W_PS 2 4 0 POLICY RESEARCH WORKING PAPER 2440 Lessons from Uganda on Lessons from Uganda on preparing a national strategy Strategies to Fight Poverty for poverty reduction, with stakeholders participating. John Mackinnon Uganda's experience contributed substantially to Ritva Reinikka the design of the Poverty Reduction Strategy Papers. The World Bank Development Research Group r i Public Economics September 2000 POLICY RESEARCH WORKING PAPER 2440 Summary findings Countries receiving debt relief under the Heavily extensive use of existing data and research about Uganda Indebted Poor Countries initiative will be among the first to refocus a range of public policies and interventions to benefit from the new World Bank-International relevant to poverty reduction. Second, the government's Monetary Fund approach to strengthening the impact on approach was highly participatory, with central and local poverty of concessional assistance in low-income governments, the donor community, nongovernmental countries. The new approach features a more inclusive organizations and civil society, and academics invited to and participatory process for helping recipient countries contribute. Third, the government was quick to translate develop poverty reduction strategies. From these the plan into its budget and medium-term spending strategies, joint Poverty Reduction Strategy Papers framework. Public expenditures on basic services were (PRSPs) will bring together the country's own priorities significantly increased after adoption of the PEAP in and Bank-Fund assistance to the country. 1997. In Uganda, such a strategy has existed for several years. Mackinnon and Reinikka discuss the general Uganda was one of the first low-income countries to characteristics of a poverty reduction action plan, prepare a comprehensive national strategy for poverty drawing on Uganda's experience; discuss what is known reduction using a participatory approach. Indeed, its about poverty in Uganda and identify shortcomings in experience contributed substantially to the design of the the data; examine the macroeconomic and fiscal policies PRSPs. Uganda's top leadership is heavily committed to that were considered most important to poverty poverty reduction. Formulation of Uganda's Poverty reduction during the participatory process; discuss the Eradication Action Plan (PEAP) in 1996-97 was the delivery of public services, especially those that directly executive branch's effort to make that commitment and affect the poor; and highlight problems associated with vision operational. land issues, including problems with access to credit and Mackinnon and Reinikka draw lessons from the financial services and with the security of productive drafting of Uganda's PEAP. First, the plan made assets. This paper-a product of Public Economics, Developnment Research Group-is part of a larger effort in the group to study public service delivery and poverty and to translate the results into poverty reduction strategies. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Hedy Sladovich, room MC2-609, telephone 202-473-7698, fax 202-522-1154, email address hsladovich@worldbank.org. Policy Research Working Papers are also posted on the Web at www.worldbank.org/research/workingpapers. The authors may be contacted at jwm1962(aaol.com or rreinikka@worldbank.org. September 2000. (55 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should he cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countPtes they represent. Produced by the Policy Research Dissemination Center Lessons from Uganda on Strategies to Fight Poverty John Mackinnon* jwml962@aol.com Ritva Reinikka** Rreinikka@worldbank.org This paper benefited from close collaboration with members of and consultants for the Task Force that prepared the Poverty Eradication Action Plan (PEAP) for Uganda in 1996-97, including Margaret Kakande, Abel Ojoo, Fred Opio, and Germina Ssemogerere. The Task Force was led by Mary Muduuli and included representatives of the central and local government, nongovernmental organizations and civil society, donor agencies, and academia; Emmanuel Tumusiime-Mutebile was an unfailing supporter of this effort. The PEAP, which provides the basis for Uganda's Poverty Reduction Strategy Paper (PRSP), was revised during the first half of 2000. Useful comments on an earlier draft were received from James Adams, Robert Blake, Lionel Demery, and Mathew Verghis. Hedy Sladovich edited the paper. * At the time of writing, John Mackinnon was a Research Officer at the Centre for the Study of African Economies, University of Oxford, as well as consultant to the Poverty Eradication Action Plan process. ** Development Research Group, World Bank. Contents 1. INTRODUCTION ..................................................... 1 2. WHAT IS A POVERTY REDUCTION ACTION PLAN? ................................ 2 3. KNOWLEDGE ON POVERTY AND SHORTCOMINGS OF DATA .............................. 4 3.1 HOUSEHOLD EXPENDITURE DATA ..................................................... 4 3.2 PARTICIPATORY POVERTY ASSESSMENT ................................................. :.7 3.3 OTHER SHORTCOMINGS .................................................. 9 4. MACROECONOMIC AND FISCAL POLICY .......... ........................................ 10 4.1 THE EFFECTS OF REFORM ON POVERTY ................................................. 10 4.2 AN AGENDA FOR MACROECONOMIC AND FISCAL POLICY ......................................... 12 4.2.1 Trade policy ................................................. 12 4.2.2 Tax reform ................................................. 13 4.2.3 Spending on basic services ................................................. 14 5. PUBLIC SERVICES ................................................. .. 14 5.1 THECASEFORPUBLIC SERVICES: MARKET FAILUREAND EQUITY . ................................................ 15 5.2 PUBLIC EXPENDITURES .................................................. 16 5.3 PRIMARY EDUCATION ................................................. . 17 5.3.1 Returns to education ................................................. 17 5.3.2 Service delivery in education ................................................. 19 5.3.3 Incidence of expenditure ................................................. 20 5.4 PRIMARY HEALTH ................................................. . 22 5.4.1 Public health services ................................................. 23 5.4.2 Incidence of expenditure ................................................. 24 5.5 AGRICULTURAL EXTENSION .................................................. 25 5.6 TRANSPORT .................................................. 25 5.7 WATER SUPPLY ................................................. . 27 5.8 LEGAL AND POLICING SERVICES ................................................. 27 5.9 REFORM: DESIGNING AND COSTING A BASIC PACKAGE ...................... ........................... 28 5.9.1 Delivering the basic package ................................................. 30 5.9.2 Cost reduction in service delivery ................................................. 31 6. LAND, CREDIT AND SECURITY OF ASSETS ............ ...................................... 32 6.1 EXISTING LAND RIGHTS .................................................. 33 6.2 THE EXISTING DISTRIBUTION OF LAND ................................................. . 35 6.3 ECONOMIC CONSEQUENCES OF THE EXISTING DISTRIBUTION ..................................... 37 6.4 WHAT LEGAL FORM SHOULD LAND RIGHTS TAKE? ............ ...................................... 39 6.4.1 The impact offormal titling on poverty .................................................. 39 6.4.2 The form of title .................................................. . 40 i 6.4.3 Who should get newfreehold? ...................................... 42 6.4.4 How should existing leases be dealt with? ...................................... 43 6.4.5 Mechanisms of enforcement ...................................... 43 6.5 AREAS FOR FURTHER PUBLIC ACTION ......................................... 44 6.6 THE LAND ACT ......................................... 46 6.7 CREDIT AND SECURITY OF ASSETS ......................................... 46 7. CONSENSUS AND CONTROVERSY ......................................... 47 7.1 MEANING OF POVERTY: GROWTH AND DISTRIBUTION ......................................... 48 7.2 MACROECONOMIC POLICIES ......................................... 48 7.3 PUBLIC SERVICES ......................................... 49 7.4 ACCESS TO ASSETS AND CREDIT ......................................... 50 7.5 WHAT HAS THE PEAP ACHIEVED ? ......................................... 50 BIBLIOGRAPHY .......................................... 51 ii 1. Introduction The World Bank and the International Monetary Fund are currently developing a new approach to enhance the poverty impact of their concessional assistance in low-income countries. Those countries receiving debt relief under the Highly Indebted Poor Countries (HIPC) initiative will be among the first to benefit. The focus of this new approach is to help recipient countries build more effective poverty reduction strategies through a participatory and inclusive process. These strategies are then expected to form the basis for a joint Poverty Reduction Strategy Paper (PRSP), which brings together the country's own strategy and Bank-Fund assistance to the country. In brief, the PRSP is an effort to help mainstream poverty reduction in the recipient country's public policy. As such, PRSPs should start from existing government strategies and build on them. This paper considers a case where such a strategy has existed for several years, Uganda. Uganda is one of the first low-income countries to prepare a comprehensive and participatory national strategy for poverty reduction. Indeed, its experience has inspired the design of the enhanced HIPC initiative and the PRSP approach. The top leadership in Uganda is highly committed to poverty reduction; the formulation of Poverty Eradication Action Plan (PEAP) in 1996-97 was an effort by the executive branch of the government to make this commitment and vision operational (Republic of Uganda 1996, 1997). This paper draws lessons from the drafting of the PEAP. The process had three specific features. First, the Plan made extensive use of existing research and data on Uganda to refocus a whole range of public policy and interventions on poverty reduction. Second, the government chose a highly participatory approach, with input from central and local governments, the donor community, NGOs and civil society, and academia. Third, the govermment was quick to translate the plan into its budget and medium-term expenditure framework, achieving significant reallocations of public expenditure toward basic services since the adoption of the PEAP in 1997. The purpose of this paper is to highlight the two main themes that emerged and were debated during the 18-month long participatory process: (1) the poverty impact of macroeconomic and fiscal policy; and (2) the delivery of public services. In addition, the paper focuses on a third theme which is important to poverty reduction and was nationally debated in a parallel process but outside the PEAP: (3) poor people's access to productive assets, particularly land. For each theme, the paper assesses the informational constraints policymakers typically face in low-income countries, and gives some examples of how to overcome them. The rest of the paper is structured as follows. Section 2 discusses general characteristics of a poverty reduction action plan, based on the Ugandan experience. Section 3 summarizes the state of knowledge on poverty in the country as well as shortcomings of the data. Section 4 looks at macroeconomic and fiscal policies that were considered most important to poverty reduction during the participatory process. Section 5 focuses on public I service delivery, particularly basic services that directly affect the poor. Section 6 deals extensively with land issues, highlights problems in access to credit and financial services, and considers security of productive assets. Within each section we consider the existing state of knowledge and public policy, the reforms debated and prioritized during the PEAP process to address the needs of the poor more effectively, and issues of implementation. Finally, section 7 concludes by highlighting the main controversies that arose and experience in forming a consensus on priorities for poverty reduction. 2. What Is a Poverty Reduction Action Plan? In poor countries with a central public policy objective of reducing poverty, optimal policy would factor poverty impacts into all public-spending appraisals. In practice, policy typically falls far short of this description. A poverty reduction action plan is essentially a device to make policy more focused on and effective in reducing poverty. It will usually have implications for policy in all sectors and for the allocation of resources across sectors. Within the whole range of public expenditure, some services reach the poor more directly than others; for instance, services from village clinics and primary schools probably benefit poor households more directly than services from universities and hospitals. It is not necessarily the case that the most direct poverty-reducing expenditure is the most effective in reducing poverty. To model the effects of primary and tertiary education on the poor properly would involve general-equilibrium considerations, comparing the direct income benefits of primary education with the effects of tertiary education on employment-generation and public services, and factoring in the effect of increased primary education on the labor market. However, in the existing state of knowledge, fiscal priorities have to be guided by a more approximate judgement. In Uganda's case, a judgment was made that direct poverty-reducing services were underfunded and needed particular attention. Specific financial mechanisms (such as a poverty action fund with the budget) were then set up to direct funding toward them, with dramatic effects. This judgment is likely to be repeated in many countries because the political voice of the poor is weak. If the concerns of a poverty reduction plan are restricted solely to the provision of direct poverty-reducing services, or limited to social sectors, it suggests that the poor need schools and clinics but not roads, markets, or technology. This is obviously not true. Hence a poverty reduction action plan should start with a broad focus and only identify the areas that require higher priority after assessing both the needs of the poor and the prospects for economic growth. This very breadth raises major issues of coverage. For instance, where the sectors of highest economic potential differ from those in which the poor are engaged, there is a short- term, equity-growth tradeoff regarding sectoral emphasis. This tradeoff is less extreme in Uganda because most of the country has considerable potential for agricultural trade. The 2 proportion of the population to be targeted is also an issue. In Uganda, t can be argued that mass service delivery is a necessary precursor to successful targeting in Uganda, given the extent of poverty at present and government's limited capacity to target the poor. Although there is no point in preparing a poverty reduction action plan unless poverty reduction is a high priority for the government, there will always be government objectives other than poverty reduction, such as political stability, which may limit the degree of redistribution that can be achieved. Moreover, some govemments may be committed to poverty reduction but hostile, in principle, to redistribution. To be effective, policies cannot stop, as much of economic analysis does, with the simple desire to subsidize or deliver a particular good or service, but must attend to the process of translating this intention into actual delivery. The process of service delivery may be as binding a constraint as the shortage of resources. While those doing the technical analysis need to interact with government, donors, NGOs and civil society more generally, the reformulation of public policy is bound to be informationally constrained. Nevertheless the very process of identifying information needs can help focus people's minds. In different forms, each of these concerns arises in each sector we examine. The process of refocusing public policy on poverty in Uganda coincided with a number of other important initiatives. First, extensive macroeconomic and marketing reforms, combined with some privatizations over the previous five years, had greatly improved the climate for doing business and the incentive to export. Second, a new constitution, promulgated in 1995 after several years of public consultations, culminated in the Constituent Assembly. A presidential election in 1996, won by the incumbent, led directly to major policy changes, most notably a public commitment to provide free primary education for four children in every family; this resulted in a strong positive response by households. Third, an ongoing process of decentralization continued to transfer decisionmaking to the district level. Fourth, although a civil service reform was in progress, the more difficult institutional reforms that affect, for example, service delivery and law enforcement, lagged behind. Fifth, government initiated a number of reforms which included: (1) the introduction of the medium-term expenditure framework into public budgeting; (2) the preparation of sectoral plans in transport, health, education, and agriculture; (3) a reform of the country's agricultural extension service; (4) the preparation of a land law after a national process of consultation going back several years; and (5) the introduction of a highly publicized and public scheme for providing microcredit. This highly participatory process involved several stakeholder groups, but the formulation of the PEAP did not include direct consultation with the poor. Initially, the process was conceived as a means for addressing concerns of the vocal NGO community and the government about the best approaches to poverty reduction. In particular, NGOs did not see much linkage between poverty reduction and the structural adjustment policies pursued by the government. The PEAP was, therefore, an effort to build consensus among these parties on the widely differing views on national priorities and, through its several thematic working groups, to serve as a means for "civic education" on matters of economic 3 policy and its linkages with poverty. Given the positive experience from the participatory process, government subsequently launched direct consultations with groups of ordinary people on poverty issues (Republic of lJganda 1999). This initiative has contributed to the revised PEAP, which was completed in the first half of 2000, and has become part of an evolving strategic framework (Republic of Uganda 2000a). 3. Knowledge on Poverty and Shortcomings of Data When the PEAP process was initiated in 1996, macroeconomic data showed high growth rates for the previous decade (real GDP growth rate of 6.5 percent; 3.8 percent in agriculture, 11.9 percent in industry, and 7.9 percent in services, on average). Nevertheless, there was a widespread perception that this growth was neither translating into measurable improvements in the living standards of the majority of Ugandans, nor leading to any major reduction in poverty (see UNDP 1997). This perception was one of the reasons for initiating the PEAP. 3.1 Household ecpenditure data The process of preparing the PEAP revealed major informational gaps on poverty. First, adequate household survey data were rLot available to provide a more objective basis for assessing the impact of the economic reform program on poverty and inequality over time. The 1989 household budget survey, which was conducted for obtaining a new base for the consumer price index, used a different sampling method from the first nationally representative household survey of 1S992. This made comparison of consumption and poverty between the two surveys impossible (Appleton 1996a). Furthermore, survey data collected after 1992 were not available for analysis due to lack of statistical capacity to clean and archive such data, which prompted a concerted effort by the Bank and the government to build capacity in this area. Consequently, five consecutive surveys became available in 1999 covering the period of 1992-97/98, each consisting of a large sample of households (typically 5,000; 10,000 in 1992) and communities (typically around 500; 1,000 in 1992). These surveys offer much better data for analyzing changes in poverty and inequality in response to economic reforms. Despite the lack of time series data during the drafting of the PEAP, the 1992 integrated household survey provided a useful cross-sectoral poverty profile (World Bank 1996a). According to the profile, poverty in Uganda is mainly rural, both because the urban share of the population is small and because the rural areas are poorer than the urban ones. Poverty also has a strong regional component; it is much higher in the North and, to a lesser extent, in the East, than in the West and Central regions (there is no "south" because the southern parts of all three regions border Lake Victoria). Sectorally, cash crop farmers were as poor as the rest of the population in 1992. Households with little access to land were poorer; and most rural households depended mainly on agricultural production either for 4 subsistence or for the market. Food was a substantial proportion of expenditure throughout the expenditure distribution, suggesting how poor the country really was after the collapse in incomes in the 1970s and 1980s.' The household survey data for 1992-97/98, which became available in 1999, filled the long-time gap in poverty data. The surveys show that mean incomes did indeed rise significantly during that period, combined with an overall improvement in inequality as measured by the Gini coefficient (Appleton 2000a). This overall improvement reflects both the sharp increase in coffee prices in 1994, which was passed on to producers under the newly liberalized regime, and some change in economic structure as the share of households in cash cropping increased. A decomposition analysis shows that almost one- half of the improvement in poverty during this period was due to improvement in the conditions of cash crop farmers. Using an absolute poverty line based on calorie needs given the typical diet of the Ugandan poor and minimum nonfood requirements, figure 1 shows 56 percent of Ugandans were poor in 1992, and 44 percent were poor in 1997/98 (Appleton 2000a). These relatively high figures illustrate the low base from which Uganda's recovery has started and that mass poverty still exists. For the purposes of comparing poverty trends over time, however, the result of falling poverty was found to hold, whatever poverty line was chosen. The fall in poverty-more than 20 percent-is substantial, given the relatively short interval of five years, and is explained mostly by income growth and only to a limited extent by changes in the distribution of welfare (figure 2). The reduction in poverty was not uniform across all regions or economic sectors, or the same for each year. The incidence of poverty fell most in the Central region and least in the Eastern region, and economically, households engaged in cash crop farming, manufacturing, and trade fared particularly well. Poverty declined in every region between 1992/93 and 1997/98; nationally, it also declined between every survey. The only group to experience worsening poverty over the survey period was households that reported a nonworking head. Such households could be headed by the elderly, children, or disabled. This finding strengthens the general impression that while economic recovery has created opportunities that many households were able to seize, transfer mechanisms for targeting the households that are not able to do so, do not work. While private remittances are important (for example, more than 20 percent of household income for the poorest quartile in 1992), alone they are inadequate to lift these vulnerable households out of poverty. ' For the poorest 86 percent of households the share of spending on food was around 68 percent in 1992, for the next 10 percent of households it was 65 percent, and for the rest (only 4 percent of households) it was 51 percent (World Bank 1996a). S Figure 1: Poverty in Uganda, 1992 and 1997/98 80 72 70 60 86 0 59 O~~~~~~~~~~~~~5 542 50 4 40 X92 . 97198 30 28 2= 20 10 national rural turban central western eastern northern Source: Appleton 2000a. Figure 2: Growth and redistribution decomposition, 1992-1997/8 120% 2% 2 * 100% 3 c ~~~~~~~~~~~~~~DResidual 2 ; 0% 2 0% 41 ~~~~~~~~~~~~~~~~-2% -20% Pp pipi Source: Appleton 2000a. 6 Appleton (2000a) finds a greater rise in living standards for the lower (poorer) deciles. The rise in consumption per adult equivalent is 27 percent at the bottom decile, 23 percent at the second decile, and 21 percent for the third poorest decile in 1992/93- 1997/98. Disaggregating into rural and urban areas, the distributional pattern in rural areas is very close to that in the country as a whole (though the mean is consistently lower). Using the median expenditure, Appleton shows that urban living standards have risen faster than rural ones. As in rural areas, there is some tendency for urban consumption to rise more at the lower deciles. The finding that urban living standards have risen is unsurprising, given the national account indicators of strong nonagricultural sector performance and visible progress in the major towns. Nevertheless, it is interesting that the rise in rural welfare is comparable with and even more consistent than the rise in urban areas. 3.2 Participatory poverty assessment The second shortcoming in poverty data in 1996 was the lack of qualitative or nonmoney metric information on poverty. It was not until after the PEAP process was completed and the government had gained some capacity in and exposure to participatory methods, that it was able to address this constraint. A national participatory poverty assessment is currently being carried out in collaboration with donors and NGOs. The Uganda Participatory Poverty Assessment Project (UPPAP) is a 3-phase process, in which the perspectives of the poor are being brought into national and district policy formulation, planning, and implementation. The first phase consulted people in 36 rural and urban sites in nine districts. Local people described poverty as a lack of the means to satisfy basic material and social needs, as well as a feeling of powerlessness (Republic of Uganda 1999). Poverty is viewed as nonuniform, complex, multidimensional, cyclic and seasonal and is described by more than one hundred indicators. For example, a poor household is a household without productive assets, income, and necessities. A poor community is a community without adequate basic services and infrastructure, livelihood opportunities, or without adequate security. Certain groups are perceived as more vulnerable to poverty than others: the landless, casual laborers, women, widows, widows and widowers with children, orphans and neglected children, the youth and the elderly, the chronically sick, and the displaced and refugees. While considerably extending these concepts of poverty beyond private consumption, these finding are surprisingly consistent with the quantitative household survey evidence. When assessing poverty trends, however, perception data collected by the UPPAP are at odds with the quantitative survey evidence discussed above. The UPPAP report concludes: 7 "Through analysis of long-term trends in poverty, many local people felt that poverty was worsening in their communities... Local people reported more movement into poverty than out of it." (Republic of Uganda 1999) The difference between the participatory findings and the surveys can be interpreted in a number of ways. McGee (2000) offers the following hypotheses. First, the participatory surveys use a different reference period, so that poverty may be being compared with the period of the 1960s. Secondly, the participatory survey is not based on a representative sample. A third explanation may be that the income gains have been concentrated on households which can grow cash crops; for the majority of households who do not grow these crops, the income gains have been significantly smaller. Fourthly, the introduction of cost-sharing might have increased the needs of households for cash; in principle, the CPI should pick this effect up. Fifthly, the participatory results confirm the claim that has been made elsewhere inside and outside Uganda that increased incomes from cash crops may sometimes come at the expense of reduced area cultivated for food, increasing local food prices or directly reducing the availability of home-produced food. Finally, the national statistical surveys do not adequately capture insecure areas.2 Where insecurity prevents the collection of data, the area affected is omitted from the sample for the relevant year and for any year to which is compared. If increases in insecurity are correlated over time with increases in poverty, then this procedure will tend to bias the comparison in favor of the year which is experiencing greater insecurity, to an unknown degree. The number of internally displaced households and refugees has recently been estimated at over a million, compared to a population of somewhat over twenty million, so this effect is by no means negligible. Certainly, the participatory surveys found that various forms of insecurity were central to the experience of poverty. All of these hypotheses deserve fiurther investigation. However, rather than in poverty trends the strength of the participatory approach is probably in its exploration of fine- grained aspects of welfare which quantitative studies will not identify (though, once identified, they can often be studied quantitatively). For instance, the surveys in Uganda shed light on the relative strengths and weaknesses of different legal institutions and different levels of law enforcement, and on the factors which deterred people from using medical services (among which the unpredictability of costs was as important as the level of costs). 2 The four districts excluded from the quantitative analysis due to security included 6.9 percent of Uganda's population. The omission raises mean consumption by 1.8 percent in the 1992/93 household survey (Appleton 2000a). 8 3.3 Other shortcomings The third informational gap was agricultural production data. Apart from the 1992/93 integrated household survey the subsequent surveys did not adequately cover households' production, which is the backbone of the economy. In particular, smallholder agriculture suffers from inadequate information about production, input use, and so forth. This gap in information will only be remedied when the national household survey is completed in 2000. Table 1: Social Indicators 1988/89 1995 Life expectancy 49 43 Infant mortality 119 97 Immunization (%, 12-23 months) 31 49 Malnutrition (%, stunting) 43 39 No Education (%) Female, 15-19 years 21 19 Male, 15-19 years n/a 7 Net enrollment age 6-10 (%) 47 64 Gross enrollment P7 (%) n/a 45 Males n/a 55 Female n/a 35 Literacy (%/O) 54 61 Female 45 49 n/a not available. Note: Literacy and enroHment figures are not fully comparable between years. Source: Demographic and Health Survey, 1995; Integrated Household Survey 1992/93; World Bank Atlas. Fourth, it was difficult to construct exact data on the budgetary commitments and actual spending regionally or on poverty-related programs. This informational gap both reflects and contributes to problems in public sector management. Given such systemic difficulties in public information systems, a diagnostic survey was carried out simultaneously with the PEAP process to explore actual financing of primary schools and health clinics across the country. As discussed below, this survey proved useful in pinpointing specific problems policymakers could tackle. Similarly, there are shortages of data on the output of some critical government services. Where there is data, different sources often conflict as demonstrated by the divergent trends between official primary enrollment statistics and survey-based data (for details see section 5). A great deal of work is needed to improve the amount, type, and quality of public spending related data collected. Thanks to the 1995 demographic and health survey (DHS) and a similar survey in 1988/89, information on health-related human development indicators was available for the PEAP process; however, several other social indicators were not available or the data were 9 of poor quality (particularly school enrollment data). The DHS estimated that child mortality was 176 per thousand in the early 1980s and 147 per thousand in the early 1990s. Levels under 200 were obtained in the mid-1960s, but the decline in incomes and health services during the period of economric and social decay in 1971-85 prevented further progress until the mid-1980s. Child rnortality partially reflects educational level and the information households have about health (Mackinnon 1995). In addition, AIDS mortality has been high, so that life expectancy has fallen dramatically in recent years (see Table 1). According to data in the 1995 DHS, a vigorous public information campaign successfully increased awareness of the causes of HIV infection, and prevalence has begun to fall nationally. 4. Macroeconomic and Fiscal Policy Major components of macroeconomic reform in Uganda have been fiscal stabilization and trade, foreign exchange, and financial liberalization. These policies have been complemented by privatization of public enterprises and reforms in taxation and public expenditures. 4.1 The effects of reform on poverty The reforms have increased economic growth, although per capita income has not quite yet regain its pre-1970 levels. Macroeconomic policy is clearly the strongest element in Uganda's recovery to date. In the PEAP process, which focused on building consensus on the importance of maintaining macro stability and analyzing key linkages between economic policies and poverty, the main issues were as follows: Price stabilization. Since price stabilization was achieved in 1992 inflation has been in the single digits. The key instrumenrt for maintaining macroeconomic stability is fiscal restraint. Conquering high inflation is unambiguously beneficial to the poor, whose assets are disproportionately in cash and who typically do not have the means for hedging against inflation by shifting into real assets. The effects of the inflation tax on poverty can be thought about in two ways. The first is the static incidence of the tax. While the poor are likely to have low ratios of financial wealth to income, they also have high ratios of nonindexed assets in their financial assets, making the overall incidence uncertain. The second is the dynamic effect of the tax. Unlike other taxes, inflation taxes specifically tax savings, thereby discouraging financial accumulation and perpetuating poverty. Relative price changes. The impact of the relative price changes needs to be assessed. The objective of the liberalization policy has been to increase the relative price of traded exports with respect to food cr'ops and nontradable goods. The intention was to reverse the dramatic decline in the production of export commodities during the 1970s and 1980s. This policy has been successful. The group that has most clearly benefited from this policy is households that grow traditional export crops, particularly coffee. In the 1992 10 survey, such households were about as likely to be poor as others, and the poor derived as large a share of income from such crops as the nonpoor. Findings from the 1997/98 survey, however, suggest that households growing traditional export crops (coffee, tea, cotton, and tobacco) are now better off than the average (Appleton 2000a). Increased trade and export of maize and beans, which are new as export crops, provide some potential to reach households in the poorer Eastern region. The increased relative price of traditional cash crops is reflected in a rise in their share in the total value of national crop output from 7 to 13 percent in 1992 and 1995, respectively (Okidi 1999). The largest increase of 20 percentage points (from 13 to 33 percent between 1992 and 1995) occurred in the Central region, which is a coffee growing area, followed by the Eastern region where the share of cash crops increased from 7 to 12 percent. A substantial shift away from subsistence agriculture (from about 35 percent of GDP in 1985 to about 20 percent of GDP in 1999) toward marketable agriculture also occurred. At the household level, the average share of marketed output to total output increased from 19 to 22 percent nationally during the 1992-95 period (Okidi 1999). Cash crop production of cotton, groundnuts, and sesame, the most suitable crops for the North, declined. As a result, the sales-output ratio declined in the North by 3 percentage points during the same period, which is consistent with the finding that the region had the lowest regional growth rate of mean household income. However, preliminary evidence from the 1999/2000 national household survey indicates that cotton growing has recovered since then. Proximity to tarred roads and a bank (or moneylender) significantly increases the chances that a household sells at least 20 percent of its crop as output. The distributional effects of liberalization, therefore, have been generally neutral or positive across Uganda, except for the North, which is the poorest region. Public services, such as agricultural extension and transport infrastructure, play an important role in ensuring that the poor can participate in the expanding sector. In the North problems in service delivery have been made more difficult by conflict. One important caveat is the extent to which commercialization in Uganda benefits men at women's expense. Men are more likely than women to control the income from cash crops. A study in Arua found that women disliked having to work on tobacco and felt that they would be better off without the crop (Harmsworth 1991). In some cash-crop growing areas health workers observe that malnourished children are likely to come from cash-crop producing households. Nevertheless, the existence of gender inequality and the control of income by men does not necessarily exclude women from pursuing new opportunities to generate income; this depends on the way the household responds. Mackinnon (1998) examines household behavior and finds that women may benefit relatively more from increases in domestic productivity than increases in cash crop prices; nevertheless, the claim that they actually lose in absolute terms from increases in cash crop prices is much stronger. During the PEAP discussions, it was felt that female bargaining power could better be promoted by improving girls' and women's access to public services and ownership rights 11 than by reducing retums to the most promising economic option facing many rural households. Movements in the real exchange have, during some periods, restricted improvement in the position of traded goods. Uganda's external position was strengthened by number of events: (1) a major inflow from the global coffee price boom of 1994 (including an inflow during 1994 into Rwanda through Ugarnda); (2) an increases in coffee production, following the coffee boom in 1994-95; (3) a strong recovery in remittances from abroad boosted by increased confidence; (4) and aid flows, which remained the same or increased slightly in real terms.3 The appreciation in combination with poor weather did significant damage to some producers of noncoffee exports, particularly maize. This induced severe distress in the Eastern part of the country. Poverty trends during the period suggest a worsening in the East during 1994 (Appleton 2000a). The possible conflict between the effect of aid on the real exchange rate and the promotion of better incentives for traded goods remains a policy dilemma in many African economies, but its empirical quantification is usually difficult (Elbadawi 1999). 4.2 An agenda for macroeconomic andfiscalpolicy The benefits of promoting agricultural exports for growth are considerable, and the most important priority for poverty eradication is to support this growth and ensure that the poor can benefit from it. The agenda for macroeconomic and fiscal policies laid out in the PEAP reflects this vision. While macroeconomic policy was largely endorsed as is, fiscal policy regarding taxation and spending for public services was reevaluated. 4.2.1 Tradepolicy As mentioned earlier, trade liberalization has been a cornerstone of the Ugandan economic reform program in the past decade, resulting in one of the most open trade regimes in Sub- Saharan Africa. Import liberalization started in earnest in the mid-1990s. The PEAP reiterated government's commitment to reduce the antiexport bias and refrain from protecting a few jobs in urban areas at the expense of the majority of rural consumers. Given that Ugandan producers are price-takers in the world market but domestically have (at least some) market power, the final burden (incidence) of import duties largely falls on exporters of coffee, cotton, tea, and fish, which are often produced by poor households. By 1999, non-COMESA import duties had been reduced to 15, 7 and 0 percent (consumer goods, raw materials and capital goods, respectively), while regional duties are even lower.4 All remaining import bans have been abolished. However, a recent firm survey indicated 3 Subsequently, Uganda's external public debt service obligations became lighter, following the Highly Indebted Poor Country (HIPC) initiatives in 1998 and 2000. 4 Except for a 10-percent surcharge on a large number of goods that compete with imnports from the neighboring countries which maintain higher regional tariffs on such goods. 12 that the duty drawback system for exporters functions poorly (Reinikka and Svensson 1999). Efforts are being made to find new ways (using commercial banks, for example) to improve the duty drawback system. 4.2.2 Tax reform The rebuilding of government's revenue base from an almost total collapse has been another important feature of Uganda's economic recovery. Institution building for tax administration was given priority. As a result, domestic revenue has increased from 7.6 percent of monetary GDP in 1986 to 14.7 percent in 1998. Tax policy and incidence of taxation, however, had received much less attention. When the PEAP was being drafted little information existed on actual tax burdens (that is, who was affected and by how much). Therefore the PEAP mainly emphasizes key principles of poverty-sensitive tax policy, recognizing that such a policy needs to address both efficiency and equity. Priority was given to reforming the existing tax system to broaden the income base, phasing out discriminatory exemptions, and making the tax system more progressive. In recent years government has indeed taken a number of measures to achieve a more efficient and equal tax regime. Exemptions have been reduced, a VAT was introduced in 1996, and an income tax reform was implemented in 1997. Subsequently, more information on the incidence of taxation has also become available. A study based on the 1992 integrated household survey found that a reform to replace the sales tax with a VAT would make the poor better off (Chen, Matovu and Reinikka 2000). When selectively implemented by a zero rating on the commodities consumed by the poor, the VAT is found to be a progressive tax. Import duties are also found to have become more progressive over time, indicating that the use of differentiated duties tends to be more regressive than more uniform rates. Fuel is generally more highly taxed in Uganda than in the neighboring countries. Although lower than other petroleum products, taxes on paraffin tend to be very regressive. In the PEAP the government committed itself to reviewing fuel taxation. Similarly, export taxes are regressive, particularly the stabilization tax on coffee which was re-introduced in 1994 but later abolished. The most important local tax is the graduated personal tax (GPT) which is assessed and collected as a presumptive tax on almost all tax subjects in the rural areas and the self- employed in the urban areas. According to a 1994 survey, GPT accounts for between 18 and 40 percent of locally raised revenue for municipalities and 60 to 85 percent for the districts. It is almost certainly regressive as everyone with an income below U Sh 840,000 (US$560) per year pays the same amount, eliminating any correction for income. The reform of the GPT was called for in the PEAP, to be focussed on reducing its regressiveness and simplifying its administration and collection, which is at present arbitrary and costly. Little progress has been made in this area to date. Finally, before 1997 small firms did not pay regular income taxes unless they opted to do so, but following the 1997 income tax reform, a presumptive tax of up to 1 percent is now levied on their gross turnover. Since the presumptive tax is imposed on the 13 gross receipts without any adjustments, small fimns are neither entitled to the generous initial allowance for investment in machinery, nor subject to any restrictions regarding writing off business expenditure. As a result, the marginal effective tax rate for small firms is lower than that for large and medium-sized firms on all other assets but machinery (Chen and Reinikka 1999). Most small firms also pay the municipal property tax, which constitute most of their actual tax burden. 4.2.3 Spending on basic services At the time of drafting the PEAP it was estimated that only 23 percent of total recfrrent public expenditure (including recurrent: components of donor projects) was spent on four basic services delivered mostly by local governments (primary education, primary health, agricultural extension and feeder road rnaintenance). The government made a commitment to increase the share of expenditure to 50 percent, but no target date was set for achieving this goal. While the Constitution calls for three types of intergovernmental grants, only two were operational in 1997: the block grant and the conditional grant. Equalization grants for poorer districts are now being introduced. The regional inequality in Uganda provides a good argument for such interventions, but because political divisions reflect regional ones, it has been important to develop a technical set of criteria for these grants. In 1998, in the context of the medium-term expenditure framework the government introduced a Poverty Action Fund which is ultimately intended to include all spending on directly poverty-reducing public services. This includes government spending, donor budget support, and some donor projects in cases where these are well integrated into the government's sectoral plans. Government and donors are now focusing on the share of spending going to the programs included in the PAF, which has increased substantially since 1997. As a share of government-funded expenditure, PAF programs are expected to increase from 23 percent in 1998/99 to 33 percent in 2002/03. This spending is more carefully monitored, with strong civil society involvement; 5 percent of the PAF is devoted to accountability. Government ministries seeking increases in funding now recognize that the most promising line of attack is to argue that their spending should qualify for PAF; this ensures that decisions about fiscal allocation involve discussions about poverty impact. Under the terms of the decentralization program the districts now deliver most of the services considered as high-priority for the poor. The fiscal implications of delivering these services are discussed in the next section. One important dimension is the regional profile of expenditure. 5. Public Services Rather than examining the whole range of public services, the PEAP process focused on services that interact directly with the lives of the poor and hence are priorities for poverty oriented public spending. Six major services were included: 14 * primary education * a basic package of medical care * agricultural extension * transport * water supply * basic legal services. While the desirability of people getting each of these services is widely accepted, who should provide them and how are controversial. We therefore consider the following aspects: 5.1 The case for public services: market failure and equity As terminology in this area differs, it is necessary to define a few key concepts For example, inefficiency in public service delivery was taken to mean that there are services which people value at more than their cost but which are not being delivered because of market and/or government failure. Equity refers to the total distribution of welfare across society, including the utilization of public services, but also the distribution of real income. In this sense equity can be improved not only when the poor get better access to a service from which they were previously excluded, but also when the price the poor are paying is reduced for a given service. In Uganda, government spending is seriously constrained, and asset inequality is difficult to address directly. Hence public spending needs to be soundly based on at least one of these criteria, and preferably both. The standard method of prioritizing different forms of public spending is by comparing rates of return. In principle (for instance in the framework of Stern and Drbze 1990), such calculations should include income distributional effects, the extent of substitution between public and private sectors; and an estimate of external effects, but in practice most estimates neglect all three aspects of the impact of public services. For poverty reduction, this point is critical. For instance, tertiary education does very little for poverty reduction directly even if it has high private returns, because its returns go overwhelmingly to the nonpoor. Indirectly, tertiary education may have very significant benefits for poverty reduction if graduates contribute to better public services or employment generation, remit their incomes to poor households, or contribute to the communities they come from. But there are no estimates available of these externalities. Moreover, the forms of benefit that are produced by public services are diverse and hard to compare. Consider, for instance, the valuation of income streams, lives saved by health and education, labor and effort saved by improved water supplies, and births prevented by birth control. Measurement of these competing benefits, given the current state of the art, is sensitive to assumptions that are quite difficult to justify, even approximately. Hence, the intersectoral allocation inevitably depends on judgment as much as quantitative analysis. Within sectors, the implications of quantitative analysis are often more clear-cut. 15 In practice, in the existing state of knowledge, analysis can typically establish the following: (1) that there are positive returns to some services, (2) the extent to which the public or the private sectors are already delivering the service and their respective efficacy, and (3) the incidence of expenditure within the sector. In the latter the benefit is proxied by its cost. In the next sections, we will establish the Ugandan information base by sector in these three aspects. 5.2 Public expenditures Public expenditure data in Uganda suffers from many problems: coverage, misclassification of recurrent and capital expenditure, administration versus facility distinction, lack of disaggregation by appropriate program, and lack of actual spending data due to decentralization. Coverage has been greatly improved by the integration of donor-funded spending in the Public Investment Plarn. The development budget includes much recurrent spending but an attempt has been made to disaggregate it. The recurrent-capital distinction is important not because recurrent spending is "consumption"-much recurrent spending can be understood as a form of investment, including health, education, agricultural extension, and infrastructure maintenailce-but because recurrent spending represents the kind of spending needed to keep a given public service that is often underprioritized functioning reliably. The distinction between administration and service delivery is not clearly made in the accounts and, as noted above, it turned out in 1996 that much spending intended for schools never passes beyond the level of the district education office. The development budget is disaggregated by program, but the recurrent budget is not detailed enough for this; payroll data, for instance, do not clearly distinguish staff by category, making the functional classification of spending unclear. Finally, the Ministry of Finance was relying on a "shadow budget," assuming what the districts should be spending on basic services, rather than using either budgets or actuals fromn the district levels. As a result, the spending which is most directly related to the lives of the poor is the spending which is known least accurately. A lesson of the experience vvas that specific requests for existing budgetary information need to be made early in the process. The categorizations of interest for poverty reduction are not necessarily available and may be laborious to extract, but the request for information helps to focus attention on the subsectors of importance. The returns to commissioning specific studies on aspects of public expenditure can be high. For example, the results of expenditure tracking produced a change in the administrative mechanisms used to transmit funds to schools within one year. It is sometimes suggested that the best way to assess the extent to which needs are being met, is to add private spending to public spending and compare it to some norm. There are two potentially serious problems with this approach. One is that the cost structure and efficiency of spending may vary widely between public and private sectors. The other is 16 that the incidence of private and public spending should be quite different, if the sector is really delivering services to the poor. In Uganda, for instance, the proportion of poor people among users of public health services is higher than that among users of private services at the primary level (Appleton and Mackinnon 1995). Where private expenditures are highly unequal, a volume of expenditure, which is in aggregate adequate, will not ensure that everyone's needs are met. One form of private sector spending, however, may play a critical role in supporting the access of the poorest to services: NGOs. While data are not available at an aggregate level, it is important to realize that the public sector figures include some projects under the government's investment plan in which money is channeled from donors to NGOs. 5.3 Primary education Even though enrollment is an apparently simple concept to define and measure, in the mid- 1990s the gross primary enrollment rates defined by the household survey and the administrative records were about 90 and 70 percent, respectively. This remarkable discrepancy was partly because the mean number of days reported missed during the past 30 days in the household survey was about 2.5, roughly 10 percent of the intended days of school; partly because the administration's figures did not include private schools; and partly because there were other (unspecified) problems with the administrative system which recorded primary enrollment data. The latter was confirmed by the school survey undertaken as part of the PEAP process (Reinikka 2000). Namely, while official enrollment figures remained stagnant throughout the first half of the 1990s, the survey of schools showed an increase of 60 percent in primary enrollment in 1991-95 (see details below). 5.3.1 Returns to education In primary education sources of market failure include the following: (1) the inability of parents to capture returns from their children's education, (2) the difficulty of borrowing on the strength of future incomes, (3) externalities, and (4) a possible misperception about the value of education. For instance, it has been suggested that people underestimate the benefits of parental information about health, which can be promoted by education. However, a women's needs assessment survey in Uganda found that mothers wanted their daughters educated because it would make them 'better mothers'. In Uganda, Appleton and Balihuta (1996) have studied the effects of education on productivity. In contrast to most studies in Africa, the data relate to rural households and incomes from farmiing and productivity, not merely wages earned in urban employment. Returns were positive though not spectacularly high. One unusual feature of this study was to estimate externalities, which found externalities to be larger that the direct benefits of education. Mackinnon (1995) examines health outcomes and finds that parental education has a very powerful impact on child survival. 17 A more recent analysis of the determinants of household earnings suggests that for each year of education, earnings increase by about 4 percent (Appleton 2000b). This is a modest benefit although it may still imply a high rate of return if the opportunity costs of attending primary school are low. Perhaps the most surprising finding is that education has similar proportional productive benefits in all three income-generating activities: farming, non-farm self-employment, and wage employment. This is what one might expect if human capital was allocated efficiently across activities. However, it does run contrary to the common belief-and much supporting evidence-that education is rewarded more in wage employment and brings only small returns in farming. Appleton suggests that the relatively small returns to education conventionally found in agricultural production functions may partly arise from a failure to properly control for the input of labor to farming. The results imply that education may bring tangible benefits to the poor in Uganda, who are typically not wage employees. It is worth noting, in this context, that the general equilibrium effects will also differ across sectors. Assuming diminishing returns to a single factor of production, an expanded supply of primary education will tend to drive down returns to education from wage employment by compressing the educated-uneducated differential. But it need not reduce returns to education in cash crcip production or food production (though it might have some effect through food markets). The abundance of agricultural land in Uganda suggests that if education does raise productivity, there should be abundant opportunities for farmers to exploit these returns. Appleton also estimates the extent to which education brings returns by reallocating labor rather than by direct productivity benefits. Once again, the results are somewhat surprising. It was expected that a principal indirect benefit of secondary education would be increased access to wage employment. However, in the decompositions, this benefit was wholly offset by the loss of income associated with withdrawal from farming and non-farm self-employment. Indeed, the combined entry effect and labor allocation effects of secondary education were estimated to be mildly negative. By contrast, an important channel through which primary education appears to benefit households is in encouraging entry to non-farm income generating activities and reallocating labor out of fanning. More generally, the consensus that returns to education are higher in primary education than in other levels has been recently challenged. In a number of African cases, urban workers show higher returns for higher levels of education using Mincerian coefficients (see Mincer 1974 for the definition). Nevertheless, translating Mincerian coefficients into rates of return requires cost information, and assumptions about life expectancy and the future behavior of earnings. For example, Mason and Khandker (1997) give high estimates for labor foregone by children attending school in rural Tanzania. However, survey evidence from Uganda does not confirm this finding (Appleton 2000b), but shows that school fees are the main reason for both never attending school and dropping out (over 50 and 60 percent of respondents, respectively). Contrary to evidence from 18 Tanzania, the proportion of the lowest expenditure quintile that dropped out because of the need to work was only 5 percent, the same as for the highest quintile (World Bank 1996b). These differences may reflect methodology rather than a difference between the two countries. 5.3.2 Service delivery in education The service delivery surveys typically examine the effects of policies or interventions on households and their demand for and perceptions of the quality of services. Inputs and outputs on the supply side, such as the flow of public funds and school enrollment fates, are left for official statistics or administrative records. However, a survey can provide a useful reality check on the supply side of service delivery when institutions operate suboptimally and official statistics are lacking or are of poor quality. The principal motivation for an expenditure tracking survey, which was carried out in the context of the PEAP, was the observation that while budgetary allocations for primary education almost tripled between 1991 and 1995, there was hardly any corresponding increase in the officially reported enrollment (Reinikka 2000). The hypothesis was that actual service delivery (output) was much worse than budgetary allocations would imply because public funds (inputs) did not reach the intended facilities. Reasons why facilities may not receive the allocated funds range from competing priorities at various levels of govermment to corruption and misuse of public funds. To test this hypothesis, budget allocations and actual spending were compared in two decentralized sectors, primary education and health care. As local government accounts were not generally available, a survey of 19 out of Uganda's 39 districts was carried out, including 250 government schools and 100 public health clinics, to collect spending data for 1991-95.5 Interestingly, local government (district) and health unit records were totally inadequate, while school records were relatively good. The school survey confirmed the hypothesis that input flows suffer from serious problems which, to a large extent, have to do with poor governance and lack of accountability. On average, less than 20 percent of funding intended for nonsalary public spending actually reached the schools in 1991-95, as district authorities kept and used most of the nonsalary capitation (per student) grant meant for schools. Similarly, at best, schools were allowed to keep only one-third of mandatory tuition fees from parents; the rest went to district education offices. There were also large variations: at the median school retention of both the capitation grant and tuition fees was zero. Compared to nonsalary expenditures, tracking teachers' salaries was complicated by the absence of disaggregated central government pay data. By and large, salaries seem to reach the schools much better than nonwage allocations, however with considerable In 1997 there were approximately 10,000 primary schools in Uganda; 15 percent were private and the rest were government schools. 19 delay. The only systematic way of misappropriating salary fumds were "ghosts" on the payroll. Previous actions by the government to clean up the teachers' payroll give some idea of the magnitude of the leakage iri salaries: close to 20 percent of all teachers on the payroll were removed as ghosts in 1993. The survey also unearthed a host of other features of service delivery that were quite unexpected but nonetheless critical to the functioning of the education services and hence to any intervention in these sectors. First, contrary to the stagnation in enrollment rates indicated by officially reported figures, the school survey reported an increase in primary enrollment of 60 percent in 15991-95. Such a stunning discrepancy indicates that official statistics cannot always be trusted. Second, although the government's share of the total expenditure on public primary education increased over time, the bulk of the burden continued to be borne by parents who, on average, contributed over 70 percent of total school spending in 1991 and 60 percent in 1995 (40 and 20 percent at the median school, respectively). Parents' contributions over the survey period continued to increase in real terms despite higher public spending. While many of the survey findings were rather troubling, there is a positive side to the story. Following the release of survey results in 1996, the government responded actively to them. For example, to increase transparency and accountability monthly transfers of public funds to districts are now reported in the major newspapers and broadcast on radio. Moreover, government policy now requires that transfers to primary education be displayed on public notice boards in each school and district center, and the Ministry of Education monitors compliance. School-based procurement has replaced the central supply of construction and other materials, and data on spending on teacher's salaries are now available at the central government level. The Ministry of Education replicated the school survey four years later. The results indicate dramatic improvements' in the flow of funds (Republic of Uganda 2000b). Instead of only 20 percent in 1995, schools received close to 100 percent of the capitation (per student) grants 1998/99 on average and close to 90 percent at the median school. This experience demonstrates that information can be a powerful tool of change in a weak institutional setting. 5.3.3 Incidence of expenditure The rationale of an incidence study is to determine the distribution of public spending, that is, the extent to which different income groups (expenditure quintiles in this case) benefit from public expenditure. This is important as effective public service delivery can contribute a great deal to poverty reduction. The incidence analysis is useful in signaling distributional problems, but it cannot provide all necessary answers on how to solve them. Hence, the next step would be to derive the determinants of demand for the services in question or other behavioral relationships required for informed policy decisions. For example, if one finds that the poor do not use public services, this may be a 20 result of household behavior (demand), or because the services provided are of poor quality (supply). Further, the benefit in this approach is measured by its cost, which obviously is less than a perfect measure of many benefits, immunization for example. An incidence analysis was carried out during the PEAP process. The study was based upon data from the 1992/93 integrated household survey (IHS) and the government's fiscal accounts to assess the extent to which public expenditure on education benefits the poor (World Bank 1996b). A summary of findings in primary education include the following: * Net enrollment rates among the bottom quintile were very low. In 1992/93 the rate was 46 percent at the primary level compared to 81 percent for top expenditure quintile. * The average education benefit per capita was very small, that is, U Sh 1,896 for primary schools. * The bottom quintile's share of the total primary education benefit was almost the same as for the top quintile because the higher enrollment rates in richer households are offset by a higher proportion of school-age children in poorer households. At the secondary level only a fraction (4 percent) of the total education benefit accrued to the bottom quintile, while the top quintile got nearly half (49 percent). The, benefit in secondary education is not at all targeted toward the poor. However, a recent update of this work provides a rural-urban disaggregation showing that the incidence of spending on rural secondary schools is very much better than that of urban secondary schools. . The level of parental contribution is high, which is likely to have two consequences. First, it keeps some children out of school. Second, it forces the poor to send their children to inferior schools because they are unable to provide textbooks and top-up teachers' salaries to the same extent as the nonpoor do. * In relation to per capita household expenditure the private costs of primary education were apparently tolerable even for the poorest households; the ratio of cost to household expenditure ranged from 13 percent in the bottom quintile to 10 percent in the top quintile. In terms of nonfood expenditure, however the poor spend as much as 40 percent on primary education. The absolute amounts of spending varied hugely between households. * At the secondary level, for the bottom quintile, private costs were equivalent to 83 percent of per capita household expenditure, a heavy burden indeed. * At both levels the high cost of education was the single most important reason cited for never attending school or dropping out. * Although sufficient data are lacking, all the indications were that the standard of education, especially the standard of education received by the poor, is very low. 21 The announcement of free universal primary education (UPE) for four children in every family starting in 1997 caused a substantial inflow of students (as many as 2.5 million children in P1). This confirms that there was a large unsatisfied demand for education, even at the levels where enrollment rates were highest. Appleton (2000b) shows that girls, poor households and children with uneducated parents benefited most. One message that emerges clearly is the difficulty of knowing exactly the true state of affairs about education, even where good household surveys exist. The PEAP consultations mostly focused on implementation issues, given the huge private response to UPE. The findings of the expenditure tracking survey focused people's minds particularly on accountability. 5.4 Primary health Uganda's health outcomes are poor. This was not always the case; in the 1960s, child mortality compared well with that in most African countries. Between the mid-1960s and the mid-1980s, however, child mortality failed to improve in Uganda, while it improved in other countries. The most recent demographic and health survey (DHS) shows a clear improvement in child health outcomes between 1989 and 1995, but AIDS has gravely increased adult mortality. Infectious diseases including diarrhea, malaria and tuberculosis form a large part of the burden of disease. The poor health performance reflects failures both in the public and private sectors. Major factors include poor economic performance, near collapse of the public health service during the 1970s, and limited information. Three elements provide good theoretical reasons why intervention is needed to improve health. First, consumer decisions about health are seriously constrained by misinformation. In Uganda's case, Mackinnon (1995) finds that parental information about health is an important determinant of child mortality, even when income and education are taken into account. The spread of AIDS was certainly accelerated by the fact that the disease was initially unknown; extensive media coverage and political leadership have, however, spread knowledge widely about the illness, as demonstrated by the figures shown in the 1995 DHS. Second, the communicable diseases, which are extensive in Uganda, involve two important kinds of externality: the possibility of infection, and the development of resistance to drugs that can be accelerated by the use of excessively short courses of treatment. For instance, many doctors now consider chloroquine almost useless for treating malaria. Third, acute illnesses such as malaria require rapid treatment. This poses a serious problem for households unless there are sources of finance that can be tapped very quickly, and distinguishes health risk from a risk such as harvest failure that typically emerges over a longer period of time. The above considerations imply that there is scope for public actions, not only in the health sector but also in primary education, to improve health outcomes. International evidence confirms that countries taking such actions have reduced mortality faster than countries trusting economic growth to do the job. There is considerable controversy, however, about the appropriate role of the state. The existing poor level of health outcomes indicates that the private sector is also not achieving the desired outcome of improving 22 people's health. Hence, in the face of the evident problems in existing health services, there is agreement that some public action is needed. 5.4.1 Public health services The quality of existing public health services in Uganda are perceived as inferior to the private sector, and there is some disturbing evidence to support this view. Household data show that the share of poor households among patients is higher in public facilities than in private facilities (Appleton and Mackinnon 1995), suggesting that public facilities are perceived as a cheaper but inferior option. Nonetheless, both poor and nonpoor in aggregate use private health services more than public services. The expenditure tracking survey carried out in health units as part of the PEAP preparation showed that clinic records of resource flows for resources were kept in-kind, if at all, and that there really were no records of inpatients or outpatients in the clinics. Through interviews at the clinics, it became clear that medical supplies, by and large, reached the health facilities, probably because they were distributed to health units directly from the center and did not go through local government. But, for the most part the survey instrument that had worked well in schools was not able to provide any substantive quantitative information for health units (Reinikka 2000). Using focus group discussion, observation, and other qualitative research methods other researchers found that problems originated within health units. For example, they found staff expropriated a large part of drugs and medical supplies and sold them onwards. The leakage was estimated to be over 70 percent, on average (McPake and others 1998). Evidence suggests that the poor, who are unable to pay for these misappropriated supplies, often do not receive even basic health services. These results indicate that the health system continues to suffer from the past legacy of poor government in the 1970s and early 1980s and that the past survival mechanisms of the health staff whose salaries used to be extremely low had not yet changed. Donors accounted for the bulk (77 percent) of recurrent expenditure on primary health in 1992/93. The government's contribution consisted mainly of salary payments and counterpart funds for donor projects. Hospitals that cater to a small urban minority absorbed as much resources as primary health services. The main areas of dispute that arose during the PEAP discussions included the following. First, some argued that the public sector should focus mainly on preventive treatments and on public information. However, the boundary between curative and preventive treatments does not coincide with that between public and private goods; inimunization is a private preventive good with positive externalities, just as treatment for TB is a private curative good with positive externalities, and inadequate courses of antibiotics are dubious private curative goods with negative externalities. Misinformation and financing problems arise in all these cases. In health, as in other sectors, evidence of poor performance is powerful but can be used to support opposite recommendations: either 23 to justify both the increased priority given to public service delivery, or to justify public withdrawal from service delivery. Second, views on the feasibility of substantial improvements in publicly delivered services diverged considerably. Remedies for the quality of public services might include community-level management, user fees, or the introduction of competition, for instance by transferring money to community health fimds rather than to facilities. Finally, there was no general agreement on the role of hospitals and other facilities in the delivery of health care. One underlying difference may be that cost-effectiveness analysis puts a high value on the life of children, who benefit particularly firom improved primary care. If adults are actually more concerned about adult survival, then the case for emphasizing curative care and possibly more tertiary care may be stronger. This is a case where technical criteria may not adequately reflect the underlying values of the society. 5.4.2 Incidence of expenditure The 1992/93 integrated household survey (IHS) data on health did not permit a similar incidence analysis of public expenditure on health as was done in education. The main problem is that the IHS data do not breakdown health services into the traditional categories of primary health care and hospital services. Still, the data provided a valuable snapshot of the extent to which the sick used both public and private health facilities. In education utilization was equated with the enrollment rates derived from the IHS. In the case of health, the analogous concept of utilization would be visits to primary health facilities as well as outpatient and inpatient visits to hospitals over a period of time. The IHS only gives a snapshot of utilization of health services at a particular point in time; that is, responses to illness during the 30-day period preceding the interview, not the utilization of particular health facilities over a period of time. Moreover, the IHS does not distinguish between hospitals and lower level health facilities. For these reasons, it was not possible to calculate unit costs and carry out the standard incidence analysis. Nevertheless, the IHS data provided other useful information on health care. Perhaps, the most surprising result is the popularity of home treatment. Of the four choices facing them, 38 percent of those in the first quintile who were sick opted for home treatment (World Bank 1996b). Even among the fifth quintile, 26 percent chose home treatment. Home treatment may be entirely appropriate, but it could also be an indication that some of the sick do not have much confidence in outside treatment, or that they are deterred by the costs or poor quality of outside treatment. Less than half (43 percent) of the sick in the first quintile consulted a modern health service provider. Surprisingly, very few of the sick consulted traditional doctors (2 percent overall) which certainly reflects underreporting. There is no direct evidence from the IHS on the reasons why the sick flocked to private health facilities rather than government ones. There is a widespread perception in Uganda that private facilities (most of them, mission hospitals and clinics) have an edge over government facilities in terms of staff, morale, equipment, medicines, and general 24 supplies. The network of private health facilities is extensive, consisting of 40 hospitals (compared to 55 for the government), 47 health centers, 42 dispensary/maternity units and an assortment of lower level facilities. For many users, higher quality in the private sector outweighs the higher prices; it is also possible that the unpredictability of prices in a public sector where health workers resort to illegal charges is a serious disincentive. 5.5 Agricultural extension Agricultural technology in Uganda has stagnated for a long time, and there is broad consensus on the need to improve it; major possibilities include more widespread use of draft power, improved varieties of seed, more diverse cropping patterns, and the use of fertilizer. The best way to achieve this, however, is controversial. Constraints on improved technology include access to markets, credit, information, possibly motivation, and labor supply. Sometimes when information is a constraint, the private sector has an incentive to provide it; there are a number of examples in Uganda, including the extensive tobacco outgrowing in the North-West region. Such schemes may to some extent depend on either dejure or defacto monopsony by traders, and they often involve credit in-kind paid out of the crop, such that farmers may have an incentive to sell to competing traders and default on the loans. Where monopsony is exercised, the price received by the farmers is a concern. Even where none of these concerns arise, the nonexcludable nature of agricultural information means that transmission of information to a farmer is likely to generate positive externalities, which cannot be fully captured by the provider of information. Nevertheless, the existing performance of the public agricultural extension service in Uganda is poor (World Bank 1996c). The evidence from three surveys in the mid-I 990s suggests that between 10 and 30 percent of households received extension services households (mainly through farmer's groups) and that the services were appreciated. As with education, the margin of error in estimates of service coverage turns out to be very wide. Also, the ability of agents to contact farmers has been severely constrained by limited expenses, fuel availability, and pay arrears; at one stage the government owed its extension agents roughly a million dollars in allowances. (We discuss the reasons for the financial problems in the sector in subsection 5.8). What we argue here is that the service is potentially valuable but faces financial and managerial constraints. 5.6 Transport The transport network declined during the years of general economic decline, and a major part of the road network needs rehabilitation. In the mid-1990s the government produced a national highway plan that uses a quantitative set of criteria to prioritize rates of return to improvements in the existing network. The PEAP process identified two major poverty issues that needed to be addressed in the existing strategy. First, the criteria used to allocate road projects raise some distributional concerns. There are good reasons to expect that a large part of the benefits to 25 projects that link urban and rural areas should accrue to rural areas. Consider the effects of reducing transport costs on the urban market for rurally produced commodities. Where the rural areas produce an internationally traded commodity and the internal trading system is competitive, theory predicts that the whole reduction in transport costs will be passed on to farmers in the form of higher farmgate prices. If the trading network is not competitive, things are more complex, but there is a good chance that the increased volume of traffic will be associated with more frequent visits by traders and therefore an increase in the competitiveness of the internal trading system. If the output of agricultural areas is nontraded, however, as is the case in many areas of Uganda, then the reduced costs of transport will be partly captured in lower urban market prices. Meanwhile, the region served by the road will gain and other competing rural areas are likely to lose. Nonetheless, n all these cases, at least a substantial part of the benefits are likely to accrue to the rural producers who are served by the road and who participate in the market. This raises two distributional issues. The first is whether rural households who do not engage in either the product or labor market will benefit. In practice, although the extent of market participation varies, a very high proportion of rural household need some cash, and must therefore be involved in markets in some way. The second concern, which is more important, is that the distributional effects on road improvements will depend heavily on their location. In some countries (for instance Ethiopia), poorer districts are distributed round the capital and it is hard to identify the distributional effects. In Uganda's case, however, the relative poverty of East and North makes a strong case for favoring these districts in distributional terms. In principle, the'criteria used in the highway road program should have taken this effect into accotmt, because the rate of return (based on a standard application of the highway planning model) was adjusted by a criterion of poverty. There are two problems with this. First, the use of existing and projected transport levels in the calculation of rates of return may lead to an underestimate of the relative returns to improvements in roads, which are initially in very bad condition and hence little used. Second, the "poverty" criterion, although they did include district income, also included criteria such as "importance of the region for food supply," which effectively double counted the rate of return and tended to fEavor more affluent regions. The second major distributional issue is that rates of return to maintenance appear to be higher than those to rehabilitation, yet many roads are not being adequately maintained. The reasons for this are matters of management and local political economy. For instance, when one part of a district has a totally decrepit road, it can be hard politically to defend putting money into the maintenance of the better-preserved roads in the district, even though this is the most efficient thing to do. Hence districts often spend money intended for maintenance of feeder roads (which is their responsibility) on road rehabilitation. The design of contracts also poses problems; one model that has worked in some highways in Africa is to allocate a stretch of road to a particular household, which is paid to keep the road maintained and is held responsible for its condition. 26 5.7 Water supply The provision of water sources is essentially a public good, being both nonrivalrous (up to a point) and difficult to exclude for those who have not paid. The returns to improved water supply take two main forms. First, clean water improves child health. The extent of these returns remains controversial; the 1993 World Development Report on health argued that the cost-effectiveness of this intervention is low relative to other child health interventions. Second, improved access releases household members', especially women's, time from fetching water. A simple basis for assessing the time benefits of water projects is the yalue of time used carrying water; studies from Africa (Malmberg Calvo 1994) find that water carrying can amount to as much as 1,000 hours per household per year, and that water transport involves more effort than other tasks involving transport. The economic value of this depends on (1) whether labor or leisure is foregone, (2) whether leisure is valued at the margin as much as labor, and (3) the theoretical model of the household (Mackinnon 1998). For instance, there may be a strong gender argument for increasing female leisure, where, as in Uganda (see Appleton 1996b), women have a heavier burden on their time than men. Neither theory nor Malmberg Calvo's data provide unambiguous evidence on whether women find carrying water burdensome (in some cases the social interaction involved in going to the well is valued), or on whether they would gain extra leisure time or find themselves working longer hours on other activities. These ambiguities provide a good reason to emphasize community participation in the design of water supply projects, in addition to the fact that only households know which sources they would be prepared to use. One of the main impacts of the participatory surveys undertaken in 1999 was to induce an immediate increase in the priority for water supply in public expenditure, which will be financed by debt relief from the enhance HIPC initiative approved for Uganda in 2000. Communities are expected to contribute only a small proportion of the capital costs and to take responsibility for ensuring the facilities are maintained. The success of maintenance will clearly be critical to the success of the government's program, which envisions 90 percent coverage by 2015 compared to a level of about 50 percent now. 5.8 Legal and policing services Although no quantitative studies have been done, microinsecurity presents very serious problems for the poor. Two major issues are the physical security of money, and the legal security of land. To protect these, people need good policing and a legal system they can rely on. The National Resistance Movement government in fact treated security and transport as two of its highest priorities when it came to power. The current revision of the PEAP has placed an increased emphasis on governance, reflecting the finding that both police and magistrates have a poor rating in a national household survey on corruption and integrity (Republic of Uganda 1998); the police have been found to have the worst reputation of all the major public services. 27 5.9 Reform: designing and costing a basic package The public sector delivers a very complex set of services, some of which have clearly defined end-users, some of which do not. Public spending in Uganda, however, has not been directed toward a clear set of outputs. Under these circumstances, the quality of services inevitably declines if the set of services to which the state is in principle committed to is too costly to deliver given the existing resource envelope. Two institutional factors have discouraged planners from thinking in terms of the recurrent costs of running basic services. First, the budget has often borne only a limited relation to outcomes, partly because of supplementary requests for expenditure during the fiscal year, and partly because the use of fiscal policy as the key macroeconomic policy instrument makes a smooth flow of expenditure difficult to achieve. Second, plans for expenditure are an input into bargaining both with the Ministry of Finance and with donors. Much important spending, while recurrent in nature, is on the development budget, which is mostly financed by loans or grants. Both bargaining processes can encourage inflation of costs, and this is complicated by the existence, in practice, of different cost structures on donor-funded projects and locally funded spending. The idea in the PEAP was to design a basic package of services which everyone in the country should have access to at a price that they could afford. Costing attempts need to be included at the stage of drawing up options, rather than merely being addressed when a certain level of service has already been committed to. The relevant notion of costing, although simple, is often misunderstood. Whhat policymakers need to know is the recurrent cost of a given level of service per head of population per year. Unfortunately, the focus of many discussions between donors and governments are capital costs of a particular activity within a service, provided to a targeted population which is often only a proportion of those who need the service, and with incentives toward inflation rather than minimization of costs on both sides. Hence the process of bargaining between donors and government can actually distract attention within the countly away from the most fundamental issues of planning. The technical aspect of costing is of very varying difficulty across sectors. In road maintenance and the provision of rural water supply, costs depend on local physical conditions; for instance, the costs of providing rural water supply in Uganda are proving higher than anticipated. The health sector is particularly difficult to cost, for three reasons. First, the actual prevalence of the major diseases is only approximately known, since the sample offered by hospital patients is not a representative proportion of the population. In the case of HIV, estimates of seropositivity are available from random surveys, but this is not the case for most illnesses. Estimates from the demographic and health and other surveys are highly sensitive to patient perceptions of normal behavior in areas such as diarrhea, malnutrition, or fever. Second, the costs of treating any particular illness are very sensitive to the prices of drugs, many of which are imported, even where optimal treatment regimes are actually delivered. An extreme example is provided by the very recent emergence of apparently life-saving regimes for HIV which cost at least several thousand 28 dollars per patient per year; a less dramatic example is the choice between long- and short- course treatment for tuberculosis, and the choice of diagnostic method in malaria. The design of a health package therefore involves costing various options for treatment of particular diseases, identifying those that seem feasible within some overall assumption about possible flows of expenditure. Third, the costs of preventive actions such as information are difficult to estimate. At the time of the PEAP, such estimates were not yet available, and therefore the best that could be done was to use $12 per capita per annum estimate from the 1993 World Development Report as a basic guide to the likely cost of the basic health package. Education and agricultural extension are much easier to cost. The most important point is that the cost of agricultural extension is surprisingly low in terms of the government's budget; generous assumptions suggest that a reasonable extension service could be provided throughout Uganda for a total of about 20 million dollars per year, a fraction of the cost of education or health packages. Currently, a more detailed costing exercise is being undertaken in the context of a restructuring of agricultural services which is allowing much more local management. A relatively approximate costing exercise revealed two things. First, some services could be delivered cheaply if they were given a slightly higher priority in the government budget and managerial problems were dealt with. Examples include agricultural extension and feeder road maintenance. The public agricultural extension is recovering from decades of stagnation, but as noted above performance has not met expectations. There is a persistent institutional problem here; while all agree on the need to improve the technology used by farmers, not all agree on the appropriate means. For instance, the Ministry of Agriculture has considered implementing a large project to distribute subsidized seed to promote the growing of maize and beans. The existing public extension service has been supported by substantial donor projects in recent years; however, the government is obliged to provide both counterpart funding for the project and to pay the wages of the extension officers. During decentralization, the number of extension assistants was drastically reduced. The fundamental problem here is that a project designed along these lines cannot function optimally unless the government at central and local levels is convinced of its merits and give it the high priority it deserves in terms of expenditures. But until the project functions optimally, it is very hard to demonstrate adequate success that will reinforce its support. Probably the main lesson of this experience relates to project design. Where projects are designed with separately funded components that are indispensable, and where one party will not step in if the other cannot pay or will not pay their share, the projects are exposed to multiple risks of failure. The justification for this arrangement is that the government needs to pay for some of the project to demonstrate its commitment. But if the government itself is not yet fully convinced, , it makes very little sense to make a project so dependent on its contribution; when the project receives low fiscal priority, all that is demonstrated is failure. A better model would be 100 percent external funding in the early 29 stages of a project, followed by transfer to the government when success has been demonstrated and the popularity of the service ensures it a political constituency. An alternative is to avoid funding any project unless it is part of a sectoral program which is within the government's medium-term resources constraint and which the government is committed to. Since 1997, iIn accordance with the PEAP, sectoral programs have become the preferred model for government and donors in Uganda; in agriculture, the recent multi-sectoral Plan for the Modernization of Agriculture is very clearly a government-led document, and agricultural services within this framework should therefore not suffer the same problems as in the past. The choice between these different approaches is not easy. If governments are already seriously committed to issues that donors consider important, then respect for a sectoral expenditure program is clearly best. But there will be cases where donors or lenders hope to convince the government of a service's value, so that aid plays a valuable pump- priming role. And then there are the cases where only temporary funding is needed: for instance, the mass vaccination for polio. External agencies should seriously consider 100 percent funding for such efforts. The existing mechanism of cost-sharing between government and external agencies on projects which the government is not wholly convinced of is a recipe for failure, and problems similar to those in Uganda have been widely reported in agricultural extension services in other African countries (see Purcell and Anderson 1997). The second thing that the review of costings revealed was that in many cases existing expenditure is clearly inadequate to achieve universal coverage. This calls for either a major reallocation within public expenditures or enabling the private sector to meet the gap, for example, by targeting a subsidy at the poor or at public goods. The PEAP called for reallocation. 5.9.1 Delivering the basic package Estimates based on the recurrent budget and investment plan in 1996 suggested that about 23 percent of total public (including donor) recurrent spending was for four basic services: primary health, primary education, feeder road maintenance, and agricultural extension. Although these figures are approximate, and the picture on the capital budget could be better, they do suggest that there is scope for increasing this share. Since revenue was growing relatively fast, total recurrent spending could rise, provided that the amount of recurrent spending funded through the Public Investmnent Plan by donors is not reduced. Given the particular importance of recurrent spending for service delivery, both government and donors should accept this as a feasible approach. Simulations showed that freezing spending on other items would allow a very substantial increase in the amount going to basic services within a few years. Under the government's decentralization program all basic services are district responsibilities. Financing an increase in these services therefore requires an increase in the spending going to the districts. It is also necessary that the districts actually spend the 30 money on the basic services, which the poor need. There are essentially two ways of achieving this; conditions imposed by central government, and bottom-up pressure from an informed electorate. While the second is more sensitive to diverse local needs, it relies on a local political apparatus with little experience in handling such substantial flows of resources. An argument for retaining some top-down control it to see government as a partial altruist; if officials at the center have no direct way of gaining directly from the funds going to the districts, there is a better chance of their acting impartially to ensure that funds really are spent on the intended purposes. It is essential that people know what services they are entitled to. In the case of education, the President's commitment to educate four children in each family from 1997 onwards was accompanied by a clear and public delineation of responsibilities between government and households. This clarity is harder to achieve in other services; one lesson may be that for a service such as health, the availability of specific treatments for specific health problems should be publicized, rather than a general commitment to providing health care. Without such precise publicity, the patient may not know whether they are being denied treatment because there is no clinical need for them, because the treatment is too expensive for the public sector to provide, because the facility has run out of drugs, or because the clinician is deliberately attempting to send the patient to his own private clinic. If the patient does not know which of these is the case, they are in no position to demand treatment or to draw people's attention to abuses in the system. Similarly, in the case of agricultural extension, it is irnportant to define some level of coverage that would represent universal access to the benefits of extension. 5.9.2 Cost reduction in service delivery Costs to the public sector can be reduced in a number of ways. First, the wage structure is currently under review. It has been necessary to raise wages for schoolteachers, and the low salaries for nursing aides have also been addressed, as well as the effectively separate levels of pay for central and local government officials. The gap between donor and government pay scales still needs to be addressed. One danger is that salaries in some areas could be pushed up more than is needed to recruit staff of adequate skill and quality, increasing the costs of delivering labor-intensive service such as education. Similarly, in the health sector government pay increases have adversely affected the NGO sector. Another difficulty is the creation of incentives for public sector employees to work in remote areas. Nonpayment of salaries has been a problem in the past, a clearly inefficient and demoralizing form of borrowing which the government has promised to eliminate. Probably more important, the high proportion of donor contribution in recurrent spending raises some management issues. At the moment, it is not desirable to end donor funding of recurrent expenditures (whether these are presented under the recurrent or development budgets) because this further squeezes funds available for recurrent expenditure. In general, donor spending should be integrated into the medium-term expenditure framework and sectoral programs designed to deliver basic services. Currently 31 the wide range of initiatives actually undermines the coherence of public action. Moreover, donor-funded projects very often observe completely different cost structures from the rest of government. For instance, project offices are rarely short of supplies, and where donors are free to recruit at their own pay-scales, there is often a very wide gap between their pay scale and official ones (for instance, donors' drivers may be paid more than senior civil servants). Even where donors try to avoid extravagance, it is quite hard for them to imagine the tightness of the resource constraint faced as a daily problem by most government departments. Governmental efficiency is certainly affected by corruption, which is hard to quantify. In some cases popular perception of corruption is actually a consequence of complex accounting systems which make the flow of funds irregular. There is almost certainly more scope for donors to urge government to remove specific officials whose corruption is established, and many Ugandans felt strongly that the donors should act more strongly on this. One important point is thalt the terms of employment for civil servants make dismissal a very difficult and bureaucratic procedure with a high standard of proof. Hence civil servants may remain suspended, and sometimes with access to their offices so that records can easily be removed, long after their malpractice is known. A more effective civil service reform program could improve discipline, but the constitutional establishment of the Civil Service Commission might merely intensify the bureaucracy involved. In Britain, the downgrading of the standard of proof required to dismiss corrupt police officers has been an important and popular part of the government's attempts to make the police service less corrupt, and this might also be popular in Uganda. The scope for targeting basic services by charging users and exempting the poor is uncertain. It would be relatively easy to charge urban households more than rural ones; different charges in different regions would also be distributionally justifiable but could be politically divisive. The ability of local service providers to identify and exempt the poor is not clear, and it is likely that such schemes will in practice exclude some of the poor from the service. Of course, this may be better than failure to deliver the service at all. T'here is scope for using more local contractors and materials in basic capital investments such as classrooms. Equally, NGOs may well be able to provide some services more cost-effectively than the state; it is proposed to establish health maintenance organizations (HMOs) to explore these options in the health sector. Local govermments need to be encouraged to consult NGOs throughout the planning process and delegate functions to them where and when this is the most efficient way of delivering services. 6. Land, Credit and Security of Assets The speed of poverty reduction depends on the access of the poor to productive assets. The most important asset for many Ugandan households remains land, and hence policy toward land has an important bearing on poverty reduction. Land policy issues are much more 32 complex than a choice between laissez faire and egalitarian redistribution. Land ownership is the outcome of a political and legal process, as well as economic forces, and the interpretation of existing land rights is contentious. Hence the definition of the legal framework is important for the welfare of the poor; and the principle of respecting property rights is not enough to determine policy where policy rights require interpretation. The PEAP of 1997 says relatively little about land issues but instead focuses on economic policy and public services. However, there was a parallel consultative process and a lively national debate on land going on simultaneously with the PEAP process as new land legislation was being drafted, following the promulgation of the 1995 Constitution which set parameters for a land policy but left the details unresolved. This section explores the issue of land extensively, given its importance to poverty reduction. It also briefly touches upon access to credit and financial services as well as the security of productive assets. 6.1 Existing land rights Traditional land rights in Uganda, as in other societies, are highly complex and diverse. Four main forms of legally recognized tenure exist, in addition to at least two important forms of unrecognized tenure. First, mailo land: this land, which includes much of the best land in the country, was vested in about 1,000 members of the aristocracy in 1900, but much of it is under a form of tenancy known as kibanja which confers use rights and security and is tradable. Rents on these holdings were fixed in 1928 at levels, which became insignificant in real terms. In 1975, however, Idi Amin's land act officially abolished tenancy, obscuring the legal status of mailo land. The result is that the relative status of kibanja and ttaka claims on mailo lands is open to dispute. So far, this has not led to an enormous amount of conflict on the ground (Kigula 1993) because in most cases the kibanja tenants have not been evicted, but there is clearly a disagreement of principle between those who would like to see the kibanja holders simply as tenants of the ttaka holders, and those who argue that the kibanja right is more valuable than a commercial tenant's right. This dispute is of acute political sensitivity because the mailo owners are Baganda; they may feel that exclusive attention to mailo without attention to other forms of inequality is ethnically discriminatory (though naturally Baganda kibanja holders may feel differently). For instance, the question 'if you are going to tax land, why not livestock as well?' is sometimes asked, with a subtext which reflects the different regional profiles of these forms of wealth. Customary tenure is the term generally used to refer to traditional land rights. These lands are individualistic to varying degrees, and in some cases had already become tradable before any legal titling was introduced. In land-scarce, highly commercialized areas such as Kisoro purchase and sale of land is very common and straightforward and in some cases freehold has been issued (Republic of Uganda 1996b); in other areas there are significant 33 restrictions on transactions.6 Also, there is often a distinction within a single land system between individually held land and common property.! Freehold tenure exists under ordinances passed in 1900 and 1901 for Ankole and Toro. By the terms of the 1975 Land Decree, however, all freehold was transformed into leasehold held by the state. Leasehold tenure is extensive in Uganda. Under the terms of legislation passed in 1967, customary tenants were protected from eviction by leaseholders. The 1975 Land Decree, however, radically changed the form of tenancy, vesting all land in the state and effectively recognizing neither customary nor mailo tenure, and reducing existing freeholds to leasehold status. Encouraged by government policy, titling of leasehold properties seems to have accelerated during this period. Leasehold and freehold titles have thus been issued on a significant part of the country, but it is not known how much. The legality of many of the land titles issued on formerly public land is unclear. Until 1975 customary owners had security of tenure on their land legally; but from 1975 to the present, they have been technically tenants on sufferance on public land. Nevertheless, several steps-including surveying and compensation-are required before customary occupiers can be evicted. Some leases have been issued without going through these steps; in others the facts are disputed. This dispute concerns both agriculturalists and nomadic pastoralists, and in some cases this form of dispute has become violent as in the Sembabule dispute described by Ddungu (1994). In addition, there are two forms of rental which have no legal recognition but are widely practiced, known as "borrowing" and "renting" land. Typically, renting is a form of commercial tenancy with some limited security; borrowing may involve some obligations in-kind and comes with very little security indeed. Typically, the landlord will prohibit the use of permanent structures or crops to protect his land rights (Muhereza 1992). Where (as very often) an owner has too much land to farm himself, it provides an alternative to wage labor, which involves lower supervision costs. In some cases the borrowing arises within an ethnic group; in others, notably in Kabarole, immigrant ethnic groups from land-scarce areas have borrowed land from local landowners (Kigula 1993). In addition to the disputes concerning mailo and leasehold land, the rights of women and children can give rise to disputes. For instance, land is traditionally meant to pass to a dead man's children, but if they are minors the land may be vulnerable. Intrafamily disputes have been recorded in Kigula (1993). According to the latest household survey of 1999/2000, land conflicts exist in about half of the communities (Deininger and Okidi 2000). Unless ways are found to establish a cost-effective way to implement the recently 6 The pattern of land scarcity and commercialization increasing the individuality of indigenous rights is well known in Africa ( Migot-Adholla, and others 1991; Bruce and Migot-Adholla, eds. 1993). ' Confusingly, individualistic customary ownership is sometimes called kibanja although the term will be used here to refer only to tenants on mailo land. 34 passed land act, these conflicts, together with other tensions, could easily threaten social stability and development. The 1995 Constitution recognizes four forms of tenure, customary, mailo, freehold and leasehold. It also, unusually, makes a specific commitment to introduce legislation to resolve the disputed relation between ttaka and kibanja tenants within two years (the issue turned out to be too controversial to settle within the Constituent Assembly). 6.2 The existing distribution of land The aggregate amount of land available in Uganda has been estimated at 18 million hectares; this would represent 6 hectares or more per farm household. Nevertheless, in 1990 the total area cultivated was estimated at only 4.6 million hectares (World Bank 1993). We can distinguish a number of possible reasons for the limited amount of cultivation: (1) limited capacity to farm a large areas imposed by technology, (2) limited access to product markets which reduces the incentive to produce, and (3) limited access for most farm households caused by the combination of severe asset inequality and poorly functioning factor markets. All three play a part; the importance of the third element is widely underestimated in Uganda. It is widely believed either that Uganda's land distribution is relatively equal, or that there is a relatively equal distribution within each region although a marked differential across land-scarce and land-abundant areas exist. Unfortunately, no aggregate Gini coefficient or other inequality measures are available either for the distribution of land or, perhaps more relevantly, for the distribution of wealth. What evidence exists suggests considerable inequality both between and within regions. We can consider inequality at the bottom or at the top. At the bottom, both large- scale data and village level studies suggest that the proportion of households with no access to land was less than 5 percent in 1992 (Appleton and Mackinnon 1995). These households include some better-off households, for example, government officials, who typically have compensating forms of wealth such as human capital. As a result, rural households that farm no land are on average better off than other rural households. Village-level studies, however, suggest that there is a poor wage-earning class, about 10 percent in a study in Lango (Opyene 1993) and 6 percent in Kisoro (Mugisha 1992). In some cases landless people migrate, for example, to fishing villages (Muhereza 1992); Appleton and Mackinnon (1995) note the relative high proportion of rural nonfarming households in Kalangala and Jinja because of fishing. But there are four ways other than the transformation of smallholders to "landless labor" in which poverty is severely exacerbated by inadequate access to land. First, many households have access to only small plots. Data from the 1992 IHS suggest that the median area cultivated per adult equivalent is 0.6 acres; for a typical household of five people including children, this translates into less than 1.5 acres. World Bank (1993) finds 35 that area cultivated is about 0.3 ha per head but varies widely across districts. Second, the amount cultivated includes land borrowed on terms that preclude the construction of permanent structures or perennial crops. Examples are found on mailo land owned by the church in Kalangala (Ssenkumba 1993) and in the practice of "borrowing" land found throughout Uganda (Muhereza 1992). Vill]age-level studies indicate that next to landless labor, the poorest group is those who have no land of their own and can only borrow land from landlords. Third, a considerable proportion of land cultivated is borrowed or rented; so many households are paying rents for a significant part of the land they use. Rental rates have not been extensively studied but Mugisha (1993) reports rents varying from 10 to 50 percent of the value of production in the land-scarce South-West. For a poor household, these rents can irnpose a very severe burden. Fourth, people may be driven into desperate actions to acquire land. This applies particularly to women who have no land rights in their own right in parts of Uganda; rural women can therefore become dependent on marital or extramarital relations with men to ensure an income for themselves and their children. Anecdotal evidence indicates that such pressures can lead to an increased incidence of HIV among poorer women. At the top, there is a significant amount of land in the hands of very few. Sampling problems, absentee residence and difficulties in interpreting outliers in large-sample surveys can make it hard to detect this phenomenon; these therefore need to be complemented by the village-level studies. In one such study iin Lango, 87.7 percent of the land in one village is owned by 2 of the 127 households (Opyene 1993). In one village in Kabale, 33 percent of the land is owned by 7 percent of the households (Mugisha 1992). In Masindi two-thirds of the land in one village was owned by one landlord (Bazaara 1992). While these cases are extreme, all the village studies reveal more iinequality than expected. Two aspects of these large landowners are worth noting. First, the source of their ownership is often past access to political power, both for mailo owners and for leaseholders. Mugisha (1993) finds that the existing inequality arises from the allocation in the 1930s by the state of titles to what was then an uncultivated, though communally used, swamp; seven households benefited of which five were either administrative chiefs or state functionaries. In Apac, of the leases issued between 1976 and 1991, 37 percent went to individual farmers and traders, whereas 54 percent went to members of Parliament, government officials, policemen, soldiers, land board members, and District Council members (Opyene 1993). The lease discussed above which led to the extreme concentration of landownership in a village in Lango was issued to a police officer in President Amin's regime. In Bunyoro, certificates of occupancy issued in the 1930s formalized the land rights of the aristocracy of the Bunyoro from whom most tenants were renting land (Bazaara 1992). Land sales have been much less important than political power in causing land inequality. The people who have most typically bought their land in Uganda are kibanja holders who have bought their kibanja rights and have in some cases gone on to buy out the ttaka owner as well; also, many people in the South-West region have traded in the active 36 land market there. However, money does enter into the issuance of titling where district land committees (DLCs) are corrupt or where the condition of issuance of the lease includes demonstration of financial capacity to develop the land (Kigula 1993). Second, large landowners are not always interested in farming. There is a significant category of absentee landlords. But there are also rural owners who are not particularly interested in farming. For instance, in the village in Masindi where one household owns two-thirds of the land, most of the land is rented out to pastoralists for grazing. Muhereza (1992) reports that most of the leases issued in Luwero went to old people who let their fields out to others. Where land is rented out landlords sometimes bring in tenants -from outside because, as one study says: "indigenous people are difficult to evict" (Opyene 1993). Since the inequality in existing holdings reflects past political power rather than market forces, it is not likely to promote efficiency. Indeed, in the mailo areas where trade in kibanja tenancies has been extensive fragmentation rather than concentration has been the pattern (Mukwaya 1953 cited in Marquardt 1994); the same seems to be true in the South-West where trade is extensive and inheritance practices encouraged subdivision. The current evolution of land distribution is even harder to ascertain than the existing distribution. However, five processes are likely to be operating. First, widows' land rights often depended remarriage to the husband's brother: AIDS has disrupted this mechanism. Male orphans' land rights although traditionally protected may also be vulnerable in practice. Second,. land is widely seen as a very attractive investment by urban residents. Third, continued population growth is increasing pressure on land, causing cultivated area to expand sometimes at the expense of soil fertility, and causing rural-rural and rural-urban migration, as well as increased of "borrowing" and "renting" of land, as small producers run out of their own land and turn to larger owners for tenancy. This process is driving up the returns to holding large landholdings and is therefore likely to increase inequality and the efficiency costs of inequality. In this sense, premature allocation of titles to "vacant" land in the 1930s, which seemed distributionally innocuous at the time, may be generating inequality in the 1990s. Fourth, district land commissions have continued to function and presumably to issue titles. The proportion of Uganda that has been allocated to large landholders is therefore probably rising constantly. Finally, the emphasis on draft power in the government's modernization strategy for agriculture is intended to increase the areas that farmers can cultivate. One consequence of this will be to increase the demand for land; it is important that land pclicy takes this into account. 6.3 Economic consequences of the existing distribution The extent to which land is a constraint on productive activities is much disputed in Uganda; many people argue that labor is the main constraint, especially at the existing level of technology. Data from the IHS shows that the amount of land owned is a significant determinant of welfare across households, and that the poor in most districts (including AIDS-affected areas and land-abundant areas) have higher labor-land ratios than the rich 37 (Appleton and Mackinnon 1995). Nevertheless, the bivariate correlation between landowner and welfare, though positive and significant., is small. Aggregate figures indicate an increase in productive area but also a decline in productive area per head. Finally, the village-level studies strongly suggest that poverty writhin the village is strongly influenced by landholding, though this involves terms of landholding as well as amount. The inefficiency caused by the existing pattern of land distribution may be considerable, and there are two main channels for this. One is inefficient allocation of land to farmers, inducing a divergence of marginal products across different farmers. The diversity of land-labor ratios across income groups suggests that the marginal product of labor is higher on small farms; if so, Uganlda is showing the same pattern as many other countries in Africa and elsewhere. Also, the marked regional divergence in the availability of land and the large areas of unused land strongly suggest that the allocation is inefficient. The inefficiency of the allocation is not corrected by factor markets, partly because large landowners are not particularly entrepreneurial, and frequently absent, and therefore do not employ wage labor as much as the allocation suggests would be needed; and because the institution of rental, though it exists, is insecure for both parties and the suspension of all rental on mailo land has reduced the supply of land for tenancy. The other way in which existing rights may generate inefficiency is that the nature of land rights may reduce or distort investment. This is a much more complex issue than generally appreciated. One channel that mLay be quickly disposed of is that land titling would enable the use of land as collateral. Evidence in other African countries indicates that the use of land to secure formal lending is very limited, in contrast to experience in Thailand (see the study by Migot-Adolla and others 1991 of Ghana, Kenya and Rwanda). One reason for this is that land rights are often enough disputed and land markets thin enough to make foreclosure unrewarding for the lender (the Thai case was atypical because none of these features operated). In Uganda's case titled land has been used as collateral, though the evidence suggests that this has usually financed investment in urban activities where returns are quicker and safer than in agriculture. In any case, it seems very unlikely that the rural poor would be interested in mortgaging their land on a large scale, in Uganda or anywhere else, if they believed that defauilt would really provoke foreclosure. Hence this argument, which has been used prominently as a justification for the titling of customary land, does not seem convincing. The second, and much more important, possibility is that the disputed and unclear status of land rights may affect investment. On the one hand, the standard argument is that insecurity reduces investment because the investor fears they will lose the value of the investment. On the other hand, if investment strengthens the investor's claim on the land, the exact opposite may occur (Besley 1996). For much kibanja land, this seems to be exactly what has happened: on mailo land in Luwero, "To increase their security in their holding, these tenants planted the greatest percentage of their holdings with perennial crops such as coffee and bananas in order to make the profitability of eviction appear costly in terms of compensation" (Muhereza 1992). T he point is that the if the ttaka holder wanted to 38 evict the kibanja holder he would then have to pay compensation for the investment. Finally, a third possibility is that the landowner may prevent the tenant from investing because he fears he will lose control over the land. This is documented in church-owned mailo land in Kalangala (Ssenkumba 1993) and widely in non-mailo cases. This inefficiency radically undermines the purpose of agricultural reform, which is to encourage small farmers to invest in commercial agricultural activities. Much discussion associates the efficiency impact of tenurial insecurity with mailo land. However, the evidence suggests that mailo land functioned relatively efficiently in Uganda until the late 1960s. Mailo land was subdivided extensively by smallholder farmers, and those farmers who bought kibanja and in some cases went on to buy ttaka as well became one of the most dynamic elements in Uganda's economic progress. Indeed, the experience is used by Feder and Noronha (1987) as a model example of a secure land system. The examples of inefficiency given above are associated at least as much with other forms of title as with mailo. Finally, to what extent does landownership drive income and expenditure inequality in Uganda? If it is accepted that land is a significant constraint on production for many households, then the impact on income follows naturally. Appleton and Mackinnon (1995) find that land does significantly affect household welfare though the bivariate correlation is low. Village-level studies confirm this impression, though the view has been expressed that the old ttaka owners include some nonentrepreneurial characters who have done nothing with their land and are now extremely poor as a result. Interregional differences, however, are not well explained by land per capita; the most land-scarce areas are not the poorest. 6.4 What legalform should land rights take? Land policy needs to begin with a legal framework that articulates the legal mechanisms available to agents. Although some commentators have argued for uncompensated expropriation of large owners in Uganda, the government has a clear commitment to restore the confidence of national and foreign investors in property rights and would therefore not consider this step. Moreover, if attempted, expropriation could very easily ignite serious ethnic conflict. Hence the feasible space of propoor policy is that which respects existing property rights. The Constitution provides a basic framework whereby the four existing forms of land ownership can gradually be converted into freehold at a locally determined pace. The parameters set by the Constitution are broad, however, and there is considerable room for both legislation and other forms of public action to affect the outcomes. 6.4.1 The impact offormal titling on poverty Although titling is not demonstrably necessary to give security or to increase investment in general (see Migot-Adholla and others 1991; and Atwood 1990), it has three theoretical advantages: security for the farmer, collateral for the investor, and ease of migration for 39 people in land-scarce areas. As we have argued above, the collateral issue is, at least for the poor, not a strong argument. The reduction of insecurity and the increased possibility of migration, however, do represent potentially significant benefits. If poor people in land-scarce areas can move to occupy land in more abundant areas, this could reduce poverty. But given the transactions costs, the poor will not be able to migrate without assistance of some form. Security provides a major argument; insecurity is widespread, and affects not only existing tenants' situation but also reduces the supply of tenantable land. Therefore, titling could in principle do some good by improving the functioning of the important rental market. The down side of titling can reduce the rights of the poor in three ways, all documented in Uganda. First, someone not previously considered to be a landlord may get title to a land which others are occupying, and proceed either to evict them or charge them rent. While safeguards have existed to prevent this from happening, they have not always been effective. For instance, the large lease given to the police officer in the village in Lango discussed above (Opyene 1993) was on communally owned land that had been used for cultivation and other purposes. The landlord assured the elders that he would allow this to continue, but has since succeeded (partly by importing tenants from elsewhere) in enforcing control over the whole area. Second, title can be declared on communally used land such as swamps or pastoralists' grazing land. Again, rights have often been overridden in Uganda's case: the Semababule uprising arose partly because of the loss of grazing rights by nomadic pastoralists (Ddungu 1994). Arguably, the titling of land by aristocracies in Bunyoro and Buganda had elements of this because although these aristocracies had traditional rights, they differed from the rights of a commercial landlord. Fonnal titling may thus confer on traditional ownership a meaning it did not have traditionally. Third, title can be declared on vacant land which at the time of the titling is not in demand; as population pressure increases, this land comes to be rented out by the person who acquired the title, and those who wish to cultivate the land then have to pay rent. An example of this process is given in Mugisha (1992); title was given to some swampy ground in the 1930s which was not occupied at the time. Now, after population growth, the holders of this title make a large amount of income as landlords. This third process, unlike the first two, is clearly legal, and accounts for the development of inequality in some areas. Since the holders of leases in Uganida are very often not farmers, the history of titling in Uganda is at best a double-edged one; titling has probably done as much to expropriate the poor as to promote development. Ironically, the much-maligned operation of mailo between 1928 and 1975 was probably the most benign case of titling in Uganda's history. 6.4.2 Theform of title The Constitution recognizes four forms of title but strongly suggests that most land will move toward freehold. However, other fonns of title are mooted including certificates of occupancy, which most people are expected to settle for initially. The first issue is the 40 merits of freehold versus leasehold. For poor smallholders, leasehold is likely to be an unsatisfactory form of title; if extensions are costly they will lack the capital to extend their lease when it runs out; if extensions are free, there is no great appeal in undermining their security by limiting the length of title. For larger holdings, however, leasehold has a considerable distributional advantage because it allows land to revert to the owners and to be reallocated at the end of the lease. We saw above that some titles granted in the 1930s are now generating considerable inequality and inefficiency. Admittedly, this argument is somewhat balanced by negative incentive effects. But the District Land Boards should at least be encouraged to consider seriously the leasehold as a form of title. The reasons for giving a separate form of title to customarily held land as opposed to freehold presumably have to do with the difference in the tradability of customary land. This is acceptable provided that customary land is really protected from the imposition on it of a higher interest. This requires both adequate legislation and effective enforcement. If freehold is to be the main form of tenure, what form should it take? Major issues here are the decisions about sale and practices associated with divorce and inheritance. Husbands' and wives' interests may conflict if the husband is very ill, if he is imprudent, and if he wishes to marry another wife. There is good reason to require the consent of immediate family to land sales for these reasons. This can be done either by vesting the land in the name of several individuals, or by constituting the family as a legal entity and vesting the land in it, or by simply writing the requirement for consent into the land legislation. A proposal that caused considerable controversy in the mid-1980s, and was physically shouted down in the Constituent Assembly, the right of women to have a specified share of household assets in the event of divorce. Bargaining models of the household strongly suggest that this would benefit not only divorced women but also the bargaining position of married women within the household. Household survey data suggest that households headed by divorced women are better off than the average household (Appleton and Mackinnon 1995); the likely explanation, however, is not that divorce is a good option but that it is a very bad option for most poor women to consider. The inheritance rights of women and children in the event of the death of a male head of household require protection. Where no will exists, customary law, in most cases, vests the land in the sons. Legislation could perhaps help by specifying rights of widows and of daughters where there is no will and no son to inherit, or where the head of household's sons expropriate the widow (who may not be their mother). Widows are markedly worse off than the rest of the population (Appleton and Mackinnon 1995; Republic of Uganda 1999). At the same, proposals to discourage the subdivision of estates among the young should be avoided; inefficiencies caused by excessively small plots can be avoided provided the siblings cooperate in production, and the proposal would significantly worsen an already unequal distribution (Republic of Uganda 1993). All of these inheritance proposals-especially those supporting female rights on divorce and inheritance-encountered considerable resistance as many Ugandan men felt that inherited land should stay in the male line of the family. As has been noted elsewhere, 41 any legislation that is too far removed from local rights may have no impact on actual behavior (see Pinckney and Kimuyu 1994 for findings in Kenya and Tanzania). While these protections are valuable, two caveats are in order. One is that parents can sometimes extract more remittances from their children through controlling the inheritance of land; this effect has been studied in Botswana (Lucas and Stark 1995) and Kenya (Hoddinott 1992). This can be good for poverty in the case of the poor elderly when land is their only asset. Second, the suspicion sometimes expressed that land markets will lead to land sales by the poor and resulting impoverishment is theoretically justified, but does not receive great prima facie support from the empirical pattern of land trade in Buganda under mailo land, nor from the fact that expropriation of the poor has been caused much more by administrative than market mechanisms in Uganda's history. The effects of the tradability of land will depend on the policy context, and whether its advantages outweigh its disadvantages will depends on the whole poverty reduction strategy. In the event, the land act included a clause requiring consent for land sales on the part of adult children and spouses, but a clause providing for co-ownership of assets by wives was dropped at a late stage of discussion among considerable substantive and procedural controversy. While the powerful lobby for the co-ownership clause has not achieved all its objectives, it has definitely put this issue on the agenda (whereas an earlier proposal that women should have a right to a specified share of household assets in the event of divorce was physically shouted down in the Constituent Assembly, as mentioned earlier). 6.4.3 Who should get newfreehold? The discussion above suggests a number of principles when considering the merits of leasehold and freehold tenure. First, land titles should not be issued to someone who is not the customary owner without the agreement of those who are occupying the land. Second, the issuance of additional large titles shoulcl be restricted except in cases where titling either generates a large volume of employment or offers benefits to smallholders.8 Where such benefits are promised, they should form explicit conditions for the lease to be valid. Such a policy resembles a flexible land ceiling on new titles, which raises different issues from imposing land ceilings on already issued titles. Third, a freehold title may be justified where land is unambiguously owned by a customary owner and rented out, but should be conditional on tenant's security at reasonable rentals. New title to large pieces of land should not be given to those who will leave it idle. It is not clear from the literature, however, how much land really falls into this category. The establishment of these principles could do much to enhance the efficiency of land use in Uganda. The appropriate balance between legal and administrative mechanisms 8 For instance, a large ranch offering facilities for slaughtering animals to smallholders with livestock; this has happened very successfully in some countries in the Pacific. 42 for achieving them is an important issue requiring further work. Many of the concerns identified could be addressed by a process of systematic titling, which makes local consultation easier than the opportunistic titling of the past. Whether or not there is a demonstrable productivity benefit, a process of systematic titling might have considerable merit as a means for protecting the land rights of the poor, which are otherwise vulnerable to the inherent abuse of opportunistic titling. 64.4 How should existing leases be dealt with? It has been suggested that all leases owned by Ugandan nationals should be converted to freehold. One argument for this is that some leases are on properties that were freehold before 1975. The merits of automatic conversion for leases which have been issued this way are not entirely clear; the policy will amount to a costless increase in wealth for the better off in a fiscally constrained economy, so something can be said for charging for the conversion. More important issues relating to the conflict between owners and occupiers can be illustrated in the following three areas. First, leases that were issued illegally should be revoked; however, legal challenges in practice may be difficult. Legal aid to the poor in pursuing such cases was proposed in the Poverty Action Plan but mechanisms are not yet in place. Grey areas may arise where land leases were legal under the 1975 Land Reform but arguably unconstitutional because they expropriated occupants of the land. Second, leases that were issued legally but are responsible for both inequality and inefficiency pose the most difficult policy problem; legal recognition of these rights may have to be accompanied by other actions. Third, although the Constitution defines mailo as a distinct form of tenure, this is controversial; many mailo owners would prefer simply to be regarded as the owners with tenants. As we already pointed out, this view is a misrepresentation of the rights that mailo owners and kibanja tenants had under the legislation in force from 1928 to 1975. It therefore seems most just to treat mailo as sui generis and to restore the terms on mailo land, including the restrictions on rent, which operated between 1928 and 1975. In many cases kibanja holders may eventually buy out their ttaka owners, but provided the system has credibility, the evidence suggests that the mailo system is compatible with efficiency. 6.4.5 Mechanisms of enforcement In designing a land law, mechanisms of enforcement are as important as the structure of land rights. If these mechanisms are remote from communities, there is a greater risk of communities' needs being ignored in the decisions. But a substantial administrative structure at local levels can be prohibitively expensive. Even if a network of local land institutions is established, the location of powers of decision is important, as are requirements to disclose applications for title in the locality of the land affected and to ascertain local circumstances before allocating title. 43 The allocation of title to applicants raises policy questions about land use. For instance, one district land board might ltake the view that smallholder agriculture was efficient and refuse applications for large leases if there were competing smallholder uses for the land; another board might take exactly the opposite view, encouraging extensive eviction of existing occupants to make way for large tenants. Given the widespread identification of 'large-scale' with 'commercial' in Uganda, there is a reason to believe that many district land boards would be uncluly inclined to favor large developers and be vulnerable to bribery. Local land use plails are to be prepared at all levels, which should provide an opportunity for land use to be debated locally, but it may still be difficult to ensure that the decisions of land boards actually respect the priorities identified by policy. Both these arguments suggest that democratic control of the land boards is crucial. Democratic control does not mean intervention (political or otherwise) in particular decisions but rather ensuring that the decisions can be audited and that clear policy directions are provided to the land boards. Also, the Inspectorate General of Government should be required to inspect the integrity, of the land boards' procedures. Dubious land allocation procedures have been a source of corruption in many developed countries and the importance of these decisions make this a ligh-priority issue within the legal system. 6.5 Areas for further public action In addition to establishing the right legal framework, various forms of public action have been proposed. The first and least controversial, is the provision of legal services to the poor to ensure that they are aware of and able to exercise their legal rights. Whatever the legal status, titling endangers the rights of the poor if they are not informed about it or if it is performed at a level of the judicial system to which they do not have effective access. Providing information is relatively cheap; providing legal services is more costly. It was strongly argued during the PEAP process that information on legal rights as well as legal services merit inclusion in the basic package of public services. However, currently even most criminal cases do not actually receive legal aid because of budgetary constraints. Second, taxation of land is a policy option. Sensitivity on this proposal, however, is considerable, not simply because of the economic interests involved but because of fear that the tax could be abused to expropriate people. In particular, some see the imposition of taxation on land as regionally inequitable because livestock is more a important form of wealth in the West of the country. Because of this concern, any tax on land would be best voted, administered, and retained at district level. Some districts might then choose to tax livestock as well.9 Third, and quite controversially, there is the possibility of using public funds to assist in resolving land issues. The significant amount of underused land in the hands of large landlords makes this potentially a valuable form of public action. However, 9 Skinner (199 la,b) notes that the international trend toward a reduction in the use of land taxes applies to centrally administered but not locally administered taxes. 44 administration is difficult and will require careful use of international experience. Major issues include: * Where should the scheme be applied? Much discussion has focused on the compensation issue on mailo land. However, the evidence suggests that mailo land is no more inefficient than the other forms of large land ownership found in Uganda. It would also be much better for political unity to recognize that the issue is broader than mailo. * Mortgaging land to secure loans made to smallholders by a land fund. Considerable appeal, in principle, exists for this option. However, the history of nonrepayment in public credit schemes, both in Uganda and elsewhere, suggests caution. Unless successful lending practices to the poor for the purchase of land can be documented, a grant approach may be preferable. * Cost-sharing. An element of cost-sharing would be desirable on incentive grounds but might make it impossible to target the poorest. * Beneficiaries. The selection of the beneficiaries then become a major issue. It is reasonable to require the purchasers to live on the land they acquire. It is not reasonable, as has been done in some countries (for example, Zimbabwe), to restrict their activities. If the plots provided are small, only the poor will choose to participate and they should then be allowed the same economic freedoms enjoyed by other Ugandans. * Increasing the price of land. The danger with this form of intervention is that it may bid up the price of land, transferring most of the scheme's benefits to the landowners. Schemes in other countries should be studied to see how this can be avoided. * Winning local support. It would be imperative to win local people's support in the areas where land is purchased. For this reason, purchase within the areas from which the poor themselves come is more promising than a scheme to resettle the poor over large distances."0 * Risk of political hijacking. Political hijacking of such a scheme could undermine its intentions, especially if the grant element is explicitly recognized. Another problem is that the scheme might well be overwhelmed by demand. For all these reasons, careful technical work is needed before the idea of a land fund in Uganda can be satisfactorily assessed. But the likely magnitude of the potential benefits (for both efficiency and distribution) merits a closer look at this option. 10 Another consideration is that the land-scarce South-West does not have malaria, resettlement to other parts of the country would be likely to cause health problems, as has been found in Ethiopia. Nonetheless, resettlement from the South-West to Ankole was very successful in the 1950s. 45 6.6 The Land Act The above discussion mainly reflects the state of play at the time of drafting the first PEAP in 1997. In 1998, a new Land Act was passed, which changes the context of the debate. The final version does not refer to mailo, effectively treating ttaka holders as owners and kibanja holders as tenants, but it does provide for security and some limit on the value of rents (2 percent of the capitalized value of land) for all tenants. For kibanja holders who have been paying no rent for twenty years, rents of this level would represent a significant negative income shock. But the legislation also provides for security of tenure for the tenant and establishes a land fund, which is to be used for the following purposes: * to compensate people who are being evicted from government-owned land; * to enable people to register laind; * to enable tenants to buy out owners; and * to enable the poor landless to buy land. However, the prioritization of these activities and the amounts of money that should be committed remain controversial. Finally, the legislation established a network of land tribunals down to the parish level. However, the costs of this network have proven greater than anticipated, and therefore a revision of the land act is under consideration. In the meanwhile, the new system is not yet functional and is creating a backlog of unresolved disputes, since the powers formerly vested in the local courts will be attached to the tribunals under the new legislation. 6.7 Credit and security of assets Before the Poverty Eradication Action Plan was developed, the government had been providing credit through the district administrations in a scheme known as Entandikwa. The difficulties associated with politically implemented public credit schemes are well known and it does not appear that Entandikwa (despite its careful design) avoided them. Similar and previous attempts also ran into problems, including a large scheme run by the Uganda Commercial Bank known as the small fanners' credit scheme. The microfinance strategy now evolving has two main components. First, all Ugandans should have access to some kind of financial institution. Deposit-taking institutions have a particular role in reducing insecurity that has afflicted Uganda in the past and continue to affect the poor; the fear of theft is a disincentive to accumulation. This objective could be met in a number of ways. The branch network of the Uganda Commercial Bank has been reduced during the privatization of the institution, causing widespread concerns about access to the financial system. Similarly, the subsequent closure of the Co-operative Bank reduced the nuniber of rural branches. Options include the revival 46 of the Post Office Bank and some support to ensure a minimal presence of the commercial bank at least in each district. Second, there is an intention to support market-driven expansion of microfinance institutions such as "village banks," as well as credit schemes of various forms, some of which (such as Foundation for International Community Assistance-FINCA) have been successful. In addition, there are various forms of credit in-kind such as outgrower schemes in tobacco. The needs for credit, while substantial, are highly specific to the particular activity, making a universal entitlement to credit meaningless. But a universal entitlement to access to the financial system is likely to be very important to the poor. Physical insecurity has been a major source of impoverishment, including both the direct effects of civil war and various specific forms of insecurity, including widespread cattle-rustling in the East of the country in the 1980s and the 1990s. Conflict continues to afflict the North of the country very seriously. While efficient use of military resources-which consume a substantial proportion of the government's own recurrent spending (2 percent of GDP)-is a major issue, it is outside the scope of the PEAP process. What was considered relevant, however, is the effect of insecurity on the asset holdings of the poor. Where assets have been destroyed by insecurity, there is an unusually strong case for subsidizing their replacement; because households previously had these assets there is good reason to assume that (provided security is restored) the assets are economically viable. Traditionally, people in the North and East of Uganda depended on animal traction for extensive farming. Cattle-rustling deprived farmers of oxen plough as a key productivity enhancing technology. Some support has been provided to households in the East for restocking cattle, and a national program of restocking has been introduced, where cattle are effectively treated as revolving credit in kind. Each household is expected to provide the community with one offspring from the animal they receive so that another household can benefit. According to household survey evidence, the Eastern region experienced an increase in the number of communities with oxen to rent from 31 to 72 percent between 1993 and 1995, while in the North continued insecurity reduced this share from an already low level of 5 percent in 1993 to zero in 1995. 7. Consensus and Controversy In this concluding section, we offer some thoughts about the process of building consensus on poverty reduction. We focus on the four principal themes: the meaning of poverty; macroeconomic policy, public services, and assets. While the PEAP was being drafted, there were extensive consultations between government, donors, academia, NGOs, and other members of civil society. Another important feature was its close link to the available resource envelope, government budget and the medium-term expenditure framework. This 47 has made it easier for the government to reflect polices they agreed to in its actual spending programs. 7.1 Meaning ofpoverty: growth and distribution The controversy over the meaning of poverty continues in Uganda, and this has some policy implications. Because the technical issues involved in measuring an essentially utilitarian definition of poverty are well known (see Appleton 1996a) and very important in getting the measurement of poverty right, most of them do not raise strong controversy. The use of a household-based expenditure measure and the appropriate price indices are sometimes questioned. For instance, people may argue that someone with six hundred cattle but very low expenditure cannot be poor. In principle an expenditure measure should incorporate all aspects of consumption, but in practice this is very difficult to achieve. Despite the use of a national consumption basket instead of a regional one (given that regional staple foods are different), the general regional profile which the expenditure measures present coincide very well with most people's impression of the regional profile of poverty. A more substantial area of disagreement lies in the limitations of utilitarianism. Discussions of poverty in Uganda very often involve discussions of poor motivation and alcohol consumption; issues which most technical advisors and economists avoid because they conflict with the usual assumptions of exogenous preferences and liberalism about personal preferences, which form the core of neoclassical economics. In this context, technical economic advice inevitably becomes involved in fundamental questions about social values, which are important but conaplex to address. The extent to which a more equal distribution of income should be a goal of the government is controversial in Uganda as in other countries. A poverty action plan cannot avoid distributional issues, since any method of equalizing redistribution which does not reduce total income will tend to reduce poverty. Therefore, the political context will usually make equalizing reforms less easy to achieve than growth-enhancing reforms, except where specific regional or other sectoral interests which are important to the government coincide with the needs of equality. 7.2 Macroeconomicpolicies The benefits of liberalizing export crop production are widely understood in Uganda. Protection of industrial production, which also contributes to an antiexport bias, remains more controversial both inside and outside government. One problem here is that while the error of discrimination against exports is fairly established, a case can still be for protecting "infant industry." The future distributional consequences of the export-led strategy are ambiguous. In 1992 export crop producers were poorer than the average Ugandan, by 1995 they were better off. The extent to which such a strategy will be propoor in the future will therefore depend on the extent to which the poor are able to enter the expanding sectors in the coming years. Finally, the gender impact of export crop production remains controversial. We would agree with those who argue that gender issues are best addressed 48 through access to assets and education rather than by discouraging the best income-raising opportunity available for many Ugandan households. 7.3 Public services The most important decision was the selection of the package of universal basic services-given mass poverty and lack of capacity to target-and the design of the package. The most difficult thing here is to limit the commitment to the feasibly affordable; for instance, the costs of delivering basic health services could easily become prohibitive unless the treatment regimes are kept very simple. The purpose of defining and publicizing a basic package is undermined if the package is not in fact affordable (which also, of course, depends on the balance between private and public finance and the extent of targeting). Defining what is affordable will be difficult given the fact that the costs of services are not widely known. For example, agricultural extension is often dismissed as expensive, whereas the large-scale provision of seed-proposed as a superior alternative-is likely to be much more costly. In some cases, the importance of basic services for the poor may be controversial. But in most cases the main controversy revolves not around the need for the services but the role of the state. Observed generic failure to deliver services can justify either increases or reductions in public funding. A second controversial area is the scope for charging and, where services are charged for, the possibility of exempting the poor. In Uganda's case, the principle of user charging in health has been fiercely criticized by NGOs. There is no overwhelming reason for the optimal rate of subsidy to be 100 percent, even given substantial informational failure, externalities and distributional concerns. It is in fact very difficult to set charges sensibly in isolation from the design and costing of a basic package; there is also a very strong case for allowing some local discretion in the levels of charges and in exemption schemes as both the appropriate level of fees and effective methods of targeting are likely to vary. This policy has been followed in the health sector; central guidelines on charging exist, but the decision is local. Much the same applies to the role of NGOs in delivering services. Publicly financed services can be delivered through health maintenance organizations and equivalent organizations in other sectors, and the decision whether to use these suppliers should be a local one. The basic package of services deserves to be settled at the national level for various reasons. The newly formed local governments are more likely to function effectively if their tasks and responsibilities are well defined and if people have a clear idea of what it is reasonable to demand from those they elect. In practice, the division of labor between the center and local governments has not been agreed causing problems to service delivery. An indication of this are falling immunization rates. 49 Finally, the acceptance of a governmental responsibility to deliver basic services was greatly enhanced by the 1996 election campaign; the government switched its emphasis, quite suddenly, from advocacy of private responsibility to the delivery of free public primary education. In Uganda's case, the shift of emphasis is to be welcomed, as the performance of existing services has been poorer than the economic performance. 7.4 Access to assets and credit Although not part of the PEAP process, land issues were hotly debated in Uganda at the same time. Land remains an extremely controversial issue, partly because mailo landlords represent an ethnic as well as an economic interest group, and partly because Uganda's very turbulent history suggests that the government should not alienate any major part of its coalition. Moreover, land issues cannot be benignly neglected in view of the potential for the revival of commercial rights on mailo land and of the ongoing titling of land in much of Uganda. The public provision of credit has been seen as a major part of poverty reduction in Uganda. The difficulties encountered which such schemes in the recent past, however, as well as the very public agony of reclaiming bad debts from former credit schemes has encouraged a more cautious and perhaps more realistic consensus to emerge. 7.5 What has the PEAP achieved ? The discussion above has focused on the state of debate at the time of the drafting of the first PEAP in 1997. The PEAP has been revised in 2000, and a summary of the revision has been used as the Uganda's Poverty Reduction Strategy Paper (PRSP). This process has provided an opportunity to assess what has been achieved in the three years since the original PEAP was adopted. The PEAP provided an agreed list of directly poverty-reducing services to be prioritized in public expenditure decisions. The mechanism of the Poverty Action Fund, which is an integral part of the budget and the government's delivery system, allowed these priorities to be implemented in a publicly accountable way. Meanwhile, sectoral strategies have been prepared in several areas including transport, education, and health, as well as the Plan for the Modernization of Agriculture. The PEAP was reviewed in the Poverty Status Report in 1999 and was thoroughly updated in 2000. More than anything else, the PEAP has provided a 'grammar of argument' in which the principles of poverty reduction have entered debate in all sectors. It has arguably been able to play this role because it was not too precise. Rather than anticipating the contents of sectoral strategies, it provided a set of principles which could guide these strategies. During the discussions of the PEAP and the PRSP in 2000, one issue which emerged was an ambiguity in the term 'costing'. Everyone agrees, in principle, that programs should be costed. But the term has two senses. One approach is to observe the existing resource envelope and establish a list of activities feasible within it. This 50 approach is useful for realism and play an important part in defining a realistic medium- term expenditure framework. But there is no guarantee that the list of activities will be closely related to the ultimate objectives. The second approach is to define a set of objectives and work back to a resource need; this can then be reconciled with realistic projections of resources either by increasing the available resources or by phasing the implementation of programs over a number of years. Ultimately, these two exercises need to be reconciled, but they need to be conceived distinctly in the first instance. Often costings produced in sectoral plans represent a mixture of the two approaches, being resource-constrained and hence inadequate for the stated objectives. The demand for programs to be costed must respect the distinction between the two approaches and recognize that both tasks are conceptually and practically substantial. In brief, the PEAP process in Uganda has * produced a clear but flexible set of expenditure priorities which have been reflected in the medium-term expenditure framework and annual budgets; * by being not too specific, allowed sectoral strategies to evolve; * enormously increased civil society involvement in policymaking and public spending; and * promoted debate both within and outside the government. Bibliography Appleton, Simon. 2000a. Changes in Poverty and Inequality in Reinikka R. and P. Collier (eds.), Uganda 's Recovery: The Role of Farms, Firms and Government. Regional and Sectoral Studies, World Bank, Washington, D.C. Forthcoming. Appleton, Simon. 2000b. What Can We Expect from Universal Primary Education? in Reinikka R. and P. Collier (eds.), Uganda 's Recovery: The Role of Farms, Firms and Government. Regional and Sectoral Studies, The World Bank, Washington, D.C. Forthcoming. Appleton, Simon. 1996a. "Problems of Measuring Changes in Poverty over Time: The Case of Uganda 1989-92." Institute of Development Studies Bulletin 27(1):43-5 5. Appleton, Simon. 1996b. Women-Headed Households and Household Welfare: An Empirical Deconstruction for Uganda. World Development (U.K.); 24:1811-27, December. Appleton, Simon, and Arsene Balihuta. 1996. "Education and Agricultural Productivity: Evidence from Uganda." WPTS/96-5. Center for the Study of African Economies. Oxford University, Oxford. si Appleton, Simon and John Mackinnon. 1995. Poverty in Uganda: Characteristics, Causes and Constraints. Background Paper for the Uganda Country Economic Memorandum, The World Bank, Washington, D.C. Atwood, David A. 1990. Land Registration in Africa: The Impact on Agricultural production. World Development (U.K,.) 18:659-71. May. Besley, Timothy. 1995. Property Rights and Investment Incentives: Theory and Evidence from Ghana. Journal of Political Economy (U.S.); 103:903-37. October. Bazaara, Nyangabyaki. 1992. Land Policy and the Evolving Forms of Land Tenure in Masindi District, Uganda. Centre for Basic Research Working Paper No 28,, Kampala. Bruce, John W. and Shem E. Migot-Adholla. 1993. Searching for Land Tenure Security in Africa. Kendall/HuntPublishing Company, Iowa Chen, Duanjie, John Matovu, and Ritva Reinikka. 2000. A Quest for Revenue and Tax Incidence in Reinikka Ritva and Paul Collier (eds.) Uganda 's Recovery: The Role of Farms, Firms and Government. Regional and Sectoral Studies, The World Bank, Washington, D.C. Forthcoming. Chen, Duanjie and Ritva Reinikka. 1999. Business Taxation in a Low-Revenue Economy: A Study on Uganda in Comparison with Neighboring Countries. Africa Region Working Paper Series Number 3, World Bank, Washington D.C. Ddungu, Expedit. 1994. The Other Side of Land Issues in Buganda: Pastoral Crisis and the Squatter Movement in Sembabule sub-district, Kampala, Centre for Basic Research Working Paper No 43, Kampala. Dreze, Jean and Nicholas Stem. 1990. Policy Reform, Shadow Prices, and Market Prices. Journal of Public Economics (Netherlands); 42:[1]-45. June. Elbadawi, Ibrahim A. 1999. External Aid: Help or Hindrance to Export Orientation in Africa? Journal of African Economies 4:578-616. December. Feder, Gershon and Raymond Noronha. 1987. Land Rights System and Agricultural Development in Sub-Saharan Africa. World Bank Research Observer 2:143-69. July. Harmsworth, J. (1991), "The impact of the Tobacco Industry on Rural Development and Farming Systems in Arua, Uganda"' in Haswell, M. and D. Hunt (eds.), Rural Households in Emerging Societies: Technology and Change in Sub-Saharan Africa. Berg, Oxford. Hoddinott, John. 1992. Modelling Remittance Flows in Kenya. Journal of African Economies (U.K.) 1:206-32. August. Kigula, John. 1993. Land Disputes in Uganda: An Overview of the Types of Land Disputes and Dispute Settlement Fora. Mimeo, Makerere Institute of Social Research, Kampala. Lucas, Robert E.B. and Oded Stark. 1987. Labour Mobility and Intrafamilial Income Transfers: Theory and Evidence from Botswana. Discussion Paper, Development Research Department, World Bank, W'ashington, D.C. 52 Mackinnon, John. 1998. Can Robust Pro-Female Policies Be Identified When the True Model of the Household Is Unknown? Centre for the Study of African Economies, Working Paper Series (U.K.); WPS/98-16:1-23. June. Mackinnon, John. 1995. Health as an Informational Good: the Determinants of Child Nutrition and Mortality During Political and Economic Recovery in Uganda. Centre for the Study of African Economies, Working Paper Series (U.K.) 95/9:1-30. March. Malmberg Calvo, Christina. 1994. Case Study on Intermediate Means of Transport: Bicycles and Rural Women in Uganda. Environmentally Sustainable Development Division, Technical Dept., Africa Region, The World Bank, Washington, D.C. Marquardt, Mark A. and Abby Sabina-Zziwa. 1998. Land Reform in the Making in Hansen, Holger Bemt and Michael Twaddle (eds.) Developing Uganda. James Currey, Oxford. Mason Andrew D. and Shahidur Khandker. 1997. Measuring the Opportunity Cost of Children's Time in a Developing Country: Implications for Education Sector Analysis and Interventions. Human Capital Development Working Paper 72, World Bank, Washington, D.C. McGee, Rosemary. 2000. Analysis of Participatory Poverty Assessment (PPA) and household survey findings on poverty trends in Uganda. Mimeo, Institute of Development Studies, Sussex (UK) and Department for International Development (DFID, UK). Migot-Adholla, Shem, and others 1991. Indigenous Land Rights Systems in Sub-Saharan Africa: A Constraint in Productivity? World Bank Economic Review 5:155-75. January. Mugisha, Robert. 1992. Emergent Changes and Trends in Land Tenure and Land Use in Kabale and Kisoro Districts, Centre for Basic Research Working Paper No 26, Kampala. Muhereza, E. Frank. 1992. Land Tenure and Peasant Adaptations: Some Reflections on Agricultural Production in Luwero District, Uganda. Centre for Basic Research Working Paper No 27, Kampala. Okidi, John A. 1999. Regional Growth Disparities and Household Economic Performance in Uganda; Economic Policy Research Centre, Research Series No. 17, Kampala. September. Opyene, James Ebong. 1993. Forms of Peasant Agricultural Co-operatives in Uganda: A Case Study of Lango. University of Dar es Salaam (Tanzania) and Institute of Development Studies, Sussex (UK). McPake, Barbara, Delius Asiimwe, Francis Mwesigye, Matthius Ofumbi, Peter Streefland, and Asaph Turinde. 1998. The Economic Behaviors of Health Workers in Uganda: Implications for Quality And Accessibility of Public Health Services. London School of Hygiene & Tropical Medicine, Department of Public Health Policy Departmental Publication No 27, London. Mincer, Jacob. 1974. Schooling, Experience and Earnings. Columbia University Press: New York s3 Pinckney, Thomas, and Peter Kiko Kimuyu. 1994. Land Tenure Reform in East Africa: Good, Bad or Unimportant? Journal cf African Economies (U.K.); 3:1-28. April. Purcell, Dennis L. and Jock R. Anderson. 1997. Agricultural Extension and Research: Achievements and Problems in National Systems. World Bank Operations Evaluation Study, Washington, D.C. Reinikka, Ritva. 2000. Recovery in Service Delivery: Evidence from Schools and Clinics in Reinikka Ritva and Paul Collier (eds.), Uganda's Recovery: The Role of Farms, Firms and Government. Regional and Sectoiral Studies, The World Bank, Washington, D.C. Forthcoming. Reinikka, Ritva and Jacob Svensson. 1999. Confronting Competition: Investment Response and Constraints in Uganda. Policy Research Working Paper 2242, World Bank, Washington, D.C. Republic of Uganda. 2000a. Revised Volume I of the Poverty Eradication Action Plan (PEAP). Second draft. Ministry of Finance, Planning and Economic Development, Kampala. May. Republic of Uganda. 2000b. Republic of Uganda. 2000. Tracking the Flow of and Accountability for UPE Funds. Ministry of Education and Sports. Report by International Development Consultants Ltd., Kampala. February. Republic of Uganda. 1999. Uganda Participatory Poverty Assessment, draft report. Ministry of Finance, Planning and Economic Development, Kampala. Republic of Uganda. 1998. National Integrity Survey. Inspector General of Government, Kampala. Republic of Uganda. 1997. Poverty Erad.ication Action Plan: A National Challenge for Uganda, Volume I. Ministry of Planning and Economic Development, Kampala. June. Republic of Uganda. 1996a. Background to the Budget 1996-1997 and National Development Strategy 1996/97 - 1998/99. Ministry of Finance and Economic Planning, Kampala. June. Republic of Uganda. 1996b. National ]Food Strategy: A Response to Overcome the Challenge of Poverty and Growth. Ministry of Finance and Economic Planming, Kampala. July. Republic of Uganda. 1993. Economics of Crop Production. Agricultural Secretariat, Kampala. Skinner, Jonathan.1991a. Prospects for Agricultural Land Taxation in Developing Countries. World Bank Economic Review 5:493-511. September. Skinner, Jonathan. 1991b. If Agricultural Land Taxation Is So Efficient, Why Is It So Rarely Used? World Bank Economic Review 5:113-33. January. Ssenkumba, John. 1993. The Land Question and the Agrarian Crisis: the case of Kalangala district, Uganda. Centre for Basic Research Working Paper No 34, Kampala. 54 United Nations Development Program (UNDP). 1997. Uganda Human Development Report. Kampala. World Bank. 1996a. Uganda: The Challenge of Growth and Poverty Reduction. A World Bank Country Study. Washington, D.C. World Bank. 1996b. Access to Education and Health Care in Uganda. Macroeconomics 2, Africa Region and Poverty and Social Policy Department, Washington, D.C. September. Processed. World Bank. 1 996c. Performance and Perceptions of Health and Agricultural Services in Uganda. A Report Based on the Findings of the Baseline Service Delivery Survey. Economic Development Institute and CIET International, Washington, D.C. World Bank 1993. Uganda: Agriculture, Washington, D.C. 55 Policy Research Working Paper Series Contact Title Author Date for paper WPS2416 The Swiss Multi-Pillar Pension Monika Queisser August 2000 A. Yaptenco System: Triumph of Common Sense? Dimitri Vittas 31823 WPS2417 The Indirect Approach David Ellerman August 2000 B. Mekuria 82756 WPS2418 Polarization, Politics, and Property Philip Keefer August 2000 P. Sintim-Aboagye Rights: Links between Inequality Stephen Knack 37644 and Growth WPS2419 The Savings Collapse during the Cevdet Denizer August 2000 I. Partola Transition in Eastern Europe Holger C. Wolf 35759 WPS2420 Public versus Private Ownership: Mary Shirley August 2000 Z. Kranzer The Current State of the Debate Patrick Walsh 38526 WPS2421 Contractual Savings or Stock Market Mario Catalan August 2000 P. Braxton Development-Which Leads? Gregorio Impavido 32720 Alberto R. Musalem WPS2422 Private Provision of a Public Good: Sheoli Pargal August 2000 S. Pargal Social Capital and Solid Waste Daniel Gilligan 81951 Management in Dhaka, Bangladesh Mainul Huq WPS2423 Financial Structure and Economic Thorsten Beck August2000 K. Labrie Development: Firm, Industry, and Asll Demirguc-Kunt 31001 Country Evidence Ross Levine Vojislav Maksimovic WPS2424 Global Transmission of Interest Jeffrey Frankel August 2000 E. Khine Rates: Monetary Independence and Sergio Schmukler 37471 the Currency Regime Luis Serv6n WPS2425 Are Returns to Investment Lower for Dominique van de Walle August 2000 H. Sladovich the Poor? Human and Physical 37698 Capital Interactions in Rural Vietnam WPS2426 Commodity Price Uncertainty in Jan Dehn August 2000 P. Varangis Developing Countries 33852 WPS2427 Public Officials and Their Institutional Nick Manning August 2000 C. Nolan Environment: An Analytical Model for Ranjana Mukherjee 30030 Assessing the Impact of Institutional Omer Gokcekus Change on Public Sector Performance WPS2428 The Role of Foreign Investors in Debt Jeong Yeon Lee August 2000 A. Yaptenco Market Development: Conceptual 38526 Frameworks and Policy Issues Policy Research Working Paper Series Contact Title Author Date for paper WPS2429 Corruption, Composition of Capital Shang-Jin Wei August 2000 H. Sladovich Flows, and Currency Crises 37698 WPS2430 Financial Structure and Bank Asli Demirguc-Kunt August 2000 K. Labrie Profitability Harry HLuizinga 31001 WPS2431 Inside the Crisis: An Empirical Asli Demiirg0c-Kunt August 2000 K. Labrie Analysis of Banking Systems in Enrica Detragiache 31001 Distress Poonam Gupta WPS2432 Funding Growth in Bank-Based and Asli Demirguc-Kunt August 2000 K. Labrie Market-Based Financial Systems: Vojislav Maksimovic 31001 Evidence from Firm-Level Data WPS2433 External Interventions and the Ibrahim A. Elbadawi September 2000 H. Sladovich Duration of Civil Wars Nicholas Sambanis 37698 WPS2434 Socioeconomic Inequalities in Child Adam Wagstaff September 2000 A. Maranon Malnutrition in the Developing World Naoko Watanabe 38009 WPS2435 The Evolution of Poverty during the Asep Suryahadi September 2000 P. Sader Crisis in Indonesia, 1996-99 Sudarno Sumarto 33902 Yusuf Suharso Lant Pritchett WPS2436 Safety Nets and Safety Ropes: Sudarno Sumarto September 2000 P. Sader Who Benefited from Two Indonesian Asep Suryahadi 33902 Crisis Programs-the "Poor" or the Lant Pritchett "Shocked?" WPS2437 Quantifying Vulnerability to Poverty: Lant Pritchett September 2000 P. Sader A Proposed Measure, Applied to Asep Suryahadi 33902 Indonesia Sudarno Sumarto WPS2438 Measurements of Poverty in Menno Pradhan September 2000 P. Sader Indonesia: 1996, 1999, and Beyond Asep Suryahadi 33902 Sudarno Sumarto Lant Pritchett WPS2439 State-Community Synergies in Monica Das Gupta September 2000 M. Das Gupta Development: Laying the Basis for Helene Grandvoinnet 31983 Collective Action Mattia Rornani t.. . ,1 s- /