WPS 4409 Policy ReseaRch WoRking PaPeR 4409 Reducing the Transaction Costs of Development Assistance Ghana's Multi-Donor Budget Support (MDBS) Experience from 2003 to 2007 Carlos Cavalcanti The World Bank Africa Region Poverty Reduction and Economic Management Department November 2007 Policy ReseaRch WoRking PaPeR 0- Abstract This paper examines whether the structure of the Multi- rules that allowed all the participating agencies to have Donor Budget Support (MDBS) in Ghana evolved over a role in the decision-making process, and, in doing so, time to minimize transaction costs commonly found in encouraged these agencies to increase the share of their accessing and delivering development assistance in multi- contribution coming through the MDBS, rather than donor settings. While the MDBS was expected to reduce through large projects and off-budget disbursements. the transaction costs involved in dealing with multiple There was less success in reaching a settled view on how development agencies, it created three additional sources to reduce so-called measurement costs, however. While of transaction costs: coordination failures, the costs of the group of development agencies made several attempts collective action, and measurement costs. The answer to overcome the difficulties in measuring progress in the that emerges from this paper is that the structure of the program supported by the MDBS, it was not able to MDBS evolved to mitigate these transaction costs. The reach consensus on the extent to which the monitoring problems associated with coordination was addressed by of the program should rely on outcome indicators. The delegating the policy dialogue to sector-specific groups Government did not favor the use of outcome indicators, aimed at reaching agreements over a narrower set of issues and some development agencies placed greater emphasis and amongst a smaller group of participants. Also, the on maintaining a dialogue around policy actions aimed at MDBS reduced the cost of collective action by devising reaching the desired outcomes. This paper--a product of the Africa Poverty Reduction and Economic Management (PREM) 4 Division of the Africa PREM Department--is part of a larger effort in the department to document the recent experience with development policy operations in multi-donor settings. Policy Research Working Papers are also posted on the Web at http://econ. worldbank.org. The author may be contacted at CCavalcanti@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Reducing the Transaction Costs of Development Assistance Ghana's Multi-Donor Budget Support (MDBS) Experience from 2003 to 2007 Carlos Cavalcanti1 1World Bank Senior Economist and co-chair of the MDBS from 2004 to 2007. Comments from Mats Karlsson, former Country Director for Ghana, and David Kuijper, Senior Advisor for the Dutch constituency at the Board of the World Bank (EDS 19) and co-chair of the MDBS from May 2004 to May 2005, are gratefully acknowledged. The views expressed in this paper are those of the author and do no reflect the views of the World Bank or its Board of Directors. The usual disclaimers apply. I. Introduction The Multi-Donor Budget Support (MDBS) has been in operation in Ghana for the last four budget cycles (2003-2007), bringing together ten development partners (DPs)2 that provide general budget support to assist the Government of Ghana's (GoG's) efforts to implement the Ghana Growth and Poverty Reduction Strategy (GPRS). The MDBS was set up in early 2003, soon after the first edition of the country's poverty reduction strategy was approved by Parliament, and in the context of the Government's decision to join the Highly Indebted Poor Countries (HIPC) initiative. The overarching objective of the MDBS is to provide more harmonized assistance for Ghana to implement the GPRS, accelerating growth and poverty reduction, as well as reduce the transaction costs of accessing and delivering development assistance so as to focus the dialogue on policy actions.3 To meet this objective, DPs have worked to improve co-ordination and to harmonize systems, with the purpose of reducing the cost of accessing aid by Government in terms of negotiating, reporting, monitoring and assessment. While there has been good progress over the past four years, it has not been without tensions. The experience of the last four years suggest that there are three sources of tensions that arise in negotiating, monitoring and enforcing agreements within the MDBS framework. The tensions within the Government of Ghana, where the Ministry of Finance and Economic Planning (Ministry of Finance hereafter) needs to balance the needs of the national budget with the demands of the several sector ministries that implement the policy actions supported by the MDBS program. The tension between the headquarters and the field offices within each development partner agency, which I call the vertical balance. And the tension within the MDBS between different participating development partner agencies, which I call the horizontal balance. The tensions that have arisen in the workings of the MDBS appear to mirror some of the transaction-costs found in the economic literature. These transaction costs reflect the costs of contracting under uncertainty, and the difficulties that arise in settling 2 African Development Bank, Canada, Denmark, France, Germany, the Netherlands, Switzerland, United Kingdom, the European Commission and the World Bank. Observers included: representatives from the Embassy of Japan, and from FAO, JICA, JBIC, UNAIDS, UNICEF, UNDP, WFP, and USAID. 3 The objectives of MDBS, as set out in the Framework Memorandum signed between the Government of Ghana (GoG) and the Development Agencies in May 2003, were to harmonize Development partners' policies and procedures in order to minimize transaction costs for the Government of Ghana. This included: (i) agreeing to common benchmarks against which performance is assessed; (ii) improving dialogue between Development partners and the GoG; and (iii) basing funding commitments and disbursements on the achievement of agreed targets. Other potential advantages include: (i) the reduction in competing demands on the GoG from Development partners; (ii) the reduction in tied aid; (iii) the promotion of GoG accountability for service delivery; (iv) the facilitation of a broad-based discussion on reform agenda; and (v) improved DP co-ordination. In striving to achieve these objectives, the MDBS also aims at increasing the predictability of donor flows. This will benefit overall economic development and the implementation and efficiency of programs in priority sectors. -2- disagreements and enforcing contracts. In the case of development assistance provided in a multi-donor context, these transaction costs fall under two broadly defined headings: coordination failures and measurement problems. In addition, there are the costs associated with collective action that may prevent agencies from contributing to a multi- agency fund because of concerns about possible free-riding. The question this paper aims to answer is whether the MDBS structure did evolve to minimize these costs and, therefore, fulfill its stated objective of reducing the transaction costs in accessing and delivering development assistance. The answers that emerge from this paper are that the structure of the MDBS evolved to mitigate coordination failures by delegating the policy dialogue to sector-specific groups that aimed at reaching agreements over a narrower set of issues. Also, the MDBS reduced the cost of collective action by devising rules that allowed all the participating agencies to have a role in the decision-making process, and, in doing so, encouraged all agencies to increase their contribution to the MDBS. There has been less success in reaching a settled view on how to best measure progress in the implementation of the program supported by the MDBS, however. While the MDBS attempted to overcome this measurement problem by agreeing with the Government of Ghana ahead of time on the indicators that would be used to ascertain whether a policy action was completed or not, it has not been able to reach consensus on whether to rely primarily on so-called outcome indicators. At the heart of these disagreements regarding the use of outcome indicators was the fact that the Government did not favor the use of outcome indicators and the evidence that some development agencies had a preference in directing the dialogue toward policy actions aimed at reaching the desired outcomes. This focus on policy actions leading to outcomes does not appear to have detracted the program supported by the MDBS from achieving important results during these first four years. The first set of more visible results came from the contribution of the MDBS toward macroeconomic stability. As highlighted in recently completed ODI briefing paper on the MDBS experience in Ghana,4 the MDBS resources helped the Government of Ghana reduce the fiscal deficit and lower the domestic debt to GDP ratio, while raising public expenditure on poverty reducing programs. The overall fiscal deficit in Ghana shrank from 4.4 percent of GDP at end-2003 to 3.0 percent by end-2005, providing scope for an increase in the share of credit allocated to the private sectors thanks to lower domestic debt to GDP ratios. The share of private sector in total domestic credit rose to 57.5 percent by end-2005, up from around 47 percent at end-2002. This increase in credit to the private sector is attributable to a large degree to declining public sector deficits, which allowed the public debt to GDP ratio decline 11 percent at end-2005, down from 24 percent in 2002. Meanwhile, the spending on poverty reducing programs reached 10.5 percent of GDP at end-2006, up from around 5 percent in 2002, accounting for just over one-third of overall Government expenditures. Moreover, the budget execution rate for most of the poverty reducing programs achieved higher than expected levels of budget execution, rising in those cases where additional funding became available during the year to over 100 percent.5 4Overseas Development Institute (2007). Budget Support to Ghana: A risk worth taking? Briefing Paper number 24, July. 5The sources of these data are several IMF staff reports. See the reference section at the end of the paper. -3- While 2007 marked the first deviation from the path of fiscal consolidation that began in 2002, with the overall fiscal deficit estimated to have risen to 7.5 percent of GDP, up from 3.0 percent at end-2005, some of the deterioration in fiscal accounts reflected difficulties the country was facing in dealing with the consequences of the rise in international crude oil prices. As a result, the domestic debt to GDP ratio, which was the cornerstone of fiscal policy during the first years of implementation of the MDBS supported program, rose to 14 percent by end-2006. The overall trend in public debt remained firmly downwards, nevertheless, with the central Government debt to GDP ratio falling to 42 percent in 2006 from 70 percent in 2005, reflecting the impact of the Heavily Indebted Poor Countries (HIPC) debt initiative and the Multilateral Debt Relief Initiative (MDRI). There were other important results in the Government's program supported by the MDBS, with progress reported in increasing access to health and education services. The share of supervised maternal deliveries rose to 54 percent by 2005, up from 49 percent in 2002, with particularly marked increases in the deprived regions (Central, Northern, Upper East and Upper West regions). There was also an increase in gross primary enrolment (GPER), with the national GPER reaching 92 percent by 2006, up from 81 percent in 2003. The latter result is one of the country's most important development successes, creating demands for new school buildings and additional teachers in the deprived regions. Lastly, there was progress reported in strengthening public financial management ­ the cornerstone of budget support, with improvements in budget coverage, timelier external auditing of public accounts, and in the implementation of the new public procurement and the new internal audit acts. The budget coverage was broadened to include more information on internally generated funds and donor grants. External audit reports became timelier, with the annual report by the Accountant General being submitted to Parliament within less of 12 months of the closing of the accounts. There was also progress in implementing the new public procurement law, with increased coverage provided by the entity tender committees and the tender review boards, and, under the auspices of the new internal audit agency law, the number of Government ministries, departments and agencies submitting internal audit reports began rising. As a result, Ghana's public expenditure management has improved since the time of the Highly Indebted Poor Countries (HIPC) completion point and it currently meets 8 of the 16 HIPC public expenditure management benchmarks, up from 7 benchmarks in 2004 and 1 benchmark in 2001. Using the Public Expenditure and Financial Accountability (PEFA) performance measurement framework, Ghana's PFM system is performing at an average standard, and in some respects it score above the average. The results achieved in the implementation of the program supported by the MDBS were possible, to a large extent, because of the ownership of the reform program by those responsible for implementing the program. The Government's ownership of the program evolved in three distinct phases. In the first phase Government led the preparation of the Ghana Poverty Reduction Strategy (GPRS), while the development agencies seized on this opportunity to design a new instrument for development -4- assistance in Ghana ­ the Multi-Donor Budget Support (MDBS). The second phase involved developing sector strategies aligned with the GPRS that would provide operational guidance for the implementation of the GPRS, as well as focus to the dialogue with the development agencies. Sector strategies completed during this phase include, inter alia, the Power Sector Reform Strategy, the Private Sector Development Strategy, the Education Sector Development Strategy, the Food and Agriculture Sector Development Strategy. The third phase is ongoing and builds on the two earlier phases. It involves the Ghanaian Government taking a stronger leadership in the dialogue, drawing on these strategies to define the content and focus of the policy actions supported by the MDBS. The rest of this paper elaborates further on the principal hypothesis being examined in this paper, namely that the results achieved by the MDBS (some of which are described above) were in part due to the ability of the MDBS structure to evolve, minimizing the transaction costs usually found in accessing and delivering development assistance in multi-donor settings. It begins with a brief overview of the literature on transaction-costs economics, attempting to provide a unified framework for understanding the transaction-costs that usually arise in managing development assistance. Next, it endeavors to build a bridge between the body of theory reviewed in Section II and the MDBS experience by examining the coordination failures, the costs of collective action, and measurement problems that appear to have shaped the evolution of the MDBS in Ghana. Within the discussion of the evolution of the MDBS, the paper examines the issues at stake in the so-called vertical and the horizontal balances, mapping the preferences of different clusters of development agencies. The paper closes with a summary of the main conclusions. II. Do Transaction-Costs Matter for Development Assistance? One main advantage of setting up a multi-donor structure such as the MDBS is reaping the benefits of the harmonization of development agencies' policies and procedures with a view to minimize transaction costs faced by the Ghanaian authorities in accessing development assistance, as well as the costs faced by the development agencies in delivering this assistance. By reducing the fragmentation that results from negotiating, monitoring and enforcing multiple agreements with development agencies, the MDBS aimed at delivering benefits of reduced costs and more time available for dialogue and implementation. To better understand whether these transaction costs matter, this section reviews the literature on transaction-costs economics to examine whether transaction-cost theory can shed light on the practice of development assistance. Coase (1937) created transaction-costs economics by shifting the focus of organizations from technological possibilities and the maximization of some market objective (e.g. profit) to transactions and to the management of relationships. The latter were termed transaction costs and encompassed the cost of deciding, planning, arranging, negotiating, monitoring and enforcing agreements between two or more -5- parties. Transaction costs also include losses resulting from inefficient group discussions, plans, arrangements, and inefficient group responses to changing circumstances, as well as the imperfect enforcement of agreements. The central tenet of transaction-costs economics is that the drive for efficiency and effectiveness should lead to organizations that economize transaction costs. The basic tenet of transaction-costs economics could not be confronted with evidence in this very general form, however. To make specific predictions from the theory of transaction-costs economics, it was necessary to first recognize the costs characteristically associated with specific transactions, and identify how the circumstances caused these costs to vary. Williamson (1985) proposed a framework within which Coase's transaction cost theory could be made more specific and operational. Williamson's theory was based on the analysis of the costs of contracting, and was predicated on the central tenant that any contract calls for the future delivery of a good or service, or the future performance of work. However, contracts could never specify precisely what actions were to be taken, since the parties in a contract can never perfectly anticipate all possible contingencies that might affect the costs of performing as promised, or even their ability to do so. Even for circumstances that could be anticipated, it was often more economical to respond when the need arose than to plan in advance for every foreseeable contingency. Drawing up contracts with too many fine distinctions might simply increase the likelihood that emerging events would fall into areas of ambiguity or overlap, leading to disagreement that will have to be resolved after the fact. Enforceable contracts could be made contingent only on information that all parties could share and could be verified by an outside party. What are the consequences of this incompleteness of contracts in Williamson's framework? If planning and contracting were complete and costless activities, parties to a contract would determine in advance and in detail the best possible actions for every contingency that might arise, and the contracts would specify those actions to be taken and would provide the appropriate incentives. Because planning and contracting consume real resources, and because perfectly explicit and freely enforceable contracts cannot be written, Williamson (1985) posits that contracting parties need to content themselves with an agreement that frames their relationship ­ one that fixes general performance expectations, provides procedures to govern decision making in situations where the contract is not explicit, and outlines how to adjudicate disputes when they arise. The difference between simple and complex contracts, as ways of organizing transactions, lies primarily in the institutions they specify for governing the relationship when unforeseen circumstances arise in the contract. Uncertainty about what circumstances would prevail when future actions were to be taken was the primary factor that made complete contracting impossible. Rigid contracts that recognize few contingencies are too costly because it is almost impossible to determine which future actions would be appropriate when there is uncertainty. Extremely flexible contracts are as costly as extremely rigid contracts, however. -6- Extremely flexible contracts are necessarily open to different interpretations and thus pre- empt effective renegotiations. There are good reasons, according to Coase's hypothesis, therefore that parties will agree to contractual arrangements that minimize the costs that arise from either extremely rigid or extremely flexible contracts. Later contributions to the theory of transaction-costs economics emphasized the bargaining costs of short term contracts These bargaining costs, if high enough, would lead to more complex governance structures designed to reduce the costs of reaching, monitoring and enforcing agreements. This literature on the transaction costs resulting from bargaining represented an important critique to the earlier literature on transaction-costs economics, which focused primarily on the implications of contract incompleteness. The argument was that, if the costs of negotiating short term contracts were sufficiently high, regardless of the level of uncertainty about future events, it would be preferable to place the agreement under a broader contract that would define parameters for settling disputes. The critique played an important role in shifting the focus from the costs and benefits of incomplete contracts to the relative advantages and disadvantages of more complex contracts or governance structures, defining a new research agenda focused on bargaining costs. What are `bargaining' costs that call for more complex governance structures? These costs were defined as the cost of monitoring and enforcing an agreement, as well as delays and failures to reach an agreement. For instance, when subjects in an experiment are asked to divide a sum of money, say 10 dollars, they should have little difficulty agreeing to split the sum equally without costly delays. However, when the division is more complicated, so that symmetry does not focus the bargainers' attention on the obvious solution, posturing, haggling, and disagreement is more likely, as each party seeks to stake out reasonably sounding position that yields a larger share of the available resources. There are many real life examples of how bargainers can fail to agree on an efficient solution, and, as demonstrated by Nash (1950), the bargainers might end up with nothing by demanding everything for themselves. Examples of why more complex governance structures are required to eliminate coordination failures that lead parties to settle for a less efficient solution are also found in budget support programs, especially once there is competition amongst development agencies interested in supporting the Government's development program. Suppose, for instance, a bargaining situation involving one recipient country and at least two development agencies. The example could involve a Government seeking funding for a development program, while the two development agencies willing to provide support have different policy agendas. The rules for this `development' game are as follows. If the Government proposed program does not match what the development agencies are interested in supporting, no agreement is reached and each party receives a zero payoff. Otherwise, the Government seeks support from the development agency placing the least demands. Almost all the possible outcomes in this auction version of the `development' game lead to an equilibrium situation, since, just as in the case of a competitive market, the Government will be able to reach the best possible agreement with the least possible effort. The rationale behind this outcome in -7- the `development game' is that, when there is competition among development agencies, and they are obliged to make simultaneous offers, the competition reduces the scope for disagreement with the Government. This limited scope for disagreement among agencies led to what in the MDBS parlance became known as an `unambitious development program'.6 The literature on bargaining costs classifies these costs under two broadly defined headings. The first class of bargaining costs result from coordination failures, such as the one in the example above. These costs arise when individual agencies pursue consistent, self-interested strategies, leading, in the absence of coordination, to sub- optimal outcomes. One example of a sub-optimal outcome resulting from negotiations within the MDBS is reaching agreement around a minimal common agenda, or an `unambitious development program'. A second source of bargaining costs result from so-called measurement costs. According to Williamson (1975) measurement costs arise in two instances: when information is asymmetrically distributed across parties and can be equalized only at great costs, or when it is costly to apprise an arbiter the true information in a dispute between two equally well informed parties. Enforceable contracts can be made contingent only on information that all parties will share and can be verified by an outside party. A unified perspective aimed at bringing together these two strands of the literature on transaction-costs economics appears to provide a useful framework for examining the transaction costs in accessing and in delivering development assistance. In an environment where contracts are necessarily incomplete, the contractual arrangement for development assistance lies somewhere in the range between extremely rigid and extremely flexible contracts. Similarly, to mitigate the risks of engaging in a longer term contractual arrangement (due to uncertainty), development assistance can settle for either a series of short term contracts (e.g., individual projects), or one large contract (e.g., budget support) that would be agreed at the highest possible level. The value of larger contracts, such as budget support, is useful to appreciate because of its importance in gaining broad support for policy reforms by elevating sector- specific issues to the Ministry of Finance and, ultimately, to the Cabinet, by linking agreements on these reforms to budget support operations from which the entire Government stands to benefit. These two dimensions of contracts for development assistance are presented in Figure 1 below. Project support is seen as lying somewhere in the southeast quadrant (point A), as individual projects are usually very detailed in nature, with rigid covenants and rely on a series of smaller interventions rather than one encompassing support program. The transition of development assistance from projects toward budget support, with all its associated benefits of lower transaction costs, is designed to move development assistance toward the northwest quadrant in Figure 1 (point B). The move 6For the record, it is important to note that two other factors appear to have accounted for some of the instances of `unambitious development programs' in the case of Ghana: the fact that a significant share of the MDBS' annual disbursements were contingent on the Government achieving the agreed policy actions, and the Government's related low confidence in the capacity of certain implementing agencies. -8- is not without costs, however, as tensions associated with uncertainty and bargaining costs (due to coordination problems and to the need for a new measurement system) build up. It is not uncommon therefore for development agencies to argue for a movement from the southeast corner to the northeast corner, with one large, but still a relatively rigid contract (point C). Figure 1. Dimensions of Contracts for Development Assistance One large contract B C More Flexible contracts More Rigid contracts A Several small contracts What Figure 1 cannot explicitly capture is the one important cost to be overcome when moving into a large, multi-agency contract ­ the cost of collective action. Olson's (1965) seminal work on the costs of collective action noted that members of an organization who share a common interest will not necessarily further their common interest. There is always an incentive to free ride on the efforts of other members when collective action is required to secure a common good. The fact that a goal is common means that `no one member of the group is excluded from the benefit brought about by its achievement'. Members might face therefore an incentive to not incur their share of the costs unless there are private benefits to be accrued. In the case of the MDBS, one way to overcome the cost of collective action, and avoid the resulting problem of free-riding, was to ensure that development agencies would commit to remain as a member of the group through several rounds of budget support, and agree in principle to raising over time their contribution to the MDBS. Repeated rounds would allow the costs of coordination to be shared amongst the participating agencies by ensuring that the costs of common activities were shared amongst all, and by rotating the development agencies in the leadership of the MDBS and the leadership of sector-specific groups. Rising contributions to the MDBS would ensure that MDBS membership entailed a financial commitment and that the overarching objective of scaling-up development assistance was indeed met. -9- III. The MDBS experience with devising mechanisms to mitigate the costs of coordination, collective action and measurement This section draws on the literature summarized in the previous section to review the MDBS experience from its inception in 2003 to today. It begins by considering the costs of coordination both inside the Government of Ghana and amongst the development agencies that participated in the MDBS, examining how the decision to delegate responsibility to sector-specific groups contributed to increase the predictability and ownership of the policy dialogue and, in doing so, played a part in scaling-up the development assistance channeled through the MDBS. It turns next to the costs of collective action and the risk of free-riding, suggesting that the rotation of development agencies in the leadership of the group, as well as in the leadership of sector-specific groups, helped mitigate the risk of free-riding, while at the same time raising the possibility of collusion among agencies. It then examines the problems that emerge in measuring progress in the implementation of the program supported by the MDBS, and how a definite solution to mitigate the so-called measurement costs has not yet been found. This section closes with a summary of the main conclusions: identifying how agencies clustered around key contentious issues, and suggesting a new equilibrium that the MDBS appears to be converging toward. Coordination costs The transaction costs literature on coordination failures highlights the potential pitfalls that arise in a setting involving one recipient country and several development agencies. While the example captures the essence of bargaining when there is one counterpart in Government and more than one development agency, the MDBS experience meant increasing manifold the potential for coordination failures arising from disagreement and haggling among the parties involved. An agreement between the Government of Ghana and the MDBS agencies involved several sector ministries within the Government of Ghana, as well as several development agencies. The former problem of coordination within Government was mitigated because the nature of the budget support arrangement placed the Ministry of Finance at the center of this contractual arrangement between the Government of Ghana and the MDBS agencies. The Ministry of Finance had final say on the Government's side on the support provided through the MDBS, since it was the best placed to understand what could be achieved with the implementation of the budget and could play an arbitrating role between the sector ministries. The Ministry of Finance's leading role should not conceal, however, the efforts required to coordinate the actions within the Government of Ghana, which itself justified the establishment and continued existence of the MDBS as a coordinating framework. While the Ministry of Finance had primary authority over the budget, it was not uncommon for sector ministries to agree separately with the development agencies on the general tenor of policy actions supported either by the MDBS or by separate projects funded by individual MDBS agencies. To mitigate the tension between the sector ministries, the Ministry of Finance, and the development agencies, the MDBS -10- organized itself into separate sector-specific or thematic groups, with a primary responsibility for carrying out the dialogue at the sector level (i.e., identifying the policy actions to be supported by the MDBS and monitoring their implementation). In parallel, the Ministry of Finance established within the ministry an MDBS secretariat with pillar leads to accompany these sector discussions from the Government's perspective and inform the Ministry of Finance about developments at the sector level. It was the responsibility of the core group within the MDBS, upon advice of the sector-specific groups, to elevate to the Ministry of Finance the policy issues that were deemed relevant to be included in the policy matrix. The MDBS secretariat at the Ministry of Finance and the MDBS core group aimed therefore at facilitating coordination and at improving the flow of information to those outside the sectors or thematic areas by providing a needed interface between the sector strategies and the implementation of the budget and, as a result, the budget support program. The organization of the MDBS with a core group of members and several sector- specific groups was an evolution from its original set up, which envisioned only the core group of members taking responsibility for the entire policy dialogue, including the negotiations with the Ministry of Finance on the contents of the policy matrix. The more complex governance structure for the MDBS, with sector-specific groups and the core group of members, on the one hand, and the Ministry of Finance and its team of pillar leads that liaised with the sector ministries on the other hand, emerged to make coordination across the ministries within Government and the several development agencies possible. In doing so, it allowed the MDBS agenda to be informed by a broader dialogue, preempting last minute decisions about issues to be included in the policy matrix and mitigating the risks of a sub-optimal outcome resulting from last minute agreements around a minimal common denominator (i.e., an `unambitious development program').7 This broader dialogue also permitted the more active participation of all the sector ministries and the MDBS agencies, as well as agencies outside the MDBS that participate as observers, broadening the ownership of the program. The evolution of this more complex governance structure between 2003 and 2007 was not without tension, however, as it happened against the backdrop of a debate amongst the MDBS participating agencies on whether the MDBS should focus exclusively on so-called cross-cutting issues (such as governance, public sector reform, decentralization, and public financial management), or should take a broader perspective across several sectors or thematic areas. Without going into the merits of this debate, the main reason why the MDBS evolved toward sector-specific groups appears to have been that these sector-specific groups facilitated the dialogue by spreading it across several agencies and, in doing so, encouraged greater ownership of the agreed policy actions within Government and among development agencies. In addition, 7 For instance, in 2005 the MDBS carried out a 6 week-long mission, with meeting in the mornings amongst the participating agencies and in the afternoons with a team from the Ministry of Finance, trying to reach an agreement over a policy matrix for the following year. The dialogue was limited to these two teams and time limitations (and sheer exhaustion) led to what was perceived as a minimal common agenda for both sides. The Government was understandably cautious about committing more than it could deliver in terms of action to be reviewed in 12 months. -11- there are two other reasons why the broader perspective appears to have prevailed. The structure organized along sector groups allowed the Ministry of Finance the opportunity to have a budget-wide perspective of Government activities. This wider perspective contributed toward a better coordination of activities within the Government, thus, reducing the first set of coordination costs identified above. It also permitted development agencies the opportunity to reduce the cost of delivering development assistance by offering the opportunity to increase their contribution to the budget support by either providing assistance to targeted sectors in the context a policy dialogue carried out through the MDBS, or by fitting individual projects into broader programs agreed under the MDBS umbrella.8 The potential pitfalls from coordination failures and the option of mitigating them by giving greater responsibility to sector-specific groups can be presented in a simple two-by-two matrix format, where the premium is placed on a prior sector dialogue. Consider the case where the Government has the first mover's advantage by defining whether it engages, or not, in the sector dialogue, otherwise the Government and the development agencies need to settle for what is agreed only between the Ministry of Finance and the MDBS core group at the final stages of the negotiations (i.e., no sector dialogue). Engaging in the sector dialogue has a cost primarily for smaller development agencies, since it involves staff time and dispersion of focus. It is understandable therefore that most agencies either agreed to go along with decisions taken by other development agencies in their dialogue at the sector level (a decision that resulted either from a twinning arrangement between development agencies, or from the decision to participate in the group as a silent partner), or propose that the number of sectors to be considered for MDBS support be limited (or a combination of both). A schematic description of the possible outcomes of negotiations where there is the option of engaging or not in a sector dialogue is presented in Table 1 below. One should note that that preferred outcome appears to be when the Government of Ghana and the development agencies jointly agree to engage in a sector dialogue, since it ensures that what is finally agreed between the Ministry of Finance and the MDBS core group resembles, at least partially, what was discussed at the sector level. However, it is important to note that neither the Government of Ghana, nor the development agencies might chose to engage in the sector dialogue. The Government of Ghana might chose not to engage in the sector dialogue because the Ministry of Finance does not agree with what is emerging from this sector dialogue, believing that what is being proposed is incompatible with the overall budget strategy. The development agencies might also choose not to engage in the sector dialogue because their participation is costly in terms of staff time and resources, and makes monitoring from outsiders (primarily headquarter offices) difficult by broadening the range of sectors included under the MDBS. 8 This approach prepared the ground for the evolution of the result-based Ghana Partnership Strategy. Based also on the Ghana Growth and Poverty Reduction Strategy, Consultative Group meetings over the period 2005-07 progressively developed an Overall Ghana Partnership Result Matrix that later became know as Results, Resources and Partnership. -12- Table 1. MDBS Negotiations With and Without Sector Dialogue Government Development Agencies of Ghana/ Ministry of Engage in Sector Dialogue Don't Engage in Sector Dialogue Finance Engage in Yes, Yes ­ leading to Yes, No ­ what is proposed at the sector dialogue is Sector proposals that reflect an agreeable to the Government of Ghana but not agreeable Dialogue agreement either close to or to the development agencies, who chose instead to either identical to what was agreed reach an agreement separately with the Ministry of at the sector level. Finance, or to exclude the sector altogether from the MDBS policy matrix. Don't No, Yes ­ meaning that No, No ­ meaning that both the Ministry of Finance and Engage in what is agreed with the the development agencies prefer an agreement among Sector Ministry of Finance is final themselves rather than involving sector ministries and Dialogue and there is no recourse to sector-specific groups of development agencies. Among the dialogue at the sector the reasons that might lead to this outcome are attempts level. This might be by sector ministries to attract resources from the budget preferred by the Ministry of support with policy actions that are not envisioned in the Finance if what is proposed budget, and an assessment by the development agencies by their sector ministries is that the sector dialogue is too costly in terms of time and not compatible with what is resources. planned in the budget. Notwithstanding these potential disagreement with what might emerge from the sector dialogue, it was understandable that the MDBS evolved toward the northwest corner of the options outlined in Table 1, strengthening the group's focus on the sector-specific dialogue and, in doing so, encouraging the Government to do the same. This move toward the northwest corner of Table 1 was important because it settled, at least temporarily, the debate about whether the MDBS should support a narrow or a broad policy matrix. The decision to keep the policy matrix broad encouraged as a result twinning arrangements between agencies, such as the twinning arrangement between the Netherlands Embassy and the UK Department for International Development (DFID), whereby the Netherlands Embassy represented both agencies in the health sector-specific group, while DFID represented the two agencies in the education sector- specific group.9 Also, as the policy dialogue was increasingly delegated to the sector- specific groups, agencies began finding that the leadership of specific sector groups was at least as important as a seat at the table of the MDBS core group. Another important dimension of the MDBS organization along sector-specific groups was that they opened the group to more active participation by agencies outside the core group of MDBS agencies.10 The reasons for greater participation by 9It is important to note that there were other factors that contributed to allowing these twinning agreements among development agencies to work in health and education. DFID considered the health sector dialogue in Ghana to be relatively robust to delegate the dialogue to the Netherlands Embassy, given that the sector group had been established in 1997. In the meantime, the Ministry of Foreign Affairs of the Netherlands was willing to consider the option of relying on DFID's education team in Ghana, since it had earmarked a substantial amount to be spent on supporting education programs worldwide, 10These agencies that participated as observers in the MDBS meetings included several of the United Nation agencies, including the FAO, the UNAIDS, the UNICEF, the UNDP, and the WFP, as well as the USAID, JICA, and several NGOs, such as the Carter Center. -13- outside agencies in the MDBS sector meetings were two-fold. These other agencies brought sector-specific interest and knowledge, which contributed to the sector dialogue. Also, the status of observers provided some of these agencies the opportunity to check whether they would like to join the MDBS as full-fledged members. The French Development Agency (AFD), for instance, sat in the MDBS meetings as observers for one year before formalizing its decision to join. USAID was an active member in several sector-specific groups, but did not join as a full-fledged member because of a mandate by the US Congress that did not allow the pooling of its funds with other agencies. The next question that arises is whether the MDBS' more open, and inevitably more complex, governance structure led to the problem of free-riding identified in the previous section. In the case of the MDBS, free-riding would happen whenever a participating agency would continue to enjoy the benefits of membership while not making an effort to increase development assistance flows. We examine below how the structure of the MDBS evolved to avoid this problem. The Costs of Collective Action and the Free-Rider Problem The literature on the costs of collective action identifies the risk that members of a group that share common interests might not necessarily further the common interest, preferring instead to further their individual interest, free riding on the efforts of other members. According to Olson (1965) the logical solution to this problem is to impose participation rules that ensure that members are not able to free ride on the efforts of other members. As we shall see in reviewing the MDBS experience from 2003 to 2007 there were efforts to establish participation rules that created incentives for members to stay in the group and to increase their contribution to the MDBS, as well as to increase their influence within the group and to reduce participation costs by partnering with other members. While the de-concentration of the MDBS activities to sector-specific groups was designed to help overcome some of the coordination problems inherent in the original MDBS structure, it also entailed a greater financial commitment by the MDBS members by requesting participating agencies to share the costs of running the group, and by encouraging agencies to scale up their contribution to the MDBS as a way to ensure greater predictability and ownership in the policy dialogue. Figure 2 below, with information on financial contributions by agencies during the 2003 to 2006 period, plus projections for 2007, appears to confirm the notion that a more open structure for the operation of the MDBS could induce participating agencies to scale up their contribution. All the bilateral development agencies11 increased their contributions, with some of them making the decision to join the MDBS during this period (e.g., France). The increase in the contribution by the bilateral agencies appears to have reflected primarily greater say in the decisions of the group, as well as increased leverage over the direction of sector-specific policy dialogue as a result of greater say in the policy agenda of these groups, as well as twinning arrangements amongst participating agencies. 11Canada, Denmark, France, Germany, the Netherlands, Switzerland, and the United Kingdom. -14- For example, the twinning arrangement between the Netherlands Embassy and DFID in the education and health sectors provided both agencies a greater influence over the directions of the policy dialogue in these two important areas. Figure 2. MDBS Contributions by Agency, 2003-2007 ­ US$ Million France Germany Denmark 160.0 Switzerland 140.0 Canada Netherlands 120.0 European Commission 100.0 AfDB United Kingdom 80.0 World Bank 60.0 World Bank United Kingdom 40.0 AfDB European Commission Netherlands 20.0 Canada Switzerland Denmark 0.0 Germany 2003 2004 France 2005 2006 2007 Source: MDBS Secretariat at the Ghanaian Ministry of Finance and Economic Planning. The evolution of the contributions by the multilateral development agencies12 was less straightforward, however. At first sight, it appears that the decision by multilateral agencies to scale up or not development assistance did not necessarily reflect a response to the new, more open and, necessarily, more complex structure of the MDBS. For instance, the contributions from the World Bank, which was the largest contributor to the MDBS during this period, had to be scaled down in 2007 following the decision to extend multilateral debt relief to Ghana, since, at least initially, the Bank's contribution to the Multilateral Debt Relief Initiative (MDRI) was partially funded by reducing the allocation of its concessional credits to the recipient countries. The contributions from the African Development Bank (AfDB), on the other hand, rose consistently during the period 2003-2006, only to drop to zero in 2007 because of the transition from one program cycle to another. The contributions from the European Commission were the only ones that continuously declined during this period, reflecting changes in overall funding schemes, as well as some concerns about how the agreements reached between the Government of Ghana and the MDBS were being enforced. The issue that apparently held back the European Commission (EC) from channelling additional development assistance through the MDBS was, in part, a disagreement about how to best measure progress in the implementation of the 12African Development Bank, European Commission and the World Bank. -15- Government's program. These disagreements were not unique to the European Commission. Since the beginning of the operations of the group assessing whether and when to disburse the funding committed through the MDBS was the main source of disagreements within the group and a source of tension in the dialogue with Government. These disagreements were important because for the Government any difference between the amounts committed and actual disbursements meant a shortfall in the financing of its budget. Among the development agencies, the decision about whether to recommend to headquarters the full disbursement of the amounts committed meant weighing the costs of reduced predictability in the funding of the budget against being responsive to requests that the dialogue ensure progress on specific policy actions in a manner that could be ascertained by headquarter offices. These disagreements around how best to measure progress in the implementation of the Government's program supported by the MDBS did not appear to have held back other agencies from increasing their contribution to the MDBS, however. Indeed, what appears to have happened was that agencies either increased their contribution or joined the group, allowing some of the multilateral agencies to reduce their contribution. For instance, projections for 2007 indicate that the joint contribution from the European Commission and Germany (which was a member of the MDBS in 2003 but began contributing only in 2004), as well as France (which began contributing in 2005), will exceed the EC's original contribution in 2003 by US$12 million, suggesting that the MDBS' more open structure allowed the entry of new agencies to offset temporary declines in the contributions from other agencies. Time should allow therefore the contributions of the participating agencies to be more evenly distributed over time. At present, since not enough time has yet elapsed to significantly change the original concentration of contributors, however, DFID and the World Bank still account for around 60 percent of the overall contributions to the MDBS. Also, as a way to ensure predictability of funding from the MDBS, and better enforce the participation rule, the MDBS introduced a new schedule of operations in 2006 that aimed at better aligning the funding of the budget with the decision processes within the MDBS. Under the new schedule, the development agencies would disburse their budget support on the basis of triggers defined two years in advance and assessed on the year prior to disbursing, assuming that the assessment was positive. This new schedule allowed the development agencies to inform the Government of Ghana about the amount of funds that would be provided through their budget support before the Government's budget proposal was submitted to Parliament. To implement this new schedule, however, the MDBS had to go through a transition period. In 2007 the MDBS disbursements by the bilateral and the EU were underpinned by the same set of actions that had been agreed in 2005 and assessed in April 2006. The World Bank, however, could not make two disbursements of its budget support on the basis of one assessment, so it executed a different transition to this new schedule. The Bank's transition required that for the 2007 budget cycle the Bank agree with the Government of Ghana on a separate set of triggers that entailed an intermediate step toward the triggers that would be jointly agreed between the Government of Ghana and all the MDBS development agencies in the 2008 policy matrix. -16- This effort to increase the predictability of MDBS flows is consistent with other initiatives either currently being considered, or already under implementation, to increase the reliability of development assistance. The European Commission, for instance, has launched a proposal for six-year aid contracts with the African, Caribbean, and Pacific (ACP) countries. These agreements are called `the MDG Contract' and aim at making the EC's budget support more long-term and more predictable to countries with a strong performance track record and a multi-year monitoring framework.13 Also, the U.S.'s Millennium Challenge Account employs 4 to 5 year aid contracts with pre- agreed disbursement schedules.14 And several European bilateral donors are adopting longer multi-year planning frameworks for aid budgets. While the new schedule reinforced the coordinating arrangements put in place by the MDBS, and provided a strong basis on which to increase predictable development assistance, it could not eliminate the measurement problems that were one of the limiting factors for a larger scaling-up effort. We turn next therefore to a discussion of these measurement problems, which are important to better appreciate the vertical balance between those MDBS agencies operating in the field and their headquarter offices. Measurement problems The transaction costs literature identifies two instances where measurement problems might lead to difficulties in reaching and enforcing an agreement: when information is asymmetrically distributed among parties, and when it is costly to apprise an arbiter among equally informed parties. For development assistance the former is more frequent and, hence, more important, than the latter. Focusing on only one case where measurement costs arise does not narrow the circumstances in which these problems occur in accessing and delivering development assistance, however. There were four ways in which the MDBS tried to overcome measurement problems that arose from the asymmetry in information between the DPs and the Government of Ghana, with these changes aiming at making agreements easier to reach and enforce. First, the MDBS sought to clearly define so-called means of verification for triggers in the policy matrix, which would spell out how the Government and development agencies would measure whether a trigger had been met or not. Defining means of verification for individual triggers ahead of time was meant to remove some of the ambiguity that could emerge at the time of making the assessment. Second, some development agencies tried to go a step further, agreeing with Government on outcome indicators that would provide a clearer benchmark, observable to everyone, to whether the trigger was achieved or not. According to this proposal, the outcomes would be assessed either as achieved, leading to full disbursement (indicated by a grade 1), or 13European Commission. 2007. Technical Discussion Paper on a "MDG Contract" ­ A Proposal for Longer-Term and More Predictable General Budget Support. June 19, 2007. 14http://www.mcc.gov/programs/compacts.php -17- not fulfilled, leading to no disbursement (indicated by a grade 0), or partially fulfilled because of observable events outside the control of the Government (e.g., drought, external conflict), leading to a partial fulfilment (indicated by a grade 0.5). Third, for those agencies that chose not to select outcome-based triggers, there was an effort to define baseline results for the expected outcomes of a given policy action. By quantifying baselines for the expected outcomes for a given policy action, progress could be measured against these baselines, opening scope for some form of outside scrutiny. Lastly, as an alternative to outcome indicators, the Government and agencies attempted to identify policy actions that were more closely linked to specific outcomes. For instance, the introduction of capitation grants for students attending public primary school were expected to lead to higher primary enrolment rates without setting ex-ante a target for primary school enrolment. The option of linking policy actions to outcomes allowed `policy experimentation' without tying disbursements to any given outcome. While not committing the Government to specific outcomes, MDBS support to specific policy actions that were anticipated to lead to desired outcomes provided an alternative to outcome indicators. The main difference between specific outcomes and policy actions that would lead to outcomes was that the latter used outcomes to inform the identification of policy actions and, in doing so, allowed the risk to be spread more evenly between the Government and development agencies. As discussed above, the focus on policy actions opened more scope for dialogue and experimentation without jeopardizing the funding to the budget that was being provided by development agencies. Several options were also considered for dealing with policy actions that were not completed exactly as anticipated within an agreed timeline. One option adopted was defining prior actions that occasionally differed from the original triggers. This option was a response to the fact that it was unlikely that the exact wording of the agreed policy action could be precisely anticipated 10 to 12 months in advance. One example of prior actions that modified an original trigger was a trigger referring to the introduction of a computerized financial and accounting system for public accounts ­ the Budget and Public Expenditure Management System (BPEMS) as it was called in Ghana. The agreed trigger was that, by the time of the assessment, the Government would have BPEMS operational in the headquarter offices of the Ministry of Finance and four other key line ministries (Education, Health, Road Transport, Local Government and Rural Development). By the time of the assessment, however, the Government had only completed the roll out of the new system to two of the five ministries (Finance and Health). The question then became whether there had been sufficient progress to justify continuing to support the program, or use this example to penalize the Government for failing to implement the trigger within the agreed timeline. Slightly over six months later, the BPEMS system had been rolled out to the remaining pilot ministries, plus a few others. By October 2006 the Government's new computerized financial and accounting system (BPEMS) had been installed and was operational in 8 ministries.15 The decision 15 In addition to the ministries of Finance and Health, BPEMS was operational at: (i) the Ministry of Education, Science and Sports, and the Ghana Education Services (GES); (ii) the Ministry of Transportation; (iii) the Ministry of Information; (vi) the Ministry of Local Governments, Rural -18- as to whether to consider that the program supported by the MDBS was heading in the right direction, or whether to send a signal about how the MDBS measured progress under the program, ended up dividing the group, nevertheless. To mitigate the problem of asymmetric information, and the risks of reaching agreements that reflected a minimal common denominator, the MDBS resorted to having an outside observer that could check and ensure that commitments were both credible and involve actual effort. Most development agencies relied on their headquarters to perform this function of objective, outside observers, since it created the right incentives for the field offices of the development agencies and allowed an arms- length distance between what was agreed by the field office of development agencies and the Government of Ghana. Some development agencies also attempted to rely on what the Government had committed to implement in its annual budget statement. The annual budget statement was seen as expressing the Government commitments in allocating resources, and in doing so demonstrating ownership (as well as ambition, since the budget statement is also a political document). Relying on the budget statement was also seen, therefore, as an option in overcoming the minimal common denominator problem identified above.16 Development and Environment; (v) Ministry of Communications; and (vi) Ministry of Water Resources, Works and Housing. 16The other side to this argument about having the annual budget statement to Parliament serve as a tacit committed by Government is that it might include policy actions that are neither necessarily agreeable to the headquarter offices of individual development agencies, nor a commitment against which the Government would be willing to attach budget support funding from development agencies. A case in point was the announcement in the Government's budget statement that it would set aside funding for the demarcation of land under dispute. The Government was not interested in having this action be considered as an eligible trigger for disbursement under the MDBS because it thought that it would not be advisable to include this commitment to demarcate land under dispute in the policy matrix, as it could be seen as an outside interference in what was seen as a sensitive and strictly internal matter. So, even when an action was included in the annual budget statement to Parliament, there was no guarantee that it could become an eligible trigger to guide MDBS disbursements. -19- Summing Up: Coordination, Collective Action and Measurement Costs ­ The Vertical and the Horizontal Balance within the MDBS To better understand the dynamics within the MDBS it is important to understand the preferences of each cluster of agencies, and knowing where they stood in reaching the horizontal and the vertical balance. While the overarching principle behind the MDBS is reducing transaction costs in accessing and delivering development assistance, each agency offered different headquarter perspectives on how to achieve this objective, as well as their own perspective of the future. Table 2 provides a summary presentation of where each cluster of agencies stood relative to the most contentious issues during the 2003 to 2007 period. The first cluster of development agencies placed a stronger emphasis on a policy matrix that focused on so-called core and cross cutting issues, such as governance, public sector reforms and public financial management. Their understanding was that a more focused agenda was in the interest of their respective institutions because a narrower agenda carried lower monitoring costs and these issues reflected the preferences of the constituencies they represented. These preferences explained in turn where these agencies stood vis-à-vis the so-called vertical balance, giving headquarter offices a greater say in the overall design of the policy matrix. The second cluster of development agencies included those that wanted to combine general budget support with an existing (and sometimes expanding) presence in certain sectors. Their preferences were usually associated with a larger presence on the ground, leading to more responsibility tilted toward the field offices. These agencies had an interest in both outcome and process-based indicators in the policy matrix. The former reflects the emphasis on measurable achievements that could be monitored by outside observers and, as a consequence, lead to increased participation of all development agencies in the MDBS. The latter reflects their interest in having sector programs in place that would allow sector-focused projects to be developed. The third cluster includes agencies that had a broad sectoral perspective that placed a premium on engaging the Ministry of Finance in the discussion of these sector specific issues. It is important to note that the division of agencies' preferences across the issues identifies in Table 2 is designed to be illustrative of the initial period of the MDBS operations, serving primarily the purpose of better understanding how the development agencies organized themselves across the issues facing the group. -20- Table 2. Agencies' Preferences in the Design of the Policy Matrix Agencies Scope of the Outcome Focus Vertical Horizontal Policy Matrix Balance Balance Cluster 1 More focused on Strong outcome The Harmonization core and cross focus, ensuring that responsibility is was seen as costly cutting issues, such policy actions agreed tilted toward in terms of as governance, with the Government headquarters, as headquarters public sector of Ghana achieved reflected in the monitoring, reform, mutually agreed greater emphasis leading to the decentralization outcomes, and on outcome proposal for a so- and public therefore could be indicators that called `a la carte' financial more monitored by could be approach in management, outside parties. monitored by selecting triggers which are seen as outside parties. in the policy central to the matrix. This `a la effectiveness of carte' approach budget support was designed to operations. allow the outcome- based approach to co-exist with other visions for the policy matrix by giving each agency the freedom to select those policy actions that better suited their preference. Cluster 2 A blend of both Some outcome focus The Strong emphasis core and cross blended in with some responsibility is on harmonization cutting issues, as process based tilted toward and a negative defined above, actions, such as field offices, reaction toward the with some sector broad policy with proposed `a la issues (such as frameworks agreed at harmonization carte' approach. natural resource the Cabinet level or among agencies Strong interest in management, between the Ministry seen as preserving the field agriculture, and the of Finance and important in office primacy water & sanitation specific sector achieving the over the process of sector), which ministries. These overarching goal negotiating the accommodated the frameworks were to of reducing policy matrix. areas of specific lay the basis for transaction cost interest of sector specific in managing individual projects (e.g., Policy development development Memorandum assistance. agencies. between the Ministry of Finance and the Forestry Commission; the National Water Policy, the National Decentralization Policy). -21- Table 2. Agencies' Preferences in the Design of the Policy Matrix (cont.) Cluster 3 Broad-based Willing to accept The responsibility Strong support to policy matrix, some outcome is tilted toward harmonization, encompassing both indicators in the field offices, with either by acting as core and sector policy matrix, harmonization a `silent partner' specific issues, provided they were among agencies throughout the aimed at reflecting not used to seen as an process, or by a broad underpin important engaging actively development disbursement achievement in the in the MDBS. agenda and the decisions. process of requirements for reducing the its effective transactions costs implementation. of managing development assistance. While by mid-2007 the MDBS policy matrix still remained broad-based, focusing on a wide range of sectors, there were some important developments that were emblematic in indicating the new directions of these vertical and horizontal balances. The 2007-09 policy matrix agreed with the Government in May 2007, for instance, included 35 indicators or policy actions, many with more than one entry, covering a total of 15 thematic areas. In the meantime, the MDBS was gradually moving away from being a concerted effort between Government and the development agencies and closer to a so-called `division of labor' that allowed greater agency responsibility over the policy dialogue in a given sector or thematic area. This `division of labor' allowed field offices to ensure headquarter offices that individual contributions to a common budget support did indeed translate into results that could be claimed as their own. Also, this new `division of labor' helped some agencies reach a vertical balance, whereby greater field office autonomy came as a result of a tacit understanding that field offices would have a stronger influence over the policy dialogue in a given sector or thematic area. Concluding Remarks The main question this paper set out to answer is whether the MDBS structure evolved to minimize the transaction costs usually found in assessing and delivering development assistance and, in doing so, fulfil the overarching objective it aimed at accomplishing in the first place. The paper begins by providing a brief overview of the literature on transaction-costs economics, identifying those transaction costs that appear to be most relevant for development assistance in multi-donor settings. The paper reviews next the evolution of the MDBS structure during the first four budget cycles, assessing whether changes that were made to the structure of the MDBS reflected the need to reduce the costs of coordination, collective action and measurement. While it is not possible to identify a counterfactual to the Ghanaian experience that could show that without the changes made in the MDBS structure during the 2003-07 period it would have become less effective, the findings of this paper suggest that the MDBS became increasingly more important, delivering higher volumes of development assistance each year, and playing a more central role in the policy dialogue. These developments suggest that the Government of Ghana and its MDBS partners were able to recognize the -22- transaction costs involved in accessing and delivering development assistance, and devise changes to minimize these transaction costs. Three main findings emerge from the paper: · The structure of the MDBS appears to have evolved to mitigate coordination costs by giving greater responsibility to sector-specific groups that were in charge of agreeing, monitoring and, to an increasing extent, assessing the performance against agreed actions in the MDBS policy matrix. This was an evolution from the original set up that envisioned most of the dialogue being conducted between the Ministry of Finance and a core group of representatives from the development agencies participating in the MDBS. The arrangement that gave greater prominence to the sector-specific groups in defining the agenda for the dialogue at the sector level was also important in building greater ownership within the Government of Ghana and amongst development agencies. · The operations of the MDBS also appear to have evolved to help mitigate some of the costs of collective action. Leadership positions in the MDBS rotated regularly among the development agencies, and the greater responsibility given to sector groups encouraged ownership and elicited higher contributions from participating agencies by broadening participation and giving a greater say to individual agencies in shaping the dialogue at the sector level. This increased sense of ownership amongst MDBS agencies resulted from a `division of labor' within the group that allowed field offices of the development agencies to take greater responsibility for a share of the policy dialogue and, in doing so, claim some of these results as their own vis-à-vis their headquarter offices. · While the MDBS made several attempts to mitigate the problems arising from the need to measure progress in the implementation of the program it supported, a settled view has not yet emerged. There continues to be tension between Government and partners, and amongst development agencies, over how to reach a balance between setting specific indicators that can be measured at the time of review missions and maintaining an open dialogue with Government around policy actions that lead to anticipated outcomes. The latter is important because they allow for more policy dialogue between Government and development agencies, as well as some degree of policy experimentation on the part of Government. While the MDBS attempted several options, such as defining ahead of time the means of verification for the actions identified in the policy matrix, establishing baselines against which to measure progress, and relying more on Government commitments in the annual budget statement to Parliament, neither of these option appear to have led to a permanent solution to the measurement problem. Given the time and effort involved in settling this issue, measurement remains as one of the most enduring transaction costs in accessing and delivering development assistance. -23- References Alston, Lee and William Gillespie, 1989, "Resource Coordination and Transaction Costs: A Framework for Analyzing the Firm/Market Boundary," Journal of Economic Behavior and Organization, April: 1-22. Coase, Ronald, 1937. "The Nature of the Firm," Economica, ns, 4; 386-405. Eggertsoon, Thrainn, 1990, Economic Behavior and Institutions. Cambridge, Mass.: Harvard University Press. European Commission. 2007. Technical Discussion Paper on a "MDG Contract" ­ A Proposal for Longer-Term and More Predictable General Budget Support. June 19, 2007 Holmstrom, Bengt and Paul Milgrom, P., 1991, 'Multitask Principal-Agent Analyses: Incentive Contracts, Asset Ownership, and Job Design, Journal of Law, Economics and Organization, Spring, 24­52. IMF, 2007, Staff Report for the 2007 Article IV Consultation, Washington: International Monetary Fund, April. IMF, 2006, Ghana: Staff Report on the Sixth and Final Review Under the Three-Year Arrangement Under the Poverty Reduction and Growth Facility, Washington: International Monetary Fund, October. Ménard, Claude (ed.), 2005, The International Library of the New Institutional Economics, edited by London, Edward Elgar Publishing. Milgrom, Paul and John Roberts, 1990, `Bargaining Costs, Influence Costs and the Organization of Economic Activity', Perspectives on Positive Political Economy, edited by James E. Alt and Kenneth A. Shepsle, Cambridge: Cambridge University Press, 57-89 Nash, John, 1950, The bargaining problem, Econometrica, 18: 155-162. Olson, Mancur, 1965, The Logic of Collective Action. Cambridge, Mass.: Harvard University Press. Overseas Development Institute (2007). Budget Support to Ghana: A risk worth taking? Briefing Paper number 24, July. Williamson, Oliver, 1975, Markets and Hierarchies: Analysis and Antitrust Implications, New York: Free Press. Williamson, Oliver, 1985, The Economic Institutions of Capitalism, New York: Free Press. -24- Williamson, Oliver and Scott Masten (editors), 1994, Transaction Cost Economics, London: Edward Elgar Publishing Co. World Bank, 2004, Second Poverty Reduction Support Credit, Washington: International Development Association, June. World Bank, 2005, Third Poverty Reduction Support Credit, Washington: International Development Association, July. World Bank, 2006, Fourth Poverty Reduction Support Credit, Washington: International Development Association, May. World Bank, 2007, Fifth Poverty Reduction Support Credit, Washington: International Development Association, May. wb22579 C:\Documents and Settings\wb22579\My Documents\Ghana\MDBS-PRSC-Lessons Learned-November 2007.doc 11/20/2007 11:57:00 AM -25-