PRSPs AND BUDGETS A SYNTHESIS OF FIVE CASE STUDIES Rosa Alonso, Lindsay Judge and Jeni Klugman 1. Introduction This paper synthesizes the findings from a series of case studies on the interaction between the PRSP process and the budget. The five studies, Bolivia, Burkina Faso, Cambodia, Tanzania and Vietnam aim to assess the extent to which public finance management and budget allocations reflect the principles and content of the PRSP, hence providing insights into progress in PRS implementation. The cases also shed light on whether the PRSP process itself has fostered more accountable, efficient and pro-poor budget processes and allocations as of 2003. 1 Public finance management and the PRSP process can mutually re-enforce each other. Public sector reforms and reforms aimed at improving the pro-poor focus of public expenditure have often suffered from unsupportive political and institutional environments, which the reforms have failed to address. Absence of champions and processes to foster the proposed reforms have often led to weak implementation and sustainability of reforms. The PRSP process, with its focus on data and information for evidence-based policy-making, open and participatory policy-making processes, poverty results and country-led donor coordination, alignment and harmonization has the potential to significantly improve the pro-poor focus and general accountability of budgeting processes. Also, by taking a “process” or “repeated-game” approach to reform with a focus on changing incentives, it has a greater chance to work than the “single-game” approach with high stakes of public sector reform projects, especially in unsupportive political environments. 2 The reverse is also true. The PSRP process necessitates sustainability of its key principles and objectives through the implementation phase via budgetary processes. Unless budget processes themselves use, produce and publicize quality data, become more open to representatives of the interests of the poor within and outside government, lead to more pro-poor resource allocation and service delivery, and donors adjust their policies to long-term, reliable support to country-driven priorities established in PRSPs, the latter cannot be successfully implemented. PRSP processes and public finance management, hence, need each other. The cases confront a number of methodological challenges. First, in some countries and sectors, lack of appropriate data constrained the extent to which the research questions could be fully answered. Second, the PRSP remains a relatively recent innovation in all the countries studied and we recognize that many of our findings are preliminary, and require additional confirmation over time. Third, any assessment of the value added of the PRSP approach needs to be cognizant of the initial conditions in country, both to avoid ascribing successes to the PRSP which pre-date its existence, and to temper expectations about what the approach can deliver in a relatively short space of time given the starting point of each country. To address this last challenge, the case studies explicitly acknowledge the pre-existing situation in-country and try to assess the value added of the PRSP process. The four countries studied have a number of common features. For example, except for Vietnam, they are all, relatively small-sized countries with a majority of the population in poverty, a stable macro-economic 1 Lindsay Judge and Jeni Klugman, The PRSP and Budget Linkages: The Case of Bolivia, 2004; Rosa Alonso and Jan Walliser, Fiscal Aspects of PRSP Implementation in Burkina-Faso; Jeni Klugman and Robert Taliercio, The PRSP and Budget Linkages: The Case of Cambodia, 2003; Rosa Alonso and Robert Utz, Fiscal Aspects of PRSP Implementation in Tanzania, 2003; Rosa Alonso and Hoa Ngo, Vietnam’s CRGS and the Budget. 2 On the dismal performance of public sector reform projects in Sub-Saharan Africa, see B. Levy and P. Engeberg, Building State Capacity in Africa. The paper finds that more technical, narrower, public expenditure management reforms had a much higher chance of success than broad public sector reforms, especially in unsupportive political environments. 2 situation, minimum public expenditure management “basics” in place, a high degree of aid dependence (though with important differences from country to country), a strong partnership between a technocratic government team and the donor community, an active civil society, a weak Parliament and a good pre- existing basis for donor coordination. Likewise, they all face severe budget constraints: in Burkina Faso, for example, fears about rising levels of indebtedness have lead to IMF-imposed caps on spending, Cambodia has an exceptionally low tax-to-GDP ratio and Bolivia has run large budget deficits in recent years and has faced political and social upheaval as it attempted to address them. Finally, and perhaps most importantly, all five countries share a high-level political commitment to addressing poverty, although the extent to which this commitment permeates throughout government agencies varies from country to country. The five countries, however, also display many distinctive features. Bolivia and Cambodia, for example, both suffer from high degrees of political fragmentation, which in Bolivia has manifested itself as civil unrest on a number of occasions in the last two years. Burkina Faso, Tanzania, and Vietnam, on the other hand, benefit from more stable political systems and an inherited commitment to pro-poor policies from socialist governments. Tanzania and Bolivia both have a recent track record of participatory processes, compared with Burkina Faso, Cambodia and Vietnam which had relatively closed policy-making systems prior to the advent of the PRSP process. Likewise, Tanzania, Bolivia and, to a lesser extent Vietnam, have a history of ongoing PEM reform dating back to the 1990s, although, arguably, gains from these reforms have been far greater in Tanzania. Finally, there are obvious temporal differences: Tanzania and Burkina- Faso both completed their first PRSP in 2000 and are now close to finalizing their second strategies and Vietnam and Bolivia produced their PRSPs in 2001 though, in Bolivia, a change of government and political disruption has slowed its implementation to date; and Cambodia completed its strategy only in late 2002, and therefore remains at an early point with respect to PRSP implementation. This paper is structured in four parts. The first section looks at the production and availability of data, and whether it has usefully informed budget processes in our sample countries. The second section considers budget formulation processes and assesses whether the PRSP process has contributed to making them more open and participatory, both within and outside government. The third section provides an overview of actual budget allocations and the extent to which expenditures have become more aligned with PRSP priorities in each country. Finally, the paper concludes with a section on the role of donor coordination and alignment with PRSP priorities as well as on the predictability and variability of aid flows. 2. Data Production, Availability and Use The case studies analyze the extent to which the focus of the PRSP process on the production of good quality data is leading to the availability to policy-makers of more timely, more disaggregated, better quality data on poverty and social indicators to support the budgeting process. They also examine whether the PRSP-led focus on country ownership, sector-wide approaches and general budget support is contributing to improved availability and dissemination of fiscal data within government, to Parliaments, donors and civil society. The orientation of the PRSP process to the achievement of poverty outcomes can also provide an added incentive for results-based, or at least results-oriented, budgeting. The case studies, thus, also aim to ascertain to which extent this added incentive is enhancing the results-orientation of budgeting processes. Data Production The production of relevant, readable and timely data has an important bearing on the policy-making process, as well as on government accountability for policy implementation. This section considers the 3 extent to which the PRSP process has fostered an improvement in the coverage, disaggregation and timeliness of poverty and fiscal data and in poverty analysis. Greater coverage and disaggregation of poverty and fiscal data. Generally, the case studies do show a PRSP-supported increase in the coverage and disaggregation of poverty and fiscal data and poverty analysis. In Tanzania, for example, the PRSP process has contributed to improvements in data on development expenditure and ongoing efforts will greatly improve the ability to consistently identify and track PRS priority sector expenditures in the budget. In Tanzania and Burkina-Faso, the PRSP process has also added impetus to the elaboration and use of disaggregated poverty data, for instance through the establishment of gender disaggregated targets in the education and health sectors. In both Tanzania and Vietnam, recent and reliable poverty data is publicly-available and used to elaborate high-quality studies. This improved data coverage facilitates monitoring progress in PRS implementation and, more generally, in the evolution of the quality of public expenditure. In all five countries, the PRSP process has underscored the problems caused by the unavailability of adequate expenditure classifications for policy analysis, monitoring and evaluation. In particular, in all case study countries, monitoring is rendered difficult by the incompleteness of the functional and program classification of budgets and the lack of systematic and consistent identification of PRSP expenditures. All the countries studied organized budget information using multiple classifications but generally each retained only one classification for widespread use by all agencies throughout the budget process. Bolivia and Cambodia, for example, use line item classifications for most budget functions while Burkina Faso employs administrative categories as well as a program budget, but only the former is used for budget execution and monitoring purposes. While using one simple classification by all government agencies has some clear advantages in terms of simplicity and control, there are also shortcomings to this system. In particular, they do not allow a country to match spending allocations with expenditure priorities at the budget execution stage beyond the very broad level of sector headings. While all countries in the study seem to have recognized the weaknesses of such an approach, only Tanzania is successfully utilizing other classifications, including economic, programmatic and functional formats as well as disaggregating spending by geographic region and gender. 3 The successful introduction and use of these classifications in Tanzania is probably due to a longer history of PFM reforms, a strong focus on poverty-reduction by both the MF and the PRSP unit and pressure from the donor community and civil society. Experience from Bolivia and Burkina Faso in particular would suggest that, in other contexts, donors efforts to encourage the production of budget classifications which would generate information more amenable to policy analysis have been successful. These classifications, however, do not yet seem to be fully used by either government or donors for budget execution or monitoring purposes. Much of the information provided in the new formats, moreover, is still incomplete, classification headings are often changed from year to year thereby hindering comparisons over time, and they are rarely used to inform future policy development. It is important to acknowledge that the development and use of new information formats takes time and the Tanzanian example seems to imply that sustained production and dissemination of these classifications in conducive policy environments has good prospects for success in the medium-term. Moreover, there is some evidence of the successful use of programmatic classifications in our sample. In Cambodia, for example, programmatic categories are used for expenditures under the Priority Action 3 The introduction of GFS coding, still incomplete for the development budget, will further strengthen the quality of data available and allow its international comparability and its consistency with the integrated financial management system. 4 Program and, in Bolivia, there is some evidence of their effective use in the context of SWAps in the health and education sectors. Similarly, in Burkina-Faso, the education sector has developed the first full- fledged programmatic budget with a medium-term horizon in the context of an education sector program. The development of these “islands of excellence” has been noted in other studies, and usually occurs in the context of programs which give them a degree of financing predictability within a longer-term time- frame than a single budget year. 4 Clearly, with the right conditions and support, countries which have traditionally organized their budgets by line item can generate information in more useful formats, although, in contexts such as Cambodia (and, to a lesser extent, Burkina-Faso), it could be argued that they lead to the development of parallel budget systems which do not tackle the core problems of the PFM system. The need to monitor budget execution for PRS monitoring purposes has also emphasized the importance of the comprehensiveness of budget data. An area where the PRSP process seems to be leading to clear improvements is the integration of external assistance. The general move toward budget supported fostered by the PRSP process per se leads to the integration of external assistance into regular budgetary processes and budgets. Outside the area of budget support, however, the degree of progress in capturing aid in budgets has very much depended on government leadership and the capacity of external financing units in Ministries of Finance. In Tanzania, great strides have been made in integrating donor financing into the budget both due to a move toward budget support as well as through better recording of in-kind and direct project support into government accounts. This progress has been achieved chiefly owing to the role played by the external assistance unit of the MoF, which led efforts to harmonize donor reporting requirements and developed a comprehensive database of donor finance in fy02 which was subsequently used to establish the resource envelope for the fY04 budget. On the other hand, in Cambodia, no one agency has responsibility for coordinating donor inputs and, consequently, much external finance goes untracked. In some other countries, even when external aid is recorded, it is often compiled in formats that are incompatible with the rest of the budget. In Burkina Faso, for example, few donors separate out capital and current expenditures, and many provide funding directly to the local level, circumventing all systems designed to capture comprehensive information on spending patterns. Similarly, information on budget execution at the local level is without exception weak and is usually captured, if at all, by primitive systems that suffer from significant time lags. As a result, a large percentage of expenditures remain untracked in any meaningful sense, especially in highly decentralized fiscal contexts such as Bolivia. Tanzania provides an example of ongoing efforts to tackle this problem, as it publishes local government expenditures on a quarterly basis, although to date only at an aggregate level. 5 A further difficulty noted in some cases is that a significant amount of spending is off budget and is never reported at any point throughout the budget cycle. In Vietnam, for example, budget analysis is hampered by its lack of comprehensiveness and, in particular, its exclusion of the significant quasi-fiscal activities of State-owned Enterprises (SOEs) and extra-budgetary funds. Data Availability The focus of the PRSP process on evidence-based transparent and participatory policy-making and on budget support as the preferred aid modality requires broad availability of information for all stakeholders. Our case studies provide some evidence that the PRSP process is indeed contributing to 4 John Roberts, ”Managing Public Expenditure for Development Results,” ODI Working Paper 203, Feb. 2003. 5 There is no easy solution to this problem. In situations where efforts are being made to extend FMIS to local levels, some PFM experts have questioned whether such complex and rigid systems as FMIS make sense in low capacity environments. In particular, there is little encouragement for local governments to pilot systems such as SIGMA in Bolivia which are not sufficiently flexible to allow agencies to vire across budget lines. Hence, improving incentives for regular reporting by local governments is probably a better alternative. 5 improved availability and dissemination of strategy, poverty and fiscal data within government as well as to Parliaments, donors and civil society. There is no doubt that the introduction of the PRSP in our five countries has motivated additional interest in collecting poverty data and social indicators and, to a lesser extent, carrying out benefit incidence analysis and program impact evaluations. At the same time, the PRSP process is also highlighting the important remaining obstacles in this area. In particular, in order for information to be usable, it needs to be available in formats that are readable and facilitate analysis, allowing the tracing between inputs, outputs and results. Less progress has been made in this regard but pressure for its provision is mounting, especially from civil society and the donor community. Improved availability and dissemination of information on government’s strategic priorities. The PRSP process seems to have improved the extent to which government strategies and policies are available and actively disseminated to the public. This is true even in countries such as Tanzania where much data and information was already public before the advent of the PRSP process. For instance, key documents such as the pre-PRSP National Poverty Eradication Strategy were only available in their technical version and in English. The PRSP, on the other hand, was made public in summary form and in Kiswahili. Moreover, in all countries reviewed, the PRSP process has been accompanied by an extensive dissemination campaign by both government and civil society organizations. In Vietnam, the CPRGS was widely disseminated, including in summary form to local government, communities and civil society as well as through radio and television. Similarly, in Burkina-Faso, the PRSP process has fostered debate about the country’s development strategy and has created an expectation that the PRSP should be linked to sector strategies and budgets. The stakeholders most engaged in this debate in Burkina-Faso, however, are donors and, to a lesser extent, government ministries, civil society and the private sector, with the National Assembly still playing a rather marginal role. Greater availability of and demand for budget execution data. The amount of budget information that is currently available publicly varies considerably between our five case study countries. On the one hand, much budget information is disclosed in Tanzania as part of its participatory public expenditure review (PPER) process, with this exercise providing a forum for stakeholders to engage in discussion on the available data as well as express their need for additional types of information. In Burkina-Faso, on the other hand, public availability of budget data is more limited. In particular, although budget documents are published in the official gazette and posted on the Ministry of Finance’s website, difficulty in obtaining budget data was identified as a problem in the CFAA and the ROSC and was confirmed by our interviews with members of the donor community and civil society. In Cambodia and Bolivia, the readability of data is a problem, as web interfaces provide the public with access to budget information, but in formats that are difficult to interpret. Moreover, without an understanding of the budget process, much of the information that is disseminated is impossible to interpret. For example, in Bolivia, central government health and education expenditures appear very low online, mainly because transfers of funds to regional government to pay teachers and health personnel salaries are not reflected under the relevant line agency budget. Without this kind of background information, non-technical users would struggle to extract useful conclusions from the existing information systems. This is, to a large extent, the situation in all our five countries. Information availability on how government funds are actually spent is critical to monitoring the consistency of budgets with government strategies as well as to curb misappropriations. The focus of the PRSP process on data production, dissemination and budget and outcome monitoring has contributed to progress in these areas. For example, significant improvements have been made in generating and publicizing good budget execution data in the countries considered in this study and pressure for further improvements is mounting. The introduction of integrated financial management systems in Bolivia and Tanzania make real time information on actual expenditures available, while in Cambodia, execution reports are produced on a monthly basis and aggregated annually. In Vietnam, the MOF produces 6 provisional monthly, quarterly and annual fiscal data on government operations shortly after the end of the reference period and final data for the year are posted on the MOF’s website since 2001. Less progress, however, has been made in the dissemination of within-year data or year-end preliminary data. In Burkina-Faso, for example, within-year data is not shared by the MOF with other ministries, the donor community or the public despite the fact that it was a condition for budget support. This is one of the ongoing discussions between the Burkinabe government and the donor community and it is indeed likely to result in greater publicity of budget execution figures. A similar situation exists in Vietnam where only final budget execution data are shared with the public and only after the lengthy six-month period it takes for its production. The 2004 PER, however, recommends that preliminary budget outturns be published and, given the eagerness of the Vietnamese government to make progress in the area of transparency in PFM and the donor community’s emphasis on this area before considering a move to budget support, progress on this front is likely. Overall, therefore, despite the magnitude of obstacles still ahead, the PRSP process and its accompanying move to budget supports appear to have resulted in efforts to improve the production and availability of budget data to government ministries outside finance as well as to the public. Linking Spending with Results Evidence from the cases studies suggests that the PRSP approach has increased government focus on assessing the impacts of expenditures on poverty reduction results. For monitoring purposes, good information on inputs needs to be complemented by tracking of outputs and intermediate indicators. The PRSP process has encouraged all the countries studied to spell out more clearly both targets and indicators that could be tracked to demonstrate progress towards their poverty reduction goals. A number of common shortcomings, however, are also observed. First, the links between some indicators and poverty reduction outcomes is somewhat tenuous, and should be spelled out more clearly. Second, it is not always clear which institution has responsibility for monitoring each indicator, nor how the collected information should be used to inform future policies or allocations. Third, indicators have often been selected without taking account of government capacity to collect relevant information. Moreover, there is a general over-reliance on survey data while routine administrative data remains weak and under- exploited. The PRSP process has clearly fostered greater emphasis on monitoring and evaluation. However, much of this effort remains at an incipient stage with few effective systems yet being operational. Tanzania is the best example in our sample of a solid monitoring and evaluation system with strong links to the PRSP and budget processes while, in Bolivia, the agency tasked with tracking results has yet to become fully functional. In all five countries, results information in most ministries is still collected in an ad hoc manner. In all the countries studied, the Ministries of Education and Health were the most advanced in the establishment and monitoring of targets and indicators. This good performance seems due to the fact that these ministries often have sectoral programs with associated monitoring frameworks, that they are priority sectors for their respective poverty-reduction strategies and that it is easier to define and monitor meaningful targets and indicators in these sectors than in others (e.g. productive sectors). While valuable information on impacts is increasingly being collected on a systematic basis, weak coordination among collecting agencies and insufficient use by policy-makers means that their full potential is not yet utilized. Moreover, relevant information is often not available at appropriate points in the budget cycle and line ministries have little guidance on what information they are expected to compile as part of their budget reporting. That said, there is some emerging good practice. In Tanzania, for example, poverty data is being used for the determination of fiscal transfer formulas to local government while, in Cambodia and Burkina-Faso, innovative mechanisms for the joint monitoring of sectoral results frameworks have been developed including an annual review of sector performance reports. In Cambodia, 7 moreover, these reports – for education and health -- feed in to the production of revised financing frameworks for the MTEF. The results-orientation of donor budget support programs seems to be increasing the attention paid by the Ministry of Finance to the links of budget inputs with outputs and intermediate indicators. For example, in 2003 in Burkina-Faso, the failure of the MF to provide a budget allocation for the Ministry of Basic Education to be able to produce routine data on school enrollment led to missing a key benchmark with the European Union. Partly as a result of this failure, the variable tranche of the EU’s budget support program was not disbursed, causing strain in budget management. This crisis, however, seems to have significantly increased the awareness of Ministry of Finance officials of the importance of budgeting for results and of documenting and accounting for those results in a timely manner. In order for increased data availability to result in improved policy-making it needs to be used by government, Parliament and civil society. Encouragingly, the studies also showed that the PRSP process has motivated new interest in how funds are spent and what results they are having. In Burkina-Faso, for example, the growing participation of civil society in the PRSP process has created expectations of increased involvement in other aspects of policy-making, including in the budgeting process. Non- governmental organizations are particularly keen and increasingly pressuring government for, at least, improved availability of budget information and, at most, increased involvement in the budgeting process. Our case studies, however, also point to the need for capacity building to increase the budget literacy of civil society (as well as Parliament and government officials). Although greater budget literacy should increase the ability of stakeholders to hold governments accountable, a recent survey of monitoring and evaluation systems carried out by GTZ finds that the main channel for increased accountability is through greater political pressure on government by stakeholders (a channel that is not exclusively dependent on “capacity”). Over time, it is expected that this greater demand for information both from within and outside government will lead to changes in how budget data is presented as well as disseminated as well as to greater consistency be als, targets, indicators, outputs and inputs, the PRSP process has enhanced the results-orientation of the tween strategies in budgets. In Tanzania, for example, by emphasizing the need for links between go PPER process and there is now greater pressure by civil society organizations to link outputs from the PPER to the budget and the PRS cycles. Emerging Policy Lessons Our studies, thus, suggest that the PRSP process is fostering the increased production and use of strategy, poverty and fiscal data that provide the incentives for greater results-orientation and enhanced government accountability. It also suggests that further improvements are needed in order for the full potential of the process to bear fruit. For example: • The production and publication of comprehensive, sufficiently disaggregated, readable and timely budget data needs to be further improved, and effective incentives or sanctions developed to ensure that all levels of government account fully for their spending. At the local level, this is likely to require capacity building as well as streamlining of reporting formats. • More work is needed to develop appropriate budget classifications to allow government to track inputs in a way which usefully informs the policy cycle. In particular, the elaboration of full functional classifications and the expansion of programmatic budgeting would allow agencies to 8 strengthen the strategic and pro-poor focus of their resource allocations. Once program budgets are elaborated, they should be tracked and used by government, donors and civil society. • PRSP expenditures need to be properly and consistently identified in the PRSP, PRSP progress report, the MTEF, the budget and executed budget data in order to allow proper monitoring of their evolution. • Information on inputs should be complemented with more systematic efforts to develop useful indicators and collect information on results. This would need to be complemented by capacity building within line agencies to interpret monitoring results and use these results to inform relevant analytical work. Moreover, line agencies could be required to report key performance data as part of their budget submission. Similarly, capacity-building for sectoral experts at the Ministry of Finance is needed to allow the ministry to properly interpret the performance data submitted by line ministries. • There is a need to stimulate better understanding of the budget basics by the legislature and civil society groups. In addition, budget information could be presented to potential users in formats such as pamphlets and periodic newsletters, thereby reaching a broader audience who do not have access to web technology. Events such as the annual presentation and discussion of the government’s budget organized by civil society groups in Tanzania is a good practice that would be very useful elsewhere. 3. Budget Formulation and the PRSP—Has the PRSP Process Increased the Openness and Transparency of the Budget Process? The PRSP process aims to foster greater openness and transparency in policy-making, which is expected to be valuable in at least two senses. First, openness/participation and transparency are of intrinsic value, as they increase the welfare of citizens by increasing their access to information on how they are governed and they ability to participate in government decision-making. Second, it is assumed that openness and participation will eventually lead to more efficient, accountable and pro-poor decision- making due to the involvement of the poor and other representative groups in monitoring and, even, in directly informing government decisions. While the ways and extent to which this potential is realized varies enormously across countries, one would expect to see the emergence of both a more strategic approach and greater participation in budget formulation as a result of the PRSP exercise. This section assesses the extent to which this has indeed been the case in our four countries. It looks at the processes associated with preparing annual budgets and, where applicable, medium term expenditure frameworks (MTEFs), and how these have been linked to, and informed by, PRSP processes. Specifically, the roles played by the Ministry of Finance, sector ministries, civil society, local government and the legislature in deciding resource allocations are examined, and the main synergies and key disconnects between the budget formulation and the PRSP process are reviewed. The Importance of Initial Conditions in Public Expenditure Management There are significant differences in the initial PEM conditions in the countries in our sample. Whereas Burkina-Faso and Tanzania underwent significant reforms in their PEM systems in the 1990s, these reforms are more recent in Vietnam and only incipient in Bolivia and Cambodia (Table 1). The differing degrees of soundness of PEM systems has a significant impact on the extent to which the budget process is a meaningful process for resource allocation. Hence, it also significantly affects the ability of this 9 resource allocation process to become increasingly strategic as a response to the PRSP process. Tanzania and Burkina-Faso have sound PEM systems and associated budget processes that have become more strategic with the advent of the PRSP process while this impact has been weaker in Bolivia and Cambodia. Countries with a stronger degree of social cohesion and government control (Burkina-Faso, Tanzania and Vietnam) have also done much better at imposing fiscal discipline on the PRSP process than countries with higher degrees of social fragmentation and weaker government control (Bolivia and Cambodia). Table 1: Key Characteristics of Budget Formulation Protected priority expenditures Separate preparation of capital History of inaccurate aggregte Medium Term Expenditure Fragmented line ministry High level of off budget History of incremental and current accounts High level of fiscal O&M as residual decentralisation responsibilities expenditures Framework projections bugdeting Bolivia Burkina Fas o Cambodia Tanzania Improvements in Public Expenditure Management and the PRSP Process Dual budgeting and costing. An important weakness of the budgetary process in Bolivia, Cambodia and Vietnam is that these countries have not integrated the preparation of capital and current budgets. In Vietnam, moreover, the current and capital budgets are prepared not just by different units but by different ministries. As a result, disconnects are observed, and forward projections for current expenditures are often inappropriately low. In Cambodia for example, necessary allocations for operations and maintenance (O&M) are typically underestimated, treated as a residual and suffer from large cuts in cases of revenue shortfalls. As a result, the sustainability of earlier investments is compromised. Moreover, high levels of off budget donor-funded capital expenditures exacerbate this problem. The PRSP process, with its focus on results, is placing increased pressure on governments to use an integrated planning process as the only way to assess sector program costs to achieve specific targets. Costing has received great impetus under the PRSP process and, in Tanzania and Burkina-Faso, for example, the costing of the PRSP continued after the initial estimates of the paper. In Vietnam, however, the impact has been smaller as the PRSP process seems to play a less important role in government processes, including budgeting. There, it is the incipient building of an MTEF that has more potential to bring about the integration of the budget process. Strategic planning. Tanzania, Vietnam and, to a lesser extent, Burkina-Faso seem to have a more strategic approach to planning and budgeting than the other three countries in our sample. In all three countries, this strategic approach has also translated into the protection of expenditure priorities in times 10 of budget crunch. Moreover, in Tanzania, the PRSP process led to the establishment of a system of quarterly releases to PRS priority sectors which has freed priority ministries from the strictures of the cash budgeting system. Beneficiary ministries and local authorities report this system has greatly enhanced their ability to plan and manage their available resources. In Cambodia and Bolivia, on the other hand, mid-year expenditure increases or cuts were routinely made across the board without any attempt to make strategic choices by identifying priorities. Some good practice, however, was observed even in these two countries, especially in the health and education sectors, which often receive a degree of protection due to the existence of sector-wide approaches (SWAPs). The greater the reliability and protection of budget allocations, the greater the incentive for line ministries and other stakeholders to invest time in good budget preparation. The reverse, of course, is also true. Namely, if line ministries know that cash flows to the sector will not reflect the budget, they are unlikely to invest time in the preparation of budget submissions. This was especially the case in Bolivia, where the budget is not a regarded as a binding document and hence few could see the value of engaging substantively at the budget formulation stage. In order to increase the strategic orientation and reliability of budget allocations, Burkina Faso, Cambodia, Tanzania and Vietnam have, to varying degrees, introduced MTEFs in recent years. There has been good progress on some fronts: in Cambodia, for example, a rolling three-year expenditure plan has been developed and, in Burkina Faso, 2003 saw the introduction of bottom up planning for a sectoral MTEF in the education sector. Similarly, Vietnam is currently working on the introduction of MTEFs in four priority sectors. However, there is still much distance to go before effective MTEFs are securely in place: in Burkina Faso they are undermined by lack of financing predictability both because of inaccurate growth estimates and large levels of undisclosed donor finance; in Tanzania, links between the annual budget and the MTEF remain weak; while in Cambodia, the MTEF is not yet comprehensive, currently excluding vital parts of the government’s budget such as the medium term wage bill and external finance. In addition, there is no guarantee that the MTEF will reflect PRSP priorities: the 2004-2006 MTEF for Burkina Faso, for example, demonstrated no discernible alignment with the PRSP. Improved ability to forecast and manage public expenditure in PRS priority sectors. In Tanzania, the establishment of a system of quarterly releases to PRS priority sectors has freed priority ministries from the strictures of the cash budgeting system. Beneficiary ministries and local authorities report this system has greatly enhanced their ability to plan and manage their available resources. In Burkina-Faso, on the contrary, the special treatment of HIPC resources leads to a complex accounting system separating regular budget resources, HIPC resources and consolidated budget. It also fosters a dual-accounting and dual-responsibility mentality according to which poverty-reduction activities are funded by donors through specific funds while the rest of the budget does not necessarily have to focus on these objectives. Role of Various Actors—Are Budget Processes Becoming More Open and Participatory? Ministry of Finance The Ministry of Finance (MoF) has a critical role to play in ensuring links between PRSPs and budgets. A traditional weakness of previous development strategies was that they remained strategies on paper, with extremely weak links to either sector strategies or budgets. Strategies were prepared by and “owned” in other parts of government while Ministries of Finance typically elaborated and implemented budgets with little consultation to line ministries and little regard to strategic considerations. Hence, ownership of the PRSP by the Ministry of Finance is critical to its successful implementation. Moreover, by shaping the process of budget preparation and implementation, the MoF plays the key role in determining whether the PRSP principle of “participation” in policy-making, including by line ministries, Parliaments and civil 11 society organizations is reflected in the budget process. Conversely, we would expect the Ministry of Finance to be more likely to “own” the PRSP when it has been closely involved in its elaboration. In our sample countries, the MoF was generally engaged in but did not have a leadership role in the elaboration of the PRSP. In both Bolivia and Burkina Faso, although the unit responsible for the PRSP was initially housed in the MoF, it was subsequently moved to ministries focused on planning or external finance. Likewise in Cambodia and Vietnam, the PRSP was managed by the planning ministry, while in Tanzania it was the responsibility of the Vice President’s Office. An institutional home other than the MoF does not preclude strong links between the PRSP and the budget, as is evidenced by the case of Tanzania where the budget and the PRSP processes are increasingly aligned. Even in Burkina-Faso and Vietnam, where the PRSP and the budget processes were housed in different ministries, budget allocations show an increasing alignment with the PRSP. In the case of Burkina-Faso, the fact that donor budget support operations take the PRSP as its basis plays an important role in focusing the Ministry of Finance on the strategy. However, experience from our other cases suggests that there is a risk that the PRSP remains divorced from key allocation decisions if the institutional links are not strong enough. In particular, we observed that the PRSP was sometimes seen only as having traction on external finance (Bolivia) or the investment budget (Cambodia). Similarly, in Burkina-Faso, the government proposed to focus donor monitoring not on the PRSP but on “the parts of the budget financed by donors.” Although some substantive congruence is observed between budgets and PRSP priorities, in practice the processes are usually quite distinct. In Bolivia, Burkina Faso and Cambodia, different timetables were followed and as a result, useful inputs from both processes (such as costings from the budget or strategic priorities from the PRSP) were unavailable at critical points in both cycles. Tanzania, however, suggests that practice can be improved over time with its successful alignment of the preparation of its PRSP Progress Report with its annual budget cycle Line Ministries Our case studies provide evidence that the iterative and collaborative nature of the PRSP process has strengthened the collaboration between the Ministry of Finance and line ministries and in both planning and budgeting. A recent study of Bolivia, for example, concluded that “the [PRSP] preparation process opened up a certain level of inter-sectoral dialogue between ministries and government institutions that normally do not have a strong tradition of cooperation...and has drawn attention to need for inter-sectoral coordination at both a local and central level”. 6 Likewise, in Tanzania, the PRSP motivated the resurrection of sector working groups as part of the PPER exercise, a key forum for dialogue between the Ministry of Finance, line ministries, donors and civil society. 6 Entwistle, Janet, Country Ownership of Poverty Reduction Strategies: The Experience of 5 Countries, forthcoming, World Bank 12 Fostering Intra and Inter-Ministerial Cooperation in Burkina-Faso In Burkina-Faso, the PRSP process is improving the dialogue within ministries during budget elaboration. Traditionally, budget funding requests from line ministries to the Ministry of Finance were elaborated by their financing departments (DAFs) without much input from the planning departments (DEPs). This practice significantly impaired links between sector strategies and sector budgets. The focus of the PRSP process, related sector strategies and donor financing on results, however, requires strong intra-ministerial coordination. This focus on results has led to a noticeable improvement in the coordination between planning and financing departments, in particular in priority sectors, in order to ensure enhanced coherence between planning and financing strategies. According to the Ministry of Finance, this improved coordination has increased the quality of the financing requests prepared by line ministries and the strategic orientation of budgets. Coordination between line ministries and central ministries is also improving. This improvement is also a consequence of the focus on results of program budgets, MTEFs and the PRSP process. The elaboration of program budgets and MTEFs has led to the organization of inter-ministerial workshops to discuss and, in principle, decide on proposed allocations for ministries. This inter-ministerial workshops include a key group of ministries –the Ministries of Economy and Development, Finance, Basic Education, Health and Agriculture. According to Ministry of Finance officials, this inter-ministerial dialogue has improved since the advent of the PRSP process in 2001 and the PRSP has formed the strategic framework for the 2002-2005 MTEF.1 R. Alonso, The PRSP Process and the Budget in Burkina-Faso However, a strategic approach to budgeting by line agencies continues to be undermined by some key practical issues. In Bolivia and Cambodia, for example, ministries do not receive their budget envelopes from the MoF in time to prepare realistic budget submissions. Timetables are often out of sync and key analytical inputs from line agencies are not available at the appropriate point in the budget cycle. For example, in Tanzania before the alignment of the PPER and budget cycles, much sector analytical work undertaken for the PRSP, MTEF and PPER remained divorced from the annual budget process and similar disconnects continue to be observed in Cambodia. As a result, much of the good analytical work now being undertaken by line ministries is not well exploited, and points to the need to rationalize roles and processes. Local Government Although practice varies between countries, local governments have typically not been significant participants in the PRSP elaboration process, with consultations focusing on grassroots communities and non-governmental organizations. As a result, a tension is observed: the PRSP is informed by local level inputs from civil society and is aggregated at a national level, but many of its policies (especially in the area of basic service delivery) need to be implemented by local governments who did not play an important role in the process. This disconnect reduces the chances of buy-in to the strategy from local government and principal-agent problems arise in the implementation phase. Similarly, local government appears to have little leverage over the preparation of the national budget, despite the fact that decisions at this level can have serious impacts for local authorities. For example, in Tanzania, a number of local taxes were abolished in the fY04 budget with little consultation with local government. This lack of responsiveness is sometimes underpinned by practical issues: in Bolivia, for example, the format for budget preparation at the local level is not standardized, resulting in numerous incompatible submissions to national government. Moreover, there is rarely a process which facilitates local government-line ministry dialogue and, as a result, sector plans are elaborated without the input of 13 those who may execute much of their content. In Cambodia, for example, the MoF has parallel negotiations with the line ministries and local government, resulting in divergences between sector plans and actual spending. In Tanzania, Bolivia and Vietnam, the three countries studied with the most extensive levels of decentralization, there does not seem to be evidence that the PRSP process has so far fostered closer collaboration between central and local government. In Tanzania and Vietnam, for example, there seems to be a disconnect between the budgeting processes at the local and national levels. Whereas budgeting at the national level aims to link to PRS priorities and sector strategies, local government proceeds in a bottom-up approach starting at the village level and there is no guarantee that both sets of priorities will match up. Moreover, local government is impaired by a limited knowledge and understanding of the strategic priorities of the PRSP. Overall, despite its importance, the inter-face between planning and budgeting for strategies and policies at the central and local government levels is not as closely coordinated as it should be. Moreover, local government is not very much involved in the central government’s budget process which is of concern, in particular since local governments depend heavily on resource transfers from the center and decisions on local governments’ own resources are also made at the central level. Finally, there does not seem to be a forum where local and central government can discuss the role of local governments in the elaboration and implementation of sector plans. In Bolivia, the PRSP process catalyzed a significant devolution of resources to the local level, invigorating the decentralization process. In particular, the National Dialogue Law (2001), a result of the PRSP process stipulates that HIPC II resources should be transferred to the municipalities according to a pro-poor formula, and spent on health, education and other “poverty related” expenditures. As a result, the relative size of municipalities expanded by about 40 percent, to around 13 percent of total expenditure, while prefectures have maintained their share. On the other hand, decentralization seems to have led to a reduction in the share of expenditures being allocated to PRS priority sectors. Indeed, an analysis of municipal expenditures undertaken in our case study found that health, education, housing and sanitation and rural roads all suffered drops as a share of the total municipal budget between 1998 and 2001, despite being designated as “strategic priorities” in the PRSP. This development appears to cast some doubt on the “pro-poor” impact of decentralization; this question, as well as the potentially poor fit between national priorities and actual expenditures at the local level in decentralized countries, warrants further attention. Civil Society Engagement Many have acknowledged that the PRSP has increased civil society participation in policy dialogue and that it has provided many stakeholders with an opportunity to influence the setting of poverty reduction priorities. 7 Although pressure and expectations for increased participation are building up quickly, the actual progress in increasing civil society participation in the budget process in Burkina-Faso, Cambodia and Vietnam is very small. In Bolivia and Tanzania, on the other hand, the process is far more advanced both due to the existence of a more active and well-organized civil society as well as a more receptive government. As a result, the Mecanismo de Control Social in Bolivia and the NGO Policy Forum in Tanzania represent the views of their members in budget discussions with government. In Burkina-Faso, Cambodia and Vietnam, the scope for CSOs to engage in technical budget discussions and hence wield influence in the budget process is still very limited. Although NGOs tend to have stronger knowledge of the sectors in which they are directly engaged, their capacity in the public finance and macro-economic areas tends to be significantly weaker. Moreover, in countries where there is much informality throughout the budget cycle, such as Bolivia, entry points for participation are often hard to 7 See, for example, David Booth, 2004, ODI 14 identify and, the build-up of societal demands can be hard to control. In fact, the case of Bolivia provides a stark example of the need for clear “rules of the game” to govern CSO engagement in the budget process if macro-economic stability and social conflict are to be avoided. Tanzania, on the other hand, provides a welcome example of how CSOs can work with the Ministry of Finance, line ministries and the donor community to increase the pro-poor focus and accountability of the budget process without creating social instability. In fact, CSOs place high value on the PPER, seeing it as their main window of opportunity for influencing the budget, sharing views and forging and sustaining alliances with actors in government and the donor community. Box 1: The Participatory Public Expenditure Review and the PRSP Process, Tanzania The Participatory Public Expenditure Review (PPER) was introduced in Tanzania in 1997 in recognition of the fact that the traditional Public Expenditure Review exercise conducted by the Bank and the government allowed for scant engagement by domestic stakeholders and hence had little ownership. The PPER has three main objectives: it provides a forum for information sharing on budget issues; it identifies and undertakes joint analytical work on macroeconomic, sectoral and structural issues of interest to its members; and it facilitates the emergence of constituencies which seek to ensure the actual budget reflects poverty reduction concerns. Lead by the Ministry of Finance, the PPER team comprises CSOs and donors as well as other members of the government from, for example, line ministries and the local level. Working Groups meet bi-weekly, developing a work program and delivering a report intended to inform the actual budget. At this point, an “external evaluation” of the PPER outputs takes place. Chaired by the World Bank, all non-governmental participants such as CSOs and donors review the PPER recommendations, and the process is finalized with the publication of a report, ready to feed in to the formal budget cycle. The PRSP process has been able to build on the accomplishments of the PPER and the PRS is now the point of reference for the PPER process, which strengthens its focus on poverty-reduction and results. Moreover, in FY03, with the impulse of the PRSP process for increased dialogue in the priority sectors, PPER sectoral working groups –which had been discontinued-- were re-established. These sectoral working groups are one of the key mechanisms to strengthen alignment of sector budgets with the PRS. Source: Fiscal Aspects of PRSP Implementation in Tanzania, Rosa Alonso and Robert Utz, 2003 Role of the Legislature Although, in principle, the legislature has a mandate to approve and monitor the budget in each of our five countries, in practice, its role in budget formulation was limited in all cases. A number of practical constraints in their role were observed. First, the budget was often presented to parliament with very little lead time, and thus there was insufficient opportunity for parliamentarians or any specialist committees to analyze and review the budget proposals. Second, there was an evident lack of technical capacity among Parliamentarians to understand the budget. Third, in countries where the budget was not perceived as a binding instrument, as in Bolivia, the incentive to engage in a serious discussion was minimal as all recognized that the executed budget was unlikely to bear much resemblance to the approved version. Moreover, there was little evidence that Parliaments owned PRSP priorities. In fact, there is no evidence that, where parliamentarians did engage with the budget and demanded amendments, the resulting allocations were more in line with the PRSP. In Burkina Faso, for example, analysis showed that over the last five years, Parliament-introduced changes to the budget have consistently increased resources to non- priority sectors while reducing priority expenditures. Interviews with high-ranking members of Parliament’s Finance Committee revealed that a perception that basic health and basic education are 15 already provided for by the donor community and, hence, are over-endowed with resources. Similar regressive amendments were also observed in Tanzania. Despite these shortcomings, the PRSP process does seem to be generating both an increased interest in poverty issues by parliamentarians and a growing awareness of the critical role they could potentially play in budget formulation. In Cambodia, for example, the Parliamentary Commission on Banking and Finance has spent time visiting poorer areas of the country and sensitizing its members to key PRSP themes. Likewise, in Bolivia members of the legislature made clear that they would like support to build their capacity to analyze and influence the budget. Similarly, in Vietnam, the degree of involvement of individual Parliamentarians in the budgetary process and the breadth and transparency of discussion on budget allocations have increased over the past two years. The members of the National Assembly of Vietnam are increasingly questioning ministers during their appearances at the legislature and proposing amendments to proposed legislation, including budgets. For example, members of Parliament have raised questions on the efficiency of capital expenditure as well as regarding governance and corruption issues. Moreover, during the last budget cycle, the National Assembly significantly increased the policy focus and allocations devoted to HIV-AIDS. This progress is regarded as particularly important by those who have noted that the emphasis to date in the PRSP has been on civil society’s role with respect to the budget, and that this has sometimes lead to a sidelining of elected representatives. Emerging Lessons Overall, the PRSP process seems to have fostered improved openness in budget formulation and appears to be facilitating better dialogue between all stakeholders, both within as well as outside government. Moreover, there is also a growing recognition that both national and local budgets need to be better connected to the government’s strategic priorities and goals and greater scrutiny of the links between the former and the latter. However, ensuring that the principles of the PRSP process and, in particular, openness and transparency, further inform the budgeting process will depend on continued support of the PRSP process by governments, civil society and the donor community. If the PRSP process becomes institutionalized and its principles are increasingly espoused by all stakeholders, there will be significant potential for continued improvement in budgeting processes. Moreover, strengthening of the links between the budget formulation and the PRSP process are contingent upon clarification of both processes and improved synchronization of both cycles. In countries with weak budget processes, the budget as an instrument needs to gain credibility in order to increase the coherence between elaborated and executed budgets and hence ensure the predictability of financing to spending units and enhance the incentives of all stakeholders to engage in the budget process. This is clearly a challenging agenda which depends on governance and public expenditure management reforms. Without progress on this front, the possibilities for linking budget allocations to strategic priorities will remain limited in the countries where it is most needed. Some specific suggestions include: • Foster the joint work of planning and financing units within line ministries (or consider merging them). This is automatically done by increasing the results-orientation of budgets, as these units will need to cooperate in budget planning in order to ensure the delivery of results. • Foster dialogue between line ministries and the Ministry of Finance. Again, this is fostered by increased results-orientation of budgets (fostered by MTEFs and PRSPs) as well as by lending instruments that focus on results (such as budget support). • Enhance the capacity of the legislature to engage on budget issues and foster dialogue between governments and Parliaments by encouraging governments to submit readable and realistic budgets to Parliament with proper time allowed for discussion. 16 • Consider the establishment of working groups along the model of Tanzania’s PPER with the assignment of channeling dialogue among stakeholders as well as examining links between PRSP, MTEFs, sectoral strategies and budgets. • Hold budget hearings or develop other forums for civil society and other stakeholders to express their opinions on planned budgets. Such arrangements, however, should be based on clear and agreed rules of engagement to ensure expectations are managed and the scope for unconstructive conflict is limited. The model of publicity and discussion forums of annual budgets in Tanzania hosted by research institutes with good ties to both government and civil society is the best example of such hearings in our case studies. On the other hand, the expectations created in the case of Bolivia proved dangerous and should carefully be avoided. • Align PRSP and budget timelines. Both the strategic priorities of the PRSP and results from monitoring need to be available at a time which is relevant to the annual budget process for both inter and intra- sectoral allocations as well as medium term adjustments. Over time, PRSP progress reports should be timed so as to inform the budget process and could usefully become background papers to national budget documents. • Streamline and integrate the multiple budget processes such as the PPER, MTEF and annual budget in Tanzania or the PIP, MTEF and annual budget in Cambodia and Vietnam, and work towards the integrated preparation of capital and current accounts in Cambodia, Vietnam and Bolivia. • Strengthen links between the PRSP process and the MTEF through its integration to the MoF’s institutional framework and sector budget/planning cycle routines. Ensure MTEFs reflect the strategic priorities of the PRSP and, conversely, that PRSPs take as their basis the overall budget envelopes determined by MTEFs. The policy orientation of forward expenditure planning requires a range of new skills, such as policy analysis, costing and programming. Experience from Africa suggests that MTEF’s may be more effective where existing departments are complemented by new units who can fulfill these functions in order to provide the support necessary for implementation. 8 • Communicate budget ceilings and overall government priorities to all spending units within a reasonable time to allow for planning to be based on a realistic budget envelope. In addition, review the format of budget submissions and standardize the template for all spending units, and introduce incentives or sanctions for the use of this framework. • In cases where macroeconomic stability is a particular concern, consider negotiating a fiscal pact with spending agencies which would ensure greater predictability of financing but at the cost of agencies recognizing that their bargaining opportunities diminish. 4. Accounting for expenditures and results The PRSP approach places strong emphasis on monitoring and on accounting for results. The key mechanisms for fiscal accountability lie with the legislature, civil society, the donor community and 8 Philippe Le Houerou and Rob Taliercio, “Medium term expenditure frameworks : from concept to practice - preliminary lessons from Africa”, World Bank Africa Region Working Paper Series No. 28, 2002 17 internal and external audit institutions. This section aims to assess whether the PRSP has given added impetus to these institutions, as well as increased the accountability for government spending decisions to a broader range of stakeholders. It begins with a survey of the problems encountered in our case studies during budget execution and then turns to review the performance of various actors in enhancing accountability. The role of the legislature is considered along with that of auditing bodies; civil society’s contribution to ensuring accountability is then assessed; and the section concludes with an overview of the role played by development partners in requiring governments to account for both expenditures and results. Initial Conditions in Budget Execution and Impact on Accountability The technical ability (and political will) of governments to execute budgets as planned varies greatly among our sample countries with an ensuing impact on accountability. In some of our sample countries (such as Tanzania), gaps between planned and executed budgets are relatively small, in others (such as Vietnam), executed expenditures have consistently been above projections while in others (such as Bolivia), gaps have been large, negative and highly variant across ministries. In Bolivia, for example, while the government has largely maintained aggregate control on spending, expenditures by ministries in 2001 varied from only 20% to over 300% of planned allocations. Deviations on this scale demonstrate the failure of the budget to perform as a planning tool, and the disconnect between planned and executed budgets seriously undermines the former as an instrument for implementing the government’s strategy and policies. It also discourages any attempt at building in participation and accountability at the budget elaboration stage, as stakeholders rightly recognize the lack of binding character of the process. The reasons for such large deviations are numerous and vary across countries. Revenue forecasts are often inaccurate and expenditure levels typically need to be downwardly revised over the course of the fiscal year. Poor revenue forecasts are due to a number of reasons ranging from high volatility of prices and quantities of crops and weak technical capacity in macro-economic forecasting and revenue administration agencies to political pressures for high expenditure levels. Governments often deal with such uncertainties by establishing cash budgeting systems. Tanzania and Vietnam, which have not tended to overestimate revenue and hence have not suffered from serious cash flow problems, have a system of quarterly transfers --to PRS priority ministries in Tanzania and to all ministries in Vietnam. Burkina-Faso, Bolivia and Cambodia, on the other hand, typically over-estimate revenue and expenditure levels and need to use cash budgeting as a way to ration available funding over the course of the fiscal year. While this system facilitates control, it undermines the ability of line agencies to plan and undermines the overall credibility and accountability of the budgeting system. In some cases, these weaknesses are jeopardizing PRSP implementation. For example, in Cambodia the education strategy contained in the PRSP has been lauded as comprehensive and coherent, outlining a range of excellent actions from decreased user fees to increased maintenance spending and teacher salaries. However, slow and unreliable cash flows to the Ministry of Education have undermined implementation and, as a result, many of the promises of the strategy have yet to be fulfilled. Similarly, in Burkina-Faso, the combination of the expenditure cuts typically implemented at the mid-year budget review and the cumbersome budget execution procedures, lead to large gaps between budgeted and executed amounts, including in PRS priority sectors. Another reason for gaps between budgeted and executed expenditures is limited absorptive capacity. In Burkina Faso, for example, the Ministry of Finance was reluctant to increase resources to key social sectors as it felt that low execution rates in those ministries demonstrated constraints on their absorptive capacity. Staff from the line agencies, however, complained that delays in transfers and excessively tight control by the MoF on all transactions constrained their ability to spend effectively. 18 Finally, decentralization can lead to important variance between national strategies and actual implementation at the local level. As discussed in Section 3, local government often has the powers to spend their funds in ways that are at variance with the approved budget. While in Tanzania the government maintains a degree of control over expenditures at the local level through conditional grants, it is expected that, as decentralization proceeds, its ability to influence spending at this level will diminish. In Bolivia, efforts have been made to ensure that municipalities spend funds in line with overall national priorities. As Box 2 shows, however, this has not been without its challenges. Box 2: Incentivizing pro-poor spending at the local level, Bolivia In 2001 as a result of the PRSP process, various social funds in Bolivia were reformed and merged to create two consolidated funds. It was envisaged that one of these, the FPS, would play a critical role in ensuring the implementation of the PRSP through spending executed by the local level. It was envisaged that the FPS would set an allocation for each municipality based on a progressive formula, and municipalities could then apply for grant finance up to this ceiling so long as they could show that they had sufficient counterpart funding to implement the project. Thus, the FPS was expected to provide incentives to municipalities to spend their resources in a way consistent with poverty reduction objectives that are encoded in the PRSP. In practice, however, the FPS has been subject to many problems. Its procedures are complex and often contradictory, reflecting the various requirements of the contributing donors. Many municipalities often lack the capacity to navigate its systems efficiently. As a result, disbursement rates are low – in 2002, for example, only 54% of the agency’s budget was executed. The problems encountered in accessing FPS support have an additional negative impact on municipal finance: HIPC resources remain unspent as they are held back for counterpart finance. In addition, the grants actually made by the FPS are not aligned particularly well with the PRSP and they do not follow a clear progressive pattern. Source: PRSP and Budget Linkages: The Case of Bolivia, Lindsay Judge and Jeni Klugman, 2004 Overall, our studies point to the contribution of the PRSP process to a strategic approach to budget execution but also underline the need to continue to improve macro-economic and revenue forecasting and budget execution systems if accountability is to be strengthened and the potential of the PRSP process realized. Legislature and Audit Bodies The key mechanism through which the government is expected to account for expenditures is the legislature. However, in most of our case studies, the oversight role played by Parliament was limited, and its supervision of accounts perfunctory. In part, this stems from weak capacity of members of Parliament to conduct budget analysis, but also suggests a lack of incentives to fulfill this valuable role. That said, there is some evidence that the PRSP process has stimulated the interest of parliamentarians and that they are now demanding capacity-building to help them undertake more detailed budget analysis. Effective auditing of accounts would assist legislators with such analysis, but line ministry audits are usually weak and untimely. Even in Tanzania, which met 8 out of 15 benchmarks of expenditure accountability set in the HIPC tracking system, line agencies’ internal audits did not meet the standards. In Burkina-Faso, an external audit office has been created, but it is not yet functional. In Bolivia, on the other hand, the independent auditor was regarded as technically competent, although its ability to scrutinize resource use across government was constrained by its own lack of budget. However, the reports it did produce were not passed on to Parliament, and it has few powers to enforce its own findings. 19 In short, both the role of Parliament and other formal auditing mechanisms remain weak in our sample countries (as in most low income countries) and the PRSP process does not seem to have made a visible impact on them. Civil Society Oversight Precisely because of the implicit recognition of the weakness of formal accountability mechanisms, the PRSP process bets heavily on the accountability brought to bear by civil society organizations and external partners. In fact, the PRSP process has provided new opportunities for CSOs to hold governments accountable, especially through participatory monitoring and evaluation mechanisms. In Tanzania, for example, the PRSP process has added new strength to the role played by CSOs in the PPER process, which now focuses on strengthening links between the PRSP, sector strategies, MTEFs and budgets. In addition, CSOs are seeking to strengthen their budget analysis skills and taking an leading role in public expenditure tracking surveys and in assessing service delivery. While expenditure tracking at the local level is generally poor with weak sanctions for local governments who do not report in a timely and comprehensive fashion, in Bolivia local level groups called Comites de Vigilancia have been given power of oversight with respect to municipal budgets. In particular, they have the ability to request that central government freeze transfers if misappropriations are detected. However, as discussed in the previous two sections, civil society ability to engage in detailed budget analysis is still limited, both by capacity and lack of user-friendly budget information. CSOs have more successfully found a role in monitoring the PRSP more broadly, and in scrutinizing whether the government has delivered on its action plan. According to a recent study of monitoring and evaluation systems, it is this kind of general political monitoring that is most effective in increasing government accountability. 9 Finding the appropriate balance between government and civil society scrutiny, however, is difficult. For instance, some have suggested that, in Bolivia, the balance has tipped too far in favor of CSOs, with the result of marginalizing core government oversight functions. It is also clear that CSOs, auditors and legislators do not collaborate effectively in any of our studies, and that there are clearly unexploited synergies between the work of all with respect to monitoring government policies, expenditures and impacts. Accountability to External Partners In the context of traditional project support, external finance is treated separately from the budget and accountability for donor funds is a relatively simple accounting exercise, albeit with negative consequences for the long-term development of government systems. However, as the importance of direct budget support increases in the context of the PRSP, both accounting for, and isolating the specific impacts of donor contributions becomes more difficult. Because money is fungible –especially when provided as general budget support--, the PRSP process has focused on general accountability in public finance management and on monitoring of performance. This is an enormous improvement upon the previous narrow focus for accountability of moneys spent on a myriad of projects. As a result, the focus of both government and donors is increasingly on government systems and overall results. This is a strong finding in our case studies. In Burkina Faso, for example, there is evidence that the move to budget support has promoted a more results-oriented culture in the Ministry of Finance, while in Tanzania the government is using its PRSP Progress Report to good effect in accounting to both internal and external stakeholders. In Vietnam, the government used the 2004 PER to show progress towards improved transparency and accountability in order to increase the willingness of donors to provide budget support. 9 GTZ, Governance and Democracy Division, “National Monitoring of Poverty Reduction Strategies/PRSPs,” 2004 20 Progress on harmonization of PFM and monitoring requirements, however, is still limited. In most countries, donors continue to use their own targets and performance benchmarks despite the emphasis placed by the PRSP process on homegrown points of reference. In some cases, however, there is progress. In Cambodia, for example, the Consultative Group (CG) has developed benchmarks, most notably in areas of special concern to major partners. While largely limited to policy actions as opposed to the outputs or outcomes picked up in the PRSP, some question whether it is optimal to have separate accountability mechanisms for donors in the face of limited capacity and a nationally defined set of actions set out in the PRSP. More recently, a joint donor-supported PFM strengthening program in Cambodia has been developed. Emerging Lessons It is clear from our studies that the PRSP process has led to increased scrutiny of budget expenditure as well as greater focus on monitoring service delivery and overall accountability. However, in countries with poor budget execution, improvements in the PEM systems are needed in order for the budget to be the key instrument of strategic policy implementation it is meant to be and to allow the exercise of proper accountability in its scrutiny by both internal and external actors. Given this, a number of suggestions are made: • It is important to improve budget execution at both the central and provincial levels through better macro-economic and revenue forecasting, and improved cash management leading to more reliable and strategic budget allocations to line agencies. • Increase the capacity of all stakeholders –internal and external audit, the legislature and civil society organizations to enable it to better understand the budget process, analyze information and scrutinize accounts more effectively. Improved capacity for performance analysis is also sorely needed to increase capacity to exercise accountability for results, starting in Ministries of Finance. • Synergies among various accountability agents/mechanisms should be explored, in particular between auditing bodies, Parliament and civil society. 5. Costing and Consistency with Overall Macro-Economic Framework Ideally, a PRSP should provide a comprehensive cost estimate for the programs it contains, including disaggregation by sector and economic classification. 10 This costing should be consistent with a realistic overall macro-economic framework. In addition, the paper should indicate which programs are priorities providing the government with guidance as to where cuts should fall in the case of revenue shortfalls and which additional programs should receive funds if growth and revenues surpass expectations. This ideal, however, had not been realized at any point prior to the PRSP process in most PRSP countries and, hence, it is not a realistic expectation that it should be fully included in a first PRSP. Moreover, in some countries (such as Bolivia or Burkina-Faso), a common shortcoming of the first PRSP was that it took a “new project” approach and focused on costing incremental activities (linked to HIPC funding). Taking account of initial conditions, therefore, what is interesting to ascertain is whether the PRSP has added to the impetus for better program costing and prioritization within a consistent macro-economic framework. In all our five countries, the PRSP indeed came closer than previous initiatives to a comprehensive costing of government objectives and programs. More importantly, as a consequence of 10 See JSA Guidelines available at http://poverty.worldbank.org/files/jsa%20guidelines.pdf 21 the PRSP process, a focus on costing sectoral strategies needed for the achievement of established targets continued after the initial paper was elaborated. For example, in Cambodia, the PRSP built on progress made through health and education sector strategies resulting in comprehensive and well-costed programs for these two key areas: the PRS process succeeded in further raising the profile, and highlighting the gains, in these important areas; in Bolivia, important efforts were made to link poverty reduction objectives with financing requirements including the programming of capital and some current expenditures; and in Tanzania, although the costing estimates in the initial PRSP were not particularly strong, subsequent PRSP Progress Reports have refined and expanded those estimates. In Burkina-Faso, on the other hand, challenges to sound costing continue, as the second PRSP includes a much- expanded list of priorities without a concomitant estimate of their costs. There is less evidence of an effective inter-action between the PRSPs’ needs-based approach and macro- economic fiscal programming. Ideally, there should be mutual feedback between PRSPs and macro- economic projections, with the former affecting the latter in medium-term plans (e.g. leading to plans to increase revenue collections to fund needed expenditure in poverty-reduction programs) and the latter prevailing over the former in the short run (e.g. revising targets, programs and expenditure levels included in PRSPs to reflect actual resource availability). In our case studies, however, we find little evidence of this inter-action. In particular, the domestic revenue side has been treated as a hard constraint in all the PRSPs reviewed, with the focus heavily placed on using PRSPs as tools to increase donor funding. Similarly, although hard budget constraints as reflected in MTEFs and budgets have tended to prevail over PRSP needs in the short run, there has been no well-defined feedback loop into adjusting PRSP targets and expenditure programs. The issue of the proper fit between PRSPs and macro-economic –and in particular, revenue and expenditure—projections is one that deserves further attention if identified PRSP needs are to be taken seriously while macro-economic stability is preserved. 5. Pro-poor Budgeting—Has Resource Allocation Become More Aligned with PRSP Priority Sectors? Assessing whether any budget is progressive is inherently complex given the methodological challenge of isolating the impact of particular programs, and the significant time lags between spending and likely poverty outcomes. In several of our countries, this challenge was more acute due to lack of information for both inputs and outcomes. Consequently, in this section we limit our assessment to ascertaining whether spending allocations have been re-aligned toward PRS priority sectors and do not seek to pursue whether expenditures have translated into more pro-poor outcomes. We also do not seek to ascertain whether PRS priority sectors are actually pro-poor in the countries studied. Although this approach does have its limitations, we are comfortable with this definition of ‘pro-poor expenditure” since the sectors identified as priorities in most countries PRSPs are those that benefit incidence analysis typically finds to be most progressive (e.g., primary health, primary education, rural roads) and, equally importantly, they are the sectors perceived by the poor themselves to be of most benefit to them, as determined through consultative processes. Ascertaining the degree of alignment of budgets with PRS priorities is also a valuable exercise in determining whether PRSPs are strategic documents that find their reflection in actual budget allocations or whether, like many previous development strategies, they are vague declarations of government intent. We begin by considering the extent to which the PRSP in each of our countries provides direction for realignment of expenditures. Second, we analyze the actual allocations to a number of priority sectors in recent years and look at expenditure projections, assess whether allocations as a share of the budget have increased, and where possible, whether current expenditures have been aligned with capital to any 22 significant effect. Third, this section looks at whether there is any evidence that allocations have shifted in favor of poorer regions in our case study countries. The PRSP as a Strategy for Budget Allocations Each of our five countries PRSPs identified priority sectors (e.g. education and health care) and, in some cases, core priority areas (e.g. primary education and primary health care) and indicated that allocations should be redirected to these sectors and areas. Although data difficulties were encountered in all countries (details provided in each case study), we were nevertheless able to provide a good view as to the evolution of resource allocations and the degree of consistency of these allocations with PRSP priorities. Sectoral Allocations All our five case studies found that, since the advent of the PRSP, there has been a significant increase in funding in absolute real terms for PRSP priority sectors. In Burkina-Faso and Tanzania, for example, real per capital allocations to PRS priority sectors and core priority areas are estimated to have double between fY99 and fY04. Figure 1: Tanzania--Real Per Capita Expenditures in Priority Sectors 3 5 ,0 0 0 .0 0 3 0 ,0 0 0 .0 0 Per capita public expenditures 2 5 ,0 0 0 .0 0 2 0 ,0 0 0 .0 0 P r io r it y B r o a d P r io r it y N a r r o w 1 5 ,0 0 0 .0 0 1 0 ,0 0 0 .0 0 5 ,0 0 0 .0 0 - 99 00 01 02 03 04 05 06 Y Y Y Y Y Y Y Y F F F F F F F F 23 Figure 2: Burkina-Faso--Real per Capita Expenditure in Priority Sectors and Core Priority Areas 18000 16000 14000 12000 10000 8000 6000 4000 2000 0 1998 1999 2000 2001 2002 Priority Broad Core priority The distribution of increases across priority sectors, however, varies greatly. The education, agriculture/rural development and transport sectors seem to have benefited the most from expenditure increases. Health care, on the other hand, seems to have received a much smaller increase although it is hard to ascertain due to the move to vertical off-budget global programs in the sector. Figure 3: Tanzania--Per Capita Spending by Priority Sector Between FY99 and FY03 16000 14724 14000 T.Sh. (at constant FY03 prices) 12000 10000 FY 9 9 8000 7029 6205 FY 0 3 6000 5130 4000 3091 2846 1742 1864 2000 709 612 636 465 237 0 0 HIV/AIDS Education Health Agric. R&E Roads Judiciary Water 24 Figure 4: Burkina-Faso Real Per Capita Spending in Priority Sectors FY99 to FY02 (In Million CFAF) 8000 7000 6714 6095 6000 5000 4729 4000 3186 3000 2712 1828 2000 1000 0 Health Education Rural Development 1999 2002 Prioritization within sectors (to core priority areas) has also varied significantly. Primary education has tended to benefit the most from the increases in education sector spending, receiving, for example, 80 percent of the total sectoral increase in Burkina-Faso. Primary health care, on the other hand, has not always received priority within the sector and, in some cases, it has even lost relative ground. In Burkina- Faso, for example, the share of the total health budget devoted to primary health since the advent of the PRSP decreased from around 56 percent in 1999 and 2000 to 46 percent in 2001 and 2002. In all our five countries, in fact, the increase in budget allocations to health care and within health to primary health, have been quite modest, especially by comparison to the progress made in (primary) education. These varying in trends (which are reflected in different progress in outcomes) could be related to the different role played by the international community in both sectors—the Education for All initiative versus vertical global health programs—as well as to the recent evolution of sector policies –with increased role of state provision in education and reduced role in health--. Although answering these questions falls well beyond the scope of this paper, our case studies do provide some room for optimism in recent progress in the education sector and raise concerns in the lack of progress in health. Finally, rural roads seems to be the most under-funded core priority area. In Tanzania and Burkina-Faso, for example, the sub-sector only received 4 and 3 percent of sectoral allocations respectively. Absolute increases in expenditure, however, can be the result of a general increase in resource availability and do not necessarily represent a change in priorities. Hence, in order to ascertain whether these sectors have received priority through the budget allocation process, our case studies also examine relative shares of the budget devoted to PRS priority sectors and compare expenditure increases in priority and non- priority sectors. Cambodia presents the most positive picture. As Table 2 below shows, health, education, 25 agriculture and rural development have all benefited from an increased share of current expenditures between 2000 and 2002 while the share designated for security and defense is expected to decline. 11 Table 2: Share of Current Expenditures, Cambodia (by Percentage) fY00 fY01 fY02 fY03 fY04 fY05 Health 8.4 9.5 10.9 12.0 12.5 12.9 Education 14.0 16.2 18.2 18.5 19.5 20.0 Agriculture and rural development 2.9 4.6 3.8 3.4 3.8 6.4 Security and defence 37.1 30.1 24.7 22.0 20.8 19.4 Note: Figures for fY03 onwards are projected In Burkina-Faso, we also find evidence of government prioritizing PRS sectors as the share of priority sector expenditure in total discretionary spending 12 seems to have increased gradually since 1998 (see figure 5 below) with a slight setback in 2001 due to a delay in the disbursement of foreign aid and the concomitant decrease in primary education and primary health expenditure. Figure 5: Burkina-Faso--Share of Priority and Non-Priority Expenditure in Total Discretionary Spending 100% 90% % of total expenditure 80% 70% 60% 50% Non Priority 40% Priority Broad 30% 20% 10% 0% 1998 1999 2000 2001 2002 In Tanzania, on the other hand, priority sectors do not seem to have consistently received priority in budget allocations. Whereas percentage expenditure increases for priority sectors and core priority areas exceed increases in non-priority expenditures in FY00 and FY02, the situation is the reverse in FY01 and FY03. 11 These findings, however, must be seen in the light of two considerations--the broad definition of priority expenditures and the post-conflict nature of Cambodia’s situation. Regarding the first consideration, we need to take into account that, in Cambodia, a proportion of the education budget is spent on sport (not strictly part of the core priority area). Regarding the second, we must make allowance for the fact that re-allocation is relatively easier in a stabilizing post-conflict situation such as Cambodia’s (where the share of the budget initially allocated to security was unusually high) than in more stable situations like those of our other sample countries. 12 Discretionary spending is defined as government’s own resources plus general budget support. 26 Figure 6: Tanzania--Percentage Change in Priority and Non-Priority Expenditures % c h a n g e in p r io r ity a n d n o n p r io r ity s e c to r e x p e n d itu r e 6 0 .0 % 5 0 .0 % 4 0 .0 % 3 0 .0 % 2 0 .0 % Pr io r ity Na r r o w Pr io r ity B r o a d 1 0 .0 % No n - Pr io r ity 0 .0 % FY 0 0 FY 0 1 FY 0 2 FY 0 3 FY 0 4 FY 0 5 FY 0 6 - 1 0 .0 % - 2 0 .0 % - 3 0 .0 % It is often argued that one of the main advantages of budget support is that it can result in a more balanced distribution between capital and current expenditures as opposed to the heavy emphasis on capital expenditure caused by projects. Our case studies found no consistent evidence of increased expenditures on operation and maintenance since the introduction of the PRSP, though an incipient trend seems to be emerging in some countries. In Burkina-Faso, almost the whole increase in priority sector spending over 2000-2002 is due to increases in development/investment spending, with only very small increases in recurrent spending. This is a worrying trend as some of the direst needs (in particular in the social sectors) are in non-wage current expenditure (operations and maintenance and school and health center materials). Similarly, in Cambodia, the PRSP does not yet seem to have helped ministries plan their O&M requirements more strategically, although there is some evidence of good practice in education and health. In Bolivia, on the other hand, the share of the budget directed to the National Roads Authority, which is responsible for the maintenance of the main and trunk roads, has indeed received increases in recent years. Finally, the PRS process seems to be contributing to increased prioritization in the implementation of budget cuts. Four out of our five case studies show strong evidence that, as expenditure cuts are effected, PRS priority expenditures have been protected. In Tanzania, protection of PRSP priority sectors from budget cuts is official policy and part of the benefits entailed in the definition of being a ‘priority sector.” In Burkina-Faso, we found the government substituted own funds to (partly) compensate for aid shortfalls and hence, to the extent possible, protect overall allocations in priority sectors. In Vietnam, revenue over the past four years has consistently over-performed and, hence, there have been no shortfalls. However, during the East Asian crisis, the government clearly protected allocations to what would later be PRSP priority sectors, whose share in the budget increased during the crisis (as other sectors’ shares dropped). In Cambodia there were not any budget cuts during the period under review. In Bolivia, on the other hand, budget cuts do not seem to have been made with strategic objectives in mind. 27 Regional Targeting of Expenditures In three of our case cases it has been possible to consider the regional distribution of expenditures and come to some preliminary conclusions on whether funds have become more progressively allotted in geographic terms since the advent of the PRSP. In Cambodia, an analysis of health and education expenditures per capita and provincial poverty data showed that in 4 out of the 14 poorest provinces funding was above the national average, but in the remaining 10 it was either average or below. Given the historical nature of budgeting, however, this is not surprising as increases currently accrue to those wealthier areas with higher numbers of facilities and personnel. Bolivia provides similar conclusions. Although increased municipal expenditures for housing and sanitation did map broadly to poorer prefectures, no straightforward progressive relationship could be determined in the health and education sectors, despite the fact that HIPC funds devoted to these sectors are allocated according to a pro-poor formula. In addition, a recently introduced social insurance scheme for maternal and child health was also criticized for poor targeting as it provides universal coverage for a range of secondary as well as primary healthcare services. However, more positive examples of successful targeting can be found: in Burkina Faso, for example, a new primary education development fund focuses on the 20 provinces with the lowest enrollment rates, while transfers to new local councils in Cambodia are based on a formula that factors in a measure of need. Moreover, the Tanzanian government has used data from the 2002 household budget survey and census as well as analysis from its poverty monitoring system to elaborate a poverty map as well as a fiscal transfer system that takes account of regional diversity in poverty and social indicators. Similarly, the government of Vietnam plans to use a study of the geographical allocation of expenditure based on the 2003 poverty map to continue increasing the resources allocated to poorer areas in the coming years. Overall, it seems fair to conclude that although the geographical distribution of expenditure is not being systematically taken into account when budgets are programmed, there is a general trend toward increased presence of poverty considerations in geographical allocations than was the case prior to the PRSP process. As Box 3 shows, international experience suggests that further progress can be made by explicit including considerations on poverty impacts in budget submission guidelines in order to ensure that line ministries consider the consequences on marginalized groups and regions. Box 3: Poverty criteria as a guide to budget allocations, Uganda In recent years, the Government of Uganda uses seven criteria to assess the poverty focus of Budget Framework Papers submitted by line agencies, namely (i) addressing needs of the poorest 20%; (ii) addressing gender inequalities; (iii) addressing geographical inequalities; (iv) HIV/AIDS; (v) environment; (vi) empowerment of communities; and (vii) addressing inter-sectoral linkages. However, in order to increase focus and reduce workloads for Sector Working Groups the Guidelines have been amended and simplified over time. For the financial year 2002/2003, the questions were narrowed down. As a result, line agencies had to ask themselves whether they had addressed the needs of poor people in their sector; addressed geographical inequalities; and designed measures to empower the poor people in their budget submissions. Moreover, a Poverty Eradication Working Group reviews and make recommendations on the overall allocation of resources and intra-resource allocations within sectors. It also considers other budget policies that have an impact on the poor and which sectors qualify as “priorities” (to be under the Poverty Action Fund), as well as advises the Sector Working Groups in applying the PRSP crosscutting principles to the Budget Framework Papers. Source: The PRSP and Budget Linkages: The Case of Cambodia, Jeni Klugman and Rob Taliercio, 2003 28 EmergingLessons A key goal of the PRSP process was to encourage increased expenditure in pro-poor priority sectors and core priority areas. All our five case studies find strong evidence that the process has been successful in this regard and that there have been strong real expenditure increases in PRS priority sectors and core priority areas. There is also evidence (in Burkina-Faso, Tanzania and Vietnam), that these sectors do receive protection in times of budget crunch, bearing testimony to real prioritization by governments. Our case studies also show that, although poverty considerations are not consistently considered in determining the geographical distribution of expenditures, there is a trend toward greater consideration of these issues in all our five countries. These are very important achievements. Some priority sectors have been more consistently well-endowed (e.g. education, transport, agriculture) than others (health) and some priority areas have received much larger increases (primary education) than others (primary health and rural roads). These divergences seem to stem both from the behavior of the international community as well as from the true degree of ownership of priorities by governments. There is less consistent evidence about whether these prioritization will be sustained in the medium term. In Cambodia, medium-term projections do show a continued trend toward increased budget shares to RPS priority sectors. In Burkina-Faso and Tanzania, however, MTEF projections for 2004-2006 reveal a stagnation and even a decline in the relative shares devoted to PRS priority sectors. This trend is worrying, as resource requirements in PRS priority sectors as established by careful costing work show important shortfalls in Tanzania. Similarly, in Burkina-Faso the new PRSP contains an expanded list of priorities which can only lead to reduced allocations to original PRS priority sectors and this is reflected in the MTEF. While data limitations make any analysis at this stage preliminary, our findings point both to important achievements (especially for such a short period of time) as well as to the need for greater congruence between resource allocation and the PRSP. This could be facilitated by: • Avoiding the over-expansion of the list of priority sectors and core priority areas included in PRSPs. The sectors, areas and programs that are considered priorities should be informed by the messages emerging from PRSP consultations and ratified by Parliamentary approval. • Defining which specific programs and expenditures will be considered priority spending and consistently identifying them in PRSPs, MTEFs and budgets. The process of identification should include agreement on the criteria for assessing whether a particular program should be regarded as a priority. Ideally, the identification of specific PRSP priority programs should be led by the MoF but would involve substantive line agency engagement as well as feedback from civil society, development partners, and possibly parliamentary review and approval. • Developing transparent, technical criteria for the prioritization of major spending proposals across sectors (which could be complemented by sector-specific guidelines) for use in the annual and medium term budgeting processes (as is currently done in Uganda). Since the extent to which effective prioritization can take place significantly hinges on the existence of program-based budgeting, we would like to stress the need to expedite progress on this front. • Re-examining existing programs in terms of consistency between identified priorities for poverty reduction and resource allocations and complement this with other kinds of analysis, such as cost- effectiveness analysis. 29 • Using priority criteria to guide capital spending, thereby increasing the transparency and accountability of public resource allocations. • Using the results of the poverty mapping work to check proposed allocations and executed expenditures against regional measures of deprivation. 6. Donor Coordination and Alignment with PRS Priorities and Predictability and Variability of External Financing As noted at the outset of this paper, four of our five countries are highly aid dependent (Vietnam being the exception). In Cambodia, for example, external financing exceeds domestically-mobilized public resources by about 30 percent while in Bolivia and Tanzania in 2001-2002 the share of the capital budget that was externally financed amounted to 49 and 90 percent respectively. In countries with such high aid dependency ratios, the cooperation of the donor community is essential in order for the PRSP approach to be successful. In particular, aid coordination to avoid the fragmentation of expenditures and systems reporting, the alignment of aid instruments with PRSP priorities and the predictability and low variability of financing are essential to allow governments to implement their PSRPs. Our case studies examine the extent to which the donor community has delivered on these fronts. Efforts to improve partnership between government and donors often pre-date the PRSP. In Tanzania, for example, the Tanzania Assistance Strategy and the PPER exercise were in place before the advent of the PRSP while, in Bolivia, the New Relationship Framework launched in 1999 sought to ensure alignment with government priorities as well as manage donors in a more coordinated manner. This final section considers whether the PRSP has provided additional impetus to such initiatives, whether it has proven a useful tool for governments in managing external finance better to deliver on its pro-poor priorities, and how this potential could be exploited further. Donor Alignment and Aid Modalities Most donors in our case study countries have indicated their commitment to supporting PRSP priorities and their willingness to align their assistance accordingly. In some cases, there is evidence that this is indeed occurring. In Cambodia, for example, a move away from funding core government functions and economic development to supporting the social sectors has been observed for some years, and there is evidence from the health sector that external finance is distributed progressively in regional terms. However, as discussed in the forgoing section, in all our five countries, PRSP priority areas are so broadly defined as to make alignment a relatively easy task for donors at least in nominal terms. In Bolivia, for example, the ministry responsible for external finance estimated that 99 percent of donor funding was aligned with the PRSP pillars in 2002. However, not all activities under each pillar could be regarded as pro-poor and this figure also obscured the fact that aid was not ideally distributed between the various priority sectors. Further analysis of donor programming for the periods 1998-2002 and 2003-2006 in Bolivia was more illuminating. This analysis showed that few donors were prepared to commit even new project support to PRSP priority sectors which the government regarded as under-funded, and generally tended to crowd into more popular areas. While this lack of flexibility can be explained in part by historical commitments to specific sectors or issues, it clearly militates against closer alignment between the PRSP and external finance. The significance of this lack of donor alignment with a country’s agreed priorities becomes even more apparent when one considers the requirement of domestic counterpart funding for many donor-funded 30 projects. Estimates from the 2003 Bolivian budget suggest that almost 20 percent of internal finance is linked to foreign financing. Thus, where donors do not program explicitly to deliver on PRSP priorities, the domestic budget is also directed away from these vital activities making overall alignment of the budget to PRSP priorities extremely challenging. The PRSP approach places a premium on moving towards aid modalities that minimize transaction costs for the recipient country as well as give the government more control over how funds are spent. Therefore, another way to ascertain the degree of alignment of donor financing with the PRSP is to trace the evolution of budget support. A move towards budget support, however, is only feasible and desirable when a number of preconditions providing confidence in a government’s ability and willingness to manage resources effectively are in place. These conditions typically include a stable macroeconomic situation, strong country ownership of the PRSP with clearly articulated structural and sectoral policies and programs and evidence of momentum toward implementation; good progress in “getting the basics right” in public expenditure management; and confidence in the area of governance and public finance management transparency and accountability. As much of the forgoing analysis will have shown, these conditions are not always in place in our sample countries, hence hindering progress with respect to moves towards budget support. In addition, it has not always proven easy for donors to agree on the shared conditionalities, monitoring frameworks and procurement plans needed to facilitate budget support and basket funding arrangements. In Burkina Faso, for example, tensions were observed between various donors with respect to which triggers should be used to unlock funding, while in Bolivia, the performance of the social funds testifies to a lack of progress with respect to harmonizing procedures. However, practice suggests that with time, donors can reach compromises and that the benefits for the recipient country of streamlined aid modalities are significant. Our case studies on Tanzania and Burkina-Faso, however, provide some evidence that the PRSP approach has indeed increased donor confidence in these governments’ ability to manage resources for poverty reduction, as is confirmed by large increases in the amounts of external assistance provided in the form of budget support in recent years. As a result, between 1999 and 2002, program grants and loans as a percentage of GDP witnessed significant increases in both countries, rising from 1.8 to 3.2 percent in Tanzania and from 2.9 to 4 percent in Burkina-Faso. In Bolivia, we also observed both increased donor commitments to fund via budget support as well as indications that the number of projects would decline in the period 2003-2006 while in Tanzania SWAps have been established in health, education and agriculture and several thematic basket funding arrangements are in place. At the same time, however, as of 2003, there seems to be a decline in the share of budget support in overall grants and loans in both Tanzania and Burkina Faso as some large projects in the infrastructure and rural development sectors get under way. Predictability of External Finance In the aid-dependent countries of our study, timely and predictable disbursements of donor commitments are critically important for effective public expenditure management and PRSP implementation. Our case studies show mixed performance in this area. In Tanzania, for example, the PRSP has given additional impetus to government efforts to collate information on donor funding, and there has been progress in capturing in-kind transfers, direct support to local governments and funding from NGOs. Conversely, in Burkina Faso, there appears to have been little progress on developing a comprehensive overview of donor finance and, thus, the understanding of the full parameters of the budget is still limited. From our case studies, it seems that the PRSP process provides additional impetus and rationale for to recipient country governments to solicit improved provision of information on aid projections and disbursements, 31 but that only when the units responsible for external financing are strong enough to provide leadership (as in Tanzania) is this potential actually realized. Moreover, even when approached by an external financing unit providing specific details on the information needed, the frequency of the provision and the format necessary, development partners have problems in supplying such information, especially for projected expenditures. Few donor countries themselves have medium term budgeting and hence are unable to make firm commitments beyond the annual horizon. As a result, efforts by recipient countries to develop MTEFs are undermined as resource envelopes can not be estimated accurately in the absence of foreign financing commitments. Donor flows seem to be extremely volatile. Evidence from Tanzania shows high year-to-year variability in donor flows, which prove to be much more volatile than domestic resources. Similarly, in Burkina- Faso, we find that foreign aid has varied significantly from year to year, compounded with poor predictability in the flows as HIPC aid delayed for almost a full year from 2000 to 2001. We found no clear-cut evidence as to what types of aid were more volatile. In Tanzania, for example, program aid had historically been more variable than project aid, but the situation was reversed in 1998/99 when major bilateral donors switched to budget support. Even within a year, aid flows are hard to predict, which significantly impairs the ability of governments to properly plan and execute budget expenditures. Analysis from Bolivia suggests that, in 1998-2002, the best performing donors disbursed only 70 percent of their committed aid while the poorest performers could only provide 33 percent, with the large multi-laterals tending to fall into the latter category. Similarly, evidence from Burkina-Faso and Tanzania finds important gaps between commitments and disbursements. Figure7: Tanzania--Projected and Disbursed External Assistance 200 Tshs. billion A c tu a l 100 C o m m it t e d 0 Q1 Q2 Q3 Q4 Q1 Q2 FY 0 2 FY 03 Evidence from Burkina-Faso also pointed to irregular flows throughout the fiscal year with a tendency for external financing flows to bunch during the last quarter of the fiscal year. 32 F ig u re 27. B u d g et S u p p o rt: Q u arterly D isb u rsem en ts 80% 70% % Disbursement 60% 50% 40% 30% 20% 10% 0% 1 2 3 4 Q uarter 1999 2000 2001 2002 Average 96-02 Emerging Lessons This section has highlighted the significance of donor financing for PRSP implementation and improved PEM. In particular, the above analysis points to the need for: • Greater self discipline on the part of donors to ensure all aid is on budget and aligned first and foremost with recipient governments’ rather than donors’ objectives. • Synchronizing aid disbursements with recipient countries’ budget cycle to support good planning and public expenditure management. • Improved forward planning on the part of donors allowing them to make better estimates of their contributions over the medium term. • Improvements in channeling aid through the treasury and in reporting external finance in the budget. • Accelerated progress on harmonizing procedures. • More realistic common understanding between donors and recipient countries on conditions for aid disbursement to help minimize gaps between committed and disbursed amounts. 7. Conclusions Evidence from our five case studies suggests that the PRSP process is having a significant impact on budget processes and budget allocations. The accomplishments are particularly impressive given the 33 difficult nature of the changes envisaged and the short period of implementation. Moreover, the PRSP approach attempts to affect results by influencing incentives and, in particular, through affecting the environment in which policies are made and the actors involved in policy-making processes, an approach which bodes well for sustainability but does not necessarily yield immediate results. In particular, our case studies show the following trends: Accomplishments • Greater production, use and dissemination of strategy, poverty and fiscal data and increased expectations for the data to inform policy-making; • Greater dialogue between planning and financing units within ministries and between line ministries and ministries of finance; • A more active involvement of civil society in budget elaboration and/or budget monitoring processes; • (Significantly) increased allocations to PRSP sectors and core priority areas in real per capita terms; • A move toward general and sector budget support and greater recognition of the need for donors to align aid with recipient country priorities. Shortcomings • Weak evidence that data and monitoring and evaluation results systematically affect policy- making; • Weak role of Parliaments in the PRSP process and weak ownership of PRSP priorities by Parliaments; • Uneven increases in resource allocations across priority sectors (with education, transport and rural development fairing well and health fairing less well); • Uneven increases in resource allocations within priority sectors (with more consistent evidence of prioritization of primary education than primary health and very poor performance in rural roads); • Continued high levels of donor financing off-budget, poor forward aid commitments, poor predictability and high variability of aid within-year. Overall, our findings are highly encouraging as the PRSP process in a short period of time has succeeded in difficult areas where traditional public sector reform initiatives had failed. In particular, we find evidence that it has increased the transparency, openness and pro-poor character of budgeting processes. We also find that these reforms have a much larger impact when implemented in a context where the public finance management basics are in place. Hence, the types of reforms the PRSP process encourages are supplements not substitutes for improvements in basic public expenditure management. The PRSP process has also contributed to highlighting areas where important gaps still exist and has created expectations of further progress. Although some observers consider this to be a shortcoming, in our view, it is one of its strengths, as greater pressure for plans to fit budgets, for greater transparency, 34 participation and accountability are the engine for further progress in the future. The process has also created greater awareness of the need for capacity-building in line ministries, Parliaments and civil society if they are to be well-informed actors in budget processes. This is also to be celebrated. Further progress on all these fronts, however, is needed and highly dependent on the continued ownership of the PRSP process by recipient governments and donor community alike. wb214506 N:\Alonso\Fiscal-CaseStudies\synthesis paper.draft 3 jan 7 04.doc January 18, 2005 2:23 PM 35