultations and policy dialog with these groups. Improving the competitiveenvironment: Continued privatization o f state owned enterprises and reduction in state ownership, as well as openness to international investment and trade will increase competition and contestability, improving the efficiency o f firms in the private sector. It will be essential to impose hard budget constraints on state owned and strategic enterprises to facilitate restructuring and the shedding o f resources. This will require the elimination of all explicit and implicit subsidies including all price subsidies, payment arrears and budget offsets in the energy sector, tax: exemptions,publicly guaranteed loans, and preferential access to inputs such asforeign exchangeand credit. The forced restructuringof enterprises will be accompanied by some social costs associated with the destruction o fjobs and higher unemployment. To reduce social costs, public expenditures will need to be reallocated from these explicit and implicit subsidies, to a strengthened social safety net, for example through the development o f labor redeployment programs (providing income support to displaced workers, and job information and counseling, and training, for example to assist in reintegration into the labor market). The most important form of social protection, however, will come from ensuring an environment that is conducive to the entry and growth of a new private sector, especially small and mediumenterprises. The establishment of hard budget constraints to enterprises is essential to level theplayingJield for theprivate sector and SMEs. The experience o f countries such as Poland has shown that overall social costs could be lower if the destruction of jobs associated with serious enterprise restructuring, is accompanied by a quick increase inthe demand for labor from a newjob creating private sector. In addition to reduced state ownership, the reduction of regulations and explicit and implicit subsidies, improvement o f the competitive environment will require a reduction in the extensive intervention of the state in production, prices and inputs, through the Production and Trade Associations, elimination of "material balances", improved effectiveness of the StateAnti-Monopoly Committee. Reduce costs of labor and labor market rigidities and enhancing labor mobilitv: Decreasing taxes on labor (including payroll taxes, in the context o f reform of the social protection system), could reduce the non-wage costs o f labor, decrease informality and increase the demand for labor, particularly those with lower levels of skills. It will also be important to reduce the role of the Government in wage setting, permitting more flexible arrangements negotiated between workers and firms. In this context limiting the influence of the Common Tariff Scheme will be essential to reducing segmentation in the labor market, and permit firms to reward 54 workers based on productivity. In the longer term, this will also contribute to investment in education and skills that are rewarded inthe labor market. As discussed inthe next chapter, to reduce rural poverty, particular attentionwill need to be paid to the above actions to generate employment opportunities outside of agriculture in rural areas. In addition, measures should be taken to increasethe flexibility and ability of workers to move to where jobs are created within and across regions. Barriers to the mobility o f labor between regions will need to be better understood and eliminated. This includes an examination o f the impacts of restrictions into Tashkent city, and reduction of these restrictions. Other measures include improved information on employment opportunities and the encouragement o f privatejob placement agencies. It will be important to understand and address constraints outside o f the labor market, such as in the housing market (for example to increase the role of the rental market, introduction of a mortgage market, etc.). Ensuring relevance of skills: In order for measures to increase labor demand and supply to have an impact on employment outcomes, it will be essential to ensure that the skills being demanded in the labor market are available in the work force. As discussed earlier, the private sector i s demanding higher levels o f skills and the skills beingoffered by the majority o f labor force participants are not in sufficient demand. Moreover, the decline in investments in highereducation is o f concern. Inthis context it will be essential to understandand reverse the declines in higher educationthat have been occurring for two decades now. Moreover, as seen in chapter 6, levels of investment arefalling at all levels of education, which will have serious consequencesfor growth and employment in thefuture. Inadequate availability of employment opportunities and rewards, lack o f transparency in labor market opportunities, and the perceived irrelevance of skills attained in the education system may be important contributing factors to the declining investments in education. Other possible factors behindthis trend and some measuresto address these, including reforms in the formal education and training system to provide the more flexible skills that are in demand in a market economy, are discussed in Chapter 6. In the short term, the skills o f workers who have already entered the labor market might be enhanced by on thejob training by firms and selective and carefully designed and targeted training programs, with substantial private sector participationinprovision. Address regional, gender and ethnic inequalities in the labor market: For growth to be inclusive and result in broad based improvements in living standards, it will be essential to reduce inequalities in labor market opportunities. Reducing barriers to labor mobility will increase the ability o f workers to avail of better opportunities in other regions. Poor employment opportunities in particular regions may themselves be a function, however, o f a number o f factors including very different endowments o f human and physical infrastructure, production structures, or differences in policies and institutions across regions. A full analysis of the sources of these differences is beyond the scope o f this report although the evidence o f regional disparities is consistent in other chapters of the report. For example, increasing equalities in human capital endowments, such as educational attainment will be an important measure to increase equality in labor market outcomes. Understanding and addressing the sources o f regional differences in labor market opportunities will be important for broad based improvements in living standards. Also important would be actions to improve the participation of women in the labor market (example increasing their ability to access higher education, increased availability of child care opportunities), and an improved understanding of the sources of unexplained differences in labor market earnings between men and women, and between different ethnic groups followed by appropriate measures to address these. Increases in the demand for labor will lead to improvements in labor supply, as the opportunity cost of not participating in the labor market increases. Reductions in some of the disincentives to supply, such as those for early retirement, particularly for women, through reform inthe pension system will also help improve labor supply. 55 Chapter4 IMPROVINGRURAL INCOMES: THE ROLE OF AGRICULTURAL POLICIES Two thirds of Uzbekistan's population, approximately 15 million people, live in rural areas. The poor are largely rural: seventy percent of Uzbekistan's poor (amounting to approximately 4.5 million people) live in rural areas. The Southern Economic Region (consisting of Kashkadarya and Surkhandarya) and Khorezm are poorer than average and also have a larger than average share of the rural population. But the rural population are spread across all oblasts. Thus policies that improve incomes in rural areas will result in broad based improvements in living standards. Agriculture accounts for nearly a third o f Uzbekistan's GDP approximately 40 percent of its exports, almost a third of employment, and nearly half of all employment in rural areas. Improving productivity in the agricultural sector can have substantial impact on growth and incomes, particularly in rural areas. State intervention in agricultural prices, production and inputs constrains agricultural growth below its potential. Higher prices, greater flexibility in choosing production, and improved incentives are essential to improving agricultural productivity, and therefore incomes. Improvingaccess to key inputsand improving incentives for their efficient use is also important to increase agricultural productivity. Against the backgroundof substantial extraction of resources from the agricultural sector, equitable access to household plots has been an important source of income and consumption for the rural population. Moving forward, it will be important to contain unintended increases in inequality inaccess to land that could come out of the process of formation of private farms. Beyond the direct effects, improving agricultural incomes inmany countries appears to have been particularly effective inreducing rural poverty because o f demand spillovers to local markets in which the non-farm rural poor have a large stake. Thus combining policies that improve agricultural productivity with actions to improve the business environment for private SMEs could expand employment opportunities for the rural poor including those who might be laid off from agricultural enterprises. Structureand Trends inAgriculturalOutput Over 60 percent of Uzbekistan's output inagriculture comes from the crop sector, of which cotton is the most important crop, followed by wheat, and then vegetables. The remaining output in agriculture comes from the livestock sector. Agriculture in Uzbekistan encompasses approximately 4.1 million hectares o f crop land (ploughed or fallow) and an additional 360,000 hectares of perennial crops. About 82 percent of these lands are irrigated, amounting to less than 0.3 hectares per rural inhabitant. After falling inthe first half of the nineties, agricultural output increased steadily92 since 1998, and by 2000, the value o f output in agriculture had recovered to its 1991 value. The recovery represents a recovery in both agricultural output as well as in the livestock sector. Within the crop sector, however, the production of cotton declined steadily until2000, reflecting both declines inareaunder cultivation (of about 16 percent), but also declines in yields (of about 23 percent.). The production of grain, however, has increased and in 2000, was about twice the production of 1991. The increase in grain production represents an increase in land allotted to grain in line with the objective of increasing self-sufficiency (50 percent), but also an increase in yields (38 percent). Overall production of major products in the livestock sector (meat and milk) have stayed more or less constant throughout this period while feed production and eggs have fallen. 92 Since 1997, agricultural GDP has been growing at about 4.6 percent per annum according to official statistics (and approximately half of that rate according to alternate IMF estimates). 56 The overall trends in output also conceal the difference by type o f farms and the expanded role o fthe dehqan farmers (household plots), who by 2000 accounted for almost 65 percent o f the output o f the sector (about 40 percent o f crop output nearly 90 percent o f livestock output). The dehqan sector's increased share in crop output reflects both more land but also higher yields o f this sector. At the same time, the output o f the former state and collective farms in crop output declined to about 48 percent (from over 80 percent), with a more dramatic share drop in share of livestock output (to approximately 10 percent). Drop in share o f crop output of the agricultural enterprise sector (mainly "shirkats" or cooperative farms) reflects both a decline in land but also lower yields. By 2000, new private farms accounted for about 5 percent of the total value of agricultural output. Key Agricultural Policies :Objectives and Directions Uzbekistan's agricultural policy since independence has the following stated objectives: (i)produce cotton in a manner that allows the country to stabilize export revenues; (ii)achieve self-sufficiency in wheat production; (iii)redistribute revenues earned in agriculture to finance industrialization; and (iv) improve rural standards o f living. To achieve it objectives, the Government o f Uzbekistan has pursued a gradual and regulated approach to reform, maintaining control over the production and sale of "strategic" crops such as cotton and wheat that are crucial to its revenues and program o f import substitution. The following discussion summarizes the key institutions, policies and practices used to achieve the Government's objectives in the agricultural sector. . Production,Marketing and Pricing Policies- StateProcurement of Wheat and Cotton: TheState's Role The state's role in determining the structure o f agricultural production is significant. Production targets for cotton and grain cover about 60 percent o f the country's arable land. The state also imposes a number o f marketing requirementsfor cotton and wheat (through procurements known as state orders or "purchases for the "state needs"), and sets the prices which it pays for its purchases. State purchasesare made by the state marketing enterprises, UzDonmasulot and UzKhlopkoprom. Cotton and wheat are priced on a "cost plus" basis with no relationship to either world/regional market prices or exchange rate. State orders for wheat occur under a two-tiered pricing system, with a 25 percent o f planned production procuredat the state order price and an additional 25 percent procured at a highly administrative price, called the "negotiated" price (which in the past three or so years have been set at approximately 20-25 percent premium over the state order price). When contracted sales volumes for a particular farm (including sales at the procurement price as well as negotiated price) are not attained, then other farms in the same district are made to sell more than their contracted amounts to compensate for lower than planned volume o f purchases. These extra volumes for the other farms are bought at the negotiated price. Since actual grain production is typically well below the target level nationally (with the recent exception o f the remarkably good harvest o f 2002), the state usually ends up purchasing about 80 percent o f all grain output and state sales account for about 90 percent of all grain sales. The remaining grain in excess o f the procurement target is retained for on- farm use or for sales in local markets. Inrecent years, the average price paid by the state for grain purchases has been approximately 40 percent of free market prices, defined as prices for domestically freely traded grain. Inthe case ofcotton, the state order is set at about 30 percent ofplannedproduction, and producers who meet the production targets, hypotheticallyhave the rightto sell the residualoutputto the state marketingboardat a "negotiated price" and to receivethe option to purchase back cotton by-products (cotton seed oil and oil cake) at very low prices. However, ifthe production target for cotton is not attained, the farm must sell all cotton to the state at the procurement price and may not sell any cotton at the higher price. Thus, since production targets are often not met, actual state procurement at the state order price is much higherthan the formal state order, and are higher than those for grain. In effect, "negotiated prices" are not different from the state procurement prices and distribution of by-products to farms rarely takes place. State purchases cover 57 virtually the entire cotton crop. Inrecent years, the average price paid by the state has been approximately 50 percent of the border price calculated at the OTC (over-the-counter) exchange rate9394 Thus, the pricing mechanism for cotton and grain transfers resources from farms in two ways: by setting unreasonable production target levels, which guarantee virtual monopoly o f the state over cotton and grain output; and by setting prices for state purchases well below world market levels. Products other than cotton and wheat are not subject to explicit production quotas or procurement orders. However, public ownership continues for major processing plants for meat and milk, and local governments in some cases require processors to provide milk and meat under preferentialterms to schools, hospitals, and other local institutions with special status. This implicit price control in the livestock sector depresses the prices that processing plants can pay for raw materials with the result that most meat and milk i s sold on local markets fully by passing the processing sector. Bureaucratic regulations and steps associated with export o f fruits and vegetables restrict trade, althoughthese were temporarily simplifiedto some extent in 2000. Agricultural Inputs The state is actively involved in the allocation and distribution o f fertilizer, seeds, agricultural chemicals and fuel. Off-farm inputusage for crops under state production (cotton and grain) are plannedand managed by the Ministry of Agriculture and Water resources. Inthis context, the Government also runs a costly program to equip machine tractor parks. Supply of inputs is primarily by state suppliers and the private sector is relatively underdeveloped inthese areas. There are implicit subsidies, however, to agriculture, associated with the provision o f imported inputs for agriculture at the overvalued official exchange rate, and the VAT exemptions for fuel and fertilizers. The state pays virtually the entire cost o f maintenance and operation o f the irrigation system and explicit budgetary outlays for agriculture are almost entirely for irrigation. There is additionally a substantial implicit subsidy associated with the irrigation system through the subsidized pricing of electricity for pumping.The Ministry of Agriculture and Water Resources manages activities of regional and district irrigation departments, which supply water to farms. Water is supplied on the basis of contractual agreements with shirkat and private farmers and trade of water i s forbidden by law. Dehqan farmers inthe household sector are also included in the planning o f water deliveries, although priority is given to shirkats and to farmers' cotton and wheat fields. Creditpolicies State marketing agencies inthe cotton and grain sectors, provide for producers part o f the value o f contracted cotton and wheat as prepayment (up to 50 percent inthe case of wheat and 80 percent inthe case o f cotton), and there is an interest rate subsidy associated with this pre-payment, which is internalized in the contracting relationship between producers and state marketing agencies. The subsidy element may be partially or fully offset by late payments for products to producers. Since 1999, these farm advances are provided in centrally directed tranches through the Fundfor State Agricultural Purchasesunder the MOF (at an interest rate o f only 4 percent) and is one o f the few options for working capital available to farms. In addition debts and arrears o f agricultural enterprises are routinely written off. Land Reform and Farm RestructuringPolicies Since 1991, a number of changes have been introduced into the structure and organization of agricultural enterprises, in an effort to create a stronger sense o f ownership on the part of the work force, through the conversion of all state enterprises into various forms o f non-state enterprises. Thus, in the first part o f the 93The exchange rate usedhere is the one that applies for foreign exchange sales for consumer goods imports, which lies between the official exchange rate on the one hand and the exchange bureau ("cash") rate, and black market rate on the other. 94In 2001-2002, however, reflecting the continued fall in world market prices, this has increased to about 70 percent. 58 nineties, nearly all state farms or sovkhozes were converted into kolkhozes or collective farms, as well as (to a lesser extent) other economic entities such as "shirkats" (cooperatives), private livestock farms, leasehold farms, andjoint stock companies. After an initial expansion inhousehold plots in 1991, the acreage of land in the household sector expandedvery slowly (approximately 4.5 thousand hectares per year). Rural enterprise restructuring in Uzbekistan gained momentum in mid-1998 under the Presidential "Program for Development of Economic Reforms in Agriculture during 1998-2000". The goals of the program, supported by the new Land Code, the Law on Agricultural Co-operatives ("Shirkats"), the Law on Private Farmers and the Law on Dehqan Farmers -are essentially three-fold: (i)maintain the present human and physical infrastructure o f the large farms in order to stabilize the cultivation of "strategic commodities" (cotton and wheat); and to ensure continued employment, minimum living standards, and social stability - until other sources of employment have been developed; (ii) improve the incentive and institutional to framework for increased output by former collective farms by restructuring them into shirkats; and (iii) to strengthen the legal and enabling environment for private farms (larger than 10 ha.) and for small dehqan farms. Thus this period saw an acceleration ofthe transformation ofthe collective farms into shirkats or cooperative f m s . By 2000, approximately half of the kolkhozes or collective farms were transformed into shirkats or cooperativesg5, and by 2001, the process was virtually completed. Shirkats consist of extended family production unitsor pudrats inwhich production is based on contractual arrangements between the pudrats and the shirkats. The period since 2000, andespecially 2001, has also seenthe liquidation of unprofitable shirkats and their conversion into private farms. Approximately 155 (6-7 percent o f total) consistently unprofitable shirkatsg6 were liquidated, by the end o f 2002, and their land used for the creation o f new private farms. Accordingly both the number o f private farms and the area under these farms increased steeply after 1998 and again more recently since 2000. By March 2003, there were more than 80,000 private farms encompassing about 1 million hectares (or 25%) o f arable landg7. This period o f farm restructuring has seen some destruction o f jobs in the enterprise sector from the transformation o f the collective farms9' into shirkats (approximately 150,000 workers), and the liquidation of shirkats (approximately 250,000 workers), although thisjob loss has beenkept as low as possible to minimizethe social costs associatedwith restructuring. The experience to date: Agricultural policies and farm productivity We now examine some of the implications of the policies discussed above, and in particular, their relationship to income and living standards. Implicit Taxationof Cotton and Grain As discussed above, the government exercises a high degree of intervention, particularly in the cotton and grain sectors, to redistribute income from agriculture to develop industriespromoting import substitutes. This redistributionto provide resources for industryis central to the Government's development strategy. Implicit taxation ofthe agricultural sector takes place through price and marketing policy in cotton and wheat, and the overvalued exchange rate. Table 1 shows the extent o f taxation from these two sources in 2000. It 95Comprisedof roughly 10 groups of 10-20pudrats (extendedfamily units) each, with each group coordinated by abash pudratchi (literally "head of the pudrats"). 96 This process is usually preceded by a period of "sanation" (usually two-years) during which the state introduces external management of the farm, restructures the farm's debt and overdue payables to input suppliers, aims to strengthen financial discipline, as well as dispense with excess assets, stocks, and labor force. If the farm is still unprofitableafter this, it is fully transformedintoprivateand dehqan farms. 97Incomparisonthere were 3.3 millionhouseholdor dehqanfarms encompassingabout 522,000 hectares. 98mainly those working insocialinfrastructure as budgetswere transferred to local governments 59 also shows the extent of the different subsidies provided to the sector, to mitigate the adverse impact o f the price and marketing policies on agricultural growth. The outflow o f resources from the sector through the implicit taxation o f cotton and grain totaled almost 550 billion soums in 2000 (or roughly $800 million). Most of this (80 percent) was through the taxation o f cotton, withjust under 90 percent coming from the loss of income to producers of cotton and grain arising from their surrender o f foreign exchange at the official over-valued official exchange rate99. At the same time, a sizeable amount o f subsidies are returned to the sector through fuel, irrigation water, fertilizer, machinery service, and credit subsidies. The largest subsidy (250 billion soums) comes from the irrigation water supply through budgetary outlays for maintenance and operation, and particularly through the subsidized price of electricity for pumping. Also substantial, were the credit subsidies (approximately 180 billion soums in2000) through working capital advances at subsidized interest, but more significantly through debt write-offs. Table 1-CalculationofNet Implicit Taxation in2000 Implicit Taxes on Grain and Cotton 1million soums 545,032 ~ ~~~~ ~ ~ Implicit Taxes of Grain million soums 112,975 Taxationthrough Price Gap million soums - 5,63 1 Taxationthrough ForeignExchange Gap millionsoums 118,607 Implicit Taxes on Cotton millionsoums 432,057 Taxationthrough Price Gap million soums 60,27 1 Taxationthrough ForeignExchangeGap million soums 371,786 (Subsidies I million soums 1 462,666 I Total Taxes Less Subsidies 82,367 Total Taxes Less Subsidies as Share of Agricultural GDP 8.5% Total Taxes Less Subsidies as YOof Total GDP 2.6% In total, net taxation of the agricultural sector in 2000 was roughly 80 billion soums or $120 million, which amounts to over 8percent of agricultural GDP,andjust under 3percent of total GDP. This is lower than the level o f net taxation in 1996-1998 which ranged between 3 and 5 percent o f GDP, primarily because of the reduction in the difference between the prices paid for state procurement and the international prices. Nevertheless, the taxation of the cotton and grain translates into lower incomes for the rural population that derives its incomes from these sectors. This is discussed below. At the same time, the input subsidies 99Exporters of cotton, which is considered a "centralized" commodity, are subject to a 100 percent surrender requirement. 60 provided to these sectors, for example, the substantial debt write offs, limits the availability o f credit and inputsto other producers and constrains their productivity. The analysis for this report was carried out in 2002, and draws on administrative data, household and farm surveys for 2000/01. Since the preparation and during the discussion of this report, the implicit taxation of the agricultural sector was further reduced in 2001,which representsconsiderable progress. Taxation, which currently takes place primarily through exchange rate policies, was still significant, although subsidies balanced them off in 2001, and when debt write-offs are included, there was a net transfer of resources to the sector. However, agricultural enterprises, particularly those involved in cotton, still incur losses'", while price distortions combinedwith mandatory cropping plans leadto inefficient output mix and land use. Emergence of three majorfarm types The combination of policies has resulted inthe emergence ofthree major farm types: farm enterprises (mainly 1863 shirkats and about 221 joint stock companies), private farms, and dehqans (household sector), which differ in management and administration, production, pricing and land use arrangements, as well as input arrangements. Farm enterprises and private farms account for the majority o f the production o f the strategic crops, Le. cotton and grain production. In 2000, for example, agricultural enterprises accounted for 81 percent o f cotton production, with private farms accounting for the reminder. Inthe case o f grain production, agricultural enterprises accounted for 66 percent, private farms for almost 15 percent, and dehqans for 19 percent. Accordingly, cotton and grain are the primary crops produced by agricultural enterprises and private farms (withjust under 80 percent o f their land area devoted to these two crops), while dehqan farms have devote on average under 40 percent o f their landto grain and have more varied cropping patterns. Despite the replacement of Soviet era collective and state f m s by shirkats, these enterprises are still subject to many o f the same centrally planned mechanisms (output targets, state supplied inputs, mandated planting times, etc.) and strongly resemble the farm enterprises they replaced. The main achievement o f shirkats compared to the collective farms is their organization into smaller work units to improve monitoring and use of labor inputs. In view o f their role in production o f strategic crops, "private" f m s are also subject to planned production quotas. They do, however, have greater flexibility o f labor usage and to some degree on the usage of other inputs. Dehqan farms, on the other hand are the most "liberalized" and make their own production and inputsdecisions for the most part. The three types of farm differ inthe average land area. Shirkats vary in size, usually between 700 to 2000 hectares, with the average size being about 1500 hectares. They are, however, divided into smaller production units or pudrats, the average size of the land farmed by each pudrat being approximately 15 hectares. The new private farms are a minimum o f 10 hectares, with an average size of approximately 20 hectares (in 2000). In contrast, the maximum legal size for a dehqan farm plot is 0.35 hectares, with the average size approximately 0.17 hectares. The private farms created encompass on average about 40% more land per able-bodied farm worker compared to shirkats and are therefore less intensive intheir use of labor. All agricultural land is still technically owned by the state and is not tradable or transferable. Land userights for dehqan plots are life long and can be inherited, while leases of private farms are legally for a minimumo f 10 to fifty years (with 65 percent being for 10 years). Pudrats, on the other hand enter into land lease arrangements (usually 3 to 10years) decided at the general meetingo f shareholders. I00In 2002, in part due to a superior grain harvest, the financial situation of farms was reportedly better than in the previous 3 years when the Government undertook massive debt write-offs. In January 2003, the Government has approved a new regulation on pricing policy to be applied for the 2003 cotton crop which puts a cap on the number o f deductions (ginning, transportation and other fees) and establishes a more transparent pricing structure. 61 Relativeproductivity of dijjferentfarm types Table 2 shows the relative shares of the three different farm types intotal value of output and total land. The first row is the share o f total monetary value of output attributable to that type o f farm, the second row is the share of total land attributable to that type of farm, and the third row i s the ratio of the two numbers. It gives a measure o f the relative productivity o f the farm types. A ratio equal to one would implythat that particular type o f farms share o f output is equal to its share of land input. A number less than one indicates that this particular farm type is less efficient because it i s not producing as much output given the amount o f land that is being used for production. While a number greater than one indicates that the farm is relatively more productive because it i s producingmore output with less land than the competing types o f farms. Thus Table 2 suggests relatively low and comparable averageproductivity of agricultural enterprises andprivate farms andsignificantly higher relativeproductivity of dehqanfarms. Farm Enterprises PrivateFarms Dehqan Farms Share of Total Value of Crop Output 48% 9% 43% Share of Total Land 72% 16% 12% Ratio 0.67 0.56 3.58 Table 3 shows the relative yield differentials in the case o f the production o f a single crop, wheat, between dehqans and agricultural enterprises, and between private farms and agricultural enterprises. It shows that yield differentials vary significantly across regions. On average, however, and in most regions and for all years, yields o f dehqan farms are higher than those o f agricultural enterprises (the only exceptions being Andijan and Jizzak in 1998. The yield inthe case of wheat i s nearly 40 percent higher on average for dehqan farms, reaching twice as high in some regions. Further, the yields o f private farms and farm enterprises are quite close on average (again with significant variation between regions). Once again, Jizzakh stands out as an exception with significantly higher yields of private farms (and higher than dehqan farms). The same pattern, i.e. similar yields on average o f agricultural enterprises and private farms, holds inthe case o f cotton. Lundell,Kropp, and Shamsiev (2002) examine yield differences for a variety of crops usinga Farm Survey of over 1200 private, shirkhat and dehqan farmers in five districts 2000/2001 (Rural Enterprise Support Project Survey, henceforth E S P Survey) and find that dehqan farmers achieved greater yields across all types o f crops compared to private farmers or shirkats. What is behind these differences? Yield differences may be due to a variety o f different factors, including quality and availability o f inputs, their intensity o f use, knowledge, etc. Also important are incentives and the flexibility available to farmers. A full analysis is beyond the scope o f this report and has substantial data requirements. Nevertheless, basedon the consistent pattern of higher farm yields of dehqan farms, it is fair to make two important conclusions. First, the greaterflexibility available to dehqanfarmers to set their mix of output to maximize returns, as well as greater incentives (higher prices and incomes) are important reasons for their higher productivity. Inthe case o f shirkats, the farmers do not directly benefit from higher yields as any dividends are shared among all pudrats, including less productive and unprofitable ones. Second, smaller size of land holding is not necessarily associated with lower land productivity- indeed dehqan farms which have the smallest land size, have the highest yields per acre. 62 11 Tashkent 169% 126% 168% 110% 160% 100% 144% 111% 12 Fergana 161% 95% 172% 99% 170% 94% 157% 101% 13 Khorem 111% 91% 106% 92% 118% 80% 119% 77% 14 Nationalaverage 137% 110% 132% 101% 137% 98% 151% 98% Inputs, CreditandIrrigation As discussed above, the production, marketing and pricing policies reduces flexibility, and lowers profitability and productivity particularly for farmers in the cotton and grain sectors. Farmers also report other important constraints to productivity, such as financing problems, input supply, insufficient irrigation. Credit: Inadequate access to finance remains an important constraint to agricultural productivity"'. High interest rates, the availability o f lease financing equipment, banks lacking the money to lend, complexity of bureaucratic procedures and difficulty o f forming a business plan, lack o f collateral, and the need for special connections and informal payments to bank officials are frequently reported barriers to financing in the agricultural sector102. The state marketing agencies for cotton and grain provide working capital advances (at a subsidized interest rate) in centrally directed tranches through the Fund for State Agricultural Purchases under the MOF. For most farmers, this is the only source o f working capital available. However, most of this advance takes the form of accounting adjustments between different suppliers o f inputsand very little cash is made available to the farmers. Since the transfers take place between various (mostly state) agencies involved in input supply, the arrangement has the additional impact of precluding private sector input suppliers. The state intervention in banks undermines their credibility, while the use of banks to channel resources to state input supply and purchasing agencies limits the availability o f credit in rural areas. According to the RESP Survey shirkats and private farms (particularly those growing cotton and wheat) are the main clients of banks in the rural areas lo' Although access to finance is an important impedimentto business in all sectors, financingconstraints were ratedas more severe by firms in the agricultural sector in the BEEPS Survey. The Ijtimoiy Fikr Survey, which includedprivate and dehqan farms in its sample reports financing constraints as most significant constraint. Unfortunatelyno breakdown of constraintsfor farmers was provided I O 2BEEPS Survey (See McLieshand Tracy, 2002) and E S P Survey *see Thurman andLundell, 2001). 63 (apart from state input and purchasing companies). Not surprisingly, few farmers make long term investments into their farms. Irriaation: Despite the large state subsidization o f irrigation, access to water remains a problem. Table 4, based on the RESP Survey for five districts shows that the majority o f farmers report that they lack sufficient irrigation and that this holds across farm types, although over 90 percent o f farmers have an irrigation system inplace. Thispointsto potentialproblemswiththe existingirrigation infrastructure. Shirkats PrivateFarms Dehqans Sufficient 33% 18% 28% Insufficient 64% 77% 60% No Irrigation 3yo 5Yo 2% Poor maintenance o f irrigation and drainage systems have ledto wasteful water delivery, and about 50 percent of water i s lost in di~tribution"~. At the same time, lack o f water charges contributes to wasteful use. Water use per hectare is 12,000m3(including amounts used for leaching). The latter figure compares with 9,000- 10,000m3/ha in countries such as India, Pakistan, and Egypt, which are not known for efficient irrigation.'04 The serious deterioration and threat o f breakdowns of the irrigation infrastructure, in addition to contributing to inefficient use, increasesthe vulnerability o fthe rural populationto the impactso f periodic droughts. Input SUPP~V:As discussed above, the off-farm inputs for planned production (cotton and grain) are specified inthe production plan, which limits choice inthe volumes, quality and delivery timings o f inputsto these farmers. The high state participation in input supply and the limited options for input urchase, however constrain the volumes and quality o f inputs available for all agricultural production"? Direct provision o f inputs or their subsidization also limits incentives for efficient use o f inputs. For example, the program to equip machine tractor parks has slowed down a switch from capital intensive methods o f production inshirkats and private farms producingwheat. Employment,IncomesandRuralWelfare The rural labor market Agriculture accounts for a third of all employment but about 49 percent o f employment in rural areas. More than half o f the bottom three quintiles in rural areas are employed in the agricultural sector. Thus the rural population and especially thepoor is highly dependenton agriculture: rural non-farm employment outside of the state and large enterprise sector accounts for less that 10 percent o f rural non agricultural employment in rural areas, with the self employed accounting for nearly half of these. The high dependence on agriculture and the limited availability of non-farm employment opportunities, combined with much higher population pressure in rural areaslo6has manifested itself in higher rural unemployment and underemployment, greater youth unemployment, and greater seasonality of employment opportunities. Rural areas are also characterized by greater informality o femployment arrangements (Chapter 3 and Murrugarra and Tracy, 2002). 103 Estimates range from 40% (MAWR) to over 60% (TACIS). The latter figure is based upon an in-depth study of farms in Uzbekistan. See: TACIS, Improving the Productivity of Water in Central Asia: (Tashkent: TACIS, 2000). 104 Republic of Uzbekistan Irrigation and Drainage Sector Study vol. 1(Tashkent: World Bank, 2000), pp. 6-7. 105 Input supply as a constraint is reported in the RESP Survey (see Thurman and Lundell, 2001), as well as in the Ijtimoiy Fikr survey. Access to water was onthe top. lo6 The share of rural populationincreased from 60 to 63 percent between 1991 and 2000, reflecting higher declines in fertility rates in urban areas, greater external migration from urban areas. In the first half o f the nineties, intemal migration was to rural areas, although this was reversed inthe latterhalf. 64 As noted above, agriculture saw an acceleration inrestructuringsince 1998 which was accompanied by labor shedding in agricultural enterprises. Although the restructuring was accompanied by a sharp increase in private farms these farms use much less labor (private farms employ about 0.66 workers on average per hectare compared to 0.90 workers per hectare on shirkats). Thus employment in agriculture fell in each year between 1998 and 2000 (Table 5). The data inTable 5, however, do not include labor effort inthe household sector, which has become more significant and is difficult to estimate. Labor effort in dehqan farms i s mostly informal ,drawing intensively on the participationof family member^''^. Thus inthe Family Budget Survey, only a small fraction of the employed (about 1 percent) report working in dehqan plots as their primary employment. 1998 1999 2000 2001* Employment('000s) 3,467 3,220 3,093 3,207 Percentagedecline from previousyear -1.9 -7.1 -3.9 3.7 Despite this restructuring, average wages in agriculture have remainedstagnant relative to overall wages and employment in agriculture is characterized by low wages and a high incidence of wage arrears. According to both official statistics as well as the Family Budget Survey, labor income in agriculture is about half of overall labor earnings'" makingagriculture the lowest paying sector. Agriculture i s also characterized by the highest incidence of wage arrears (with 62 percent of workers reporting arrears compared to 45 percent on average), and the hi hest incidence o f workers earning below the minimumwage (12 percent compared to 6 percent on average)1f .9 Since the majority of formal agricultural employment"' i s in agricultural enterprises and private farms, these wages reflect the low earnings from farms heavily engaged in the production of cotton and grain. The high levels o f state intervention in the production of strategic crops, and the associated implicit taxation translates into lower productivity and farm earnings for the majority o f workers inthese enterprises. Data on gross profit margins from agricultural enterprises shows that these enterprises have been incurring losses, particularly from the production of cotton. Welfaregainsfrom higher grain and wheatprices Using data from the RESP Survey, Lundell, Kropp, and Shamsiev (2002) simulate the impact on the gross incomes o f cotton and grain farmers from an increase in the price cotton and grain to world market prices111. Figure 1presentsthe results o fthis simulation inthe case of cotton price liberalization for shirkats and private farmers. lo'World Bank staff estimates of employment inhousehold farms is substantially higher than official estimates in the Balance of Labor resources. While the BLR still suggests an overall fall in agricultural employment between 1996 and 2000, informal estimatesby World Bank staff suggest that employment stayedabout the same. logOfficial labor earningsrefer to "monthly wages" inthe public sector, althoughsince 1999, the official series includes wages on collectivefarms. Labor earnings computedfrom the FBSare hourly wages. 109Murmgarraand Tracy (2002). 'loEstimates based on official statistics are about 90 percent, self reportedemployment status in FBS suggests about 70 percent, with another 10 percent inprivatefarms. The RESP Survey is not a nationally representative survey and does not contain adequate information on consumption or expenditures. The authors use a welfare forecasting equation from the FBS to forecast consumption in the RESPdata set. The predictedconsumptionis then put into welfare quintilesbasedon FBS welfare quintiles. 65 Figure 1 Cotton Price Simulation - 2500 - 140% 1 2 3 4 5 Welfare Quintile 1 IncomeChan e CottonFarmers (thousandsof Soums) 1 -% Change infn&ne It shows that an increase in cotton prices to world market prices will increase gross annual incomes of all cotton farmers, with the increase in absolute terms approximately equally distributed across the welfare quintiles. However, since cotton comprises a larger share of the agricultural income o f farmers from the lower income quintiles, cotton price liberalizationwill have a relatively greater benefit for those at the bottom of the distribution. A similar pattern, although with smaller magnitudes, i s observed in the case of shirkats and private farmers from liberalizationof wheat prices (Lundell, Kropp, and Shamsiev, 2002). In interpreting these results it is important to remember that these simulations are for farmers in the RESP Survey only. Unfortunately we do not have nationally representative data that will permit the examination at a national level. Second, this looks at the first order impact o f increasing the prices and does not take into account the impact of eliminating production plans, or of production responses o f farmers to the increased prices. The role of householdplots As noted above, household plots now account for over 60 percent of total agricultural output (over 40 percent o f crop output and about 90 percent of livestock output). As noted in Chapter 2, access to agricultural land (primarily in the form o f household plots) is widespread in rural areas : approximately 97percent of rural households have access to at least oneplot, poor and non-poor alike. The poor also have slightly more land area available to them (0.19 hectares of usable land compared to 0.15 hectares for the non-poor). The data does not permit an evaluationon the quality o f land which could differ across different households indifferent regions. Ownership of livestock i s also common inrural areas althoughthe poor are slightly less likely to own livestock :76 percent o f the rural poor own livestock compared to 81 percent of the non-poor. The non-poor are also more likely to register their farms: 2.3 percent o f the non-poor inrural areas register compared to 1.1 percent of the poor. Not surprisingly,then householdplots are an important source of income andfood consumptionfor the rural population, and especially the poor. Lundell, Kropp and Shamsiev show that income from household production comprises the largest share o f income for rural households (31 percent compared to 24 percent from labor income). Thurman and Lundell (2001) confirm the relatively greater role played by household plots as a source o f income, particularly for the shirkat farmers inthe RESP districts. 66 Murrugarra and Storozhuk report the importance o f household plots in food consumption: more than a quarter o f the food consumption of rural households comes from non-purchased food compared to only 7 percent for urban households. Interestingly and somewhat surprisingly, non-purchased foods account for a higher share of the total food consumption o f the better off households: they account for about 30 percent of the total food consumption of the highest quintile, compared to 21 percent for the lowest quintile. In particular, the poorest quintile derives 27 percent more o f their consumption o f wheat and wheat products from purchasescompared to the 18 percent for the richest quintile. Thus the expansion of household plots has provided the rural poor with an important safety net inthe context o f low agricultural incomes and has likely had an important impact on the living standards o f the rural population including the poor, preventingmany of them from falling into poverty. Growinginequity in access to land? Against this relatively equal distribution of household plots is the concern o f growing inequity in access to land that is resulting from the process of formation of private farms. As discussed earlier, the private farms are substantially larger than dehqan farms. There is increasing evidence that the better connected are relatively better off are more likely to be awarded land for private farming (Thurman and Lundell). Since the private plots quite large on average, especially compared to household plots, this process could result in growing inequality in opportunities. Further, as seen above the inequality in land size goes beyond distributional concerns as the evidence supports greater efficiency and higher productivity o f smaller land holdings o f dehqan farmers. Conclusions and PolicyRecommendations The above discussion has showed that the substantial state intervention in agriculture in support o f its development strategy may have come with high costs in terms o f agricultural productivity and growth, and therefore in terms o f living standards for the rural population. Thus the objectives o f the government's agricultural policies are only being partially achieved. Reflecting the erosion in cotton production, cotton exports have been slowly declining in volume and cotton's share of exports dropped from 50 percent in 1993 to about 28 percent in 2000. While wheat production has increased to almost self-sufficiency levels, the opportunity cost of the Government's strategy for self sufficiency inwheat are high. The trade-off has been lower areas to cotton and to neededfodder and fertility replenishingcrops. The net transfers from agriculture have supported the Government's industrialization strategy but as we saw in the previous chapter, that strategy has not been associated with adequate creation of jobs, while it has resulted in lower incomes in agriculture. The government intervention in agricultural production and inputs reduces flexibility o f farmers to select the most efficient output and input mix, while lower prices for outputs and subsidies for inputs reduce incentives for efficient production. The experience in China showed that early reforms in agriculture through the abolishing o f collectives, introduction o f household responsibility system, and reduction in mandatory deliveries o f output to the state were important factors behind the substantial growth, and increased incomes for the poorer segments o f the economy and the substantial reductions inrural poverty. The evidence from Uzbekistan also suggests that the impact on poverty reduction o f reduced state control o f production and increase inprices to world market prices could be substantial. Against this background, the dehqan farms have played a crucial role in alleviating poverty in rural areas and have beenthe most dynamic farm type. The potential o f this small farm sector as a source o f employment and incomes can be exploited by improving the operating environment for these farms for example through addressing constraints to inputs and markets. In this context, the dichotomy between the private farms with much larger land area, and the smaller dehqan farms is difficult to understand while the resulting inequity of land access could have highcosts. The following policies are recommended for broad based improvements in rural incomes. 67 Increase the equitv of land reforms: Judging by the objectives o f land reform stated by the Government, it sees the shirkats as an intermediate state and recognizes that private and dehqan farms are the future foundation for a more efficient agriculture. The last three years have seen a doubling o f land area under the private farms sector and only gradual increases in dehqan land. The transformation of shirkats into private farms could be made more transparent and equitable in a variety o f ways, thus improving the ability o f private and dehqan farms to provide an effective safety net for those who lose out from the restructuring of agricultural enterprises. In the future transformation of shirkats, the Government should consider reducing the legal distinctions in land size between the private and dehqan farms, and award greater consideration to potentially more productive altematives in the dehqan sector, permitting expansion o f household plots for shirkat members. Advisory support, assistance in preparation o f business plans, etc. could result in more equal opportunities for all farmers to benefit from the privatization. The Governmenthas begun to do this under the Rural Enterprise Support Project. Moving forward, it would be important to create a legal and institutional framework for eventual marketing o f land use rights and consolidation as land i s transferred to those who derive the highest marginal benefit from it. Give the Rural Population Greater Freedom in Selecting Crop Structure and Reducing State Purchases: This amounts to the gradual reduction of the imposition of production quotas for cotton and grain and reduction o f state procurement quotas. Reduce state role in purchase o f agricultural products by increased privatization in product markets, for example cotton ginneries. If cotton production targets were abolished, farms would be free grow the amount o f cotton they want, which would make it necessary for the ginneries (eventually privatized) to contract with them on more profitable terms. Reducing Implicit Taxation of Cotton and Grain: Bringingprocurement prices in line with world market prices for cotton and grain, and removingremaining distortions through foreign exchange policies, i s essential to increasing farms profits and improving incentives in agricultural enterprises. Price liberalization in the cotton and grain sectors could have a significant impact on rural incomes. Simulations suggest significant gains in incomes for cotton and grain producers, with a greater relative benefit to poorer farmers. Within shirkats, incentives for efficient production could be improved through greater financial autonomy and responsibility and the exclusion of unprofitable pudrats from sharing inoverall shirkhat's profits. Reduce input subsidies and encourage Private sector urovision: The reducedrole of the state in marketing and pricing o f cotton and grain should be accompanied by reduction o f state subsidies for inputs, and efforts to increase the role of the private sector (example privatization o f input suppliers). The replacement of provision o f working capital through centrally directed tranches, with an arrangement, for example that permit farms to use some o f their output of grain and cotton as collateral for working capital and purchase their own inputs could be one option. Elimination of debt write offs and implicit energy subsidies would additionally impose hard budget constraints on enterprises, forcing effective restructuring, while leveling the playing field for small farm and non-farm enterprises, and improvingtheir access to inputs. The government should gradually introduce effective charges for water use. To this end, it would be important to create the legal basis for organization of farmers into self-governing water user associations, install (for example on a pilot basis) tamper proof water flow measurement infrastructure, and prepare the legal basis for the introduction of volumetric water charges. Gradual introduction o f an explicit land tax could also accompany the reduced implicit taxation of the sector. Since the preparation and discussion o f this report, some changes are currently planned or underway in this regard. The Cabinet of Ministers is considering the introduction of water charges in selected regions with an initial charge at 10% o f cost recovery level, and a view to gradual increase. Every region has now established a model water users association which provides seminars to all other interested farms on a regular basis. 68 Improve the business environment in rural areas, for smaller farms, but also non-agricultural SMEs: The experience of other countries, including China and Vietnam, have demonstrated the importance of the generation of employment opportunities outside the agricultural sector for reducing poverty and vulnerability inthe mediumto long term. The measures discussed above to improvingagricultural incomes could have a broader impact on reducing rural poverty through the demand spillovers to local markets in which the non- farm rural poor have a large stake. Inaddition, however, it will be essential to expand and improve non-farm employment opportunities for the rural poor, by taking actions to improve the business environment for private SMEs in rural areas. Measures to addressthese have already been discussed inthe previous chapter. 69 Chapter5 IMPROVING CAPABILITIES PROTECTING HEALTHSTATUSAND PREVENTING IMPOVERISHMENT HealthStatus and Epidemiology As with other countries in the region, the collapse in incomes in the early years of the transition was accompanied by a worsening o f health outcomes. The recovery since 1995 has seen improvements in health outcomes; for example, the Government reported infant mortality had dropped over 20 percent in five years to 18.3 per 1000 live births in 2001. Official data also show improvements in life-expectancy, with life expectancy 73.1 years for women and 68.2 years for men in 2001. Nevertheless, the overall levels and general trends in many health indicators continue to be o f concern. Despite recovery to pre-transition levels by 2000, life expectancy remains low compared to EUlevels. Despite recent improvements, infantmortality, under five mortality and maternal mortality continue to be o f concern. Alternate estimates of their incidence suggest that these are higher than suggestedby official statistics. For example, the infant mortality rate based on the Demographic and Health Survey (1996) is 76 percent higher than the rate of 24.7 per 1000 live births reported inofficial statistics for the same year' l2 Table 1: SelectedHealthIndicators TB Incidenceper 100,000 States; Official data from MoH Note: *age-standardized death rate. Maternal mortality rates are notoriously inaccurate across the world. '12Although not available for citation at the time this report was going to press, preliminary indications from the 2002 DHS are that there have been significant improvements in the nutritional status of children under 3 however, the substantial discrepancy between official statistics and survey data remains. 70 Uzbekistan suffers from a double burdeno f disease, and the incidence of morbidity and mortality is both from infectious diseases characteristic o f poor countries, and non-infectious diseases, characteristic of developed countries and most transition countries in the Europe and Central Asia region. Death rates from cardiovascular and circulatory diseases increased until the mid-nineties, possibly as a result of the psychological stress associatedwith transition, and remain high.113At the sametime, communicable diseases remain a problem in Uzbekistan. There i s improving vaccination coverage against tuberculosis, pertussis, measles, diphtheria, tetanus and poliomyelitis. There has not been a single case of poliomyelitis for several years. Incidence o f these diseases is often lower than in neighboring republics. At the same time, acute respiratory infections among children remainthe primary cause o f death and morbidity. And, Uzbekistanhas in recent years experienced several outbreaks of preventable infectious diseases, including tuberculosis, diphtheria, viral hepatitis, and recently typhoid. Some infectious diseases are clearly increasing. The tuberculosis incidence rate has increased by nearly 30 percent since 1993, reaching 56.1 per 100 000 population in 1998. The incidence of AIDS is now higher on a per capita basis than in other neighboring Central Asian republics. The Demographic and Health Survey (DHS, 1996) in Uzbekistan reported high child malnutrition, with stunting (chronic malnutrition) at 31 percent, wasting (acute malnutrition) at 5 percent, and underweight children at almost 19 percent. Iodine deficiency disease for school aged children, a serious disease easily prevented by iodizing salt, is 18 percent. Iron deficiency anemia is found in 61 percent of children under 3 years, when rapid periods o f growth are occurring. Also significant is the incidence of anemia among women with 60 percent of women ages 15-49. Poor nutritional outcomes especially among children have highfuture costs as they retard child development, and eventually educational attainment and productivity. It should be noted, however, that the 1996 DHS was carried out soon after the lowest point in Uzbekistan's transition recession. Although not available for citation at the time this report was going to press, preliminary evidence from the 2002 DHS does show significant improvementsinchildren's nutritional status. These average indicators hide significant differences in health and nutritional outcomes between rural and urban areas, different geographical regions and between low income and higher income households. Infant and child mortality rates show some differences between income groups, although these differences are not as severe as in some other countries' 14. There are significant differences in nutritional outcomes, among children and women, however from poor and non-poor households (Table 2). For example, incidence o f stunting and underweight children is substantially higher among the poorest quintile as is incidence of low body mass index among mothers. Similarly higher fertility rates among poor women are indications o f potentially greater reproductive healthproblems among the poor. Official and DHS data also confirm the significant regional disparities in epidemiological profile. Anemia rates are especially high inthe Aral Sea region, in Karakalpakastan. Similarly, the hightuberculosis rates in recent years in Karakalpakastan were reported to be over 110 cases per 100,000 population, and multi- resistance tuberculosis is increasingly a challenge. In general incidence o f malnutrition and infectious diseases tend to be higher in the poorer and more rural regions, although recent outbreaks of infectious diseases have been high even insome o f the better off oblasts such as Tashkent and Bukhara. 113 Data by income group and region are not always available, though Government statistics report no correlation with these indicators. The Government further reports no changes in overall incidence in chronic diseases over the last five years. '14 It should be notedhere, however, that the measureofwealth inthe DHSis basedon ownership of assets andtherefore i s not a very reliable measure of wealth, especially in a transition country. Neverthelessthe findings suggest differences by socioeconomic status and suggest the importance of monitoring nutritional and health status indicators among different segments in the population. Efforts are being made to improve the measure of socioeconomic status in the forthcoming 2002 DHS. 71 High rates of some infectious diseases (e.g. tuberculosis and AIDS) and poor nutrition, particularly among children likely reflect the crumbling health and social institutions, low incomes, poor access to water and sanitation facilities. As reported in Chapter 2, data from the Family Budget Survey shows that overall levels of access to running water and sewerage are quite low in Uzbekistan: nationally only 56 percent of the population have access to runningwater intheir own dwelling or have their own well. On the other hand 12 percent of the population relies on untreated water from rivers and canals. Government sources do report that access to safe water has improved, especially inrural areas, where access has increased to 78 percent from 62 percent in 1996. Also important determinants o f childhood nutrition and disease are maternal knowledge and awareness, often strongly correlated with education o f the mother. Low incomes and inadequate dietary intake, low access to health services, lower utilization rates o f health services and access to poorer quality o f services might be additionally important determinants of higher mortality and poor nutritional outcomes among the poor in Uzbekistan. Higher fertility rates and the associated reproductive health concerns might explain the higher anemia rates and poorer health outcomes among poor women. Quintiles PopulationPoorRich Indicator Summary Definition (*) PoorestSecondMiddleFourthRichestAverage Ratio HNP Status Indicators IMR ~eathsunder age 12months I49.5 I 43.8I 41.5 133.6 I46.8 1 43.5 I 1.058 /Per thousandbirths 5MR beaths under 5 years per I65.8 I 46.6 161.4 147.3 I50.8 I 55.2 I 1.295 khousandbirths Total Fertility Rate Birthsper woman age 15-49 4.4 3.7 3.3 3.3 2.1 3.3 2.095 Age SpecificFertility RateBirthsper 1000women age 15-19 58.0 50.0 85.0 68.0 39.0 61.0 1.487 (15- 19years) Multivariate analysis on determinants of nutritional status among women - prevalence o f moderate to severe anemia, and prevalence o f malnutrition, usingdata from the DHS 1996, provides an insight into some o f the causes of poor nutritional status of women in Uzbekistan (see Langenbrunner et al, 2002 for detailed analysis). When other factors, such as region o f residence, religion, ethnicity, household size, age and education, fertility history, economic status etc. are taken into account, the income quintile o f the household is not always a significant determinant of nutritional status. This may be due in part to the imperfect measurement o f poverty quintiles are measured in the 1996 DHS. Similarly rural or urban location does not 72 have an independent effect on women's nutritional status. On the other hand, region o f residence has an independent effect on outcomes after controlling for incomes as well as other determinants. Specifically, likelihood o f anemia amon women i s highest in Karakalpakstan while women in Tashkent region are least likely to be malnourished . These strong regional differences are consistent with the importance of a 5 regional differences in access to and quality of health care services, as well the importance of regional differences in environmentalfactors including access to water and sanitation. The significantly lower access to safe drinking water and sewage facilities poorer regions are likely to be among the most important reasons for poorer outcomes116. Other important determinants were age, for example with women aged 20-29 being significantly more likely to be anemic than younger women. Non-Muslim women were more likely to have better nutritional outcomes. While women with higher education were less likely to be anemic, education did not have an independent impact on malnourishment. On the other hand, women from larger families were less likely to be malnourishedwhich may suggest the importance o f additional resources in these families. Multivariate analysis o f the determinants o f nutritional status among children (anemia, malnutrition and stunting) also confirm the independent importance of location on children's nutritional status. Likelihoodo f anemia i s highest in Karakalpakastan relative to other regions,'17 while likelihood of stunting and underweight is highest in Kashkadarya, Surkhandarya and Novoi-Bukhara. Even after controlling for other variables, urban residence is associatedwith lower odds of beingunderweight. Children whose mothers are anemic are twice as likely to be anemic. Interestingly children whose mothers work inagriculture are more likely to be anemic while those whose mothers works outside agriculture are less likely to be stunted compared to those in households with unskilled or unemployed mothers. Equally interesting, mothers education does not have a significant impact on stunting or anemia, while children o f mothers who have higher education are much less likely to be underweight. These outcomes also shows strong correlation with age, with children 1-3 year old significantly more likely to be underweightor stunted than infants. The results on mother's education and ruraVagricultura1 variables are consistent with the findings in Ismael and Mickelwright (2001) for stunting. On the one had, they are not surprising given high levels of literacy, on the other handthey confirm the rural bias o f poverty. Despite the above, selfreported healthproblems from the Family Budget Survey are low. Only 5.6% of the population reportedthat they were sick in last 30 days. Again there are regional and income disparities. As in other countries, incidence o f self-reported illness is higher for the non-poor and urban populations, and for women and elderly. With some exceptions (e.g. Namangan where self reported illness is high), self reported illness is also higher in regions where poverty is lower. Patterns of Health Service Utilization From a policy perspective, to designactions that improve quality and utilization of health services it is helpful to examine utilization patterns to understand constraints to accessing health care by different segments of the population, and particularly the poor. The DHS shows that in 1996, immunization coverage was high, as was the use of antenatal care. Immunization coverage was higher among poorer groups, and strong equity patterns 115 The more disaggregated analysis shows additional variation. For example, while many regions have significantly higher prevalence of anemia among women compared to Tashkent region, the analysis shows that prevalence is particularly high in Karakalpakastan, Andijan and Namangan oblasts. Similarly maternal malnutrition and family planninguse differ significantly by oblast. 116Inruralareas, the quality ofdrinking water suppliesmay be further affectedby pooragriculturalpractices 117 An aggregated regional variable was used. In a more disaggregated analysis, Khorezm also has high incidence of anemia. 73 were also observed inthe use o f antenatal care. Poorer income groups seemedto do worse only on deliveries where there is more reliance on a nurse or trained mid-wife -instead of a physician - and relatively more deliveries take place at home. The Family Budget Survey provides insights into utilization o f care in 2000-01. It found that about 70 percent of those who reported health problems in the past month sought formal care. Nearly half of these benefited from home visits. Individuals in poorer households were less likely to seek care, while individuals living in rural areas were more likely than average to seek care. Interestingly, although residents in Tashkent city were almost twice as likely to report illness in the past month, they were least likely to seek care. Residents o f Karakalpakastan were also less likely to seek care. Fewer than 60% o f those who reported illness sought care (Table 3). Examining the independent effects o f poverty status, region of residence, age, gender and family composition confirms these results: the poor are less likely to seek care and the strongest correlate of health care seeking behavior is region of residence. Once region of residence is taken into account, residence in urban or rural locations are no longer significant. Thus the multivariate analysis confirms (see Langenbrunner et al, 2002), in particular that residents in Karakalpakastan and Tashkent city are significantly less likely to seek care compared to Tashkent oblast. I Table 3: HealthStatus and Care SeekingBehavior Sick in Last 30 days(%) I I fSick, Consult (YO) IIf Consulted, Home Visit (YO) Source: Family Budget Survey, 2000-01 74 Other studies provide further useful insights. Cashin (2001) found, in a sample in Ferghana region, that approximately 45% of individuals self-treated their illness with drugs before subsequent decisions about care- seeking were made, with no clear pattern across income level. About half of these subsequently went on to seek care, including visits to traditional providers. Inthe Ferghana sample, while low income individuals were less likely to seek care than higher income individuals, when they did, they sought care more quickly after the onset of illness. This supports the hypothesis that low-income individuals have a higher threshold for considering themselves ill,and when they do recognize an illness, it is at a more severe stage and requires more rapid attention. The Family Budget Survey asked respondents for their reasons for not seekingcare if ill.The most important reasons for not seeking care were "self cure" or "got better without help". More than a quarter of respondents were already following instructions from previous consultations. A significant fraction (13%) did report costs as a problem, while almost 10% reported that consultations had not been helpful. The latter two reasons are o f concern in as much as they reflect high costs and poor quality. These barriers to utilization and access to quality care are examined later. There is also some reason to be concerned about the high proportion o f self treatment, particularly through drugs. Table 4: PersonRlace of First Consultationfor Illness Richest 15.2 3.9 3.5 18.8 39.4 11.1 2.2 6.0 Location Urban 8.0 3.1 1.7 19.5 47.5 11.4 2.2 6.7 Rural 38.2 8.3 10.3 19.7 6.6 9.5 2.9 4.5 Where do individuals go for care? Interestingly a significant fraction of those seeking care (nearly half), went directly to the central rayon hospital/polyclinic complex or to the one intown, although inprinciple this level o f care is reserved for more complex cases and requires a referral. Approximately 45% of those seeking care, did, however, go to health posts or doctors in rural clinics or polyclinics or received a home visit. There is some variation in health care seeking patterns between income quintiles (Table 4), with a bias in health seeking behavior towards higher end care by individuals residing in urban areas and households in the highest income quintile, and towards primary care in rural areas. There are also interesting regional differences, although sample sizes are small"'. Data on hospital use and stay over the past year (asked for all individuals) confirms more frequent hospitalization and longer length o f stay in hospitals by better off households. Cashin (2001) also found that individuals from the lowest income quartile rely most heavily on "* For example, higher use of primary care outpatient facilities and home visits in Kashkadaryaand Surkhandarya, and higher use of higher end care in Djizzak and Novoi. While Kashkadarya and Surkhandarya have higher than average rural population, Djizzak and Novoi are not more urban than average (in fact Djizzak is more rural), suggesting independentregionalvariation in care seeking, and inhealth provision . 75 PHC facilities for first contact in Ferghana where reforms to strengthen primary health care are more advanced. Health Financingand Delivery:The inheritedsystem, policy transitionand the introductionof comprehensive reforms Partly reflected its Soviet legacy, the health system in Uzbekistan remains a large public system financed largely out o f general revenues, with a top down resource allocation process. The Government, at all levels, builds, runs, and maintains facilities and equipment, although there has been some recent privatization (see below). At the facility level, budgets were, until recently, allocated on the basis o f specific inputs such as beds and staff, and provided across eighteen line items which provide the basis for spendingthe budget. More recently, as part of financial reforms, the number of line items has been reduced to 4 for all budget organizations. Despite the relativeprotection of the social sectors in Uzbekistancompared to other countries, public funding of health care in Uzbekistanfell +om about 6 percent of GDP in the eighties to about 3 percent of GDP in the late nineties. This fall in public expenditures has been accompanied by an increase in the private financing of health care partly as a result of transition from a "free" public to a mixed publidprivate system with the introduction o f "paid" services,' l9 but in part, likely reflecting the increased significance o f informal payments. Based on the Family Budget Surveyfor 2000-01, estimates of private expenditures at the national level putthe total health care expenditure between 64% of GDP, which is closer to EUaverages. Since independence, the health system has been under transition from the old Soviet system characterized by too many facilities, staff and a bias towards specialized care. Substantive developments were initiated in 1996 with a Presidential decree on improvement in rural infrastructure, aimed at rural and poor populations. This established the basis for a progression away from the central hospital-based system towards primary health care (PHC). In 1998 a comprehensive health program was introduced, that went beyond rural health infrastructure. A central theme o f the reforms is to flatten the structure o f the state health system by shifting towards primary care over secondary and tertiary, and increasing the provision o f private care. As part o f rationalization o f facilities, the reforms plan to condense a referral chain of 6 layers to 3 levels: rural medical centers (SVP's), central district hospitals, and regional hospitals. Major aspects o f the reform include: development of PHC and shift towards greater reliance upon General Practice (GP) physicians; strengthening of emergency care; gradual introduction of fee-for-service for inpatient and tertiary care; privatization and rationalization o f facilities; encouragement of private providers; specification of exempt and benefited groups; introduction o f mandatory and voluntary health insurance; and training of health providers, in particular nurses. The 1998 Presidential decree also outlines measures for new quality standards and an improved drug supply. The decree envisages government's ultimate role as focusing on emergency and primary health care, public health (prevention and care for socially significant illnesses), and medical research and education. Several aspects of this reform, including greater financial and organizationalautonomy of primary healthcare institutions at the oblast, rayon and facility level, upgrading o f the medical profession, equipping of primary health care facilities, and closure o f redundant facilities are being supported by a World Bank project. Privatization and theIntroduction of Feesfor Services The drops in public revenues encouraged privatization and the introduction o f formal out o f pocket payments for care in public institutions. Private care is to be developed through three methods: sale o f state facilities on a competitive basis, establishment of conditions for private health facilities to be built with private funds, and granting collectives of medical workers long term leases o f government facilities with the option to The Government has been diversifying its revenue base, and has lookedto insurance, enterprises, state farms, donors, and citizens for increasing funds for the sector. 76 purchase these. The Government also issues licenses to medical doctors for private medical practice and charges permittedfor health services provided by private doctors. Although the Government does not regulate their fees, these providers are obligedto provide up to 20 percent of services free of charge to exempt groups for which the Government will reimbursethem Inaddition, the useoffees for services inpublicfacilities identified for privatization is seen as an interim step towards privatization by 2005. The majority o f facilities targeted for privatization have been SUBS(small rural hospitals with 30-60 beds) and dedicated health facilities o f health enterprises and organizations. In addition, health providers may also apply for privatization o f other facilities. These institutions are considered "mixed financed" institutions as they also provide free or subsidized services. "Private" providers in these higher level public facilities enter into quarterly contracts with the health facilities based on an expected volume and mix o f services. Providers pay the facility based on the contract and patients pay the physicians fees for services provided under the contract. Providers can generate unlimited fees for services provided under the contract. The Government does, however, regulate the fees that are charged in mixed-financing institutions and the maximum fee may not be higher than 125 percent o f the rates set by the Ministry o f Health. The rates are estimated on the actual cost o f a health service, adjusting for facility and treatment and some equipment depreciation. Facilities set their own fee scheduleswithinthese guidelines, but are required by law to publish their fee schedule. The health facility may use up to 15 percent of its fee-for service earnings, accumulated on the off-budget accounts, for salaries and incentive payments to staff, the remaining is for accounts payable and for infrastructureneeds. Paid services inmixedfinance facilities are required by law to be delivered in separate departments from free services or those services that are provided for "exempt" groups (see below and Table 5). Accounting and financial reporting for paid and subsidized services are requiredto be separate, and contracts between health providers and departments are required. Nevertheless, a possible incentive o f this scheme for the provider i s to use provider discretion to refer patients he/she sees as a pubic provider for services and tests later inthe day as a private patient. To protect vulnerable groups, to ensure free care for diseases with public health concerns and to ensure access to primary care, the 1998 decree specijies individuals, services andfacilities with exemptionsfor paymentfor medical services, pharmaceuticals, and equipment. According to the Ministry o f Health (January 2003), more than half of the population enjoy one or more exemptions from payment for one or more categories of services. All pharmaceuticals and medical diagnostics that are not specified inthe exemptions (Table 5) are considered the responsibility o f households. The exemptions generally support the priorities of the Government's reform program as specified- primary health care, emergency care, immunization, and care for socially significant diseases, for example are free. For some exemptions such as examinations and treatment for children or deliveries, charges are permitted at facilities providing paid services and of course, at private practices. It should be noted that the exempt groups as specified do not correlate very strongly withpoverty status and exemptions do not appear to be available on the basis of income or poverty status. 77 I Table 5: Rules for Free-of-Charge Medical Services 120 Categorv Descrintion Excentions/Conditions I Exempt Children (examination andtreatment) under Except for hospitalsproviding paid services Groups 17 and "teenagers" age 17-24 Benefited population groups (people with None disabilities of lstand 2ndcategories- who receive disability pensions, orphans, war veterans) Military recruits between 18 and27 years old None (`Draftees') Citizen's injuredwhile saving other's lives or None providingmedical assistancein emergencies121 Free Emergency care None Services PHC inPHC facilities and some Govemment None facilities 122 Immunizationfor infectious diseases None Deliveries Except for hospitalsproviding paid services Specialized care for socially significant None diseases(TB, psychiatric, oncological, etc) Free PHC facilities (3,875 facilities), including Government health care facilities (except in Facilities central rayon (city) hospitals, obstetric those for pay andor with paid beds facilities, child hospitals, emergency care, and others Emergencyhospitals (218 facilities) None Facilities for socially significant diseases (TB None dispensaries& hospitals, AIDS centers, oncological distlensariesetc) (367 facilities) Reforms in privatization o f provision and financing, are forward looking, and can in principle improve efficiency and quality. The introduction o f fee-based services could in principle improve accountability o f providers to patients, discourage overuse o f the health system and reduce the burden of public health expenditure. However, it is essential that these reforms are accompanied by adequate regulation and protectionfor thepoor . They need to be accompanied, however with clarity, specificity and transparency o f legislation, as well as with sufficient enforcement. They also need to be accompanied by suitable incentive structures as well as accountability mechanisms. The regulatory framework inUzbekistan for private medical practitioners are general rather than health sector specific and will need to be strengthened as the private sector expands. The analysis in the next section shows that gaps in the specification and dissemination o f policies, in monitoring and enforcement, and in incentive structures and accountability mechanisms in Uzbekistan contributes to informal practices inthe health sector which disproportionately impact the poor. 120As defined in the major acts of legislation: the 1998 Presidential Edict 2107, and the 1996 law "On Protection of Citizens' Health", as amended in 1999. Other resolutions also specify certain exemptions. For example a Cabinet of Ministers Order lists which groups do not need to pay for food at hospitals (single pensioners, etc). Resolution 532 by the Cabinet of Ministers, December 1997, "On Improving the System of Financing Healthcareand Disease Prevention Establishments" determines the categories of individuals to be exempted from payment for healthcare services, medications, and pharmaceuticals. 121 This provision is specified only in the 1996 Law on Health Protection, as amended in 1999 and not the 1998 Presidentialdecree. Specified in the Decree of President, #UP-2107, Attachment 2 78 HealthReforms to date: the ImplementationExperience Although Uzbekistan has initiated comprehensive health reforms as described above, actual progress in implementation in the reforms has not been uniform. Emergency care has been a clear priority in implementation, while other areas such as privatization, rationalization of facilities, and the introduction of health insurance have moved slowly. The detailed legislative basis to support these broad health sector reforms remains underdeveloped. Although mass media is used to cover health care reform issues, the LSA qualitative study suggests that the public have limited information and understanding at the point o f use of services on healthpolicy and reforms, and enforcement and accountability mechanisms are weak. Emergency care, Restructuring and Strengthening of Primary Health Care: the UnevenPace of Reforms The high level o f funding allocated to buildingand renovation o f emergency care raises both efficiency and equity concerns. It i s not clear that poor and more rural populations will benefit from extensive emergency facilities over strengthened PHCs, or over central district or regional hospitals. The emphasis placed on emergency care may also distract from other key reforms. There i s also concern over the overlap between emergency care hospitals and other services and facilities that provide emergency care such as first aid stations. Although primary health care strengthening and emergency care are envisaged as simultaneous reforms, anecdotal evidence suggests that the recent efforts to boost the emergency services and hospital sector have negatively affected thepublic budget resources going to thepoor and to the rural areas andfor primary and outpatient care services123 . Implementationo f some of the eficiency reforms have been impressive and are among the most significant in all CIS countries. Between 1999 and 2000, the number o f hospital beds fell from 122.1 to 53.3 per 10,000 population, inpatient admission per 100 population fell from 24.4 to 13.2 and the average length of stay for inpatient admissions fell from 14.8 to 12.1. Nevertheless, staffing levels remain bloated and inputs are still skewed towards hospital based and outpatient specialization. There is also some question over whether the closures and privatizations o f facilities are taking place as rapidly as the construction of new or rehabilitated SVPs. At the same time, there is some concern over whether the selection of facilities to be closed will be adequately aligned with the relative needs of local communities. Local politics and the wishes o f local Khokimiat, or insufficient local capacity, can sometimes be factors. Geographical differences in the SVP/primary health care restructuring program reflect some o f the concerns with the unevennessof the reforms. As of 2000, the SVPs and to a lesser extent outpatientfacilities (which will be expanded) were more prevalent on aper capita basis in regions with higher wealth while FAPs and SVAs cfacilities in worsephysical conditions with less stafi remain moreprevalent inpoorer areas. It should be noted, however, that following this earlier pattern of inequity, there has been an increase in SVPs inrecent years. Implementation of privatization and introduction of fees: the emergence of informal practices, the burden of privatepaymentsfor health care and impacts on access and utilization Although growing, the number of officially registered private medical practitioners remains extremely small as compared with the state sector. Estimates o f share of health care delivered by formal private providers range from 3 to 10 percent. On the other hand estimates of private financing o f the sector are much higher when formal and informal privatepayments are considered. 123See Langenbrunner et al. In Volume 2 of this report. The present break down of the MOH budget does not provide adequate detail, and data on expenditures on emergency care facilities were not provided. However, the program for rural primary services has been very delayed in some regions. Allocations for health since 1995 show a very big jump in funds going to "investments and major repairs". The percentageof budget going to these have increased5 fold and has gone from less than 1 percent in 1997 to 5.2 percent in2001. Additionally localgovernment hnds and budgets 79 Despite the development of formal systems for private practice, the qualitative research'24 carried out for this study suggests that informal private practice by doctors employed at public facilities is signiJicant. Health providers reported earning between 40 and 60 percent of total income from private practice. Most private practice was informal -either in that providers are unlicensed or did not have formal contracts with facilities, or in that they substituted for official services by taking place in working hours and with use o f public facilities without reimbursement. Since obtaining o f a PMP license is relatively straightforward, with no accreditation process, informal provision of health services i s likely to have been the result of other institutions - including incentive structures and a lack of accountability mechanisms such as penalties for unlicensed practice. High fixed tax obligations (regardless o f number o f private patients seen) and the formal costs o f the facilities provide incentives for doctors to evade the formal system while the opportunity to negotiate lower than official charges provide incentives to users to participate in the informal health system. The qualitative research also suggests that informal private practice is well accepted, unmonitored and completely un-penalized. The incidence of private paymentsfor healthcare is significant. On average 81percent of households with at least one consultation reported expenditures on health care (Table 6)'25. These percentages are higher than the percentages reported by Novak et al. (1996) and Sari et al, (1999) for Kazakhstan. The incidence of out of pocket payments is higher on average inthe better of regions such as Bukhara and Tashkent city and lower in poorer regions such as Kashkadarya and Khorezm. Not surprisingly, poorer households are less likely to make out o f pocket payments for health care (Table 6). These results are also consistent with the incidence of out o f pocket payments in Ferghana reported in Cashin (2001) who found that 86% paid out-of-pocket for care inFerghana including over 70% of the lowest income group. Out o f pocket payments in practice include those that are permittedby law, but informal payments are also significant'26. The qualitative survey found that more than two thirds of health users interviewed reported making informal payments of cash and in-kind goods and services. Cash payments seemed to be more common than in-kind payments in urban areas, such as Tashkent city. Both types were either requested or voluntarily given to doctors either before or after the treatments. Payments were almost always made directly to providers. Health users seem to generally know what is meant by informal payments, although there were some areas of ambiguity, such as payments for exempt services, and the distinction between informal practice (unlicensed and without a contract, inregular working hours) inpublic facilities. 124Qualitativeresearch(Nishino, 2002) was usedto supplement and explainresults from the householdsurvey data with contextualizedin-depthinformation, as well as to suggest avenues for further analysis. The research is not intended to be representative. The research is based on interviews with 14 focus groups as well as in-depth interviews with 60 health users, 49 providers and 24 Mahallah committees in four raiyons. There was a particular focus on differences betweenthe poor andnon-poor. 125This might be an overestimateof payments associatedwith a particularconsultation. On the other handthe design of the Family Budget Survey is likely to be associatedwith an underestimation of the incidenceand extent of expenditures on healthcare. See Langenbrunneret a1inVolume 2. 126 Informal payments are not technically illegal in Uzbekistan. The 1996 Law on Health Protection (as amended in 1999) permits health establishment to use several sources of funds including budget, insurance funds, earmarked funds, as well as receive voluntary donations from organizations and individuals, and other sources "not prohibited by the legislation" (article9, and legislationprohibitingpaymentsremainsunspecified). 80 payments among householdswith at lea t one consultation Households reporting heath expenditures 83.9 Rural 78.1 Karakalpakastan 82.8 Andijan 81.2 Bukhara 90.5 Djizzak 86.5 Kashkadarya 46.5 Navoi 74.2 Namangan 75.4 Samarkand 76.3 Surkhandarya 71.7 Syrdarya 74.5 Tashkent Region 85.8 Ferghana 84.0 Khorem 63.9 Tashkent City 87.4 Quintile" I Poorest 72.7 2"d 79.5 31d 82.8 4'h 80.3 Total 81.0 Box 1: Examplesof Cash iformal Payments Tashkent Djizak Ferghana Namangan -225,000 sums to doctor for -20,000 sums for tests, -15,000 sums to doctor -2,000 sums for tests, treatment (bricks); 10,000 analyses, and X-rays, as athank you, 7,200 analysis, and X-rays; soum to doctor for blood 250,000 sums to professor, sums (dress length) to 1,000 sums for cot; 2,000 transfusion; 15,000 soum to 25,000 sums for bandages nurse. sums to Tabib (traditional nursefor injections. and 10,000 sums to nurses. healer). - 30,000 sums to doctor - 10,000 sums to doctor, -50,000 sums to doctor, for the operation, 2,000 -30,000 sums for the 20,000 sums for food and cot. 12.000 sums for food. sums for X-rays, tests, operation, 25,000 sums to and analyses. donors, 2,000 sums to -10,000 sumsto doctor, - 60,000 sumsto doctor, nurse, 3,000 sums (shirt) 10,000 sums for the cot, 3,000 sums for plov to - 2,000 sumsto doctor to doctor inward. 5,000 sums for the X-rays in doctors, 10,000 sums for inAndijan, 7,000 sums - 10,000 sums for Medical Diagnostic Services, analyses, tests andX-rays. to doctor inMargilan as treatment, 30,000 sums 5,000 sums for X-rays in athank you. for medicines. emergency aid hospital. -48,000 sums to doctor, 1,000 sums for analyses, -20,000 sums to doctor, -20,000 sumsfor prenatal tests and X-rays. 10,000 sums for food. hospitalization; 30,000 sums for the operation, 15,000 -25,000 sums to the doctor. sums for medical services. iource: Nishino, 2002. Qualitative Informal payments tended to be less prevalent and lower in rural areas, where people know each other well and doctor's salaries are closer to desired salaries. Underscoring the importance o f transparency, they were reported to be less prevalent and lower infacilities and for services that are widely recognizedto be free, such as PHC facilities, emergency care and immunizations. They are also less prevalent for people with personal connections to health providers and vary according to individual negotiations. By contrast, informal payments were more likely and higher for more complex services and services where official fees are charged such as surgery, as well as for obtaining certification o f disability. As in the case of informal provision, contributingfactors to informal payments include incentives structures for patients and providers, lack of clarity or information on government policies and inadequate understanding of health rights, insuflcient enforcement and accountability mechanisms,and other institutionalfailures (Box 2) The evidence in Uzbekistan suggests that large amounts of resources bypass facilities and therefore cannot contribute to reinvestment in the system. As mentioned earlier, private expenditures estimated from the FBS put total health financing inUzbekistanon par with the levels inEuropean countries. Yet the relatively large share o f funds that fail to pass through public channels has no doubt impeded sectoral development in civil works, equipment, and supplies and pharmaceuticals. Because they are outside the financial and policy control o f the Government, informal payments and practices reduce accountability and transparency of health institutions, obscure true needs, and impede resource allocation. As the analysis in the following paragraphs shows, together with formal payments, informalpayments also contribute to poverty and impoverishment and impede access and utilization of health services by the poor. The President has taken a strong leadership position inthis area. He has acknowledgedthat informal payments are usedto secure better quality treatment, and that this unofficialsystem needsto be f0rma1ized.l~~ The Government also has demonstrated commitment. The MOH is working on improved regulation setting prices and clarifying conditions for providing free services to exempt groups and on the introduction of mandatory and voluntary health insurance to ease the burdenof out of pocket payments on households. Out-of pocket payments constitute a significant expenditure for households, constituting approximately 18% of food consumption on average (Table 7). Expenditures reported in the FBS are much lower than those reported in Cashin (2001) for Ferghana in large part due to differences in questionnaire design. Therefore the estimate o f private health expenditures and the burden o f these expenditures may be higher than these numbers suggest. The burden of expendituresfalls disproportionately on the poor: the poorest households spend 22% o f food consumption on health care. Out of pocket payments are also especially high(in levels as well as expressed as percent of household food consumption) in the poorest regions Kashkadarya, Surkhandarya, Khorezm as well as inTashkent city and Djizzak (Langenbrunner et al, 2002). Cashin (2001) .confirms the importance of financial barriers to obtaining health care in Ferghana, and finds that financial barriers are disproportionately highfor poorest households. 127See, for example "Uzbekistan to Move to Paid Health Care Soon - President" BBC Monitoring Service 28 March 2002, in which President Karimov says "(Doctors) are beingpaidby both the state and by the patient. Ifthe patient does not pay for the operation this will be done differently. Am Iright? This is taking place anyway. For this reason, what should we do in order to put an end to this? Patients should agree to pay for specialized treatment and for a complicated operation." 82 Box 2: Contributing factors to informal payments and practices in the Health Sector Incentive for patients and providers: The primary reason why patients said that they make informal payments was to secure better quality health services - payments were considered necessary to obtain quality treatment, more attention and a polite attitude. Several respondents felt that providers were less responsive and they had received poorer quality treatment when they did not make informal payments. Some health care users also observed that informal payments are a cultural norm. The opportunity to negotiate lower than official charges provided incentives for them to participate in the informal health system including the services provided by unlicensed private providers or those practicing without contracts. As in studies o f other transition countries, healthproviders justified informal payments on the basis o f low official salaries o f health providers. The average salary o f health care workers in state facilities is low, averagingjust over 16,000 soum per month as o fthe endo f 2001. There are also few performance incentives -salaries are usually paid at a fixed rate. Working more than full-time i s one option to earn more, and some doctors in the qualitative study occupied more than one position. High fixed tax obligations (regardless o f number o f private patients seen) and the formal costs o f the facilities provide incentives for doctors to evade the formal system. Lack of clarity and transparency on Governmentpolicies is an important contributor. For example, both doctors and patients are not always clear on who should receive free services from which facilities. InTashkent, for example, many providers did not mention children as constituting an exempt group. The list o f exempt drugs is not available while the definition o f exempt groups is not usually clear to health users. A similar lack o f clarity was apparent on what constitute paid services and on setting o f tariffs for paid services On the other hand, most health users know that health services are free at the primary levels and in emergency units. This explains the lower incidence o f informal payments reported at these facilities and underscores the importance o f transparency. Lack o f clarity and transparency on policies stems both from legislative gaps in their specification, as well as insufficient dissemination. It reduces the ability o f users to claim their entitlements. Inadequate enforcement and accountability mechanisms: Informal payments and informal private practice were widely accepted by health care workers as a tradition. Providers have discretion on what to charge, there are no formal oversight systems for monitoring provider charges or performance, and in interviews providers perceived there are no penalties for taking informal payments or for informal private practice. Similarly, while hospitals are requiredby law to publish their fee schedule, this rule does not appear to have been fully enforced and not all facilities visited in the qualitative study had fee tables. Although health rights are generally well specified, inadequate dissemination o f this information reduces the ability o f users to demand their entitlements or to use their knowledge o f health rights to file complaints or to negotiate quality treatment, and increases their reliance on informal payments and networks. The lack o f adequate understanding o f procedures and institutions to address complaints and insufficient confidence in the complaints system further precludes users from holding providers responsible. About half o f health users interviewed were not familiar with any complaint systemand most did not feel comfortable using it or did not believe it was effective. As inother countries inthe region, the role of civil society organizations in enhancing accountability through representations in negotiations as well as in broader government policy formulation is limited. Institutional inefficiencies such as lack o f availability o f drugs at health facilities, combined with late reimbursement from governments to pharmacies or slow government monopoly distribution systems can additionally encourage violation o f rules on drug provision and lead to informal payments for drugs. Cases were reported where pharmacies demanded additional payments or were reluctant to provide exempt drugs to patients. Inadequate competition andmarket inefficiencies, particularly in rural areas, contribute to substantial price differences and high costs, and lead to informal purchases o f drugs from non-licensed sellers. Institutional inefficiencies, for example slow re-imbursement mechanisms o fpublic subsidies to private providers could additionally contribute to informal practices. Source: Nishino (2002), Qualitative study carried out for LSA. The qualitative study also confirmed that the burden of payments was a major factor affecting utilization of health services, particularly for the poor, noted by both providers and users and that it Zed to delay or avoidance in seeking treatment. It also found that networks and social connections played an importantrole 83 in accessing treatment (to secure better services and in some cases to jump the formal levels in the system), probably accounting for some of the variation in utilization across quintiles. The higher incidence of out-of- pocket payments at higher end facilities, further contributes to the use of lower level facilities by the poor. There is also evidence of misuse o f facilities such as emergency care facilities widely known to provide free services. Poorest Second Third Fourth Fifth Total Expenditureinsoums 3899 4524 4123 5400 4898 4680 IExpenditureas a share ofhouseholdfood 22.1 19.1 15.8 20.5 14.6 17.6 consumption Source: Family Budget Survey, 2001. Calculatedfor those householdsreportingexpenditures. The most frequent payments made were for medicines which were also significant in magnitude: average expenditures were 13% of household food expenditures while median expenditures were 6% of food consumption. For those who incurred them in the past month, hospital expenditureswere significant with average expenditures at 34% of food consumption and median expenditures at about 6% o f food consumption. The qualitative research confirms these findings: many consumers complained about high out of pocket payments, especially with pharmaceuticals, the prices for which have reportedly escalated significantly over the last several years. The costs of medicines accountedfor the majority of total out-of- pocket payments - usually more than 50% and in some cases more than 90%. While expenditures on pharmaceuticals are formally permittedexcept as specified in exemptions, informal payments for drugs occur because o f limited information on policies and exemptions, shortage of drugs, or slow reimbursement to pharmacies for exempt drugs (see box). Respondents in the qualitative study, very rarely received free medicine at facilities. Although the FBS did not ask about travel costs, Cashin (2001) and the findings of the qualitative study suggest that distance to health facilities and associated difficulties with transportation could be significant barriers to utilization o fhealthcare services, particularly inrural areas. While the burden o f out of pocket payments is disproportionately high for the poor, evidence from other countries has shown that many non-poor households are at risk o ffalling intopoverty as a result o f substantial health expenditures occurring due to catastrophic or chronic illness. InVietnam, the impoverishing impact o f even routine medical expenditures was significant (Wagstaff, 2001). Several households in FBS reported health expenditures greater than 100% o f food consumption. The FBS also shows that the impact o f out-of- pocket expenses are especially severe in households with more vulnerable members such as children and individuals with chronic conditions. Households in the lowest income quintiles with children under 15 spent 28% o f monthly food consumption on health care. For households with chronically ill or disabled members thepotential impoverishment impact on non-poor groups9 o m health care expenses is signzificant. Average health care expenditures as a proportion of food expenditures were two and a half times that for the average household, and constituted between half to 60% o f food consumption for the second and third income quintiles (Table 8). These income groups and the elderly were also more likely to reports costs as a reason or not seeking care suggesting that a coping strategy from potential impoverishment is to delay or avoid care. Poorest Second Third Fourth Richest National Householdwith children under 15 17.9 21.1 15.9 11.4 15.9 17.5 Households with chronically Wdisabled 37.3 57.2 51.3 42.9 23.2 41 members 84 Coping with high health care expenses: Qualitative evidence suggests that across locations and different economic groups, the burden o f out-of pocket payments does lead to impoverishment in the case o f health crises, such as an accident or chronic illness. There are no formally established social protection mechanisms for those who faced unexpected health care expenses from serious illness or accidents. Beneficiary coping mechanisms to manage such large expenses were mostly informal (family, relatives, and friends) and were insufficient to stop people from falling into poverty. The typical ways o f coping with such crisis were, for example, depleting savings, borrowing money, selling assets, such as sheep, cows and jewelry, and reducing certain expenses, including food. No respondents reported having any health insurance. Nor were there reports of private money lendingestablishments, from which health users could borrow money. There were a few cases o f pharmacies or medical doctors informally accepting patients' delayed payments. As mentioned above, one coping strategy is delaying health care. Table 9 from Cashin (2001) provides quantitative evidence on the significance o f borrowing and delayed care due to financial barriers and shows that this was a significant barrier for the poorest but also other segments of the population inFerghana. Financial Barrier Source: Cashin (2001). Sample sizes in parentheses. Multivariate analysis on the probability o f making a payment confirm that rural residents and the poor were significantly less likely to make out o f pocket payments. On the other hand households with members reporting poor health or chronic health problems were significantly more likely to make out o f pocket payments independent o f their poverty status and all other factors. Selfreported health status (bad or chronic health) and region of residence were the strongest correlates of incidence of payment. Thus controlling for all other factors including rural residence, residents of Kashkadarya, Khorezm, Surkhandarya and Syrdarya among others were less likely to make payments. Among those who did make the payments, the poor made lower payments, and payments were significantly higher among households with members who had poor health or chronic health problems. Once again, self reported health status and region of residence were the strongest correlates of the level of payments. Conditional on being made, the level o f out o f pocket expenditures, are significantly higher in poorer oblasts such as Kashkadarya and Khorezm suggesting a very high burden for the poorest households. Levels of payment were not significantly correlated with urban or rural residence once other factors including region of residence were taken into account. The higher incidence and levels o f expenditures for "vulnerable'' households is o f concern , and provides evidence that the medically vulnerable face a higher risk o f impoverishment from health care expenses. Finally, the lower incidence and levels of expenditures among the poor support the hypothesis that private expenditures on health care are regressive. 85 Mixed evidence on quality of health care services Quality o f care is very difficult to measure and the FBS does not provide information on the quality of care. Nevertheless, useful insights are obtained from Cashin (2001) for Ferghana, and from the qualitative survey (Tashkent, Ferghana, Djizak, and Namangan). The overall picture is one o f mixed evidence of the impact of reforms on quality and differential access to quality of care by thepoor and non-poor. In terms of the general perceptions of the reorganizations of health care institutions, the results are inconclusive. Both health users and health providers perceive some positive changes, such as increased availability of new equipmentand medicines, as well as better facilities such as new emergency units.On the other hand, users across different social groups in all locations studied, expressed negative opinions of the increased cost of health services and medicines without improvements in quality sufficient to weigh against increased private costs. Inpart becauseo fthe perceptions of increased costs, poor peopletended to give more negative evaluation of the changesthan non-poor, and perceived that they were given lower quality o f service. Inthe Ferghanaregionwhere some ofthe reforms are mostadvanced, Cashin(2001) reports that 56.6% ofall individuals and only 47.7 % of the poorest individuals felt that their exam by the medical professionals was thorough enough, with a 23% point difference between rich and poor individuals. Similarly opinions about the communication and attentiveness of the provider show more significant differences across income group. Although individuals inthe sample who paid more out of pocket tended to report higher satisfaction, not all of the difference in satisfaction i s explained by level of payment. The qualitative report supports these findings: the primary reason given by patients for making informal payments was to secure better quality health services. Moreover wealth, social connections, and a socially recognized position emerged in the qualitative study as important factors for influencing the quality o f treatment received. Incidence of public healthcare expenditures If private expenditures on health care are regressive, do public expenditures improve the distribution of overall health expenses? Data is not available for a thorough incidence analysis o f health care expenditures by program. However the examination o f utilization patterns above suggests that the poor are less likely to seek care and tend to delay care due to low incomes, but also high costs, differential access and possibly poorer quality. Moreover similar reasons result in higher use by better off households o f higher end facilities, particularly in urban areas. Figure 1 shows the incidence of public expenditures on hospitals. Reflecting the utilization pattern discussed earlier, the figure shows that public expenditures on hospitals favor the highest income quintiles, and urban populations. The urban bias is primarily driven by the large expenditures on republicanfacilities in Tashkent. Figure1:IncidenceofPublicExpendituresonIn-patientCare 70.0 60.0 .cl 50.0 40.0 $ 30.0 a 20.0 10.0 0.0 Source: 2000 FBS 86 Since hospital expenditures constitute the majority o f public expenditureson health care, overall incidence o f public expenditures is likely to be regressive. To improve the incidence of public expenditures on health, it will be important to take measures to improve access of thepoor to all levels of care, including much needed hospital care. It will also be necessary to address the unevenness inthe geographic development o f primary health care facilities. Increasing access of the poor calls for adequate protection from high costs (formal or informal), and improvedaccess to quality care for the poor. Protectingthe poor from formal costs impliesthat it will be necessaryto refine exemptions for the poor and ensure that they are covered for hospital expenses. The regional differences in health outcomes and health care infrastructure and services also suggests that it will be necessary to remove existing rigidities in resource allocation and underscores the importance o f developing more equitable mechanisms o f resource allocation across regions. ProviderAccountability The analysis o f contributing factors to informal payments (Box 1) shows the importance o f health sector institutions in ensuring transparency and accountability in public institutions and in ensuring that reforms benefit all segments of the population, and especially the poor. It suggests that in addition to inadequate clarity and specificity of the legislation supporting the Government's reform program and inadequate dissemination o f policies, mechanisms that ensure accountability o f health sector institutions and providers are underdeveloped. Accountability in service delivery in the health sector is compromised by insufficient oversight and monitoring and inadequate enforcement of governmentpolicies. Although health rights are generally well defined and comprehensive inUzbekistan, these rights are often not realized inpractice. Insufficient awareness o f health rights, inadequately specified procedures and institutions to address complaints, and insufficient confidence inthe complaints system preclude health system users from holding providers responsible and from demanding their rights. Finally, the limited role of civil society organizations in enhancing accountability through representations in negotiations as well as in broader government policy formulation i s another barrier that prevents the reforms from havingtheir intendedpositive effects on the poor. The NGO sector is still nascent and dialogue on policy i s rare. Mahallahs (local community organizations) tend not to be substantively involved in advocacy for health care, although some cases were reported where mahallas assisted poor people in negotiating for free treatment. The qualitative study suggests that mahallahs could play a significant role in mediation and information dissemination- most people, particularly inrural areas reportedtrust in mahallahs inthis role PolicyImplicationsand Recommendations: Uzbekistan has embarked on a comprehensive system of health reforms, many o f which are forward looking. Nevertheless, the analysis in this chapter suggests that several factors may have prevented the reforms from havingtheir desired impacts on efficiency and particularly on equity. The incidence of poor nutrition and the infectious diseases o f tuberculosis and AIDS, and their regional concentration suggest the importance o f interventionsoutside as well as inside the health care delivery system inachieving broad based improvements inhealth outcomes. The slower pace rationalization of higher end facilities has meant that fewer resources have been released for expenditures that benefit the poor. The faster pace and approach to emergency care reforms, and the regional concentration o f primary health care strengtheningin the better off regions have further prevented the reforms from benefitingthe poor. The introduction o f privatization and fees for services, although perhaps a necessary reform, has had the unintended effect o f imposing a greater burden on the poor, and o f leaving a larger fraction of households at risk of falling into poverty. Inthe absenceof formal mechanisms for protection, many households are forced to employ counterproductive coping strategies including delaying care, borrowing, reducing essential expenditures or depleting assets. The regressive nature of private payments for health care has led to differential access in the quality of care. The emergence of informal practices and significance o f informal 87 payments in the out o f pocket costs, hurtsthe health system overall, reducing the availability of resources for reinvestment and reducing accountability and transparency in health institutions. Reflecting these developments, government expenditures on health care remain biased towards better of households. In order to ensure that the reforms have their desired effects on health outcomes for the population including the poor, the analysis suggeststhe consideration ofthe following policy measures. Measures outside the health system: The analysis o f determinants of health outcomes suggests that several factors outside the direct control of the health system could improve outcomes for children and adults. These include measure discussed inother sections o f the report to improve incomes, increase access to education of women, as well as improvements in access to safe water and sanitation overall, as well as inpoorer regions. Fortification of foods is also a cost effective way o f dealing with some micronutrient deficiencies (e.g. fortification of salt for iodine deficiency and wheat for iron deficiency). Strengthen or introduce mechanisms to protect the uoor as well as other vulnerable households from high costs of health care: The existing exemptions for disadvantaged groups do not appear to protect the poor, particularly from the expenses associatedwith drugs and diagnostics, or in-patient care. The problem i s partly one o f identification. In addition to improving the information base to identify the poorest, it would be important to re-examine the current systemo f exemptions, and replace it with one that i s better targeted to the poor. Chapter 7 demonstrates that local community groups (Mahallahs) can play a role in targeting social assistance benefits. More recently the mahallahs are beingused to target some education allowances to poor households. The Government may wish to consider whether there i s a role for mahallahs to play in targeting health protection to the poorest. It would, o f course be important to take into account the capacity of these organizations, as well as strengthen accountability mechanisms, astheir role expands. Other local solutions to identify and target the poor could be explored. Once the poor are identified, the benefits package could be extended to cover all necessary services for the poor, notjust service provided at selected facilities, such as the emergency care centers or primary health care facilities. Equally challenging i s identification and protection o f all income groups, includingthe non-poor groups from the expenses associated with chronic or catastrophic care and protecting them from impoverishment. In principle, the concept o f insurance that pools risks could be used. The 1998 decree provides for the introduction o f mandatory and voluntary health insurance through the creation o f a State Medical Insurance Fund. The experience inthe region is that (inpart due to low incomes and formal employment) such schemes have failed to protect the poor and the vulnerable, and indeedhave been associatedwith an emergence in gaps inthe coverage of services128. The 1998 decree also provided the legislative basis for private voluntary medical insurance funds. While some joint stock companies have begun selling employer based insurance, implementation of insurance remains essentially in the planning stage. The development o f supplemental insurance to cover high cost care could be one way o f bringing additional resources and innovation into the system. However, such schemes need to be designed carefully to avoid many of the typical problems such as favorable selection. They also need to be accompanied by enabling legislation and regulatory framework. Moreover they are likely to benefit only a few, at least initially, and create a two-tiered health system with the majority, including the poor dependent on the public system. The above discussions suggests that the Government should consider the use of public subsidiesto cover part or all o f the costs associated with catastrophic health expenses or longer term chronic costs. Some kind of 128Many householdswho do not have formal employment opportunitiestend to be excluded from these schemes, while other individuals or families often fall behind or avoid premium payments, facing large out-of-pocketcosts when they are sick. These schemes are typically financed by additional taxes on employment which further compromise employment opportunities. 88 targeting based on income level might also be important here. Again, decentralized application or administration through local government offices or Mahallahs could be considered. They would, however needto be accompanied by adequatemonitoring and accountability mechanisms. Take actions to reduce the incentivesfor informal practices: Introduce performance based salary structures for providers and ensure, for example, adequate share in revenues from official fees for services. Reviewtax arrangements and other costs for private providers. Address institutional inefficiencies such as reimbursement mechanisms to private practitioners and pharmacies. These will need to be accompanied by greater clarity on the rules for paid as well as exempt services, and their widespread publication and dissemination. They will also be need to be accompanied by stronger oversight, enforcement and monitoring mechanisms including performance audits and spot checks, as well as the introduction o f penalties for informal payments and informal private provision (unlicensed and without contracts), and encouraging the use o f receipts. Transparency and accountability could be further enhanced by facilitation o f public debates on informal payments and provision. As the formal private sector develops, the Government will need to consider ways to foster its development while at the same time protecting consumers and monitoring quality through efforts such as accreditation standards, reports on clinical indicators, review by external peers, selective audits, and disclosure o f outcome measures. It will be important to set incentives and maintain quality rather than impose unnecessary burdens on private providers. Analvze and address inefficiencies in the pharmaceutical market: Analyze reasons for high costs and difficult access to pharmaceuticals, with a special focus on poor and rural regions and take actions to reduce barriers. Improved information and knowledge on health outcomes and treatments could also reduce inappropriate utilization. Removing restrictions on imports will also help. Accelerating the restructuring program to imurove efficiencv and equity of resource use and ulacing prioritv on strengthening primary health care in the uoorer regions: Acceleration in the restructuring program, through continued rationalization of staffing and infrastructure and reduction o f tertiary facilities, particularly republican facilities, will release additional resource that are currently being spent on higher end facilities. In this context, it will also be important to review the organization o f care within the emergency facilities and to consolidate the provision of these services. Provide incentives to facility managers to use resources efficiently, and give them greater flexibility to use resources and purchase inputs accordingto local needs. The earlier bias in the primary health care restructuring program i s now being reversed with a greater emphasis on poorer regions. This trend should be continued. Introducing more flexible resource allocation and financing mechanisms: To address differences inaccess and quality of health services, replace the traditional Soviet based input formula with one that i s more responsive to differences in demand or needs between regions as well as within regions. Within regions, buildand use an improved information base, such as censusand surveys to facilitate the direction o fresources to poor and remote areas. Increase accountability of state institutions and empowering health users: Clarity of policies and their dissemination, strengthened oversight and monitoring, and the introduction of penalties for poor performance or inappropriate practices by providers, could be accompanied by additional measures that enable consumers to hold providers accountable, such as enhanced access to information, for example on health rights. It would also be important to provide a more readily available and routine mechanism for filing complaints, especially anonymous complaints, on local provider or facility misbehavior. This could include, for example, the use of telephone hotlines, which have been successfully introduced in other arenas. Also important would be the encouragement o f civil society organizations (such as consumer health associations) or extended outreach 89 efforts from the local Mahallahs in representing consumers in mediation as well as in health policy formulation, holding providers accountable, and Continuous Monitoring and Evaluation: Strengthen survey data sources to regularly monitor health outcomes by different segments o f the population, their utilization o f services and core family practices relevant for health and nutrition, improved data on payments made by households for health care, a more systematic and larger scale survey o f informal private provision. 90 Chapter 6 IMPROVING CAPABILITIES: PROTECTING INVESTMENTSINEDUCATION Education is an important determinant of living standards. Higher levels of education are associated with higher incomes in the labor market and higher consumption. Education, particularly o f women (particularly at higher levels) i s also correlated with lower fertility and some improvements in health outcomes, such as anemia among women and higher birth weight. In this context the gains in basic education coverage after relative stagnation inthe early years o f the transition are an important achievement. However, the still lower coverage rates, relative to independence, for all other levels, representing potential erosion in the human capital stock, is of concern, and could undermine Uzbekistan's successful transition to a market economy. There are significant disparities in educational attainment and quality between regions and the poor and non- poor, men and women, particularly at non-compulsory and higher levels, which results in unequal opportunities to benefit from market reforms and could exacerbate income inequality inthe future. Like other transition countries, Uzbekistan faces the challenge of reorienting its education system to provide the skills in demand by a market economy. Concerted efforts are needed to reverse the declines registered since independence and to step up the gains that have recently begun to take place in secondary and higher education, to address disparities, and to improve the quality and relevance of education. This is essential to ensure Uzbekistan's successful transition to a market economy and to prevent intergenerational transmission ofpoverty. Trends in educational attainmentand existingdisparities Uzbekistan has made impressive gains in educational attainment over the past several decades. Illiteracy has virtually been eliminated and official data report 100 percent literacy among both males and females in the 15-24 year old age group. Figure 1 from the FBS, shows the educational attainment o f successive cohorts o f men and women. It shows the remarkable gains made ineducational attainment by both menand women and a gradual closing of gender gaps: while about 30 percent of men and 19 percent of women who are presently 75 years or older had some upper secondary or higher levels o f schooling, approximately 92 percent of men and 93 percent of women who are currently 25-44 years old have had some upper secondary or higher levels o f schooling. Some gaps remain, however, between men and women, at teknikum and university levels of education, however. Similarly, as we saw in chapter two, although gaps have been closing between the poor and non-poor, the poor started out with lower levels o f education and gaps remain at the teknikum and university levels. Figure 1 also shows a worrying and unusual trend - recent cohorts, those who are currently between 25-34 year olds, have been investing less in higher levels of education and particularly university education. While successive cohorts o f menhave also been less likely to complete teknikum,this has not been true for women. The declining trends may reflect some out-migration, particularly during transition, o f younger people with higher educational attainment. However, the discussion below suggests that disinvestments in higher education have continued since independence until very recently. Since these levels of education are precisely those associatedwith higher labor market earnings and welfare, this trend and remaining disparities are of concern. What has been happening to educational attainment since independen~e'~'? Both absolute enrollments and actual coverage (as measured by enrollment ratios) have increased continuously in thefirst 9 grades (basic education). Table 1 shows that the enrollment rates stayed more or less stagnant inthe early transition years, but picked up significantly since 1995/96, as enrollments in basic education increased faster than the number 129 It should be noted that the lack of systematic monitoring of enrollmentratios, rather than absolute enrollments has made trends in coverage challengingto establish. Some trends inthe data reportedhere (providedby SDOS) contradict with previously reported trends inthe UNICEF, 1999,Common CountryAssessment. 91 o f 7-15 year old children in the population. By 1991/92, the net enrollment ratio was nearly 95 percent. Gender disaggregated gross and net enrollment ratios and their trends were not readily available from administrative data. Official data do however show that girls accounted for about 49% of total enrollment in grades 1-9 in 2001/02; indicating relative equality. Figure l a Figure 1b: , ~ EducationalAttainmentof Males by Age EducationalAttainmentof Females by Age 75+ 65-74 55-64 45-54 35-44 25-34 , Age in years 75+ 85-74 55-64 45-54 35-44 25-34 Age in years 0lllnerate m< 5 yrs 05-9 yrs sch 010-11yrs sch mCompleteTech OCompleteHigher Olllderate.< 5 ynO5-9 y n schOl0-11 y n sch.Complete Tech OCompleteHylher, Source: Family Budget Survey 2000/2001 Source: Family Budget Survey 2000/2001 Absolute enrollments in pre-school, secondary education, and higher education, however, all declined in the early years o f the transition. Whilethese have now begun to recover in recent years, absolute enrollmentsfor these levels remain below pre-transition levels. In 2000/0 1,absolute enrollments in pre-schools and higher education institutions, in particular, were only about 50 percent of 1991/92 levels. With the school age population continuously increasing, lower absolute enrollments have meant that a substantially smaller fraction o f the relevant age groups are now covered by the education system at pre-school, secondary and higher levels compared to independence: thus coverage rates (enrollment ratios) at these levels have declined substantially since independence (Table 1). As discussed below, however, there have been recent gains in highereducation (especially since the 2001/02 school year) and inpre-school education (especially when non- traditional forms are taken into account). Enrollment ratios in pre-school education dropped from 35% in 1991/92 to about 16 % in 1999/2000 and increased to 20% in 2001/02; still far below the 1991/92 ~overage'~'. This reflects declines in coverage in both urban and rural areas, with particularly steep declines inrural areas, from 27 percent in 1991to about 9.5 percent in 2000, and 12 percent in 2001/02. The declines in pre-school participation could be of concern because o f the important role that pre-school education can play in reducing early drop outs and improving school performance among groups most at risk of leaving school before completion of compulsory. These enrollments in pre-schools, however, may understate the actual number of children receiving some preparation for elementary school. According to the Ministry o f Education, recent years have seen the development of a network o f various non-traditional forms of pre-school education, such as educational centers, independent groups under kindergartens and schools, short term kindergartens, and Sunday schools to prepare children not attending continuous pre-school institution^'^^. When these are taken into account, the coverage of children by pre-school in 2002 was about 27.4%. If these non-traditional pre-schools are providing adequate preparation for elementary, the recent surge in non-traditional pre-school enrollment could be a significant achievement. 130 These figures reflect primarily enrolment inpublic kindergartens. 1 3 'The Ministry of Educationestimates that approximately 57% of the plannedfirst grade enrollment in 2001 and 84% in2002 had receivedsomepreparation for education in preschoolinstitutions. 92 For secondary education, both gross and net enrollment rates fell and in 2000101 were 48% and 35% respectively, compared to nearly 70% and 57% in 1991192132. The figures in Table 1 show a widening gap between gross enrollment ratios and net enrollment ratios in basic and especially secondary education during the past decade,133 which implies an increase in the number o f over-age students. With repetition rates reported to be low (0.05 percent in grades 1 to 1l), the increase in over-age students has likely been a result either o f higher than reported repetition, or late starts'34. The source o f this gap between gross and net enrollment ratios warrants further investigation. 1991192 1995196 2000101 Gross enrollment ratio, preschool 34.6% 24.3% 17.6% Including non-traditional pre-schools 27.4% Gross enrollment ratio, grades 1-9 88.8% 88.4% 97.4% Females Net enrollment ratio, grades 1-9 87.6% 87.4% 94.8% Males Females Gross enrollment ratio, all secondary 66.9% 5 1.6% 48.3% Net enrollment ratio, all secondary 57.3% 45.6% 34.8% Gross enrollment ratio, higher education 14.3% 7.5% 6.4% Table 1 shows that enrollment ratios in higher education have fallen sharply, from 14% o f the relevant age group in 1991192 to only about 6.4% in 2000101. The fall in higher education enrollment ratios presents a stark contrast to the pattern in other transition countries where higher education enrollments have grown very rapidly duringthe past decade (Figure 2). The decline might have been accelerated by the introduction o f fees and fall in incomes o f many during transition (see below), but, as noted above, the decline appears to have started before the transition, again an exception to the regional trend. It raises concerns, because it implies that the country's competitiveness is declining relative to other countries inthe region. The evidence from chapter 3 suggested that only higher (university) education commands a significant premium compared to lower levels, reflecting a demand in the market for these skills. Preliminary data for 2001102 show the start o f a reversal o f this downward trend in higher education coverage, a welcome development which needs to be better understood and maintained. 13'The recovery inabsoluteenrollments in secondaryeducation seems to have startedaround since about 1996197. 133The gross enrollment ratio is defined as the number of enrollments in a given cycle of education, divided by the total population in the age group that is considered normal for that cycle of education. The net enrollment ratio is similar, exceptthat the numerator includes only enrollments within the education cycle in question and within the age group that is considered normal for that age group. The gross enrollment ratio tends to overstate actual enrollments within a given cycle of education because it includes over-aged and under-aged enrollments in the numerator but not in the denominator. By removing these under-aged and over-aged enrollments from the numerator, the net enrollment ratio gives a better indication of the actualproportion of the age group which is enrolled in a given cycle of education. 134Possible explanations include the establishmentof 7 years as the age for entering grade 1, while the previous practice was 6 years. 93 Figure2 -EvolutionofHigher EducationEnrollmentsas a YOof the Age Group, Uzbekistanand SelectedComparatorCountries Source: Datafrom TRANSMONEE DataBase, Innocenti Center, UNICEF School Attendance versusEnrollment Official figures on enrollments in developing and transition countries, often overstate actual coverage of the education system for several reasons. The actual number o f children not attending school will be larger the greater is the extent o f over-registration. Inmany countries, there is an incentive to over-register enrollments as a means of maintaining existing schools and teaching positions. More fundamentally, data on official enrollments may not reflect the number of children actually benefiting from education to the extent that there is significant non-attendance by children who are officially enrolled in school. Non-attendance can take various forms. While a small number o f children - some in urban areas, but more typically in rural areas - may never start school at all, a more common pattern is that children may start school but not stay in school through the end o f the compulsory cycle. In some rural areas, children who are enrolled in school miss a crucial part of the school year at the beginning and end o f the school year because o f the need for children to help inthe autumn harvest or spring planting. What can we say about actual attendance in Uzbekistan, relative to registered enrollments? Unfortunately, there are difficulties in establishing the extent o f non-attendance. On the one hand, the preliminary picture from official data looks very positive: according to official data from attendance records, approximately 99 percent o f children who are enrolled attend school135. Government policy through the " Vseobuch" (education for all) system allows for implementation o f the mandatory education policy through careful monitoring o f all children in the catchment areas (or micro-districts) o f schools. Children who miss school for some length of time or who drop out o f school are followed up with and "all necessary actions" taken to returnthem to school. Data from household surveys could provide very useful informatiodverification o f attendance rates. Their availability in unit record form could be a very useful tool in linking non-attendance in children with the characteristics of the household, for example, to enhancethe understanding of non-attendance. Problemswith the attendance data from the 2000/2001 Family Budget Survey (FBS), however, make it difficult to establish the extent o f non-attendance. An initial analysis o f the attendance variable suggested that there may be a serious problem with non-attendance: the weighted average rate of school attendance for the 7-15 year age group, corresponding to the compulsory nine-year program of basic education, was only 77.8 percent -- or 78.5 percent for males and 77.2 percent for females. 135The definition o f this variable is unclear. Most likely it is an average over the year, from official attendance records providedby schools. 94 A closer look at the data, by month and region, however, suggestedthat there were inconsistencies inthe way the question was asked by different interviewers, or interpreted by different households. The only systematic pattern that could be detected was the interpretation o f this variable in three regions (co-incidentally the regions with the lowest value for this variable). In Surkhandarya and Karakalpakastan, the question "Are you presently attending school", appears to have been interpreted, for the most part, as actual attendance in the reference period. In Kashkadarya, there is no discernable pattern: however, the substantially lower average attendance rate there does raise questions about the validity o f this result'36. Inthe remaining 11 districts, there is also likely some variation inhow the question was interpreted but the pattern i s not obvious. Ifthe 3 districts with systematic patterns are excluded, the school enrollmendattendance rate for the 7-15 year old age group i s 89.4%. This is substantially lower than the official measure o f net enrollment rate: as mentioned above, this may in fact represent a mixture of enrollment and attendance, thus suggesting that there is a problem o f non-attendance, although not as severe as would appear if all oblasts were included. Interestingly, the MultipleIndicator Cluster Survey'37(or MICS) reports significantly lower school attendance for 7 to 11 year olds (73 percent) than even the national sample o fthe FBS. As a result o fthese problems that were detectedduringthe preparation o fthis report, the Statisticalauthorities monitored the data collection in the subsequent year more carefully (2002), and guided all interviewers to interpretthe variable as enrollment, rather than attendance. Although the new data were not available for use in this report, Table 2 presents the results of the regional enrollment ratios from this data provided by the Statistical authorities. As can be seen, the overall enrollment ratio of 96% is quite close to the officially reported net enrollment ratio for basic or compulsory education. Unfortunately, no measure of non- attendance or absence from school was available. Moreover a close look at this data by region and month suggests that there still may be some differences in interpretation, although less prominent. Thus, the above discussion suggests that actual attendance may be lower than enrollment ratios, but the precise magnitude is difficult to establish from the available household data for 2000/01. For the majority of the remainder o f this chapter, we conducted most o f the analysis with two sets o f data: the full national FBS sample for 2000/01 and the FBS 2000/01 excluding the three oblasts- KKP, Kashkadarya and Surkhandarya. Care needs to be exercised in interpretingthese results, for the reasons mentioned above. Inthe case of the results from the 11 oblasts, the reader should bear in mindthat li)these do not capture actual attendance but are some average of attendance and enrollments; and (ii) results are not nationally representative. and these indeed some o f the poorest regionsare excluded from the a n a l Y ~ i s ' ~ * ~aAresult,and s perhaps due to some of the particular characteristics o f the excluded regions (e.g. larger number o f children inKashkadarya), some o f the variations seen inthe data when the entire data set is used- by income deciles, age, gender, etc. are diluted or do not remain. Notwithstanding these caveats, the analysis presented here is useful for the purposes o f the present assessment, especially beyond com ulsory education, where there typically tends to be less of a discrepancy between enrolment and attendan~e''~. It i s also useful in as much as it establishes some o f the problems with the existingdata that need to be addressed in follow up surveys. The results presented here should be verified once improved data are available. 136One explanationprovided for the low attendance rates in Kashkadarya, was an earthquake during the school year, which disruptedmany classes for severalmonths. 137 137The MICS was carried out for 5,313 households in 2000 and is meant to be representative at the national level. This survey looks inparticular at school attendance inthe first five grades, and therefore has information on a narrower age group. Detailresults are reportedin Mertaugh(2002) 138Moreover we cannot be certainthat the data from the remainingoblasts are free of errors. 139 It should be noted that the net enrollment rate for secondary with the full sample actually compares with the net enrollmentrate reported in official statistics. 95 Age, Regional, Gender and Income Disparities in AttendancdEnrolIment Figure 3 plots school enrollment rates by age and gender for the smaller sample of 11 oblasts. For both boys and girls, school enrollment rises steeply at age 7, with a peak at about 8 or 9 years, staying more or less constant thereafter, before falling steeply after age 15. The general pattern is similar, with the national sample as well as with official data provided by the Ministryof Education: although in these cases for girls there is a more gradual rise to maximum school enrollment, and an even steeper decline at age 15. As mentioned earlier, another source o f attendance data is the MICS. The MICs found somewhat higher rates of school attendancethan the full sample of the Household Budget Survey for both boys and girls at ages 8, 9, and 10. Butthe much lower rates of school attendanceat ages 7 and 11 more than offset the higher rates at ages 8, 9, and 10, with the result that the proportion o f the 7-11age group which attending school isjust 73 percent (74 percent for boys and 73 percent for girls). These unexpected findings from the MICS call for further investigationto determine the extent o f this pattern of late average starting age and unexpectedly early drop- out from formal schooling. Figure 3: School Enrollmentby Age and Gender, 2000/2001 Male Female I 6 8 10 12 14 16 18 20 22 24 Age (in years) Source: Family Budget Survey 2000101(1 1oblast only) Table 2 presents school attendance rates by oblust for the 8 through 14-year age gr~up'~~using boththe FBS data from 2000/01 as well as the revised table prepared for 2002 by the Government (after introducing changes to interviews). The suggested picture on regional differences is dramatically different from the two datasets. When the full 2000/01 sample is used, with the exception o f three oblasts, Table 1shows fairly high rates of school attendance, with Tashkent city showing the highest rate. But for three oblasts, it shows dramatically lower rates o f school attendance for both boys and girls: Karakalpakstan, Surkhandarya, and, especially, Kashkadarya oblasts. As discussed earlier, there appear to be systematic differences inthe way the data were collected in these three oblasts. For 2002, however, the data show uniformly higher enrollments with much lower variation between region^'^'. As we look at older age groups, corresponding to secondary and non-compulsory cycles, regional discrepancies become sharper with the same broad pattern - highest enrollment in Tashkent City and Novoi and lowest in the three oblasts - Kashkadarya,, Surkhandarya and KKP. 140 This age groupingeliminatesambiguity about whether 7-year-olds may not yet have started schoolby the time of the survey, or whether 15-year-olds may have completed school by the time of the survey. All children 8 through 14 years o f age shouldbe inschool under the prevailingeducation law. 14' Interestingly, these findings on regional disparities are also broadly consistent with the results of the MICS, which found that the proportion of children entering first grade who eventually reach grade 5 is far lower in the region that includes Kushkudutyu and Szirkhandutyathan in all other regions (Mertaugh, 2002). 96 Source: Family Budget Survey 200012001, weighted results; 2002 tabulationsprovidedby GSS. Equality of access to education by gender i s guaranteedby the Constitution. As we have seen, Uzbekistan has made substantial progress in closing gender gaps in education. Table 2 also shows that female to male ratios inattendance among 8 to 14year olds are over 90 percent inall regions. The only exception is Kashkadarya, which also has the lowest attendance rate, when the national sample i s used. However, signifjcant difSerences still remain at higher levels of education, particular1 university education: the FBS shows that the percentage females students i s approximately 36 percent.J2 It should be noted that some gender streamlining begins to take place at the secondary level: data from the Ministry o f Education show that although total enrollments at this level are roughly equally distributedbetween males and females, women constitutedabout 35% of total enrollments in the academic lycees and 39% of total enrollments in professional colleges. But even if women are near parity with men in terms o f their educational qualifications when they enter the labor market, throughout their working lives women consistently earn less than men after correcting for earnings differentials attributable to differences in education, job experience, sectors of employment, and temporary departures from the labor market.'43 When the full sample is used, there is a consistent, direct relationship between family income'44 and school attendance at all levels of education. However, when the three oblasts are dropped, these differences disappear for basic education- not surprising since the two of the three oblasts reporting lowest enrollments were also the poorest. However, the differences remain for post-compulsory education This can be seen in Tables 3, for basic education, and Table 4 for secondary and higher education. 14* The percentage of female students in undergraduate higher education is 38%, and in graduate higher education programs isjust 30%: Report on the Status of Women in Uzbekistan.UNDP, Tashkent, 1999. 143See the discussionof the gender gap in the Labor Marketschapter. 144Usingper-capitafood consumption as a proxy for householdincome. 97 Male Female Total Poorest 68.7 68.0 68.3 2nd 76.6 77.8 77.2 31d 79.9 78.1 79.0 4" 81.4 79.3 80.4 Richest 85.5 83.3 84.4 All 78.2 76.9 77.5 ~ For post-compulsory education (Table 4), most conspicuous is the particularly high rate of attendance in technikum and university programs for the highest income/consumption quintile. For most households, education embodies elements of both consumption and investment. In Uzbekistan, as in most countries, education especially at these levels, is a normal good: families with higher incomes tend to consume more education - in terms of both frequency o f school attendance for their children, and their expenditures on education -- than families with lower income.145 Table4 Percentageof ChildrenCurrentlyPursuingSecondary and HigherLevels, - Source: Family Budget Survey 2001 data. Age groups for the four columnsare 15-17, 16-17, 16-19, and 18-23, respectively.Note: When the 11oblasts are used, the overall enrollmentrates are slightly higher. However, the general patternis similar, particularly in the case of Technikumand University. What is behindthe falling participation and the disparities in attendance, particularly beyond the basic level? The answer is complex and most likely includes a variety o f factors affecting both the demand and supply for education. On the demand side, are falling incomes, lower perceived benefits and higher costs, while on the supply side are closure of facilities, deterioration inquality especially inthe early years o fthe transition could be potential reasons. In what follows, we try to gain some insights into the importance o f some o f these factors. '45 Using data from the education module of the Household Expenditure and Consumption Survey, the estimated expenditure elasticity of demand for education is .81, implying that education expenditures are a normal, but not a superior (elasticity > 1.0) good. 98 The Education System: the Legacy, Policy Response to Transition Challenges, and the Introduction of a M o r e Active Approach Reflecting its Soviet legacy, Uzbekistan inherited a relatively well developed education system with high levels o f participation. Curricula and education planning followed the centralized Soviet model with teacher- centered and traditional pedagogy, focused on repetition o f factual knowledge imparted by the teacher as proof that students mastered the concepts in question. The financing formula for education was driven by centrally mandated norms which provided little room and no incentive for school principals and local education administrators to improve efficiency in use o f public resources or to mobilize additional financing.146 Most schools were relatively well provisioned and maintained, following norms established for application throughout the entire Soviet Union. Late in the 1980s, the situation began to unravel in the education sector, as in other sectors, and with independence in 1991, the situation fundamentally changed. The disappearance o f subsidies from Moscow, falling levels o f and shrinkingpublic revenues meant that the country had to struggle to maintain the inherited education system148. Despite the relative protection o f the social sectors compared to other countries, public fundingo f education in Uzbekistan fell from 10.3 percent o f GDP in 1992, to 6.7 percent of GDP by the end of the decade. Many educational facilities, especially pre-schools which had been financed and managedby public enterprises were closed or transferred to state control, while schools at all levels faced funding shortfalls. These closures likely contributed to some o f the dramatic declines in coverage in pre- school, especially inthe rural areas, especially insome areas (see Mertaugh 2002). At the same time, the new situation called for a fundamental change in the content and management o f education, as traditionally guaranteed employment opportunities for graduates disappeared with the closure o f enterprises and the emergence o f open unemployment. In order to maintain education programs in the situation of falling resources, the Government's initial response to these developments focused on reducing expenditure needs and diversifying sources of financing. The duration of compulsory education was reduced from eleven years to nine years, and education financing was diversified through three actions: a) decentralizing the responsibility for financing and managing most education programs from central to oblast governments, b) requiring parents to purchase textbooks which had formerly been provided by the Ministry o f Education, startingwith textbooks for grades two and 3,'49 and c) institutinga program of cost recovery for students in specialized secondary education and higher education. Cost recovery in higher education is achieved through a program of "contract" education for students who score below the designated threshold score for free entry into these programs. The fee levels for these students are meant to provide full cost recovery. There are no privateuniversities. In 1997, the Governmentadopted a more pro-active approach to transition challenges inthe education sector. It announced a new education reform program designed to reorient the education system to the needs of a market economy. The new program, called The National Program for Personnel Training, extends the duration of compulsory education to 12 years and restructures education programs at the secondary and 146These were common features o f education systems under the FSU. (For a fuller description of the shortcomings o f the inherited education systems in the transition countries, see Hidden Challenges to Education Systems in Transition Countries, Sue E. Berryman, The World Bank, 2000. 147 Uzbekistan's level of output fell for six consecutive years during the decade, reaching a cumulative decline of 18 percent, versus an average decline of 51percent for the countries ofthe former Soviet Union. (Transition: the First Ten Years-Analysis and Lessonsfor Eastern Europe and the Former Soviet Union. The World Bank, 2002. 14' By the end of the decade public education expenditures remained 19 percent below their real level in 1993. Expressedinterms of real expenditures per student, the declinewas even more pronounced- amounting to a 25 percent fall since 1993. 149Except for grade 1, where they continue to be providedfree. 99 tertiary levels'jO. A new program o f three-year acadenzic Zycea is to replace the current two-year program for complete secondary education. The new academic lycea are to be attached to universities and staffed by university faculty. In parallel with this, the country-wide network o f vocational secondary tecnikunzs is to be transformed into a new network o fprofessional colleges at the secondary level, with flexible curricula which are intended to respond to local employment needs. About 90 percent o f secondary enrollments are expected to be in these new professional colleges. Higher education programs are being regrouped into the anglo- Saxon model o f bachelor and masters degree programs, with an expansion o f programs in computer sciences, business administration and economics, other social sciences, and English language. To improve the quality and relevance o f education at all levels, the program includes a number o f actions to improve key educational inputs, including developing more modern curricula and more student-centered teaching practices, strengthening pre-service and in-service teacher training, and improving the quality of textbooks and other educational materials and the process for generation, review, manufacturing and procurement o f educational materials. The Experience to date: Costs, Quality and Relevance, and Attendance Privatepaymentsfor education As discussed above, the shrinking public expenditures for education were accompanied by policies to decentralize education management and diversify education financing, in particular through the formal introduction o f cost-recovery. Thus, extra-budgetary expenditures on education have been increasing as a share o f total education expenditures (Mertaugh, 2002). Although the FBS Survey does not provide information on costs to households, an add-on module to the main survey was implemented in September 2000, which provides information for all households with school age ~hildren'~'.Tables 5 and 6 present these results. 7 - 15 years 16 - 18 Years 19 - 22 Years Non- Non- Non- Type of Expenditure Poor Poor Total Poor Poor Total Poor Poor Total Private Education Outlays: Average Expenditures Per Child 6,886 9,506 8,850 13,403 19,034 17,865 56,865* 55,430 55,631 YOof Householdswith expenditure 87.0 91.6 90.4 92.6 93.6 93.4 42.8 81.0 71.9 Gifts/unofficial payments to staff: I%ofHouseholds with expenditure 1 3.1 1 12.8 110.3 119.3 1 17.5 117.9 1 O* 128.0 121.3 Source: September 2000 EducationModule, weighted *n 4 0 Note: Only households that reportany expenditure The majority o f households report making expenditures on education for their children- approximately 90 percent o f households made payments for their primary school age children, 93 percent for secondary school age children, and 72 percent for children o f higher education ages. In all cases, the incidence o f payments was higher by the non-poor - in other words, better o f f households are more likely to make out-of-pocket payments for education, and especially for higher education. Out-of pocket payments could include those that 150Under the new program, students will be required to complete twelve years of schooling, starting with all students who have not yet completed grade nine in the current (200112002) school year. In addition to general secondary, special secondary and vocational educationhave become compulsory. The data for one month are representativeat the national level but not at the level of oblasts. 100 are permitted by law (such as textbooks for grades 2 and above, official tuition fees, uniforms, etc.) or informal payments. Approximately 10 percent of households report making unofficial payments or providing gifts to staff, for their primary school aged children, 18 percent for their secondary school age children, and over a fifth for their older children (higher education ages). The magnitudes of these informal payments are not large, but like the incidence, they are higher for higher levels o f education, perhaps representing the increasing stakes (for example interms of access to public highereducation institutions, or o f better results on examinations) at these levels. As in the case o f health, informal payments from a survey questionnaire like this mightbe under-reported- in part because of the design o f the questionnaire, but also because individuals are less likely to report unofficialor under the table payments. Out-of pocket expenditures for education are also quite substantial in magnitude, especially at higher levels of education. The actual expenditures for the school year may be higher than reported here as many households incur expenditures over the summer months. As inthe case o f health, private expenditures on education are regressive:the non-poor spend more on education than the poor at all levels. Although the non-poor make higher payments, private education expenditures constitute a greater burdenfor poor households. Thus in 2000/01, private outlays for the school year alone constituted 8.6 percent o f food expenditures for non-poor households but 15.8 percent for the poor households. The burdenof the out-of pocket expenditures, combined with higher indirect or opportunity costs associated with sending their children to school, could be an important reason for the lower attendance rates o f poor households, at all levels o f education, noted earlier. In addition to the direct and indirect costs, poor households often face significant opportunity costs associated with sending their children to school because children in poorer households are more likely to contribute to household income or to house work and childcare. Box 1 on child labor, provides some interesting insights into some of the opportunity costs faced by poor parents and into the discrepancy between enrollment and attendance inUzbekistan. Inan attempt to address the burden of expenditures for poor households, the Government has recently introduced targeted subsidies for educational materials and winter clothes for the poorest students in grades 1 to 4, delivered through the mahallahs, or local community organizations and has begun to gradually introduce free textbooks for needy students. Table 6 shows that the most frequent and also largest payments for primary and secondary school are for uniforms, followed by textbooks. Reflecting the more dispersed distribution of educational institutions at these levels, cost o f travel and food are also important at secondary and higher levels. Tuition i s by far the most significant expenditure at higher levels although only about a fifth o f households make these payments. Highexpenditures at higher levels of education are directly associatedwith formal tuition fees introduced at these levels. The average tuition fee for a "contract" student, or a student whose score on the admission examination was not high enough to qualify for free admission. The average tuition fee for these students is about 300,000 soms per year,'j2 which i s prohibitive for poor households. 152A curious provisionof the contract student fee is that about 40% of the fee is retumed to the student as a stipend. 101 Table 6: Average ExpendituresPer Child on Educationby Type ofExpenditure, September 1999-May 2000 7 - 5 years 16 L8 Years 19 - 2 Years Average Iouseholdswith Average Householdswith Average louseholdswitE Type of Expenditure Lxpenditure :xpenditure ("A) 3xpenditurc expenditure ("A) 3xpenditurt :xpenditure (YO) Tuition 6,336 1.4 35,000* 0.5 85,611* 20.9 Textbooks and Supplies 2,813 84.8 3,689 90.2 4,790 61.4 Uniforms, etc. 5,222 85.0 8,926 54.1 13,644 54.4 Giftslunoffcial paymentsto staff: 565 10.3 2,524 17.9 3,271* 21.3 Cost of Travel 2,011 1.6 8,682 18.9 8,134 51.1 Food 2,060 31.0 5,433 43.2 12,126 54.3 Privatetutoring 3,939 6.1 14,232* 7.8 18,000* 2.0 Fieldtrips 392 36.0 1,299 31.7 1,766* 17.0 lother 591 5.1 3.310* 4.9 4,613* 7.2 The prohibitively high cost o f higher education associatedwith the introduction o f formal charges, could be an important contributor to the steep decline in higher education enrollments during the nineties, although inadequate places in higher educational establishments could also be an important reason. In principle, a large fraction of the returns from higher education accrue to the individual through higher returns in the labor market. Moreover, public higher education expenditures tend to benefit the better off who are more likely to reach higher levels. Therefore, both on efficiency and equity grounds, transferring many of the costs associated with higher education to the individual, for example through the introduction of fees at higher levels of education is appropriate. Cost recovery at higher education needs to be accompanied, however, by measures to protect the less well off, and also to address credit market failures associated with financing higher education (the returns incur later while costs are incurred earlier, and many households do not have collateral to finance higher education expenditures). One measure could be government assistance to needy students in the form of scholarships. The present system of provides public subsidies to the students who qualify on the basis o f the examination, and is therefore based solely on merit. This also tends to favor the students from better off households. Because students from higher-income families are more able to afford private tutoring and the informal fees to attend better schools, and thus to obtain higher scores on their university entrance examinations, the better off are more likely to qualify for budget expenditures on higher education. A combination o f merit and means for qualification for free or subsidized higher education would be preferable and would help address the inequities associatedwith accessing highereducation. There is a new student loan program to help poor students finance the costs of tuition for higher education students who are on contract. Students from poor households are eligible for a 50 percent reduction of interest charges on the loans. The high outlays for higher education, however relative to the average wage of 200,000 ayear, combinedwith the uncertainty associatedwith finding employment after graduation, however, may limit the effectiveness o f this program in bringing higher education within reach o f poor households. Not surprisingly, then children of households in the highest income quintile are six times as likely to attend university as childrenof households inthe lowest income quintile. 102 Box 1- Child Labor in Uzbekistan It is difficult to assess to level of child labor in Uzbekistan. National labor laws greatly restrict the employment of children. However, qualitative analysis indicates that the vast majority o f child laborers are employed in the informal sector and subsequently are not protected by these laws. Official statistics do not provide substantial data for child labor analysis. While school enrolment can be used as a proxy for child labor, it is important to note that a large number o f children who work are technically enrolled in school. Despite their enrollment, it is feared that many enrolled children miss school in order to hlfill their employment obligations. There are several supply-side causes o f child labor in Uzbekistan. The most prominent is the need o f children from poorer families to help support their household. According to the Tahlil Center for Social Research about 80 percent o f child laborers fall into this category. It i s believed that their financial contribution o f these low income child laborers makes a significant impact on the family's welfare. Other frequently sighted supply-side motivations for child labor include; demands from parents, the opportunity to increase personal consumption, and among non-institutionalized orphans and street children, survival. It is worth noting that laborers who list "increase personal consumption" as their primary motivation from gaining employment tend to do well inschool despite having ajob. Qualitative studies show that most child workers are hired informally. Children often find employment in small and micro- enterprises o f the informal sector. Even in the formal sector, most employers do not use formal labor contracts when hiring juveniles, and the child laborers are seldom registered as employees. In unregistered employment, children are usually paid less than adults and working hours and wages are set arbitrarily. Even ifthe child is formally hired, employers do not consider it necessary to observe the standards set out inthe national legislation. Children work both overtime and at night but they are not usually paid the requiredrates for such hours. According to the European Training Foundation andthe National Observatory a large majority o fUzbek enterprises do not hire individuals under the age o f 18. This is due to the desire to hire experienced employees and the existence o f laws which, if followed, make child laborers comparatively more expensive. Subsequently, there are a few key industries which employ the majority o f juvenile workers. In urban areas, children are mainly employed in the service sector (public catering enterprises, discotheques, clubs and hairdresser's saloons) and the trading sector. In rural areas, child laborers most often work in agricultural production processing plants, textile workshops, and family farms. Source: Child Labor inUzbekistan, Save the Children andTahlil, 2002 Quality andRelevance Quality is difficult to assess, both because the notion of quality is more complex, but also because data on quality indicators is not available. There is an extensive literature on school effectiveness which suggests that inputs (such as the quality of the curriculum, availability of instructional and learning materials, trained teachers, etc.) are critical in determining educational effectiveness and learning. The shrinking budgets to education were associated with a progressive deterioration of education infra~tructure'~~.While educational budgets declined, the response of educational managers focused on maintainingjobs o f teachers and other educational staff, resulting in increasing teacher student ratios at all levels of the system and falling workloads, signaling decreasing efficiency of provision. Thus, salary expenditures displaced vital expenditures to maintain and update teaching and learning materials and to maintainschool infrastructure. As a result, as in other transition countries, the quality of education and the condition of schools deteriorated in all communities except the few that were able to supplement budget financing for schools through local sources. Fundingshortages are especially severe at the base of the education pyramid - in primary and lower secondary education. As discussed in Mertaugh (2002), the actions taken to decentralize financing and management, combined with the generally higher unit costs of rural schools may have contributed to increasing inequity in the quality of schools inrural and urban areas. 153For example, 75% o f rural schools and 47% o f urban schools do not have functioning toilets. (Monitoring Learning Achievements: Uzbekistan 1999. UNESCO-UNICEF) 103 The National Program for Personnel Training, still in the early stages of implementation, is an ambitious and imaginative program aimed at improving quality and relevance at all levels of the system. The program raises several concerns about its ability to have a lasting improvement on the quality and relevance o f education, and also on possible negative impacts on costs and equity. First, it appears to be conceptualized as a discrete, once-and-for all change, rather than as an evolutionary process. It also pays more attention to the hardware requirements ofthe program, such as investmentrequirements of new facilities. Becausethe ultimate form o f many o f the changes associatedwith education reform cannot be anticipated with precision, successful reform needs to be seen as a process, with less attention to designing the ultimate changes, and more to introducing incentives, consensus-building activities, and feedback mechanisms through monitoring and evaluation of results, that would facilitate continuous improvements. Successful reform also needs to be given to "software" inputs, such as curricula, teacher training and incentives that are necessary to bring about a revolution inteaching practices (Mertaugh, 2002). The requirements o f the program in terms o f its massive investment and recurrent costs o f the secondary education reforms -the investment costs alone are estimated by the Government at US$ 4.5 billion'54-raise additional concerns. Even if financing could be mobilized to fully implement the program, there are important concerns about the equity implications of the program, since the program's focus on upgrading secondary and higher education and the associated recurrent cost requirements risk drawing resources away from basic education and further exacerbating conditions inthe initial years o f education. The channeling o f a small fraction o f secondary education students into academic lycea attached to universities may prove to be elitist and increase inequity inopportunities. The analysis o f returns to education in chapter three suggested the possibility o f a mismatch o f skills of graduates to the education system to the demands o f the labor market. While the national program attempts to address this mismatch, it emphasis on occupation-specific training at the secondary and higher education levels may prove too rigid for the needs o f a market economy. Under the global economy, skill needs evolve in a manner that is fundamentally different from the pattern observed in the past. Internationalcomparative advantage leads to a rapid succession of specialization in many markets. Specific skill needs change rapidly and unpredictably as a result of changes intrade patterns and changes in technology. Almost all workers can expect to changejobs - and, often, to change occupations as well - several times intheir working lives. The main role for universal education in this setting i s to provide the basic skills that are in high demand in the global economy and that provide the necessary basis for effective job-specific training later Job- specific training i s better, and more cost-effectively, provided in short, flexible tailored programs o f training offered invarious formats -often by the employer on the worksite. Understandingdisparitiesin attendance and reasonsfor non-attendance We saw above that, whenthe entire 2000/01 FBS sample is used, school attendance varied significantly with age, region o f residence and poverty level o f the household, and for higher levels additionally by gender. Multivariate regressions on the determinants o f school attendance'56 enables us to examine the independent impact o f these and other variables on school attendance. When the full sample is used, at the primary education level, the most significant impacts on attendance come from region o f residence: the most striking being the substantially lower likelihood o f residents o f Kashkadarya (75 percent lower), Surkhandarya (65 percent lower), and Karakalpakastan (55 percent lower) relative to Tashkent to attend school. When region 154Capacity Building in Education Finance, Project TA2948-UZB, Final Project Report, 10April 2000, page 10. 155The labor-force skills that are proving the most fundamental to success inthe global economy are skills in numeracy, problem-solving, accessing information, teamwork, and communication (including communication in foreign languages). -"Improving Capabilities: Education", in Making Transition Workfor Everyone: Poverty and Inequality in Europe and Central Asia. The World Bank, 2000. 156See noteby Abu-Ghaidain Mertaugh(2002) on determinants of attendance. 104 (oblast) o f residence is included, rural-urban location is not a significant determinant o f attendance at this level. Indeedthe inclusion o f region of residence "washes out" the impact o f household income: if region of residence is included in the specification, poor households are not significantly less likely to attend school. Age and oblast have the strongest impact on attendance at the secondary education level, although the magnitudes o f the oblast effects are smaller. In the case o f secondary education as well, the impact of household income and urban-rural residence washes out with the inclusion o f oblasts in the specification. Other interesting results are the statistically significant positive impact of education o f head, especially females (also males, but the magnitude is smaller) and the small but lower likelihood o f females attending in the case o f basic education (about 2 percent lower than males). Inview ofthe potentialdata problemsmentionedearlier, the multivariate analysis was repeatedfor the partial sample o f 11 oblasts. Not surprisingly, some o f the earlier results no longer hold. For example, for both levels of education, even when region o f residence i s excluded, poverty level o f the household is not a significant determinant of enrollment. As in the case o f the full sample, urban residence i s associatedwith higher enrollment at both levels, when region of residence is not included. However, when region of residence i s included, rural-urban location is no longer a significant determinant o f enrollment. Region of residence is strongly correlated with participation at both levels: participation levels are highest in Tashkent and lowest inAndijan (almost 24% lower for bothprimary and secondary age). For both levels, education of the household head no longer remains as a significant determinant. For basic education, girls are 2% less likely to be enrolled, when all other factors aretaken into account. How can we interpret the striking results on region? Unfortunately, no additional information on quality or access of facilities, or "price" o f schooling, are available to disentangle the impacts. Thus regional variables might be proxying for both supply side characteristics (such as distance, infrastructure, quality of schooling), etc., but also other demand side variables not included (such as differences in costs, perceived relevance relative to employment opportunities, cultural influences), etc. The lower and less significant impact of income i s partly due to its correlation with some of the variables that are being proxied by the regional variables (e.g. the poor live inregions where the distances are greater, or infrastructure is poorer, etc.) The September 2000 add-on data set provides additional insights into interpreting the results on non- attendance and the significance of regional variables. The data suggest that low demand for additional schooling might be an important reason for non-attendance -"completed schooling and not working" was the most frequently reported reason (37 percent of non-attendees). However, access and transport appear to be important constraints as well: 27 percent o f non-attendees reported "no suitable or accessible school" as the main reason for non-attendance. Next in importance was "completed school and working", reported by 21 percent of respondents (Mertaugh 2002), which suggests that there may be high opportunity costs associated with keepingchildren in school. The variations inresponses by region suggests that the relative importance o f demand and supply side constraints may vary across regions, with access and transport issues likely to be more important in Kashkadarya, Surkhandarya, and Bukhara, and quality and relevance (employment opportunities) more likely to be important in Karakalpakastan, Syrdarya, and Samarkhand. A fuller analysis o f attendance and other educational outcomes with enhanced data i s desirable for more precise policy prescriptions. The incidenceof public expenditures on education If private expenditures on education are regressive, do public education expenditures improve the overall distribution of expenditures? The answer is, unfortunately, that they do not. We saw earlier that differences in attendance or enrollment between poor and non-poor were particularly sharp for teknikum and university education. Public expenditures per student on the other hand, rise dramatically with the level of education 105 after general schools 157: the government spends three times as much on professional/technical programs compared to general secondary schools, and just under 5 times as much on higher education programs. The combined impact o f these patterns can be seen in the curve on overall incidence o f public expenditures on education in Figure 4: overall, public education expenditures are only slightly regressive, with the poorest 20 percent o f the population gets about 16 percent o f total education expenditures, while the highest income quintile gets about 24 percent15*. Breaking down the incidence by educational programs, we can see that public primary education expenditures are distribution neutral, reflecting mainly very similar absolute attendance numbers across quintiles. Since the poorer households have more children o f primary school age, but also likely greater absenteeism, successful efforts at improving attendance of children from poorer households would enable poorer households to benefit more from basic education expenditures. The incidence o f public expenditures on general secondary education i s similar, but slightly more regressive (the curve lies below the 45 degree line). The larger distance of the curves for teknikum and university education, however, shows that public expenditures on these levels of education disproportionately benefit better off households. As discussed earlier, the incidence o f expenditures at this level could be improved if scholarships or "budget status" was awarded on the basis o f means and merit, not simply merit. Figure 4: Incidence of Public Education Expenditures M-Primary - 4 Generalsecondary - --A-PTUiSPT - -X -Tekhnikum - &University -Total Finally it should be noted here that the significant regional differences in participation, for example with much lower participation in some o f the poorer regions, implies that these regions are less able to benefit from public education expenditures. Allocating expenditures on the basis o f some measure o f need rather than simply financing o f inputs, could be one way to improve the regional distribution o f expenditures. In addition, taking actions to improve the participation o f the poor at all levels o f education is essential to improving the distribution o f public expenditures, and counteracting the regressive impact o f private expenditures. 157Pre-schools have higher unit costs (by a multiple of 3) than general schools but lower than other levels of education. There is no enrollment data for preschools inthe main FBS. 158This figure is basedon the full sample. When the 11 oblasts are used, and the poorest three are dropped, the overall incidence of public education expenditures are slightly more regressive: the poorest 20% of the population gets about 14% o f total education expenditures, while the richest gets about 26%. The relative pattern between levels, however, is unchanged. 106 Conclusionsand PolicyRecommendations After several decades o f impressive gains in education, the transition posed special challenges to the education system. As in many other countries in the region, the early years o f the transition, associated with the recession saw signs o f stress in the system. Since the mid-nineties, however, coverage rates in basic education picked up and are now even higher than at independence. With available household data, however, it is not possible to establish the extent o f actual participation inthe system: the available evidence is mixed. Despite recent increases, at all other levels coverage rates remain below pre-transition levels. . While other transition countries have also experienced declines in enrollments in the nineties, the sharp and apparently more prolonged declines in higher education enrollments is unusual. There are also disparities in attendance particularly prominent at pre-school, teknikum and higher levels, between regions and income groups. Gender disparities at higher levels o f education are still significant. Unless the very recent turn-around and increases in coverage are maintained and increased, the potential erosion o f Uzbekistan's high human capital stock could undermine its successful transition to a market economy and will limit the ability o f the people to benefit from market reforms. Since education, especially higher education, is strongly correlated with improved living standards, without concerted actions to eliminate existing disparities in ability to invest in human capital, there is a risk that some segments o f the population will be left behind, contributing to greater inequality inthe future. The increase in basic education coverage is likely due to a combination o f factors, including many o f the positive reforms introduced in the system, to improve quality and attendance and to target assistance to the poor. A number o f different causes have likely contributed to the declines in enrollments during transition - inthe case o f preschools, for example the closure o f facilities that were previously managed and financed by state enterprises i s likely to be an important cause. Similarly lack o f funds led to closure o f other schools in a number o f areas. As in other countries, declining resources for education, and inadequate adjustment o f the input mix meant that many key inputs were under-funded and the quality o f education suffered. Falls in incomes combined with emergence o f formal and informal costs associated with sending children to school made it more difficult for the poor to keep their children in school. Insufficient job opportunities for graduates from the education system could also be an important contributor to declining enrollments by lowering the perceived benefits from the education system. The importance o f the different factors for attendance varies significantly by regions. The decentralization o f financing has also contributed to inequities in quality o f education between rich and poor communities and families. Public education expenditures are also slightly biased towards the better o f f and therefore have been unable to compensate for a regressive distribution o f private expenditures. Improvements in incomes and employment opportunities will increase the demand for education and therefore attendance in the long term. In the nearer term, however, it is important to ensure that education programs serve the objective o f enabling all children to access good quality education to be able to reap the benefits o f economic growth. This could be done through improving efficiencies in existing expenditures and channeling greater share o f public expenditures to improvingthe quality o f general compulsory schooling combined with targeted interventions at all levels to protect the investments in human capital o f the poorest sections o f the population in the human capital o f their children. Inthis context, and in view o f the concerns that it may be contributing to reduced flexibility o f skills, higher costs and growing inequities, a reorientation o f the National Program for Personnel Training i s called for. The following specific measures are recommended, many o f which would require substantial changes in the way in which education i s financed and managed . For example, movement from input to capitation based financing, and changes in formulae for allocating public resources between and within regions would be important changes. Strengthen monitoring and evaluation of erliicational outcomes. The analysis in this chapter shed light on many o f the difficulties with available data in assessing educational outcomes. Although many indicators are 107 monitored administratively, such as absolute enrollments, systematic monitoring of enrollment rates or coverage, repetition and drop-out, needs to be carried out, with appropriate disaggregation by gender, region, and income levels. Household surveys can be an important tool in such a system as well; however the present survey does not adequately establish levels on non-attendance, absenteeism, etc. Periodic household surveys can also be usedto monitor private costs and the burdento poor households, while special studies (such as the one recently commissioned by the Ministry on the causes o f non-attendance) could provide valuable information on specific issues. The Ministry has recently been carrying out monitoring of educational achievements through tests; this is an important step and further efforts to strengthen the tools, to rely on systematic objective national and international tests159, could provide useful feedback for continuous improvements (see below). Continue efforts to imurove aualitv and relevance,but with greater emphasis on the software elements and the urocess of reform, and re- orient stratew for secondarv education restructuring: The objective of the National Program to improve quality and relevance at all levels o f education addresses a crucial need o f the system. As discussed earlier, many crucial inputs are being supplied through the program at all levels, starting from the base of the education system. The provision o f crucial inputs and the modernization o f curricula, introduction o f student centered teaching practices, etc. are welcome and needed actions for the general education system. To ensure that these interventions lead to continuous improvements and increasing relevance of the skills provided, however, more attention needs to be paidto the software elements of reform. These should include building in incentives for change in all aspects of the program, including, crucially, financing mechanisms for education, and teacher evaluation and promotion policies. Also essential are feedback or monitoring and evaluation, for example in the form o f continuous assessment of educational outcomes and monitoring o f student learning over time, including participation in international assessments. The modernization of curricula should also be accompanied by the introduction of a process to build consensus on content among key stakeholders and to ensure ownership of reforms. Strengthening of teacher training programs calls for more extensive sessions of supervised practice teaching in actual classroom settings, while in-service training should provide more mentoring by experienced teachers, and more networking opportunities for teachers to work out their own solutions to teaching problems. The system o f teacher evaluation, promotion, and compensation should be fundamentally changed in order to reward what teachers do rather than what teachers know. At the secondary level, it is essential to reconsider the current plan to restructure secondary education into professional colleges and university affiliated academic lycea in the light o f international experience. As discussed earlier, this plan may not provide appropriate skills for future labor-market needs while entailing high investment and recurrent costs. The Government could constitute a task force to examine best practices in secondary education restructuring in OECD countries as a basis for re-orienting this strategy. Moreover, rather than impose the proportions of enrollments in secondary schools (90 percent into professional colleges and 10 percent into academic lycea), it is recommended that students and parents make these choices basedon up to date informationon the career implications of alternative choices o f secondary specializations. Effective implementation of such an approach would require the establishment o f a database and the provision of this information through career counseling and other media. Continue, monitor and build on targeted efforts to raise attendance and qua& in basic education: The introduction of targeted subsidies for educational materials and clothes through the mahallahs i s very welcome, as is the plan to provide free textbooks to 75 percent o f "needy" children duringthe 2002/03 school year and 100 percent starting from 2003/04 school year. The government should build on and monitor the effectiveness of these programs that assist the needy to meet basic education expenses. It could also consider 15' Participation in international tests such as PISA, TIMSS etc. provide information on the performance of the majority o f children, unlike Olympiads, etc. that focus on those that excel. 108 and pilot other targeted programs found to be successful in other countries in raising attendance such as targeted school lunches, scholarships to compensate for indirect and direct costs o f schooling. Step up efforts to increase investments in secondarv and higher education, and introduce targeted assistance for poor and girls in higher education: The lower enrollment rates insecondary and higher levels o f education relative to independence indicate a serious deskilling o f the labor force which will undermine Uzbekistan's productivity and competitiveness, unless recent increases are stepped up. A survey o f recent drop outs and graduates o f secondary education could provide useful insights into some o f the causes declines inhigher education. Private provision should be encouraged and student demand should drive secondary and higher educational programs. To address inequities in access to higher education, in addition to merit, consideration should be given to means or neediness, in the allocation o f public subsidies for higher education, and special interventions should also be considered to improve attendance by girls in higher educational establishments. Investigate sources of regional differences and encourage local solutions to raising educational attendance and quality. Since reasons for non-participation vary by regions, local communities should be encouraged take initiatives to improve educational results interms of both attendance and quality. For example the school calendar could be altered according to needs of local communities for seasonal agricultural inputs from their children on householdfarms. Inrural areas, where the unit costs o f schooling are typically highand schools are difficult to staff, efficient approaches to addressing access could be explored, such as multi-grade schooling, transportingteachers, etc. Consider amroaches to increasing participation of younger children: The serious reversals in pre-school education, particularly in rural areas are o f concern as pre-school education can play an important role in improving school performance, particularly among vulnerable groups. An additional concern comes from the evidence o f late participation in basic education. Options to address this include: including at least one year of compulsory pre-school (attached to existing general schools), combined with piloting o f new approachesto early childhood education targeted to poor and particularly rural areas. The recent emergence o f private and non-traditional pre-schools i s welcome and should be encouraged. At the same time, however, it would be important to set up adequate regulation or accreditation of their quality, and to ensure that they provide adequatepreparationfor primary school. 109 Chapter7 PROVIDINGSECURITY: THE ROLE OF SOCIAL PROTECTIONPROGRAMS Social protection programs can play an important role in providing greater security against life time risks, such as those associated withjob loss or old age, and protecting living standards o f the poor and vulnerable. Uzbekistan's social protection system is quite extensive. The pay as you go pensions system is quite generous, and has extensive coverage and it plays an important role inreducing poverty for pensioners as well as other individuals intheir households. Poverty reduction through the pension program, however, comes at high current and future costs. The low coverage o f unemployment benefits means that it has a very limited impact on protecting the unemployed. The majority o f expenditures inthe employment fund go to active labor market programs, most significant o f which arejob creationprograms inthe form o f subsidized credit through commercial banks. Although employment offices place a majority of registered unemployed through these programs, impact evaluations are important in assessing their effectiveness. The current system o f provision o f means tested child allowances and low income benefits through the innovative Mahallah scheme has many positive features- the child allowance program is the most efficiently targeted program. The low income assistance program, however, does not perform as well. Uzbekistan also has an expensive program o f provision of housing privileges- these programs however are least well targeted to the poor. Finally, existing social protection systems could be strengthened or special programs designed to help mitigate some of the short term social costs associatedwith reform options discussed inthis report. Overview of SocialProtectionProgramsand FinancingTrends Reflecting both its socialist heritage and recent transformation, the social protection system in Uzbekistan is quite extensive for its level o f income. The mainprograms include a pay as you go pension system; active and passive unemployment programs, and means tested poverty benefits, in particular, child allowance and low income benefits. Other programs are child benefits to mothers with children under 2 years old, special discounts/privileges on public utilities, and other privilegesto particular groups'60. Table 1 shows the relative importance o f the three main programs and recent trends in financing and beneficiaries. Government spending on these programs amounted to roughly 9 percent o f GDP in 2000. Pensions, including old-age, disability and survivor pensions, are the most significant program in terms of resources (expenditures were 7 percent of GDP) and also number of beneficiaries (2.6 million). The next highest share o f expenditures (1 percent of GDP) were on poverty benefits, the majority o f which were on targeted child allowances. Expenditures on unemployment programs accounted for under 0.3 percent f GDP in2000. The Uzbek pension system is a pay as you go system, quite similar indesign to similar systems inOECD and other transition countries. It provides old age, disability and survivor insurance to all eligible private and public sector employees,161 with old age being the largest program (by number o f recipient and program expenditures), followed by disability and survivor pensions'62. Pensions are financed from a 37.5%'63 160 A detailed description of all programs can be found inRashid and Mehra (2002). Military personnel are not included in the public pension system. Participation o f dekhan farmers is voluntary and therefore very few contribute to the system. 16' The FBS shows somewhat different composition o f the three programs than Table 1, based on Administrative data. According to FBS data, about 80 percent o f pensioners receive old age pensions; 17 percent received disability pension and 3 percent received survivor pensions. According to administrative information 259,000 individuals received "social pensions" or other benefits (for e.g., allowances paid to individuals disabled since childhood, or elders and invalids with insufficient work history to qualify for pensions). This chapter does not provide separate analysis o f social pensions, as sample sizes in the FBS are too small to permit this. 110 payroll tax, including 35 % paid by employers, and 2.5 % paid by employees paid into an independent Pension Fund. In addition, enterprises pay into the Pension Fund, 0.7 percent of the value of sales. The objective of the programs are, quite obviously, to provide income security during old age, disability or loss of a breadwinner, respectively. Disability 404.51 442.62 473.45 508.31 549.67 570.56 584.95 % of all pensioners 16.88 18.06 18.99 20.04 21.20 21.69 21.90 FASurvivorpensioners 220.41 222.78 220.71 219.96 224.19 226.48 216.35 of all 1 9.20 I 9.09 I 8.85 I 8.67 I 8.65 I 8.61 I 8.10 1 The unemployment protection program has the objective of providing temporary income support to the unemployed and assisting them in re-entering the workforce. Two main types of support are provided: passive support inthe form of unemployment benefit and early retirement pension (together amounting to 9.6 percent of the Employment Fund budget in 2001), and active support, such as job creation schemes (46.3 percent of total expenditures in 200l), and retraining programs (3.7 percent), job counseling, job training search, and public works for the unemployed (2.1 percent). The program is financed by a mandatory contributions of employersof 1.5 % of payroll'65, and is administered by local Employment Offices. 163The agency does not distinguish between tax rate for pensions and other benefits, but estimates from recent studies, basedon approximate shares of expenditures, show that approximately33% payroll tax can be attributed to the pension system. 164Since the preparationand discussion of this report, the payroll tax was recently reduced, from 39 to 37%, with 35% beingthe employers' contributionin2003. 165Revenues are collected regionally. Nearly 70 percent of funds of the EmploymentFund are redistributedwithin each region;while the remaining30 percent is centralized at the Republican level. 111 Child allowances are provided to low income families with children under 16 years o f age166while low income benefits are also means tested. The administration of both these programs is decentralized to local communities or ma hall ah^.'^^Mahallah Committees can provide benefits at their discretion: In the case o f child allowances, benefits range from 50 percent o f minimumwage for families with one child, to 175 percent o f minimumwage for families with four or more children, while the level of low income benefits can range from 1.5 to 3 times the minimum wage. Federal guidelines provide several indicators for determining eligibility o f claimants, including monthly per capita income, family composition, labor force participation o f the household members. ownership of land and durables, but the Mahallah has considerable discretion in determining a household's eligibility for benefit withinthese broad federal guidelines. The provision of social assistance benefits through the Mahallahs is a unique feature of the Uzbek Social Protection system. Inprinciple, decentralization of program administration to local communities in this form could have important benefits as it allows greater local information to be used in evaluating and monitoring claims. On the other hand, it could also have some disadvantages. For example, unless properly monitored, it could result in awards based on social norms and political necessities at the local level, while open discussion of claims during the application and selection process may also discourage individuals to apply for benefit. Other social assistance programs (in 2000/01 - the period o f the survey) included universal benefits (equivalent to 1.5 times the minimum wage) for all mothers with children under 2 years (provided through employers for working mothers and through the Mahallahs to non-working mothers). More recently, the Government (in February 2002) has taken the important step o f improving the targeting o f this program, by replacing the program o f universal benefits for all mothers under 2 with a means tested program: for non- working mothers this benefit i s now means-testedby the Mahallahs. Inadditionto cashbenefits, the Government provides a wide range of socialprivileges to selected population groups - old-age pensioners, disabled, participants in military actions, veterans of labor, some categories o f government employees The privileges include a range o f free or subsidized services and in-kindbenefits168. The total estimated annual cost o f these in 2001, were about 40.5 billion sums or 13 times the total expenditure on the means tested mahallah benefits. The most expensive entitlement is 50 percent discount on maintenance fees and tariffs for utilities (gas, water, heating and electricity) provided to teachers and faculty staff and to health care worker^.'^' These privileges, by design, are not intended to protect the poor, but to reward service in these occupations. After the preparation and discussion o f this report, the Government has recently (April 1, 2003) taken the politically difficult decision to abolish most privileges for employees and medical staff. Table 1 shows that, with the exception of pensions, expenditures on most social protection programs have fallen over time. Especially since 1998, Government spending onpoverty beneJts have declined substantially (approximately ha@ and expenditures in the employment fund have also fallen. The sharp falls in 166 From 2002 -the eligibility has been extended to children under 18 years of age if the child is at school, college or other educationalestablishment). There are about 12,000 mahallahs inUzbekistan, comprising an average of2000 individuals each. 168 These include free vouchers to spas andrecuperation centers; free food package for retirees inneed and living alone; free drugs with the doctor's prescription; subsidized services, such as reducedmaintenance fees and tariffs for utilities, and discount fees on public transport; discounts on phone tariffs; subsidized telecommunications services; free or subsidizedpharmaceuticalsand medicalservice; tax exemptions; interest free loans for housingconstruction;priority in housing provision; priority inpublic servicing, priority admittance for resthomes andresidentialcare andnursing, and so on. See noteby Kuddo inMehraandRashid, 2002. 16' Many of the benefits are financed by a service provider and thus the total costs of the program are even higher. Approximately 940,000 teachers are entitled to such benefits with total costs of around 29.3 billion sums in 2001. The same program for health care workers costs around2.0 billion sums annually, and 65,000 people are eligible. 112 expenditures on poverty benefits have been accompanied by a fall in the number of beneficiaries o f these programs. Since it is unlikely that living standards have improved significantly, this most likely reflects lower award o f benefits to eligible applicants as a result of contracting resources. While the number of registered unemployed have increased, the number o f benefit recipients has declined over time. On the other hand, the number o f pension benefit recipients have increased over the past five years, reflecting primarily a rise inthe number o fold age and disability pensioners. Figure 1: The Level of Per Capita Benefits for Social ProtectionPrograms (in Sum) 2,500 1,966 .-a 3 2,000 ';1 1,500 0) 1,000 cn 5 500 0 Figure 1 shows the levels o f per capita benefits provided by the various social protection programs. The pension system provides the largest per capita benefits, followed by the Mahalla low income benefit ,benefit for children under 2 years, Mahalla child benefit, and last the unemploymentbenefit. Inthe following sections of the paper we examine these social protection programs, in an attempt to assess their ability inmeetingtheir statedobjectives, and to protect households from falling into poverty. The Performance of Social Protection Programs Pensions Thepension systemprovides extensive coverage of the elderly population. Coverage rates are 90 percent for women of pension age (over 55 years) and 85 percent for men o f pension age (over 60 years)170. Coverage rates are higherin urban areas compared to rural areas. Tashkent city and Samarkhand have the highest share o f households receiving pensions, while Kashkandarya has the lowest share of such households. This variation in coverage across ruralhrban areas and regions i s likely the result of the labor force and demographic composition of these regions(Mehra and Rashid, 2002). Thepension systemprovides the most generous benejits of all socialprotection programs. Figure 1 showed that the pension system provided the largest per capita benefit compared to other social protection programs. The level of per capita benefit provided to households by the old age pension system was about 2374 sums. Given the link to past wages, the average old-age pension i s higher for non-poor households. Old-age pensions comprise a significant share of household consumption (nearly 40 percent), higher than any other "O The FBS data show that women receive 60 percent of all old age pensions (reflecting their longevity and higher coveragerates), and 80 percent of survivor pensions, while they receive halfof disability pensions (FBS data). The high receipt of survivor pensions by women may in part reflect the receipt on behalfo f children who have lost one or both parents,andwho are reportedlythe majority of the beneficiariesof survivor pensions. 113 social protection benefit, and the largest share (62 percent) of all benefits received by household^.'^^ The generosity o f the pension system i s also evident from its replacement rate. The replacement ratio, or average pension as a percent of average wage, (averaging about 50% between 1996 and 2002) i s uite high and indicates that while the system provides adequate protection, it may have large financial . The level of pension appears to be quite constant by age of recipient, perhaps as a result o f indexation policies. As a result of the extensive coverage and generous benefits, the pension system has reduced poverty among pensioners and the households among which they reside . The poverty rate among pensioners i s far lower than the incidence o f poverty realized by non-pensioners (Figure2), particularly the very aged not covered by the pension system. Moreover, almost a third o f all households have members that are pensioners. Most pensioners live in extended families of about 6 members with an average of 2 children and 2 pensioners , Thus individuals livinginhouseholds with pensioners are less likely to be poor (chapter 1). Thusthe poverty alleviation impact o fthe pension system i s significant. The poverty rate would be 10 percentagepoints higher inthe absence ofold agepensions.Not allpensioners have low ratesofpoverty, however. Figure2 showsthat the poverty rates for disability pensioners and survivors are significantly higher than for old age pensions. However, these rates are still below those realized for the population as a whole (27.5 percent). Figure 2: Poverty Rates by Age, and Type of Pension 45 pensloners old age disability suniwr 45 - 54 years 55 - 64 years 65 &1years - All Figure2 also shows that elderly persons without pension benefits (approximately 10%-15% o f elderly) are at a muchhigherrisk of poverty (nearly twice) than those with pensions or the average elderly person. Although the primary goal of pension systems is to enable individuals to smooth consumption over the lifetime rather than to provide protection to the poorest or most disadvantaged, all pension systems have re- distributive objectives and elements. The design o f the system in Uzbekistan, as inmany other countries with pay as you go systems (including developed countries), allows for redistribution from the rich to poor (through the use of minimum and maximum pensions), from men to women (because of longevity of women but also through earlier retirement age, additional provisions for "early retirement" ), across occupations (through early retirees in particular occupations), and from contributing to non-contributing persons. Early retirementfor particular occupations have the effect o f cross subsidizing privileged workers at the expense o f non-privileged workers as many of the individuals retiring later may have lower incomes than some o f the early retirees. While a case may be made for favoring those in hazardous work with early retirement, 17' Consistent with statutory benefit provisions, the level of disability and survivor benefit are lower than the old age pension, but still higher than any other social protection benefit.. The share of invalid and survivor benefits in total consumptionare lower, but not insignificant (23.1 and 26.7 percentrespectively) 172The minimum pension was about 36 percent of average wage in 2000, and is not particularly high according to international standards. In 2002, the average replacementrate dropped to 41.7%. 114 countervailing arguments include the political influence on risk-rating inpay as you go pension systems, and the tendency o f privileges within public pay as you go systems to proliferate. While often intentional, re- distributive elements in pension systems have the effect of weakening the linkage between benefits and contributions, and therefore dampening incentives of workers to contribute. Despite re-distributional objectives, pension expenditures are not skewed to the poor. Pension expenditures are distributeddisproportionately more towards higher income households. The bottom quintile only receives 11 percent o f total pension benefit expenditures, while the top three quintiles, or 60 percent of the population, receives 72 percent of benefit expenditures. Invalid and survivor benefit expenditures are also disproportionately received by the non-poor. This is not surprising given that pension benefits are directly linkedto wages so that the non-poor receive higher benefits than the poor. However, consistent with their significant coverage, and the provision o f minimum pensions, the poor do receive a non insignificant share o f pension expenditures. Thus, despite pension expenditures being largely receivedby higher income groups, the poor also benefit from the pension system. Thepoverty protection provided by thepension system is costly to employment and also has potentially high futurejscal costs. . Pension spending amounts to 7 percent of GDP which is highcompared to countries with similar per capita levels (Mehra and Rashid, 2002). As is the case with many other transition countries, the payroll tax to finance these are high, and substantially higher than the average for OECD countries. The pension system also provides incentives for early retirement contributing to higher levels of inactivity. . The problems of the pension system stem, inpart, form the transition from a command-and-control economy. This has led to reductions inthe effective tax base: increases in official unemployment, employers remitting on the basis of lower reported income and growth inuntaxed self-employment. Reductions inthe effective tax base in turn contribute to high payroll taxes. . The high payroll taxes required to finance the system have a negative impact on employment opportunities inthe formal sector and contribute a growing informal or non- paying sector. In addition, they contribute to tax avoidance within the formal sector, in enterprises that are within effective monitoring, b providing incentives for under-reportingof actual payments, and payments in cash or kind, "off the books"". Thus highpayrolltaxes contribute to a vicious cycle of hightaxes and low compliance. The pension system is also not fiscally sustainable inthe future. A number of factors affect the financing of the system, and therefore its fiscal sustainability. The first is the generosity o f the system as measured by the system replacement rate, or the ratio of pensions to wages. According to the latest available data, the effective re lacement rate in Uzbekistan, of 50 percent (average since 1996) i s high relative to most other countries1 .2 The Government forecasts a reduction inthis ratio as wage growth outpaces pensions growth. The second factor is the system dependency ratio'75, which (because o f the demographic structure), is favorable in Uzbekistan. The dependency ration will continue to fall until 2010, but will rise thereafter with the aging ofthe population. The third factor is a measure of compliancewhich is a ratio of collectedto owed taxes. Reflecting the informality discussed above, but also inadequate collection enforcement by the Pension Fund'76, the system compliance rate in Uzbekistan is quite low relative to other countries in the region. In 2000, only 40 percent of the wage base appears to being captured in assessing contributions (Hendricks et al, 173InUzbekistan, centralized wage setting couldhrthercontributeto informality inthe formal economy. 174The exceptions are Poland and Hungary. 175The dependency ration is about 29 percent or there are about29 pensionersper 100contributors. 176Since January 2001, collectionenforcement has beentransferred to the Tax Authority, with a view to improving the administrativeefficiency o fthe system. 115 2001).'77 Inmost o f the non-FSU countries and the Baltic states, the compliance rate i s far higher- generally on the order of 70 to 80 percent. After the transfer of collection to the Tax Authority since 2001, however, collections are reportedto have improved by over 20 percent compared to 1999. The Ministry o f Labor and Social Protection, based on macroeconomic and demographic assumptions, and therefore projections for these factors, projects that the financial position o f the pension fund will deteriorate after 2010, primarily driven by declining birth rates and the aging of the population. As a result, after 2010, the pension system will be not be able to provide current level of benefitwithout incurring major deficits (or higher taxes), thus adding to future costs o f the system. Several changes in the parameters of the pension system, can help reduce fiscal pressures on the pension system, while preservingthe ability of the pension system to provide old age security. These include the phasing out of privileges, gradually increasing and equalization o f retirement ages and longer and equalized service periods. In addition building on present efforts to improve pension administration will help increasethe quality o f services offered to pensioners while contributing to improved compliance. UnemploymentProtectionPrograms The unemployment benefit program provides 50 percent of the average wage (based on average of last 12 months) to workers with previous work experience. Benefits are provided for u to 6 months duration. Workers with no work experience receive benefits equal to the minimum wage. 17'Registered unemployed are also eligible to participate active labor market programs. The range o f such programs offered by Uzbekistan is similar to other CEE transition countries and OECD countries and includes, inter alia, job counseling and search services, training, re-training and re-qualification, job quotas and public works. However, unlike most other countries, Uzbekistan spends almost half of its Employment Fund on job creation programs which are mostly in the form o f credits on interests below market level through the commercialbanks The coverage of unemployment protection programs is low. The number o f registered unemployed has doubled between 1995-2000, and placement rates o f people who participate in the programs are surprisingly highand in~reasing.'~'Despitethis, the registered unemployedcomprise avery small share, only 8.2 percent, of total unemployed, according to the household budget survey. Incontrast, in some OECD countries such as Austria and the Netherlands, and CEE countries, such as Slovenia, the number o f registered unemployed i s actually higherthan the number of surveyed unemp1oyedl8'. Less than a fifth registered unemployed draw benefits, and the share of registered unemployed receiving benefit has declined from about 20 percent in 1999 to 16 percent in 2001. According to administrative data, the low share of benefit recipients intotal registered unemployed is the result of a large outflow of registered unemployed to jobs; and the participation o f registered unemployed in public works or training programs, both o f which pay a stipend in lieu of benefit. However, even so, the real measure o f program coverage the share o f unemployment benefit recipients as a share of surveyed unemployed, remains quite low, and much lower than inother countries (Mehra and Rashid, 2002). 17'Resultspresentedin the labor market chapter also indicatethat only 50 percent of individuals pay contributions,with the compliance rate is higher (66 percent) in urbanthanrural (42 percent) areas, for agriculture andretail sectors, and for higherrelativeto lower income groups. (Table C5, labor marketschapter) 178Severance pay inUzbekistanis not less than one monthly wage, and can be extendedup to three monthwages only if the PES cannot find a job to a registeredjob seeker within that period. In such a case, hetshe has to register within 10 calendar days following contract termination.. '79In 1995, 62.4 % ofregistered applicants were placedinjobs (153,500 out of 246,200), and in2000, 66.6 % (280,600 out o f421,400). 180 There is considerable variation among OECD countries. The lowest registration ratios are in Greece (30 percent), Japan (36 percent) andthe UnitedStates (34 percent). 116 The registered unemployed are primarily secondary school graduates. According to 2001 administrative data, the typical registered unemployed has previous work experience (70 percent), has completed general secondary education or higher (90 percent); were blue collar workers (50 percent); lives in a rural area (70 percent) and 60 percent were less than 30 years old. Registered unemployed are roughly evenly divided between males and females. As in other transition countries, very few higher education graduates are registered as unemployed. Registration rates are higher for the poor relative to the non-poor. These characteristics o f registered unemployed are similar to those o f survey unemployed in most respects. However, the latter group comprise a higher share o f males, and are equally prevalent in rural and urban areas, suggesting possible disincentives in program design for males, and urban unemployed in registering181 '**. This profile o f registered unemployed-less educated and poorer-is similar to that found in most OECD and CEE countries. The registrationprocess contains some self-targeting as this group o f unemployed is likely to be most in need o f assistancefrom Employment Services. We saw in Figure 1, that the level of per capita unemployment benefit was the lowest among all social protection benefits. It comprises only 8.2 percent o f total household food consumption, again the lowest consumption share o f any social protection benefit,'83and an insignificant share o f total public cash transfers to households Mehra and Rashid, 2002). Thus, the unemployment benefit program does not contribute to raising household income/consumptionin absolute terms or relative to other benefits. The level of benefitfor those who receive it, however, is not low relative to international norms and likely provides adequate replacement against loss of income from job According to administrative data, the average unemployment benefit i s 21 percent of average wage, which is comparable with replacement rates in other (higher income) CEE countries.lg51g6 The average rate, however, masks the much lower and most likely inadequate protection for the groups of workers who receive the minimum benefit (about 60 percent of recipients). Reflecting, primarily the low coverage of the program, but also the modest level o f benefit and its small contributionto household income, the unemployment benefit program has no measurable poverty alleviation impact. It should be noted that in most Central and Eastern European countries, with some exceptions like Poland and Hungary, the poverty reduction impact o f unemployment programs i s mild, usually due to the insignificant share o f poor individuals for the program. As mentioned earlier, the primary objective o f such programs i s to protect against income loss associatedwith the loss of ajob, not poverty reduction per se. InUzbekistan, as elsewhere, the unem loymentbenefitprogramdoeshave re-distributive elements by design: between all workers who contrib~te''~to only those who lose their jobs, between high and low income groups as a result o f minimum and maximum benefits (but also because the unemployed are typically less educated and low wage workers). An additional element o f redistribution in Uzbekistan's program comes 181The previouswork experience of surveyed unemployedis not known. 46 percent are between 16-24; 63 percent are male; and 63 percent have education of 11 years or less. About 15 percent had completedvocational training, and 12 percentcompletedteknikum. Just under 5 percent of unemployedhadhigher education. 182There are also regionaldifferencesin compositionof registered and unregisteredunemployed, although FBS sample sizes are too small (Murrugarraand Tracy, 2002). lg3 Unemploymentbenefits comprised almost a third o f all income in Poland, and between a third and a quarter in HungaryandPoland. However it formeda small share of income inBulgaria, Estoniaand Slovakia. I84 The apparent contradictionbetween this statement and the previous one can be explainedby the fact that very few registered unemployed receive the wage related benefit. Benefits for other categories are low, at 75-100% of the minimumwage. The average benefitreceivedby the household is very close to administrativedata. I86The replacement rate for benefit in Uzbekistanis much lower than replacement rates found in higher incomeOECD countries (60 percent). 187As discussedearlier, the contribution is made by employers on behalfof workers. 117 from the fact that the program also provides (albeit lower) benefits to re-entrants and new entrants to the labor market. This type of re-distribution may be towards the rich or poor, depending on the income characteristics of these individuals. Despite these elements, the FBS data shows that the share of benefit expenditures from the program are about equally received by all income groups. Nevertheless, the lack o f progressive distribution o f benefit expenditures would not be not surprising given the program benefit is linked to an individual's wage18*. Evidence from transition countries and OECD countries, is mixed with distributional impacts typically neutral or progressive, but typically, unemployment benefits are not an important tool for income redistribution(Mehra and Rashid, 2002). Generous unemployment benefits programs with highreplacement rates can impose significant disincentives to work, and have led to an unemployment tra in many OECD countries. Negative effects can also come from a long duration o f benefit provision'f though the effects of replacement rate are much less pronoun~ed'~~. Thus, benefit levels have to balance objectives of smoothing consumption and protecting living standards against providing disincentives to work. The relatively modest levels of legally mandated and effective benefit provided in Uzbekistan, and limited and reasonable duration o f benefit means that adverse work incentive effects from the program are likely to be minimal. Nevertheless, some features of the present design do pose incentive problems. One problem, i s the eligibility, by law, of youth and labor market re-entrants to the program. Inmost countries, the eligibility for an unemployment benefit is restricted to laid off workers with a minimum contribution history in order to prevent abuse o f program. The eligibility o f these groups, in Uzbekistan, and the apparent absence o f a minimumcontributionrequirement, limitsthe 'insurance' aspect ofthe program, which may reduceincentives for employers to contribute to the program. Further, low and limited benefit receipt for some workers may also reduce their incentivesfor ensuring compliance to tax payment by employers on their behalf On the other hand, there are likely to be significant administrative costs associated with the program in its current design. A recent study (Vodopivec, 2000) finds that the informal economy prevailing in many transition countries, such as Uzbekistan, makes the task o f monitoring wages and employment very difficult. At the same time, the culture o f entitlement often remains ingrained in both beneficiaries and employment office workers, so that eligibility for benefit i s often not strictly verified. Finally, weak monitoring and enforcement capacity, such as the lack of technology, resources, and often also the political will to monitor and enforce existing laws, makes this task even more difficult. The study concludes that the adoption of unemployment insurance programs more suited to the administrative capacity, information availability, and formalized markets of OECD countries may have been slightly premature in transition countries, such as Uzbekistan. As discussed earlier, the administrative data suggest a high success rate associated with the active employment programs (henceforth ALMPs or Active Labor Market Programs), including, inter alia, job counseling and search, training, re-training, and re-qualification, job quotas, and public works (for more information on these programs, see Murrugarra and Tracy, 2002, and Mehra and Rashid, 2002). Nearly half o f all those who register obtain a job through the Employment Service, and about 40 percent o f t hose applying to labor offices obtain a job within the first 10 days after getting registered as job seekers. While this result seems very positive, in the FBS data, individuals indicated dissatisfaction with the services of It should be noted, however, that limited number of observations on unemployment benefit recipients limits the robustnessofthese results on poverty alleviationandtargetingefficiency. For example in some CEE countries including Bulgaria, Czech Republic, Estonia, Hungary, Poland, Slovakia and Slovenia. I9O Vodopivec, 2000. Interestingly, international evidence does not show much impact o f unemployment benefit on increasingintensity ofjob search, improvingjob matches, or entry into regularjobs. 118 Employment Offices. Nearly 17 percent of those who applied to the labor office were offered ajob that did not suit them, while only 2 percent o f unemployed found the job acceptable. Care should be taken, of course in interpretingthese, as it is difficult to judge whether the rejection ofjobs by registered unemployed was reasonable given their qualifications or not. Moreover, although high compared to FSU standards, the participation o f registered unemployed in some active labor market programs is also quite limited (according to administrative data only 16-18 percent of registered unemployed participate intraining activities and in public works). The FBS data also confirm that a very small share of applicants-mostly youth who have completed secondary school, are offered training and re-training programs, or public works. Again, this result is difficult to evaluate as many unemployedmay not be suitable for training, or the Employment Services may have limited capacity to provide or contract out these programs. Insum, then, although the majority of the expenditures (and beneficiaries) from the employment fund go to active labor market programs (and particularly credits to businesses), available data are not sufficient to evaluate their effectiveness. The best way to gauge the success of ALMPs is to undertake impact evaluations of eachprogram. This type of evaluation helps identify gains in terms o f employment and wages o f program participants relative to a control group o f non-participants with roughly same characteristics. Impact evaluation can help address critical questions which are not possible to addresswith administrative data: such as whether or not the outcomes are different than they would have been without the program (deadweight loss), whether or not workers in subsidizedjobs simply substitute for other unsubsidized workers who might have been hired anyway (substitution), and whether or not firms with subsidized workers take business away from other unsubsidized firms displacement). International evidence suggests that ALMPs outcomes are mixed. The available research from OECD and advanced Central and European transition countries, suggests that the aggregate gains from ALMPs tend to be modest at best, but that targeted groups can potentially benefit from certain types o f well-designed programs, albeit at highcost, and they generally require substantial administrative capacity. These studies also show that public works programs do not necessarily lead to continued employment but can be rationalized on the basis o f community development or as anti-poverty measures. In some countries, they can also provide protection to informal-sector workers. The results show that the most cost effective ALMPs arejob counseling andjob search assistance. These programs, rather than costly training or re-training programs or job subsidies, should be the basic building block o f ALMP programs, and should be used where resources are scarce. It should be noted that ALMPs, evenjob counseling and services, work best in an economy with a competitive labor market, where job creation rates are improvingand there is an increase in economic activity. Finally, as discussed later in this chapter, ALMP programs are also effective, in combination with benefit packages for allaying the social costs of laying off workers and should be considered in Uzbekistan to reduce any social costs associatedwith restructuring, or as a general policy to promote economic efficiency. The FBS data provides some useful insights into the low coverage o f the unemployment programs, and low rates o f registered unemployed. The vast majority of the survey unemployed do not appeal to the labor offices- over a third of unemployed instead appeal to friends or acquaintances for assistance with finding employment, while over a fifth do not take any action. This suggests the importance o f informal networks in obtaining employment. Low registration rates may also reflect a lack of knowledge about the program or the low quality o f programs and services offered by Employment Offices. Nearly 24 percent of the respondents reported that they did not register because they had no knowledge about Employment Offices or their location; another 21 percent found Employment Services unhelpful, a further 10 percent found the jobs offered by the Employment services unsuitable; and finally 6 percent found the application process costly (in terms of reaching Employment Offices) and the registration forms complicated to complete. Thus, about 60 percent of those who did not register took that step because of lack of information and confidence about the services provided by the Employment Bureau. 119 Figure 3: Reasonsfor Not Applying to the Labor Office Complicated application Notworking for healthlother process - - 6% , reasons 1 Other 1 I 12% Their jobs do 8% I1 not interest me - Did not know 10% Canfind myselfwork 3 18% Cannotcount Unemployment i on helpfrom benefit IS low 1 labouroffice 1% 21% Source: FBS data Only 1percent of respondents reported low unemployment benefits as a reason for not applying to the labor office, despite the fact that for some groups o f workers, especially workers with no prior experience, receive very low benefits (at or below the minimum wage). However, another reason for low application and low participation from younger male workers, may be the disincentives associated with the design for these groups. For example, male workers 35 years or under (with fewer than three dependents), who register as unemployed, are required to participate in public works programs, which pay 15 percent more than the unemployment benefit for this group. This mechanisms allows self-targeting o f benefit to the poorest youth, but also likely deters many o fthe near poor younger persons from applying for benefit. The informal sector participation o f the unemployed may also reduce application and registration rates. Working in the informal sector makes it difficult to obtain the official documents requiredto claim unemployment benefit. Finally, low application and greater levels of discouragement associatedwith the program may also be aresult o f the declining budget allocations to the program, which may have induced Employment Offices to reduce the number of clients served by the program. Onlyjust over half of those who did apply to the labor offices were registered (although nearly one-fifth were still waiting for a response). This is also the case Russia where many individualswho apply do not receive benefitbecauseprogram resources are limited. SocialAssistance Benefits In what follows we examine the performance of the various social assistance programs against their stated objective o f protecting the poorest from abject poverty. In evaluating the targeting efficiency and performance o f the decentralized means tested programs, i.e. the child allowances and low income benefits delivered through the Mahallahs, it i s important to note, that inclusion and exclusion errors are not easy to determine when eligibility conditions are decentralized and varied and there is no official policy on households that should be eligible for benefit. Nevertheless, useful insights are obtained from such an examination. Particularly useful are the insightsinto the relative performance o fthe programs. Means Tested Child allowance^'^':Table 2 shows that approximately 15 percent o f households receive child allowance benefits, and the coverage is higher for poorer households. 192 Nearly 26 percent of the poorest 191 It should be noted that data on child allowances may include households who receive child benefits under 2. The instructions to interviewers did not clearly distinguish between both benefits. As the Mahallah provides benefits to children under 2 for non-working mothers, these benefits may have been included in the Mahallah program. As, by law, all households in 2000/01 (no matter level of income) were eligible for child benefit under 2, this means that the child 120 Quintile Know of % %knowledgeable YOapplicants %Rejected %Waiting Average %oftotal scheme Received applyingfor awarded for Benefit* expenditure YO Benefit benefit benefit Decision (sum) received 1 89.7 25.8 54.4 65.0 23.5 11.5 2917 28.4 2 88.7 22.0 51.3 61.5 27.3 11.2 2804 54.0 3 86.6 17.8 45.2 62.3 27.1 10.6 2452 74.5 4 84.9 12.2 40.7 55.3 32.9 11.8 2342 90.0 5 82.3 6.3 30.7 50.5 42.4 7.1 2152 100.0 Total 86.3 15.4 44.3 59.9 29.4 10.7 2599 Why do some poor households not apply for benefits? Approximately 45 percent o f the poorest income quintilereported that they did not apply becausethey were usedto coping by themselves (27 percent) or that they did not consider themselves poor enough (13 percent)'96 or becausethey obtained help from parents or relatives (5 percent). However complexity or fairness o f the application process also appears to be a significant deterrent: 30 percent of respondents from the poorest income quintile (30 percent) cite that they have difficulties in filing claims including filling out forms and providing relevant documentation) while 11 percent felt that they would be unjustlytreated ifthey applied. About 6 percent o f the poorest were deterred from applying as they would be subject to the stigma and disclosure of family circumstances associated with filing claims. Thus, at the claims stage, exclusion errors may result from household's inability to fill documents, perceived injustice by Mahallah committee, and unwillingnessto publicly expose their family's socio-economic situation. However, on a positive note almost half of all respondents responded affirmatively when asked ifthey would apply ifthey were inneed. Table 2 shows that a large proportion of applications/claims (60 percent) were accepted by the Mahallah and the poor had a much higher acceptance rate than the non-poor. Thus the award of benefits by the mahallah allowances program results may be much better than reportedhere. Thus, exclusionerror may be lower; and targeting may actually be betterthan reportedhere. This is because, by law all households(no matter level of income) are eligible for child benefitunder 2. 192 Given the short period of allowance duration (6 months) and rotation o f benefit holders, the number of households receivingthe allowance for at least 6 months accounts to 29% ofhouseholdswith children 193 The coverage rate, or benefit receipt, of the program is higher for rural (vs. urban areas), and varies across regions The highestcoverage of the program is in SurkandaryaKashkadarya, andthe lowest inTashkendSyrdarya. `94 These percentages (in Table 2,3) on award and claims are based on the set of individuals who know about the "'rogram.knowledge Ifwe take allindividualsinthe sample, these percentagesaremuchlower The rate is higher for rural (vs. urban) households, male headed (vs. female headed) households, and Uzbek vs. Non-Uzbekhouseholds There was little systematic variation inprogram knowledge across householdsby age (except youngest age group) or education of householdhead. 196 These reasons were more significant among better off households, supportingthe hypothesis of self-targetingat the applications stage, and because o fthe stronger privatenetworksfor better offhouseholds. 121 committeesfavors the poor. However, the acceptance rate o f the non-poor is not negligible. Almost half of the top quintile applicants were accepted, indicating inclusion error. Inaddition, a large proportion, nearly a quarter, o f the poorest households were rejected, indicating exclusion e ~ 0 r .Moreover, Table 2 shows only l ~ ~ very little variation in the level of benefits between consumption quintiles, with the non-poor receiving on average only a slightly lower level of benefitthan the poor. Why were some poor households rejected? Household responses on reasons for rejection o f claims support the finding that the Mahallah committee is able to screen out well off households during the process of awarding benefit. About 60 percent of households claimed that they were rejected because they failed the means test, either because (i)their household income was too high (40 percent), (ii) were employable there adult members in the household (13 percent); (iii) because they had too much property (7 percent) or had a plot that could be used better. These rejections were mainly mentioned by the non-poor than poor. However, it i s also important to note that about halfof all poor also failed the meanstest for the same reasons. Another significant reason for rejection cited, particularly by poor households, was that `the committee did not consider their applications'. The interpretation o f this reason for rejection is not clear. On the one hand, the committee may not be able to consider applications because of inadequate funds. On the other hand, the fact that poor households were more than twice as likely to report this reasonthan those inthe highest income quintile suggests the possible influence of better off households with the mahallah committees. This is consistent with the fact the poor were more likely to report the need for connections as a reason for rejection. About 11 percent o f the poor noted difficulty in the application process as a reason for rejection (wrongly filled or inadequate application forms). Thus, at the award stage, exclusion errors may stem from the lack of consideration of application-presumably because of limited funds or the need for connections, wrongly filled out applications, and wrong failure o f the means test. Inclusion errors may result from non-poor households wrongly passing the meanstest or exerting undue pressure onthe committee. Administrative Delays affect few households. About 10 percent o f eligible households who applied were still waiting for a decision. Most had either been accepted or turned down. Residents inrural areas and in regions such as Kashkadarya and Surkhandarya were more likely to report that their applications were still being processed. Multivariate analysis of determinants of probability of benejt receipt allows us to look more closely at how various characteristics of the household influence its likelihood of receiving benefits. The detailed results are reported in Mehra and Rashid (2002). They examine in particular, two sets o f determinants. The first set comprises indicators o f household welfare. These include an index of durable ownership, ownership o f agricultural assets: a plot and livestock, ownership o f transport (car, etc.), and other socio-economic variables-such as female headed household and number o f children and unemployed, which are also indicators of poverty.'98 Some o f these indicators are in fact used by the Mahallah committee to evaluate if the household should receive benefit. The second set comprises indicators of horizontal equity. The program should not discriminate on the place of residence o f a household. The probability of benefit receipt should therefore be invariant to a household's place o f residence (region, town/village, or ruralhrban area). `97 Acceptance rates were higher in rural (vs. urban) areas, and for male (vs. female) headed households. The acceptanceirejectionrate for households by age or education of household headdoes not show a systematic pattern, but the acceptance rate for non-Uzbekhouseholds is higher than for Uzbek households. These indicatorsmay only partially correlate with unobservable indicatorsof householdwelfare, but it is unlikely that all o f the association can be provided by unobservable characteristics. It should also be noted that the Mahallah committee may observe characteristics correlatedto povertythat are not providedin the survey. 122 The results shows that the probability of receiving child allowance is strongly and inversely related to household welfare (as measured by household consumption, durable goods ownership)lg9 and the number of children in a household. Other welfare characteristics o f a household (i.e. whether it is unemployed members, owns livestock, etc.) do not have any impact on benefit receipt. They also suggest that there is horizontal inequity between regions in benefit provision. Even after controlling for household welfare, likelihood of benefit receipt is significantly lower in rural areas (vs. urban areas) and in Djizzak-Samarkand relative to Karakalpakstan (the base, or omitted region). Other regions (with the exception o f Fergana Valley, where the likelihood o f receipt is greater) do not have a significant difference in the probability of benefit receipt than Karakalpakstan. To obtain greater insight into the process of targeting, we disaggregate the probability o f receipt (following Micklewright et al, 200l), into the product of the probability o f knowledge, conditional probability of claims, given knowledge, and conditional probability o f award given claims. The results from this disaggregation confirm that low income households (including those with low level of consumption, who do not own transport; and those with more children) have a higher probability o f benefit receipt becausethey have higher probabiliov o f knowledge o f the program, are more likely to claim benefits and are more likely to receive awards. Interestingly, the level o f household consumption is only weakly important in determining probability of claims or award for benefit, ifother welfare variables are included. The lower receipt of benefit inDjizzak-Samarkandderivesmainly from lower probability o fknowledge about the program and a lower probability o f being awarded benefit, while the higher probability o f receipt in Ferghana valley stems from a larger propensityto apply. The more disaggregated results also show that there are significant independent regional differences particularly in program knowledge and the likelihood of award, between regions201. Thus regional inequities may stem from differences in outreach and knowledge, propensity to claim benefits202, and the performance of mahallah committees and their ability to screen. Regional differences in probability o f benefit receipt may also result from inequity in regional distribution of financing. These issues need to be investigated further. Does the level of the benefit provide adequateprotection? In 2000, according to administrative data, as well as the FBS data, the average child benefit was slightly above 3000 sums, or about 30 percent o f average wage. Benefit levels are regularlyadjusted to the minimummonthly wage. Child Allowances comprise 14 percent of household consumption, and a higher share for the most needy households: the poorer, younger and households with less educated heads. The importance of child allowances for households is also indicated by their not insignificant share of total cash transfers (10 percent). For the poorest quintile, they account for 20 percent of total cash transfers, the second in importance after old age pensions. Thus, the child allowance benefitprovides an adequate level of protection to thepoorest groups. The level o f benefit and the process o f award suggest that it probably does not compromise work incentives. The poverty rate would be 2 percent higherifthis benefitwas eliminated. 199However, benefit receipt is somewhat inexplicably positively related to ownership of a plot (perhaps because of positive correlationbetween landownershipandpoverty-see chapter one) 2ooHowever, the probability o f knowledge is somewhat inexplicably positively related to ownership of a plot (perhaps because of positivecorrelationbetweenlandownership andpoverty-see chapter one) 20'See Mehraand Rashid for detailedresults. 202 Interestingly,only in Ferghanawas the propensityto claim significantly higher. 123 Figure 4. Targeting Efficiencyof Social AssistancePrograms ~~~~ .- . 0 10 - -- 20 30 40 50 60 70 80 90 100 __ Equality 1---t---Mahalla-Poor +Mahalla-Child -X-Child Bentit under 2 Child allowances are the best targeted of all social protection programs. Figure 4 shows that about 28 percent o f total expenditures on the benefit are received by the bottom consumption quintile, while the top 2 quintiles percent receive only 25 percent o f all benefit expenditures. As the level of per capita benefit does not vary significantly across income groups, the higher share o f expenditures going to the poor i s mainly the result ofthe better coverage o fthe poor relative to the non-poor. Low income benefit: In contrast to the child allowance program, the coverage of the low income benefit program is vevy low, although coverage is higher for poorer households (Table 3). The coverage o f benefit o f last resort would be expected to be low, and focused on the most vulnerable groups. About 5 percent of low income households receive benefit relative to about 2 percent of high income households. Table 3: Knowledge,Application and Processingof Claims: Low Income Benefit Total 1 82.1 1 3.5 I 24.9 31.2 1 55.1 I 13.7 1 1833 1 Table 3 shows that although slightly lower than in the case o f the child benefits program, Program Knowledge is widespread and varies inversely with household income, the latter a result o f greater self- targeting o f information or better targeting o f program information to the poor. Claimsfor benefit are very low (relative to program knowledge, and compared to the child benefit program), but are once again much higherfor thepoor. Why do poor households not apply to the program? The main reasons are similar to those provided by non- applicants o f the child allowance program. About 47 percent o f all poor households cite that they can cope on their own, receive help from relatives and friends, or are not poor enough to apply for benefit. Again, quite 124 similarly to the child benefits program, another 24 percent o f the poor are deterred from claiming benefit because they find the administrative requirements cumbersome while 14 percent cite the unfairness of the mahallah committee as a reason for not applying. Once again, complexity and lack of justice were seen as more problematic by the poorest. Finally, about 7 percent o f the poorest quintile cite the stigma or exposure o f household circumstances as constrainingtheir claim for benefit. On a positive note, it is important to note that the program does have a positive image, and nearly two thirds o f all those who do not apply would do so if their economic circumstances deteriorated. The award rate is very low (relative to child allowances) andfavors the non-poor. Only 31 percent of all submitted applications are accepted by the Mahallah Committee compared to 60 percent in the case o f Child Allowances. This may be the result o f the very limited resources available to the program, or because the program i s indeed meant to serve as a last resort. More problematic however, i s the fact that acceptance rates increase slightly with the wealth o f a household, a result which runs contrary to program objectives. 203 On the other hand, the mahallah committees do reject a slightly greater proportion o f better off households. The difference, then comes from the more significant fraction of applications being processed, and greater processing delays for the poorer households. In the case of child allowances and low income benefits, the mahallah committees use decentralized information to screen out the poorest. In the case of child allowances, the additional requirement is the presence of children under 16. The correlation o f the number of children with poverty is probably one of the reasons for the better and possibly simpler screening in the case of the child allowances program. This may also explain the larger fraction o f households waiting for a decision in the case of the low income benefits. What i s problematic, however i s (as in the case o f the child allowances) the significantly higher processing delays for the poorer households - households from the poorest quintile were twice as likely to report that their application was still in process- suggesting that better off households are able to speed up the process. It i s also consistent with more difficulty in discerning among the poorer income quintiles. Why werepoor households rejected? The reasons cited by households for having their applications rejected by the Mahallah Committees are quite similar to those provided for the child allowances program. About 60 percent o f households were rejected because they failed the means test for a variety o f reasons (75 percent o f non-poor, but also half of all poor). Once again, the lack of consideration of the application by the Committee was a frequently sightedreason, somewhat more problematic than in the case o f child allowances, and cited by a slightly larger fraction of the poor, and once again the poor were more likely to report the need for connections. Again, about 11 percent o f poor households claimed that rejections were the result o f the difficulty o f the application process. Thus, lack of consideration of applications-perhaps as a result of limited fiscal resources or the need for connections, difficulty of application process and potentially wrongfully failing the means test appear to be potential reasons for errors o f exclusion at the stage o f awarding benefits. As noted above, and contrary to the child allowance program, exclusion errors appear to be greater for the low income benefitprograms. Multivariate analysis of determinants of probability of beneJit receipt: As in the case of child allowances above, we use multivariate analysis to determine which factors affect the probability o f receipt, and the nature o f self-targeting in the process (see Mehra and Rashid, 2002 for detailed results). As above, two types o f determinants are included in the estimation: welfare indicators and indicators o f horizontal equity. Again, as inthe case ofthe child allowances program, the results show that the probability of receiving the low income benefit is inversely related to household welfare.204Households who do not own transport, who have more *03 There is not a systematic variation in acceptancerates across rural vs. urban areas, age or education ofhouseholdhead (Table A17) *04 Adding other welfare indicatorsreducesthe significanceandmagnitude ofthe impact of food consumption (potentiallybecausethese other variables capture some o fthe consumption impact). 125 children and, who have a higher number o f unemployed members are more likely to receive benefit. The probability of benefit receipt i s inversely related to household food consumption, but weakly so. Other welfare indicators for households (i.e. livestock, land and durable goods) do not have a significant impact on benefit receipt. The probability o f benefit receipt does not differ significantly across rural and urban areas or across most regions relative to Karakalpakstan (the base), controlling for wealth. The only exception is in the Ferghana Valley region which has a significantly lower likelihood of benefit receipt relative to the base region. To obtain greater insight into the process of targeting, we once again disaggregate the probability o f receipt into the product o f the probability o f knowledge, conditional probability o f claims, given knowledge, and conditional probability o f award given claims. The results from this disaggregation show that there is self targeting of knowledge o fthe program. Households with lower per capita consumption, those who do not own transport; and those with more children are more likely to know about the program. Low welfare households also have a significantly higher propensity to self-target through benefit claims. The probability o f benefit claims, given program knowledge, is inversely related to household wealth (proxied by household consumption, ownership of durable goods, and mechanized transport) and is positively related to family size (number o f children). Finally, benefit awards are also inversely related to low income households-with no ownership o f mechanized transport and with a higher number o f unemployed. These characteristics, rather than differences in household consumption, appear to play a part in award determination by the Mahallah Committee. The reason for the low probability o f benefit receipt for the FerghanaValley is not clear. The region has about the same knowledge o f the program than the base region, and a significantly higher propensity to claim benefits. The only explanation i s a (weak) lower probability o f award receipt in the region. On the other hand, the disaggregated analysis shows that regions differ intheir propensity to know about the program; as well as by their propensity to claim benefit, given program knowledge. There are some, although weaker regional differences, also inthe probability o f benefit award. For households who receive it, the amount of benefit is quite signifzcant. The average level o f low Income Mahallah benefit was about 4000 sums in 2000, around 40 percent o f the average wage in that year.*05 This level o f benefit i s quite high relative to the average wage in 2000 (and relative to the minimum wage) and could result in work disincentives. The benefit comprises approximately 17 percent o f total consumption of households that receive it. For households from the poorest income quintile, the benefit comprises nearly 30 percent o f total consumption. The average level o f benefit received by the household does not appear to be related to the level o f income, which is somewhat surprising, given that it is means tested (Table 3). Despite a reasonable amount o f benefit, the low coverage o f the program means that it has a very small impact on poverty alleviation. The removal o f the Low Income Benefit would only increase poverty by only .5 percent overall. Not surprisingly, the targeting eficiency of Low Income Benefit expenditures is weaker than the child allowances program (figure 4). This result holds in both urban and rural areas. The relatively small difference in coverage rates between rich and poor, and lack o f variation o f the benefit with respect to income likely explains this result. Benefits to children under 2: The coverage o f this program is only slightly lower than the child allowance program: approximately 12 percent of all households receive benefits. Coverage rates fall slightly for higher income quintiles, but not as steeply as in the case of child allowances. Because o f the correlation of these characteristics of households with the number o f children, coverage rates are higher for younger and less educated household heads, and in rural areas. The average per capita benefit is 580 sums, slightly higherthan the Mahallah Child Benefit. Given the significant amount o f benefit, child benefit for under 2 comprises important share o f total cash transfers (9 percent) received by households, particularly by younger households ~~ ~~~~ 205 According to FBS data, the average households monthly benefit is also about 4000 sums per month. 126 (who do not receive many other benefits). The benefit also comprises a significant share, about 13 percent o f total consumption o f all households, but comprises a higher share of total consumption for the poor, younger and for less educated (vs. more educated) household heads. The share of this benefit in total cash transfers received by the households; as well as total consumption are only slightly lower than in the case of child allowances and targeting efficiency is between the low income and child allowance program. In 2000/01, the year of the survey, this program was not means-tested, a change which has been introduced since 2002. The poverty reduction impact o fthe program is nearly comparable to the child allowances program. SocialPrivilegesand Utility Subsidies According to the Household Survey data, 12.1 percent of urban households, and 7.6 percent of rural households reported receiving electricity subsidies while 7.2 percent of urban households and 4.9 percent o f rural households were entitled to privileges on gas (Mehra and Rashid, 2002). The estimated coverage from official information on the list of privileged groups and average households size, suggests that almost one third of the total population may benefit from housing privileges. Not surprisingly, given their design, coverage rates are higher for the non-poor (Rashid andMehra, 2002). Figure 5. Targeting Efficiency ofUtilityExpenditures Targeting Efficiency of UtiIity Expenditures 100 90 80 70 60 50 1 40 30 ~ 0 10 20 30 40 50 -Equality PCFoodConsumption +Rent - - - 60 70 80 90 100 +Electricity +Gas ~ As noted, the benefits are generous as they cover halfo f costs. The savings reported by households from their access to the subsidies are quite significant. For gas and rent, these subsidies are roughly in the range of the per capita disability and survivor pensions benefits. The per capita subsidies are higher for the non-poor, and for urbanresidents. It i s nottoo surprisingthen, that the targeting efficiency of the subsidies for electricity and gas are disproportionately distributed (in value) towards the non poor (Figure 5). The subsidy that accrues most disproportionately more to the rich, is rent, once again because the well off are more likely to live in apartments (chapter 2). In addition to poor targeting, such subsidies create price distortions and reduce incentives for efficient use for those who benefit. In addition, housing subsidies accrue to households through low tariffs which do not cover full operating costs, and through various forms o f cross subsidization. As mentioned earlier, such subsidies through low prices are largely untargeted subsidies, that, because higher utilization of these services by non-poor, in fact tend to be biased towards the better off. Once again subsidies provided in this manner distort energy prices and therefore incentives for efficient use. 127 Most FSU countries have abolished the system of social privileges, and have raised utility prices. The President of Uzbekistan has recently approved a decree on reforms in the communal sector which anticipates an increase inutility tariffs between 2003-06 by 2 to 3 times, depending on the region. Itwould be important for Uzbekistan to phase out the system of social privileges and subsume the utility subsidies into a more transparent and targeted social assistance program. Since the preparation of this report, the Government has taken important steps inthis direction. Economic Reform and Social Protection In addition to protecting from old age, loss of job or protection from abject poverty, social protection programs can play an important role in mitigating the negative impacts on some groups o f economic reform. This report has, particularly in chapters 3 and 4, identified many key reforms that would lead to improvements inproductivity and translate into higher incomes for many inthe medium to long term. These include, for example, liberalization of the foreign exchange and trade regime, serious enterprise restructuring, price liberalization including for cotton and grain, gradual elimination of implicit subsidies, including through low prices or arrears in the energy sector. While these reforms will contribute to higher incomes, there may be some groups who will lose out inthe short term and will need some protection. We saw in chapter 4, that liberalization of cotton and grain prices through raising procurement prices and liberalization of the exchange rate regime, would benefit the majority of the rural population, especially poor farmers. However, inthe short terms some groups, for example net consumers of wheat (in the urban areas, but also in rural areas) may face higher prices. Similarly, unification of the foreign exchange regime could result inhigher prices for essential commodities consumed by the poor. The actual impact on cost o f living can be complicated to estimate as it depends on extent of price distortion (for example, we saw in chapter 1 that despite price controls, many consumer goods are resold to final consumers through market channels at the black market exchange rate)206,as well as the importance o f the good in the consumption basket of different households. Moreover, as prices rise, consumers adjust their consumption and use cheaper substitutes. To estimate the short-run impacts of a devaluation on food prices, Murrugarra and Storozhuk (2002), estimated the exchange rate elasticities o f prices for different food groups (based on two recent episodeso f devaluation). Taking into account the consumption structure and differences in the importance of non-purchased consumption for households, they find that the immediate cost of living increase will be roughly equal (at about 50 percent) for all income quintiles inurban areas, but will be higher for the amongthe poorer quintiles inrural areas. Cost o f living increases, will, as expected be higher inurban areas where there i s very little reliance on own production207.While the rural poor will also gain from higher incomes from agricultural reforms, some vulnerable groups may need short term protection from price increases. Other losers in the short run will be those groups who lose their jobs as a result o f serious enterprise restructuring that could result from the imposition of hard budget constraints on enterprises, or from the exchange rate reforms. To reduce social costs in the short run, these implicit and implicit subsidies may be reallocated to strengthening the social safety programs, including the existing unemployment protection programs and the development of labor redeployment programs to deal with mass layoffs or continued restructuring. As mentioned above, the President has announced substantial increases in energy prices, which 206The experience o fmany transition countries was that many prices had already adjusted so that impacts on cost of living from adjustment of prices of essential goods was smaller than expected. 207 It shouldbe notedherethat the magnitude of the impacts on cost of living are lower when a different methodologyis used, as inMilanovic (2002). There the author looks at the share of seven essentialcommodities, incash food expenditures, and estimateprice increases based on a comparison of the prices ofthese seven goodswith prices in neighboringKyrgyz republic. 128 is very welcome as across the board utility subsidies do not benefit the poor. To protect the poor and vulnerable from the negative impact of higher utility prices, however, it would be important to design a more transparent social assistance program to replace the existing system of privileges. ConclusionsandPolicyOptions The performance of different social protection programs in proving security is mixed. Through its extensive coverage and generous benefits, the pension system has protected pensioners from old age poverty and has reducedpoverty inthe households inwhich they reside. Nevertheless, some pensioners remainvulnerable to poverty and the elderly who do not receive pensions are at high risk of poverty. The poverty protection provided by the pension system has come at high current and future costs. Payroll taxes required to finance the program are high and contribute to a low compliance, high tax ratetrap, by reducing the demandfor labor and increasingthe degree of informality. Despite a favorable demographic structure, the gradual aging of the population will put pressure on the future financing of the pension system especially after 2010, limiting the ability ofthe system to provide effective old age support inthe future ina financially viable manner. The unemploymentbenefit program has had very limited impact on protecting the unemployed. It has very low coverage and provides little, if any poverty protection to households. The average level of benefit provides adequate protection without inducing work disincentives, but benefits are very low for the majority of recipients who receive the minimumbenefit. Other design features of the present program, may also pose incentive problems, and limit the effectiveness of the program. The majority of the expenditures of the employment fund are in fact allocatedto active labor marketprograms, and among these, mainly job creation programs in the form of subsidized credits to businesses through commercial banks. The evidence on the effectiveness of ALMP programs in Uzbekistan, based on FBS and administrative data is mixed: while Employment Offices are able to place a majority of registered unemployed injobs, the FBS data show that many unemployedare not very satisfied byjobs found throughthe EmploymentOffices and many do not find the service very helpful Evaluating the performance of the two means tested social assistance programs, child allowances and low income benefits, against their stated objectives, is complicated by the decentralization (within broad guidelines) of eligibility conditions. Nevertheless, useful insights are obtained, particularly on relative performance of the programs.Both programs are well known. The coverage of the child allowance program is significantly higher than the low income benefit program (perhaps because of lower funding levels associatedwith a program of last resort), but both groups have higher coverage of the poorest groups. The higher coverage of the poor results in higher knowledge of the programs among the poor but also self targeting in the process of claiming benefits. In the case of the child benefits program, the mahallah committees appear to be better able to screen out better off applicants, than in the case of the low income benefit program. Despite the decentralizationinthe determinationof benefit levels (and the larger number of childreninpoor households), benefit levelsdo not vary significantly by income level. Within the caveats stated above, there is evidence of errors of inclusion and exclusion at claims and award stages for both programs, and particularly for the low income benefit program. Because of the much higher coverage of the poor relative to non-poor, the child allowance program is the most efficiently targeted of all social protection programs, while the low income program does not do as well. Both programs provide adequate benefits for the poor who receive them, but the higher level of benefit relative to average wage inthe case of the low income benefit program is more likely to induce work disincentives. The relatively poorer performance of the low income program may be due to the greater difficulty associatedwith discerningthe needy (without additional characteristic based screening), but could be due to pressure on the committees from more influential local households. For both programs, there are horizontal inequitiesbetweenregions in benefit provision, resulting from differences in program knowledge, propensity to apply and likelihood of award. Processing delays are higher in the case of the low income benefits program, but in both case poorer 129 households are disproportionately affected. Nevertheless, both programs have the confidence o f many households. The coverage, and adequacy o f the cash benefits program for children under 2 is only slightly lower than the child allowances, but because it was not meanstested, it was less well targeted to the poor. It does provide a significant transfer to younger, poorer and less educated households with small children and its poverty reduction impact i s comparable to the child allowance program. Its targeting efficiency and poverty reduction impacts have likely improved further with the recent policy change that introduced means testing in this program. The recent system of social privileges in the form of discounts for utility fees and tariffs is expensive and poorly targeted. It provided significant savings to those who benefit (a large fraction of whom are teachers and health care workers). Additional, largely untargeted, subsidies are also provided to the populationthrough low prices. Both types of subsidies distort incentives for efficient use, while the system o f privileges could be incur significant administrative costs. Pensions: Reduced payroll taxes for pensions could contribute to increased employment opportunities inthe formal sector and reduction in informality in the economy setting of a virtuous cycle. It needs to be accompanied, however by other measures that could ensure the solvency of the pension system. These include parametric changes in the pension system, such as increasing and equalizing retirement ages, eliminating privileges, and equalizing years o f service requirements, which could help reduce fiscal pressures on the pension system, without compromising its ability to provide old age security. Such changes, by tighteningthe linksbetween contribution and wages and reducing re-distributive elements, could additionally increase incentives to contribute to the system. The projected deterioration in the fiscal situation of the Pension Fund as a result of the aging population especially after about 2010, means that these changes are important to prevent future deficits or highertaxes. The Government has recently taken several steps to improve the administrative efficiency o f the pension system, for example through transfer o f collectionsto the Tax Authority. The pension fund now The pension fund now only handles registration of pensioners, collection and maintenance of a data base on pensioners, benefit calculation and other client services functions. The capacity of the pension system to efficiently perform these functions needs to be strengthened. Specifically, data collection needs to be streamlined and automated, benefit calculations also need to be automated, a unified account number (that allows pensioner records to linked to tax and other benefit systems) needs to be created, methods for reconciliation o f contribution information between the tax authority and the pension system need to be developed. Some o f these changes are being piloted in Samarkand region. The lessons learned from the pilot can potentially serve to improve the administrative capacity o f the pension system nationwide, and help increase the quality o f services offeredto pensioners. Unemvlovment Protection Programs: Uzbekistan may wish to consider a move to a flat unemployment benefit program, funded from (adequate) general revenues. This would permit it to better focus its scarce benefit expenditures on the poorest unemployed, and to reduce its administrative burden. It should be noted that the `insurance nature' of the benefit may not be particularly relevant in Uzbekistan given the informal nature of the labor market, limited administrative capacity, and lack o f reliable wage information. Flat benefits, such as those introduced in Poland provide a lump sum, and therefore, progressive, benefit to eligible beneficiaries might be more suitable in Uzbekistan. There may, however be practical difficulties in establishing a suitable flat rate inthe context of the Uzbek labor market with large inequalities inwages. Ifthe unemployment benefit program is to continue in its broad present structure, as an insurance program funded from payroll contributions, some design changes are recommended. Unemployment benefits should be restricted to laid off workers, while benefits to new entrants or re-entrants into the labor market should be 130 phased out. . It would also be important to allocate adequate funding for unemployment benefits, for example by shifting the composition of expenditures of the Employment Fund away from costly micro-credit programs, and limiting them to a few effective ALMPs. Impact evaluations of existing ALMPS in Uzbekistanare recommended. To address unemployment problems in particularly depressed regions or rural locations, limited public works programs could be used; including expansion o f those already in existence but found to be effective through impact evaluations. In general, however, international experience has found that job counseling and job search assistance programs are typically the most cost-effective ALMPS, although o f course their usefulness i s greater in economies where the labor market i s competitive, job creation rates are improving, and there is an increase in economic activity. The best policies for job creation and employment growth are those that ensure that the basic institutions of a market economy are inplace, provide macroeconomic stability and an investment climate and business environmentjob creation by the private sector, and labor market policies that provide labor market flexibility-as discussed in chapter 3. Employment programs can only be effective when such fundamental policies are inplace. Social Assistance Programs: The current system o f provision of meanstested benefits through the Mahallah committees has many successful elements, but could be strengthened in many ways to improve their targeting efficiency and administration. Errors of inclusion and exclusion could be reduced through: simplification of the applications process; assistance to poor and less educated individuals in filling out applications and obtaining correct documents; greater positive publicity about fairness of the process and efforts to reduce the stigma associated with the application process. Also important would be a further examination o f the means test to determine how best it might capture the non-poor (for example, whether additional characteristics correlated with poverty such as education o f the head should also be included) and closer checking of claims at the application stage and ex post monitoring o f awards. Further examination i s needed to understand and addressthe invarianceo f benefit levels with income. Cross regional targeting could be improved through improved channeling of information across and within regions, and by examining why households in particular regions are more likely to claim and receive benefit than others. The variance in the performance o f the mahallah committees could be one reason for this variation, but the Government should examine the mechanisms for allocating social assistance funds across regions, to see ifthe linkage with needs could be strengthened. Stable and adequate financing for eligible households are essential for effectiveness of social assistance programs and in this context the declines in resources and beneficiaries in recent years are o f concern. The reasons for the higher processing delays in the case of the low income benefits program needs to be better understood and addressed. The limited financing, low coverage and inefficient targeting o f benefits to the poor by the low income benefits program calls the future of the program to question. The Government may wishto consider whether the program should be phased out. Onthe other hand, the low income program may be useful incatching all poor who are not covered by child allowances and other social protection programs. Inthis case, serious efforts need to be made to better understand why program administration and targeting are worse than that of the child allowance program, which is also administered by the same Mahallah, and addressing the issues identified above. Finally, the recent step to improve to introduce means testing into the program o f cash benefits for children under 2 years should have further improved its targeting efficiency and poverty impact. Integrating this program with the child allowance program, which provides means tested benefits to all families with children under 18, would further reduce duplication o f benefits, permit greater coverage of poor, and save administration costs. Social PriviIeges: Most FSUcountries have already eliminated the system o f social privileges. These should also be phased out in Uzbekistan. For teachers and health care workers, who are poorly paid, it would be preferable to subsume me the subsidy in the form o f improved pay. This would be more transparent, easier to 131 administer, while reducing distortions for inefficient use. Recognizingthis, on April 1, 2003, most privileges for employees, such as teachers and medical staff were abolished and replaced with cash compensation inthe amount equal to 45 percent of the minimumwage208. Alleviatinp Short Term Social Costs of Reform: In addition to the reform options for unemployment protection programs suggested above, Uzbekistan may wish to design labor redeployment programs to provide a safety net for those who are laid off duringenterprise restructuring. International evidence suggests that labor redeployment programs, which include post and pre-layoff services in combination with benefit packages (unemployment benefit, severance, etc.) can help alleviate the social costs associated with layoffs and enterprise restructuring. In many transition countries, Active Labor Market Programs, particularly job counseling andjob search assistance, but also some targeted training and retraining programs have been used inconjunction with unemploymentbenefits and severance pay to facilitate layoffs. This combination of active and passive programs might also be used in Uzbekistan to reduce any social costs of restructuring in particular industries or as a general policy. While this combination o f programs can be expensive inthe short run,they areoftenjustified becausethey promote longterm economic efficiency. As discussed above, the existing costly and untargeted system of provision o f utility subsidies through privileges, or through across the board price subsidies should be replaced with a more transparent and better targeted social protection program. Several mitigating mechanisms could be considered to compensate the poor for elimination or reduction of utility subsidies, including a) life-line tariffs for utility services with metered or relatively easily estimated consumption, where the price subsidy is restrictedto the initial block o f consumption (called the basic need level); b) a notional burden approach (most popular in FSU countries), which limits the burden placed by utility expenditures on household budgets, for example by compensating low-income households for a portion o f rent and public utility services exceeding usually 15 to 30 percent of family income (based on normatives for utility consumption, e.g. apartment space); c) earmarked cash transfers can be provided to poor households to pay part o f their utility bills; d) non-earmarked cash transfers targeted to poor households to cushion the impact o f utility tariffs increases209.Inview o f the role played by the Mahallas at present in identifying poorest inthe community for the allocation o f social assistance benefits, it mightbe rational to enhance the existingsystem and attach some additional functions while strengthening the Mahallah's capacity in provision of targeted social assistance210.Infact, since 1999,Mahalla committees are already authorized to collect taxes from citizens for uti1ities2l1. Similarly protection from cost of living increases may be targeted through the Mahallahs to most vulnerable households, on a one-off basis, or for a short period o f time. 208This importantpolicy change has recentlytaken place, as the reportgoes to publication; and after its preparationand discussion. *09 (See Reference: World Bank (2000). Maintaining Utility Services for the Poor: Policies and Practices in central and eastem Europeand the Former Soviet Union. Washington, DC.) 210Ifutilitiespaymentsare donefully andtimely, the Mahallacommitteeshavea rightto keep 10percentofcollected taxes for their own needs, inparticular for the payment of utility allowancesto the poorhouseholdsand the development oftheir own materialbase. 211Typically housingbenefitsare grantedwithin the social limit of total living space andutilities consumption standards, while their above standardconsumption will be paid accordingto the effective tariffs 132 Chapter 8: IMPROVING POVERTYMONITORING INUZBEKISTAN This report has relied heavily on the Family Budget Survey, complemented by other data sources (including administrative data on public expenditures, for example, as well as other specialized surveys) to provide the first national level assessment of poverty and living standards and to examine the linkages with Government policies and programs. Based on this analysis, the report provides several policy recommendations to improve the living standards of the population. It has also, of course, identified many issues, for further analysis and study. Moving forward, it would be important to continue to monitor poverty and living standards on a continuous basis, and to examine the impact of government actions to reduce poverty. This would provide information needed to improve the effectiveness of the Government's social and economic policies. I t will be essential to develop and continuously improve a National Poverty Monitoring System in Uzbekistan. Key elements of a poverty monitoring system include: 0 Timely and accurate annual national accounts 0 Current public expenditure statistics at adequate levels o f disaggregation, for example by programs and regions 0 Accurate consumer and producer price statistics 0 Upto date population census (the current international recommendationis every ten years) 0 Regular, population based surveys to monitor household incomes and expenditures, and regular labor force surveys. 0 Regular population based surveys to monitor non-income dimensions o f well-being including health and education 0 Administrative data on public programs (education, employment services for example) including data on clients and services provided, as well as project specific information on administrative costs, and on targeting. 0 Longitudinal studies (panel surveys) which follow samples o f the population over periods o f time to examine changes intheir status 0 Participatory poverty assessments or qualitative surveys Developing and Improving the Poverty Monitoring System NationalAccounts, Price and Budget Data: Inthe past few years, the Government took steps to improve the reliability of its national accounts and consumer price data with technical assistance from the IMF and World Bank. However many problems still remain with the current data212 which results, for example, in divergences between Government estimates of growth and inflation, and alternative estimates from the IMF, for example, as well as inconsistencies between government estimates of growth and inflation and other economic indicators. It will be essential to continue and step up improvements in the national accounts and price data, to enable more accurate assessment o f economic developments in Uzbekistan. The Government also been improving its public expenditure data, by consolidating Government operations and adopting new budget classifications. However, further improvements are required to provide more timely data on public expenditures, at appropriate levels o f disaggregation, by program and by regions. Establishment o f the Treasury under a the World Bank project would permitthe Ministry o f Finance and line ministriesto receive '12 These include, for national accounts data: lack of GDP estimates at constant prices, incorrect assessment o f the deflator, especially for retail trade, problems in calculating trade margins for retail trade, problems in calculating subsidies in constant prices, and improper coverage of the informal sector. Similarly consumer prices suffer from several problems, including recordingiimputingof prices of non-seasonalgoods, goods that are temporarily unavailable or whose sources o f supply change frequently, etc. See IMF Recent Economic Developments 2000, 2001 for further details. 133 more up-to-date and detailed information on budget execution and thus contribute to improved monitoring o f social spending. UpdatedPopulation Census: The last population census was conducted in 1989. Since then there have been several demographic change?13. A new populationcensus is being plannedfor the next few years and will be essential to providing an up-to-date picture o f the population and its structure, and can also be usedto provide other useful information at a national level. A new population census i s also essential to improving the reliability o f the sampling frame for population based surveys such as the Family Budget Survey or Labor Force Surveys. A population census may also be used in conjunction with concurrent population based surveys to conduct relatively cost-effective "poverty mapping" exercises, which could permit the design of targeted policies or programs, for example to improve access to water or to schools inpoor districts. Population Based Survevs: Familv Budget Survev, Labor Force Surveys, and Demograuhic and Health Survevs:The mainstay o f the poverty monitoring system will continue to be the Family Budget Survey. This i s the only household level data collection effort at the national level that i s expected to be continuous. Although the Family Budget Survey was not initially designed to measure poverty per se, it has been shown to be a useful tool for increasing our understanding o f poverty in the country. Strengthening the Family Budget Survey to provide regular and reliable poverty monitoring would be an essential and basic action that should and could be undertaken quite quickly. Several changes are recommended, and clearly the advantages and disadvantages of any changewould have to be weighed, but it does a ear reasonable that these could be done without endangering the other purposes for which this survey exists2 . v To improve the adequacy o f the FBS for poverty measurement and monitoring, three main changes are needed: 0 Enable the construction of a total consumption value for each household. This would permit a reliable analysis o f inequality, depth, severity, and would also provide a more robust indicator of poverty. It would also facilitate international comparisons of poverty. To do this requires some changes to the questionnaire, includingthe inclusiono f questions to permitthe estimation of the use value of housing and of durable goods. 0 Change the periods of reference for non-food goods so as to capture, not only national means, but also to capture accurately the spending of each household. 0 Expand the present, relatively limited, information on some non-income dimensions of poverty, especially education and health. While some modifications were made to include some basic questions inthe 2000-01 survey, and insome cases, ad-hoc modules were added, the analysis hasrevealed several areas where the information could be improved and where regular, perhaps annual, information would be useful. In the case o f education, the analysis revealed problems with the current Family Budget Survey in gauging true participation (enrollment as well as actual attendance) in education which needs to be addressed urgently. Additional areas for strengthening include, for example, the introduction of questions on private costs in education, improved information on utilization of health care services, as well as longer reference periods for hospital expenses, and more information on actual attendance in educational institutions (see background papers o f this report for some of the issues and recommendations). The analysis also revealed some problems with the modules on social protection programs and the way they were administered, for example on the potential confusion between the mahallah child allowances program and universal child allowances, which could be addressedright away. ~ 213 These include drops in the birth rate, for example, from 33.7 per thousand in 1990 to 21.5 in2000 (the total fertility rate dropped from 4.07 in 1990 to 2.72), and in mortality rates accordingto official statistics, resulting in a decrease in the natural growth of the population. There was also substantial out-migration of the populationespecially inthe early 199O's,particularly from the returnof Russianspeaking populationto Russia, as well as some reversemigration through the repatriationof titular nationalities from Russia and other FSU states. Net migration over the nineties, however, was negative at about 600,000 people. 214 For example, estimatingthe weights for the consumer basket. 134 Inimplementingthese changes, however, some effort is neededto determine which should be 'core' questions (included every quarter for all households) and which questions or modules could be administered only on quarter or only for a subset of households. This i s an important set o f decisions and needs to be coordinated with line ministriesto ensure the key questions are asked on a routine basis and to identify specific issuesthat can be addressedperiodically. Inthe case of health, the Government has an extensive program of data collection from health care facilities on health outcomes, as well as utilization. These are very valuable. This report, however, has demonstrated the usefulness of additional information that can be obtained from population based surveys such as the Demographic and Health Surveys, as well as the Family Budget Survey. Inparticular, these surveys, permit the inclusion o f people who do not regularly utilize health facilities, which could be one important reason for the difference observed between administrative and survey based information (for example on births at home, infant mortality2'5, etc.) The DHS surveys examine many specific health nutrition and population outcomes, as well as utilization o f services. After the DHS in 1996,a follow up DHS i s now being carried out, which has far more extensive epidemiological information, as well as other information on immunization, sanitation, hygiene and breastfeeding practices, etc. Such surveys, should become a part o f a regular monitoring system, say every 3 to 5 years. In view o f the importance o f monitoring such outcomes among poor populations, efforts should be made to improve the modules relating to household income/consumption, preferably using one that i s comparable with the improved module inthe Family Budget Survey. Chapter 3 of this report referred to the problems with labor force data, and emphasized the importance of regular labor force surveys as an essential tool to monitor and analyze information on employment, underemployment, and unemployment (including seasonal, occasional and informal employment), hours of work, etc. (see box 1 in Chapter 3). The Government has piloted a labor force survey with a view to full implementation. Full implementation should take place quickly, and regularly. The Government may also wish to consider the inclusion o f an improved labor module in the Family Budget Survey, on a rotational or regular basis. This would permit the monitoring of labor market outcomes by poverty status. Other data sources and svecial studies: The analysis of this report has also identified other areas where additional data would be important or useful. These include establishment or firm level surveys maintainedand utilized for analysis at the level o f the firm, rather than aggregated (as is the present practice). Such data can provide very useful information on job creation, destruction, firm productivity, growth, ownership, etc. which are essential to answering key questions for industrial policy development (see box 1inChapter 3). To understand better, farm productivity and employment outcomes, and the linkages with agricultural policies, farm level or agricultural enterprise surveys are important. This report utilized the RESP Survey to analyze some o f these issues. However that was a purposive survey and was not meant to be representative of the rural farm sector at the national level. Also useful would be periodic surveys, perhaps surveys of providers or facilities, on formal and informal provision o f services (such as health care) which would facilitate information on the exact size o f the private sector in health as well as qualitative surveys (such as the one carried out for this report) on informal payments in education and health care facilities. Such qualitative surveys are not presently used by the Government, but the analysis o f the health sector showed that they can be a very valuable complement to quantitative surveys and facilitate an understanding of issues that are not easily captured by quantitative data sources while identifying potential areas for investigation in quantitative surveys, perhaps at the national level. '15Another importantreason is differencesindefinition. 135 This report has also identified other special studies, such as an investigation into the decline in higher education enrollments, into regional differences in the performance of social assistance programs and into regional disparities in education and health outcomes, etc. Also important would be an impact evaluation o f the Active Labor Market Programs, which are costly and whose true (net) impact on employment needs to be assessed. There are several areas where administrative data could be further improved and it is recommended that international conventions or practices be adopted. These include, for example, official data on labor statistics from the Balance o f Labor Resources, definitions for infant and child mortality, etc. The Government is increasingly adopting internationally accepted methodologies to improve these, which are welcome and positive steps. Ensuring usefulness of the data through capacity building and widespreaddissemination Continued utilization o f data sources for poverty monitoring and analysis i s essential both to ensure that the data serves the purpose for which it was collected, as well as to obtain continuous feedback to improve the quality and relevance o f the data. Thus inaddition to improving and buildingthe data sources, a program for poverty monitoring should include (i)a program for capacity building for key users o f the data within and outside the Government; and (ii) a program for dissemination o f analysis and results as well as of raw data to motivated researchers within and outside the Government. Given the lack of tradition in the use o f unit level or micro-data, particular efforts will need to be paid to train key users within and outside Government in the use o f micro data. Much of the analysis of this report was done in close collaboration with key government officials, which permitted both a reality check on the analysis, but also permittedthe sharing o f methodologies on the analysis of unit record data for poverty and policy purposes. Moving forward, it will be important to build on this, with continuous opportunities for government officials to use and upgrade their capacity for poverty analysis. A key issue would be sufficient capacity within line ministries. Also essential will be sufficient inter-ministerial and interdepartmental coordination- for example between line ministries, core ministries (such as Finance and Ministry of Macroeconomics and Statistics), and the State Department o f Statistics, which will ensure adequate feedback and discussion between policy makingunits and data collecting units. Finally, expanding the users of this information beyond the government and including them in capacity building efforts would facilitate the availability o f new analyses by innovative researchers outside the Government that could be useful to policymakers. It would also widen the sources o f useful feedback that could further improve the quality and relevance o f the data. Most importantly, it would be one step further towards facilitating meaningful exchange with the public and with groups outside the government on poverty monitoring methodologies and analysis, but also on the progress on poverty outcomes, and on the success or lack thereof of policies aimed at improving living standards. Thus essential to a successful poverty monitoring system, would be a plan or program for regular dissemination o f key poverty indicators, as well as o f raw unit record data for poverty analysis. 136 Bibliography Beck, T., A. Demirgug-Kunt and V. Maksimovic (2002). "Financial and Legal Constraints to Growth: Does Firm Size Matter?" Processed. Berryman, Hidden Challengesto Education Systems inTransition Countries, The World Bank (2000) Blanchard, O., (2000). "Employment Protection, Sclerosis, and the Effect of Shocks on Unemployment." Lional Robbins Lecture. London School of Economics. October 2000. Broadman, H. (2000). "Competition, Corporate Governance and Regulation in Central Asia: Uzbekistan's StructuralReform Challenge." Working Paper 233 1. The World Bank, WashingtonDC. Cashin, C. 2001. Access to Health Care inRural FerghanaOblast, Uzbekistan.Study preparedby ZdravPZus. Study preparedfor USAID: Tashkent. Common Country Assessment of Uzbekistan, United Nations Resident Coordinator System, Uzbekistan (2001) Fields, G. and G. Pfefferman (eds.) Pathways Out of Poverty. International Finance Corporation. Washington DC. forthcoming. Gang, I. and R. Stuart (1999) "Mobility where mobility is illegal: Internal migration and city growth in the Soviet Union," inJournal of PopulationEconomics, Vol. 12(l), 117-134. pp. IMF,Uzbekistan, Recent EconomicDevelopments(2000) IMF, Republic ofUzbekistan, StatisticalAppendix, 2002 Japelli, T. and M. Pagano (1999). "Information Sharing, Lending and Defaults: Cross Country Evidence." Working PaperNo. 22. Center for Studies inEconomicsand Finance. Salerno. Kaufman, D., A. Kraay and P. Zoido-Lobaton (2002). "Governance Matters 11: Updated Governance Indicators for 2000/2001. Policy Research Working Paper2772. The World Bank. Washington DC. Langenbrunner et al. (2002), Uzbekistan Living Standards Assessment, Health, Nutrition and Population. Background paperpreparedfor the UzbekistanLiving Standards Assessment. Lubin, N.(1984) Labour andNationality in Soviet CentralAsia. An uneasy compromise.Princeton: Princeton University Press. Kropp, Lundell and Shamsiev (2002) Rural Welfare in Uzbekistan.. Background paper prepared for Uzbekistan Leaving Standards Assessment. MacIsaac, D., Rama, M. (1997) "Determinants of hourly earnings in Ecuador: The role of labor market regulations," inJournal of Labor Economics; Chicago; Jul 1997; Vol. lS(3). Part 2. McLiesh and Tracy (2002), "Poverty, Firms, and the Business Climate in Uzbekistan," Background paper preparedfor the UzbekistanLiving Standards Assessment Mertaugh (2002), "Education and Living Standards," Background paper prepared for Uzbekistan Living Standards Assessment Milanovic (2002), Background paper prepared for the "Uzbekistan, County Economic Memorandum, World Bank, 2003 Murrugarra and Storozhuk, "The Impact of Exchange Rate Devaluation on Cost of Living: a Preliminary Assessment". Backgroundnote preparedfor Uzbekistan PovertyAssessment (2002) Murrugarra and Tracy (2002) "Uzbekistan: Poverty and Labor Markets." Background paper prepared for Uzbekistan Living Standards Assessment 137 Nishino, Y. 2002. "A Qualitative Assessment of People's Interactions with Health Institutions: Formal and Informal Payments and Individual Histories of Coping with Economic Burdens." A Study prepared by Challenge One Associates for the Uzbekistan Living Standards Assessment, The World Bank: Washington D.C. Novak, J., Goldin V., Langenbrunner, J., Borowitz, M.,et al. 1996. Household Survey of South Kazakhstan Oblast.Preparedfor Abt Associates: Bethesda, Maryland Olson-Lanjouw, Jean and Peter Lanjouw, 2001. "How to Compare Apples and Oranges: Poverty Measurement Based on Different Definitions of Consumption", Review of Income and Wealth, Series 47, Number 1, March 2001: pp.25-42 Paci, P. (2002) Gender in Transition.The World Bank, Washington, D.C. (2002) Rashid and Mehra (2002), "Social Protection in Uzbekistan," background paper prepared for Uzbekistan Living Standards Assessment Reporton Status of Women inUzbekistan, UNDP, Tashkent (1999) Republic of UzbekistanIrrigation andDrainageSector Study vol. 1,Tashkent, World Bank (2000) Rosenberg, C. and de Zeeuw, M., Welfare Effects of Uzbekistan's Foreign Exchange Regime, IMF Staff papers (2001) Sari N., Langenbrunner, J., and M. Lewis. 1999. Out-of-Pocket Payments for Health Services, European Health2(4), pp. 28-33. Save the Children (2002) Child labor in Uzbekistan. Tashkent. Save the Children and Tahlil, Child Labor in Uzbekistan, (2002) Schiffer, M. and B. Weder (2001). "Firm Size and the Business Environment: Worldwide Survey Results," IFC DiscussionPaper, No. 43. International Finance Corporation, Washington, DC. Scott and Mehra, A Description of Poverty in Uzbekistan, May (2002), Background paper prepared for UzbekistanLiving StandardsAssessment SituationAnalysis of Basic Education inThe Central Asian Republics andKazakhstan, UNICEF (1999) TACIS, Improving the Productivity of Water inCentral Asia, Tashkent (2000) The World Bank, Making Transition Work for Everyone: Poverty and Inequality in Europe and Central Asia (2002) Thurman and Lundell (2001), Agriculture in Uzbekistan: Private, Dehqan, and Shirkat Farms in the Pilot Districts of the Rural Enterprises SupportProject UNESCO-UNICEF, Monitoring LearningAchievements:Uzbekistan(1999) Vodopivec, M., A. Worgotter, and D. Raju (2000), "Unemployment Benefits in Transition Economies: Lessons LearnedandPolicy Options." Washington, D.C. Voices of the Poor, Uzbekistan Country Report (2000), "Consultations with the Poor", Participatory Poverty Assessment inUzbekistanfor the World DevelopmentReport 2000/01; World Bank Wagstaff (200l), Paying for Health Care: Quantifying Fairness, Catastrophe, and Impoverishment, with Applications to Vietnam, 1993-98 World Bank,Capacity Building inEducationFinance, Project TA2948-UZB, Final Project Report World Bank, Maintaining Utility Services for the Poor : Policies and Practices in Central and Eastern Europe and the Former Soviet Union, WashingtonDC (2000) 138 World Bank, Transition: the First Ten Years - Analysis and Lessons for Eastern Europe and the Former Soviet Union (2002) World Bank (2003), RepublicofUzbekistan, Country Economic Memorandum 139