Delivering Project Aid in Old and New Ways: Institutions Matter1 Geoffrey Shepherd May 27, 2002 An "output-based approach" to aid (OBA) is a term to describe the practice whereby donors or governments delegate service delivery to private third parties (firms or non- profits) under contracts that tie donor or government payments to the particular outputs. The use of such structured contracts is not new. Adjustment lending ­ or program lending ­ has generally been under contracts that tie disbursements to particular policy changes. More recently, reforms in the infrastructure sectors have used a contractual framework to involve private participants and there have been various experiments in contract-based provision of services.2 The growing use of such contracts reflects the growing interest of governments and donors in results-based management (RBM). Implicitly at least, OBA seeks a better way of doing projects than in the past.3 To simplify, the traditional project vehicle of most donors ­ it can be called an input-based approach to aid (IBA) ­ consisted in an agreement on what services the recipient government would provide, with payment up front of the inputs necessary to produce the services, hence with relatively weak provisions (within the contract itself at least) to enforce that agreement. It would be simplistic to say that the input-base approach has "failed", and is certainly difficult to demonstrate this with numbers. But there has long been considerable frustration-to-disenchantment with IBA in the aid community. As a result, there are attempts to improve it through better procedures or to replace it with other aid vehicles. OBA is thus a new vehicle that shifts the focus of the project agreement from inputs to outputs (Brook and Smith, 2001, Part 1). This contractual approach in turn reflects the increasing emphasis on the role of institutions in economics, public management, development, and even (though at a slower pace) the aid business. Institutional analysis emphasizes the political and personal rules and incentives that shape economic interactions, whether through contracts or organizations, whether public or private. 1 The original version of this paper was presented at a Workshop on Output-Based Aid, organized by the World Bank and KfW in Frankfurt am Main on January 24-25, 2002. The revised paper has benefited from the comments of the Workshop participants. The findings, interpretations and conclusions of this paper are my own and should not be attributed to the World Bank, its Executive Board of Directors, or any of its member countries. 2Brook and Smith (2001) provide some examples of nationally- and World-Bank-financed schemes. In addition, the Consultative Group to Assist the Poorest (CGAP) provides performance-based aid to micro- lenders (World Bank, 2002, page 67). 3This paper generally uses the term "project" to designate investment (including technical-assistance) activities and to exclude structural-adjustment (program) lending (or credits). 1 This paper is intended to provide a background for the discussion of donor-financed OBA.4 It will do this by sketching the broader context of aid reform and public-sector management that OBA fits into and by doing this, in part, using an explicit contracting framework. · The first part of the paper will look at the changing fortunes of IBA in the recent period. The IBA project vehicle has both evolved and been partially substituted by other aid vehicles in response to changing politics, changing perceptions of the development challenge and process, and changing techniques of public management. Overall, the aid community remains dissatisfied with the results of the traditional approach. · The second part of the paper provides a relevant contracting framework by postulating what makes contracts work, how the attributes of governments constrain them in contracting, but how new models (including OBA) have emerged to get around these constraints. The available evidence suggests that these models do work for governments contracting out to private entities, but with greater facility in more- than less advanced countries, better in some services than others, and better when the contracts are designed to provide adequate information, incentives, and commitment. · In the third part, I look at the contracting attributes of donor agencies and recipient governments and the available evidence about how contracts have worked in aid projects. Information on both attributes and outcomes is as yet underdeveloped. The descriptive literature on aid and a small literature on the contractual nature of structural-adjustment lending indicate some similarities in government-private contracting and donor-recipient-government contracting: both face problems of lack of credible-commitment mechanisms, the difficulty of measuring (hence monitoring) outputs, and the generally poor quality of developing-country public administrations. However, in aid projects and structural-adjustment lending the convergence of both donor and recipient interests in maintaining financial flows and their divergence in substantive development objectives appears to lead to a frequent outcome in which recipient government fail to implement adequately the loan as contracted. · Finally, the paper sums up the implications of the public-sector and aid contracting experience for donor-recipient OBA. OBA clearly promises to improve the aid contract. But of itself, OBA does not strengthen the donor's commitment to project implementation. Nor can donors afford to ignore the role of the government in making OBA work.5 4OBA can provide a contracting model both for donor projects and for government-private-entity contracting out. The issues overlap for both cases, but also exhibit differences because of differences in the complexity of the contracts and the attributes of the contracting parties. (See Keefer, 1998, for an analysis, within a contracting framework, of the pros and cons of contracting out by governments in transition economies, but also applicable to developing economies. Similarly, Preker, Harding, and Girishankar, 1999, provide a "map", based on institutional analysis, of where and under what circumstances private participation in health can work.) 5In this paper, there is a bias towards the World Bank in the choice of evidence: this is a reflection both of where I have been able to find the more interesting analytical literature and of my own direct experience. 2 I. The Changing Fortunes of Input-Based Approaches For the World Bank, the regional development banks, and many bilateral donors, the IBA vehicle has been the workhorse of aid delivery, but it has come under a lot of pressure. Trends and problems in World Bank aid 6 In its early decades, the World Bank was built around project expertise: stand-alone projects, mostly infrastructure investments, were at the heart of the Bank's European reconstruction efforts in the 1950s, then of its efforts in the then-narrow range of independent developing countries. From the 1960s, as more countries achieved independence, the Bank was more squarely faced by the challenge of development, diversifying its projects into new sectors and shifting from an overwhelming emphasis on projects to a greater balance between projects and country strategy. From the late 1960s ­ the McNamara years ­ the Bank initiated a steep increase in lending in response to the perceived development challenge. As a result of these factors, its traditional strength in projects began to be diluted. During the 1970s the world economy fell into crisis, with unprecedented balance-of- payments problems and a growing debt crisis in the Third World. The 1980s was marked by stagnation in Latin America and decline in Africa. By the beginning of this decade, the World Bank's diagnosis had changed: traditional projects were too small and too slow to disburse and did not have the levers to touch the core policy problems (Mosley, Harrigan, and Toye, 1995). Thus structural adjustment lending ­ quick-disbursing and policy-based ­ was born. By the 1990s, conditionality was already coming under scrutiny, not for its policy content so much as for the weakness of policy implementation (for more on this, see part III below). At the same time, weaknesses in the performance of the traditional IBA vehicle had been becoming more manifest over the years. Finally, the World Bank was confronting new challenges from outside, typically from the civil-society movements of some of the richer countries, eventually with the support of their governments and aid communities. Thus, environmental concerns burst on the scene in the 1980s and gender, governance and other concerns in the 1990s. At the same time, civil society and NGOs all over the world were finding their voice and demanding a role in the development debate. (The end of the Cold War also made it easier for the new voices and concerns to be heard.) External criticism of the Bank grew considerably in the 1990s. New intellectual currents also made themselves felt. Theories of economic growth evolved, changing the emphasis from physical capital accumulation and exogenous technical change to more multi-faceted concepts emphasizing human capital among other things. New theories of public management in response to the crises faced by some 6For some more detailed background, see: Kapur, Lewis, and Webb (1997); Gilbert and Vines (2000); and Mosley, Harrigan, and Toye (1995), Chapters 1 and 2. 3 OECD countries in the 1980s introduced new techniques of results-based management (RBM). The World Bank has responded to these considerable pressures for change in various ways (for more details see Box 1): · A large effort has gone into tightening up the managerial frameworks for projects and lending. In part this has been by emphasizing "results on the ground" in preparing and reporting on projects (for instance, use of output and outcome indicators and the Logical Framework) and by tightening the evaluation process (for instance, added ex- ante to ex-post evaluation; aggregating and quantifying evaluations). Financial management of projects (procurement, disbursement, audit) has also been tightened and disbursement is beginning to be linked to project outputs (the LACI initiative). · Much effort has gone into a more strategic approach to lending and other operations. Country and sector strategies have come to the fore. The Bank is developing a pronounced poverty orientation. The Bank is seeking to position itself more as a knowledge provider than a provider of finance and physical investment. · Perceiving the partial failure of the conditionality approach, the Bank is now increasingly seeking to align its objectives with those of its client countries through the language of partnership with governments and NGOs. At the same time, it is seeking to promote greater coordination and specialization among donors. · The Bank has also sought management changes that would improve internal incentives, notably towards greater client orientation, through changes in organization and human-resource policies. · Most recently, the Bank has, in concert with the rest of the development community, been trying to mitigate the proliferation of objectives and diffusion of efforts that occurred in the 1980s and 1990s by adopting client-proposed and internationally- agreed country-level goals and targets for poverty reduction and social improvement. The World Bank is by no means alone in this ferment of experiment (Box 5.1 of Dollar & Pritchett, 1998). The regional development banks have trod a similar path to the World Bank. Thus for the development banks, the traditional project (i.e. IBA investment projects) still remains the most important aid vehicle. The bilateral donors, traditionally more reliant than the development banks on grants, have shifted from traditional projects to capacity building and support for policy reform. In the UK, for example, traditional projects are in the process of being abandoned in favor of technical assistance, support to the private sector, and support to NGOs (Mosley, 1999). Table 1 attempts to provide a synthetic overview, from the recent literature, of the many types of problem of effectiveness of traditional project aid that have been identified, their causes, their impact on projects, and the related improvement initiatives that donors have typically undertaken. 4 5 Box 1. A Decade of Changes in the World Bank: the Permanent Revolution The World Bank commissioned a report, appearing in 1992 as the Wapenhans Report (World Bank, 1992), on the effectiveness of its traditional lending. The Report found that the "approval culture" ­ securing loan approvals ­ was such a powerful staff incentive that more attention was given to the quantity of lending than its quality. This report has been followed by a plethora of initiatives aimed wholly or in part at improving the effectiveness of traditional instruments. Here is a selection of the main ones: · The "Next Steps" program in 1993 indicated 87 specific measures to be implemented to address the concerns of the Wapenhans Report. · In the mid-1990s a "renewal process" was launched to build new competencies, update systems and processes, establish new management arrangements and reorient its values and culture. · In 1996 the Quality Assurance Group was established to provide line managers with independent assessments of their work by vetting the quality-on-entry of operations . · Evaluation was also reinforced by: establishing an Annual Report on Portfolio Performance (ARPP), prepared by the Bank's operational staff; by having responsible staff evaluate their own completed operations; and by strengthening the OED (the Bank's internal auditor) and charging it with independently evaluating a representative sample of completed operations. · Sector Strategy papers were introduced in 1996. · Bank management's Strategic Compact with its Executive Board in 1997 secured an injection of extra resources in return for a promised improvement along various lines of management and performance. · In 1997 the Bank was reorganized: a matrix organization of sectoral and country responsibilities, the establishment of thematic Networks, and the decentralization into the field of many country departments aimed to increase the Bank's client focus at the same time as securing its technical and knowledge base. · In 1999 the Comprehensive Development Framework pulled together the major themes of the emerging consensus in the development community by proposing the involvement of all stakeholders in the development process and requiring improved global- and country-level partnerships. · In 1999 the Bank (with the IMF) introduced the Poverty Reduction Strategy Paper (PRSP), a new approach to reducing poverty in IDA countries based on country- owned strategies that would provide a framework for development aid, including debt forgiveness under the HIPC Program. Source: World Bank (2001, 2002) and other sources 6 The performance of IBA projects How effective or ineffective have traditional projects been? It is very difficult to say because there is no body of consistent and comparable ex-post evidence. (The situation is better in the specific area of structural adjustment loans; see part III below.) The aid literature ­ both the external studies and the internal evaluations of the donors themselves ­ seems to range from the cautiously optimistic to a dissatisfaction with the historical performance of projects. Perhaps that cautious optimism has lessened over time. Cassen (1994), in a study largely based on the experience of the 1980s, reviewed a large number of evaluation studies to make the finding that aid generally works. A study of aid to Africa (van de Walle and Johnston, 1996) found that in spite of massive aid, the continent's development had been poor. While aid could not alone take the blame for this, there was certainly a need to make this aid more effective. A World Bank study (Dollar and Pritchett, 1998) was driven by a sense that aid was not working well. The study provided a nuanced view of where and why project aid did and did not work. At the (quantifiable) macroeconomic level at least, the correlation of aid with growth was at best weak. A World Bank study of aid and reform in 10 African countries (Devarajan, Dollar, and Holmgren, 2001) found that heavy aid did not necessarily lead to good policy outcomes. While the widespread dissatisfaction with traditional projects appears to persist in the aid community, there are indicators that the managerial changes made in donor agencies since the early 1990s have led to some improvement in the performance of loans. Only a few donors have consistent time series of evaluation data to indicate this. The World Bank's 2000 Annual Report on Development Effectiveness (World Bank, 2001, Chapter 2) found a clear trend of improving effectiveness of loans since around 1993-95. "Satisfactory project outcomes" (measured, by perception, at the time of the completion of Bank disbursement on the project), having deteriorated since 1990, improved with reasonable steadiness from 1994. The perceived impact of loans on institutional development and the perceived likelihood of their sustainability over the longer term also improved steadily from the middle of the decade, although absolute levels of institutional development and sustainability were considerably lower than satisfactory outcomes).7 However, these quality improvements were notably more marked for adjustment lending than for investment lending (see Figure 1). In fact, whether weighted by number of loans or by disbursements it is difficult to see any clear improving trend in investment lending outcomes (though sustainability and institutional development showed clear improvement). Moreover, the extra efforts that went into tightening the managerial framework for lending may well have increased the cost of lending.8 7The Meltzer Report (2000) took the aggregate results of the ARDE to task because of the weighting system used by the Bank and recalculated a far higher rate of failure. But whatever the weighting system used, the ARDE results indicate an improving trend in recent years. There are also other indicators of overall improvements in the Bank portfolio. For instance, in recent years, the quality-at-entry of operations has improved, as has the proportion of projects with satisfactory supervision, while the number of projects at risk has fallen (pages 76-77 of World Bank, 2002). 8The Meltzer Report (2000) mentions the World Bank's high administrative costs. 7 Table 1: A Taxonomy of Project-Effectiveness Problems Cause of the Problem The Problem Project Outcome Improvement Initiatives 1 See large literature on policy and Poor supporting environment Project is overcome by adverse Project evaluation (1970s); program governance failures. In addition, (policy and institutional) in client political and economic forces lending (1980s); governance there are aid-dependency issues. country [D&P, vdW&J] concerns (1990s) 2 Emphasis on finance: "Approval Poor substantive design: over- Project is only additional if it brings Sector work, etc.; participatory culture" of donor agency? [W]; emphasis on finance, under- new technology or knowledge techniques (and use of NGOs) Poor donor knowledge of local emphasis on global and local (supporting policy or institutional conditions, due to lack of local knowledge; emphasis on larger reform), given the fungibility of participation [D&P] projects and physical outputs more money [D&P] than efficiency gains. 3 "Approval culture" of donor agency? Poor substantive design: emphasis Project is not sustained when Public-expenditure reform (medium- Emphasis on procurement from Part I on: capital, not recurrent, spending recurrent expenditures do not term-expenditure frameworks, etc.) governments (reflecting domestic materialize [vdW&J] commercial interests)?; Poor public expenditure management in client countries 4 Poor public administration in client Poor substantive design: Enclave projects that undermine Public-sector modernization efforts; countries; proliferation of stand-alone projects client-government cohesion and Output-based approaches to aid Response to political and [vdW&J] prove unsustainable [vdW&J] bureaucratic demands in donor countries [vdW&J]; 5 "Approval culture" in donor Poor incentives: project aims are Project implementation is poor "Results-on-the-ground" measures agencies; ill-specified and payment is for (ex-ante & ex-post evaluation; Lack of "ownership" in client inputs, not results [W] project indicators ...); Output-based countries approaches to aid 6 Domestic political agendas and High administrative costs: Fragmented aid, leading to Country coordination (CAS, PRSP); bureaucratic politics in donor fragmentation of donor efforts and fragmentation of client sector coordination ( SWAPs); donor countries weak client administrations administrations [vdW&J] coordination (CDF) 7 Bureaucratic politics in donor High administrative costs of donor countries agencies [M] Illustrative sources: D&P: Dollar and Pritchett (1998) vdW&J: van de Walle and Johnston (1996) W: World Bank (1992) (Wapenhans Report) M: Meltzer Commission Report (IFIAC 2000) 8 Figure 1. World Bank Loans: Quality-at-Exit Ratings, FY80-00 (% satisfactory) OED Quality-at-Exit Ratings FY 90-00 Adjustment Loans Weighted by operations Weighted by disbursements 120 120 FY80-89 FY90-94 FY95-98 FY99-00 FY80-89 FY90-94 FY95-98 FY99-00 97 100 100 87 84 86 81 80 76 80 73 70 71 68 tnecre 60 60 60 52 48 tnecre 62 58 60 44 P 41 39 P 38 40 40 34 34 30 19 20 20 0 0 Outcome Sustainability Institutional development Outcome Sustainability Institutional development Investment Loans Weighted by operations Weighted by disbursements 120 120 FY80-89 FY90-94 FY95-98 FY99-00 FY80-89 FY90-94 FY95-98 FY99-00 100 100 76 tnecre 80 80 72 72 72 71 69 68 65 57 60 P 46 tnecre 62 57 60 54 50 49 43 44 42 P 40 40 35 36 33 30 30 26 20 20 0 0 Outcome Sustainability Institutional development Outcome Sustainability Institutional development Source: Figure 12 of World Bank, Operations Policy and Country Services (2001) II. Contracting in the Public Sector All loans ­ whether IBA or OBA, investment or adjustment ­ are enshrined in contracts. A World Bank Loan Agreement, for instance, will, typically, set out what the "principal" expects the "agent" (the recipient government and/or its sub-units or sub-contracting firms) to do (build a bridge, provide inoculations, reinforce management, change policies, and so on), the way this activity will be financed, how the principal will monitor and audit implementation, what happens if implementation goes wrong, and how the loan or credit will be serviced. The economic literature has paid increasing attention to the attributes of principals and agents, products and services, contractual arrangements, and external conditions that substantially determine how effective these contracts will be (i.e. where principals and 9 agents get what they both want at acceptable levels of transactions costs).9 This section looks first at the attributes of contracts, then considers the particular problems present when the public sector is involved. What makes contracts work? A World Bank research study, Bureaucrats in Business (World Bank, 1995, pp. 119-112) proposed a useful framework for understanding what works and what does not in the domain of contracting. To simplify, contracts work better when they: · make information more completely available, in particular to principals; · provide incentives that induce the release of information and promote the proper implementation of the contract; and · benefit from mechanisms of mutual commitment of principal and agent to the terms of the contract. Information. Asymmetric information ­ in particular an agent's specialized knowledge of his production process and the principal's imperfect ability to observe the agent's true performance ­ allows the agent to pursue interests at odds with those of the principal. For instance, the more that the quality of the product or service the agent produces is difficult to judge, the more the agent can "deceive" the principal by producing inferior products. Several instruments to improve the knowledge available to the principal can be indicated. Generally most important, competition reveals information on costs of production by encouraging agents to lower their prices (though greater efficiency or elimination of monopoly profits). This can be through direct competition (multiple suppliers or auctions), contestability (potential competition through threats to change agents), or simulated competition (for instance, benchmarking or cross-jurisdictional competition). There are other informational considerations: 9 These issues are treated in a number of different academic traditions dotted across the social sciences which are now increasingly cross-pollinating one another. In particular, the new institutional economics (NIE) asks questions about the rules and incentives structures that make organizations and markets work more or less effectively. Within NIE, more or less, are several highly related academic fields (summarized in Part D of Preker, Harding, and Girishankar, 1999): principal-agent theory, which highlights the needs to reconcile the divergent interests, in a contractual or working relationship, caused by widespread uncertainty and uneven access to information; transactions costs economics, which has provided particular insights into the alternate governance structures of organizations (hierarchies such as ministries or firms) and markets; and property rights theory, whose main relevance to the debate on public-sector contracting has been to explain incentives as relating to who owns the "residual" rights to making decisions and to reaping efficiency gains. Overlapping with NIE, but still not fully integrated are other bodies of relevant knowledge, is a diffuse set of work, under the loose umbrella of organization theory (more the province of political scientists, sociologists, and psychologists) which provides additional insights in the areas, for instance, of non-formal arrangements within organizations, employee motivation, and the external "authorizing" environment (for a tasting try Pugh and Hickson, 1996). In addition, there is a growing literature (political economy, political science, public choice) on the particular authorizing and incentive environment for public agencies and bureaucracies provided by political arrangements and by the particular interests of politicians, interest groups, and bureaucrats (for instance, see Moe, 1990, Wilson, 1989, and Horn 1995). 10 · Obviously, contracts are better when they are more simple and more complete (i.e. covering all contingencies). · Contracts appear to be better when principals and agents have some equality of standing and knowledge: weak and junior principals will have some problem, ceteris paribus, in controlling stronger, more senior agents (and vice versa). · Where there is more than one principal, they may have greater difficulty, without cooperation, in supervising the agent. Incentives. Principals can reward and penalize agents with money and power (and by affecting their well-being). For instance: performing managers can receive bonuses, share in profits, or get promoted. Principals can empower agents to perform better by giving them greater control over resources (notably pricing, buying and hiring & firing decisions). One explanation for the power of incentives "lies in the concept of `residual claimant'; that is, the agent who has claims on whatever residual benefits remain after other claims are met has the greatest incentive to improve performance" (World Bank, 1995, page 168). Commitment. Each party needs to be convinced of the other's commitment to deliver ­ i.e. to trust the other partner ­ because it is impossible to include all possible contingencies in a contract. Contracts can, to an extent, be structured to contain internal mechanisms to commit the parties: · The successful repetition of contracts (a "game" in which each partner learns from the previous contract and, as necessary adjusts its behavior) can create trust. · Some contracts (spot contracts) minimize contingencies and others can be structured to do so in the form of "hostage" contracts where both parties commit assets for the duration of the contract. Most contracts structure a payment schedule both to create commitment (the down-payment) and to follow delivery. In addition, the principal can offer an enforceable guarantee (such as collateral). But the key to commitment in the case of "incomplete" contracts is the existence of an external party or standard with power to encourage or compel the parties to meet their promises. These external commitment mechanisms are essential in large markets and can be classified as follows: · Reputation mechanisms (word of mouth, credit and performance ratings established by professional agencies, the press and public opinion). · Courts, or other arbitration bodies, which can resolve conflicts and enforce contracts. This simple framework is not complete, for the subsequent purpose of looking at aid contracting, without consideration of two further complications. First, it is axiomatic that the quality of the external institutional and economic environment helps determine the success of the contract. For instance, price instability (inflation, fluctuating real exchange rates), poor courts and an unstable regulatory regime undermine contract fulfillment. Second, principals and agents both have other principals to answer to: firms must answer to their shareholders, just as governments must answer to voters and interest groups. As a 11 result, contracting is often a "nested game" in which principals and agents may appear to act against their own apparent, or declared, interests because they also have to satisfy their own principal. What are the obstacles to efficient public sector contracting? Both governments (whether as principal or agent) and donor agencies face obstacles to efficient contracting, though these are not completely identical. Table 2 tries to capture these obstacles and to hypothesize about the likely outcomes. It is difficult for governments to contract efficiently and in the public interest because of problems of credible commitment, divergence of objectives with donors, difficulties in measuring the quality of the services they produce, and, in the case of less-developed countries, the quality of public administration. First, governments have a problem making credible commitments. This appears to come from two sources, sovereignty and time inconsistency. Paradoxically perhaps, sovereignty makes it difficult for them to commit credibly since there is no body above them to make them keep their word. Second, more than private firms at any rate, governments cannot commit easily over time because they cannot bind subsequent political regimes.10 This leads to the problem of government as a principal because it may not be able to live up to contractual conditions such as price stability or non-interference in labor markets and as an agent because it may not live up to implementing or repayment conditions. In governments that are less democratically accountable, the commitment problem may be exacerbated by the "nested games" of interest groups (and, at its worst, corruption). Second, governments have non-profit objectives which are difficult to align with those of other public bodies, including donors. Governments do not maximize profits. They serve their principals, voters and private interests in democratically-accountable countries and more squarely private interests in less accountable countries. This need not impede contracting, as principal, with private firms, but it does impede contracting, as agent, with foreign donors. This is at the heart of the "ownership" problem which is discussed in the next section. Third, governments as agents of donor contracts, generally have incentives built into their systems of public administration (for instance, personnel and wage policies) that are too weak to encourage efficient production. Fourth, governments typically produce services which are difficult to measure. Typical of such services are public goods such as policies, external security, and policing and more 10Private firms are able to commit themselves more effectively over time (for instance to repay a loan) because their management is not subject to change-by-politics, they are subject to the threat of being forced to cease business, and external commitment mechanisms support commitment. (Thus, accrual accounting is a technique to support time consistency that the private sector uses and the public sector, with few exceptions, does not.) 12 Table 2. Public Sector as Principal and Agent: Some Problems in Contracts Source of problem and Democratically- Less- democratically- Less- democratically- Bilateral donors (as Development banks (as contracting outcome accountable accountable accountable principals in aid principals in aid governments (as Governments (as Governments (as contracts) contracts) principals in national principals in national agents in aid contracts) contracting with contracting with private entities) private entities) 1 2 3 4 5 Sovereignty Difficult to commit Very difficult to Very difficult to ??? Can commit credibly on credibly commit credibly commit credibly finance; cannot commit Time inconsistency credibly on conditionality Type of objectives Non-profit objective: Unrevealed objective: Unrevealed objective: Multiple objectives: Multiple objectives: voter welfare private interests private interests developmental, special developmental, special interests, bureaucratic interests, bureaucratic, financial sustainability Poor measurability of High proportion of High proportion of High proportion of High proportion of High proportion of outputs public or complex public or complex public or complex public or complex public or complex private goods favors private goods favors private goods favors private goods favors private goods favors private entity private entity government (as agent) agent (recipient agent (recipient government) government) Quality of personnel Poor quality Poor quality favors (clientelism and principal (donor) underdevelopment) favors private entity Hypotheses about likely Can contract where Area for contracting Poor implementation of ??? Financial objectives outcomes outputs more narrowed by challenges aid contacts because of: ensure disbursement. measurable. to contracting: credible commitment problems; Some non-financial Making credible commitment problems; conflict of objectives; objectives, excess of commitment is a corruption (private corruption (private objectives, poor moderate challenge interests); poor quality interests); ability to measurability of of personnel hide implementation outputs favor poor quality; poor quality of implementation. personnel 13 complex private goods such as certain health and educational services. This makes it more difficult for governments as principals to monitor what they contract for and easier for governments as agents to avoid full implementation of donor-driven projects or adjustment loans (see part III). Fifth, many developing countries have poor public administrations, making contracts difficult to negotiate and manage. Poor public administration also leads to the pervasive problem, which the donors have helped promote, of creating "islands" of good public administration to administer key services or to protect donor projects. Such islands are often in themselves effective, for a few years at least, but they further debilitate the rest of the public administration (see van de Walle and Johnston, 1996, Part IV). Results-based management in the public sector The impetus for a wide-ranging revolution in techniques of public management came from the dissatisfaction felt in many OECD countries with the inefficiencies and inflexibilities of the old "hierarchical" (or Weberian) model of public administration. Virtually all the OECD countries have tried to increase contestability in the provision of public services, through privatization, voucher schemes, contracting out, devolution of activities to lower levels of government, and so on. A smaller number of governments, notably including Anglo-Saxon and Scandinavian countries, has gone further in experimenting within the broad bounds of a new, results-based-management model.11 The emerging model, as it applies to public administration, can be characterized along four lines whose common theme is to borrow from the managerial methods of the private sector (See Chapter 7 of Burki and Perry, 1998). · Devolution of decision making. Reducing the burden of hierarchical rules and fostering greater discretion at lower points in the hierarchy -- operating agencies, regional agencies, sub-national governments -- allows decisions to be made effectively because decision-makers are closer to the problem and have clearer objectives. · Performance orientation. Changing the accountability relationship from an emphasis on inputs and legal compliance to one on outputs provides incentives leading to greater effectiveness. · Client-focus. Reporting to and "listening" to the clients of the public sector allows governments to understand more systematically what citizens want and to respond with more appropriate outputs. · Market orientation. Making greater use of markets or quasi-markets --typically through management and personnel contracts, competition between public agencies, inter- agency fee charging, out-sourcing, personnel contracts -- improves the incentives for performance orientation. Needless to say, the new model is controversial. Some countries, such as France and Japan, remain wary of the reforms for fear of their undermining the existing public- 11The model goes under several names, most frequently the "New Public Management", but also the "new managerialism, "performance-based management", and ­ more in vogue in the donor agencies ­ "results- based management". 14 service ethos. For the application of the model to developing countries, some argue that if political incentives do not promote democratic accountability, the new model will certainly fare no better than the old model and informality, clientelism, and corruption will remain the norm in public administration.12 Others argue the importance of sequencing: you cannot manage outputs until you can effectively manage inputs (Schick, 1998). While the extent that some countries can adopt the model is limited, there is no doubt that many can make progress in increasing the contestability of provision of public services. Contracting is part of this. Result-based management: the new types of contract New forms of public-sector contracting of public services have emerged in several distinct forms. · Performance contracts regulate relationships within governments, usually on an indicative, rather than legal, basis. · Management contracts establish the mutual obligations of governments and private managers of public assets. · Regulatory contracts establish the mutual obligations of governments and private firms operating under public-interest regulation. · Under supply contracts (also known as outsourcing or contracting out), the government contracts out to private firms, usually on a competitive basis, for the supply of public services in exchange for state reimbursement or from government- supported beneficiaries.13 The evidence on how well these contracts are working is not yet systematic, but the study Bureaucrats in Business (World Bank, 1995) examined the working of the first three types of contract across a sample of developing-country cases.14 Table 3 characterizes the attributes of the three different kinds of contract (in terms of the contractual framework established in part II of this paper) and their outcomes. The following paragraphs provide some insights from this and other studies. Performance contracts. Governments pursuing the RBM reform model have increasingly tried to create contracts inside government. These consist in "a range of management instruments used to define responsibilities and expectations between parties to achieve mutually agreed results." (OECD, 1999, page 7). These instruments include: agreements between a minister and chief executive, between a funding or policy-setting ministry and an executive agency; with state-owned enterprises; with sub-national governments. Performance contracts are largely "relational" contracts designed to clarify commitments 12Chapter 7 of Burki and Perry (1998) presents this position. Keefer (1998) takes a similar position by arguing that contracting out will not improve efficiency outcomes if governments are corrupt. 13OBA is located in this category. 14Each of the three contracts has different sectoral characteristics that can help explain success and failure. There is also a problem of small sample size and non-comparability of contracts. But for the purposes of looking at the relative contractual merits of IBA and OBA, the study is effective in showing how different kinds of contractual arrangements can systematically change outcomes. 15 Table 3. Bureaucrats in Business: Outcomes of Three Types of Public Sector Contract Performance contract Management contract Regulatory contract (government and public (government and private (government and regulated manager) manager) monopoly) Sample 12 state-owned enterprises: 20 contracts: tradable goods 7 countries: privatized utilities (hotels dominant) telecoms Information Incentive for agent to Tendency to operate in Increased competition negotiate weak, soft, competitive markets and (allowed by new excessive targets when contract to have been result technologies) or, where principal was often of auction. monopolies remained, technically weaker, more simulated competition junior. (yardstick, auctioning). Best performers had quantity and quality targets. Rewards Bonuses infrequently used Rewards closely linked to Efficiency-inducing pricing and and effects frustrated by poor performance: money regulations: price caps (RPI penalties targets. Agent's career rewards (success fees ­ x) and benchmark pricing development depended more generally better than fixed better than rate-of-return on politics. Agent given little fees); power (better regulation. managerial freedom. performance when governments did not impose other objectives such as wage and employment policies). Commitment Principal reneged on Limited use of third-party Explicit conflict-resolution promises (e.g. pricing). enforcement, suggesting rules helped. Enforcement Generally no third-party importance in enforcement was best from independent oversight of contracts. of contract length and regulators (if well staffed) repetition and firm or courts (if independent). reputation. More successful Insulating regulators from contracts where politics was important (e,g, governments signaled by law in Chile, by appeal commitment, e.g. by taking to Privy Council in on contract-preparation Jamaica). costs (rather than donor). Contract Poor: no positive impact on Good: positive impact on Good: generally positive Outcomes profits; minimal labor profitability and impact on network growth, productivity impact; neutral productivity. labor productivity, rate of total-factor-productivity return, consumer impact. satisfaction. Source: World Bank (1995), Chapter 3 on the basis of trust and cooperation, rather than legal enforceability.15 New Zealand is noteworthy for having tried to move such contracts more towards legally enforceable instruments. It appears that performance contracts, while they have not been systematically studied, have been "effective for managing relationships in a devolved management 15The change in donor language from "conditionality" to "partnership" parallels the difference between an enforceable contract and a relational one. 16 environment." (OECD, 1999, page 41), though challenges remain of setting targets, designing appropriate incentives, evaluating performance, and controlling the costs of preparing and managing contracts (Brook and Smith, 2001, Part 1). UK hospitals provide an example of reasonable success in relational contracting (England, 2000, Box 6). The World Bank study confined itself to a small sample of performance contracts with state-owned enterprises (SOEs). These were generally ineffective in improving SOE performance. The study found that, typically, it was in the interest of SOE managers to negotiate "soft" or conservative ­ hence easily achievable ­ targets, civil servants were technically too weak to monitor effectively, SOE managers were incentivated more by political factors than effective rewards or punishment for the jobs they did, they were overly constrained by government decisions (on employment policies, for example), the government frequently reneged on its undertakings, and there was no external body to provide a counterweight to this (Table 3). Management contracts. For the sample of firms in the World Bank study, private managers (often from reputed multinational corporations) managed public assets in industries operating in competitive markets. Private management improved performance more than did the public management of SOEs. Not all firms performed better, but better performance was typified by: firms operating in competitive markets; rewards linked to performance (success fees worked better than fixed fees); absence of government interference in firm personnel policies; long or repeated contracts; the importance of firms maintaining their (international) reputation; and government signaling of its commitment (notably by paying contract-preparation costs, rather than letting donors do it). Regulatory contract. Such contracts have become frequent, often following privatization, in natural monopolies (infrastructure and utilities). The World Bank study looked at the privatized telecom sector in seven developing countries. The study found privatization generally to have been beneficial, but the benefits varied according to contractual conditions. The privatized firms benefited from competition, but the best performers benefited from government-set quantitative and qualitative targets. Price-caps or benchmark pricing led to better performance than rate-of-return regulation (i.e. the better performers were forced to reveal more information about their cost structure). Performance benefited from more explicit rules about conflict resolution and from independent regulators or courts. Brook and Smith (Part 1) report that infrastructure contracts under this rubric have expanded rapidly, with generally positive results. Supply contracts. Keefer (1998) cites examples of cost savings in advanced countries, typically in sub-national-government services (refuse collection, street cleaning, road construction and maintenance, fire protection), but also in contracting out internal government support services. The World Bank study also made some general comments about why management contracts and regulatory contracts do a better job than performance contracts (World Bank, 1995, pages 168-170). These comments are equally relevant to supply contracts 17 and to the debate on IBA versus OBA. The study suggests that private firms, and to a lesser extent private managers, have a greater incentive than public managers because they can claim a greater share of (i.e. have greater "residual rights" to) the returns to good performance. By the same token, they face a greater penalty for failure. The public managers faced none. Contracts with private parties are also more prone to public scrutiny. Private contractors often try to simplify measured targets so that they can more easily defend them against public criticism and more easily resolve disputes. For public managers, on the other hand, contract complexity improves their chances of a satisfactory rating. "Commitment was the Achilles' heel of all three types of contract, although performance contracts suffered from this the most." (page 170). When one of the parties is the government, commitment problems are compounded because the private party cannot force the government to comply. Since private parties are not obliged to contract, the government has to offer some sort of bond or risk premium. When external donors pay for this, commitment is weakened because it insures the private manager's risk. Finally, where government management skills are weak, private parties can behave opportunistically. III. Contracting and the Delivery of Aid Given the widespread dissatisfaction with traditional aid projects, there is some anecdotal description of the incentives in agencies that lead to poor outcomes. A common theme is that these incentives promote a volume of lending, but do not emphasize implementation. There is also considerable anecdote on the lack of project and program "ownership" by recipient governments.16 There has been apparently less analysis of how the two sets of incentive interact. However, two research studies on World Bank adjustment lending provide some analysis, consistent with the broader anecdotes on incentives and ownership, that suggests that the donor-recipient contract is a singular and problematic contractual relationship where there is a confluence of interests to move money, but a mismatch of objectives and, as a result, poor loan outcomes. It should be understood that the discussion of poor project and program outcomes in this section and of perverse incentives is a simplification for presentational purposes, and risks becoming a misleading caricature. Aid projects and programs do of course have a substantial impact, as the assessment literature indicates, but the question addressed here is why, given the sophisticated resources that go into their preparation, is the average impact disappointing. Donor-recipient contracting outcomes: the results of two research studies Boxes 2 and 3 provide some relevant results from two research studies of World Bank structural adjustment lending, Killick (1997) and Mosley, Harrigan, and Toye (1995). 16One of the fullest and most careful analyses of donor and recipient government incentives is to be found in van de Walle and Johnston, 1996, Parts V and VI. The study covers Africa, but is relevant for other parts of the developing world. The 2000 Annual Report on Development Effectiveness, World Bank, 2001, is also remarkably clear about incentive issues in the Bank. 18 Both were concerned to explain the relatively disappointing performance of adjustment lending. To do this, they adopted a contracting framework emphasizing the attributes of the contract, the incentives of principal and agent, and the outcome of their interaction. Adjustment lending is, unlike IBA, formally structured within a results-based contract: funds are not released until the conditionalities have been met (unless there is a waiver). But apart from this, the set of incentives facing donor and recipient governments appears broadly the same for structural adjustment lending and traditional investment lending. The results reported in Boxes 2 and 3 may reflect the Bank's earlier experience of lending, particularly in the case of Mosley, Harrigan, and Toye. World Bank evaluations indicate a substantial apparent improvement in adjustment lending outcomes since the 1980s (Figure 1). Thus the main value of these studies may be to provide hypotheses about traditional investment lending, more than about current adjustment lending. The studies together tell the same broad story. The donor and the recipient both wish to promote financial flows. The donor wishes to promote policy change. The recipient accepts conditionality in order to secure the loan, but then resists or reverses it during the subsequent period, for fear of its political costs. There is thus a conflict of interests. The loans were not large enough to "buy off" governments, while political, bureaucratic and financial pressures made it difficult for the donor to punish the recipient. As the Mosley, Harrigan and Toye study put it: "The donor and recipient, therefore, have conflicting interests concerning the implementation of conditions and a common interest in spending the donor's budget." Incentives in Donor Agencies It is probably true that donors provide aid for many reasons including not only "revealed" developmental objectives ­ based on a belief that poorer countries need financial resources to develop or that they need to change policies and behavior to develop ­ but also less revealed objectives related to national, private, or bureaucratic self-interest. The evidence on the incentives and arrangements that may tilt donors towards one objective or another is not very well developed, but a persistent theme from the literature is that lending has continued to be driven by an "approval culture" that reflects a set of incentives to donor-agency staff to maximize the volume of new lending without taking adequate account of implementation risks.17 The empirical outcome, as the two studies reviewed above indicates, is that project and programs quality suffers. The World Bank's 1992 Wapenhans Report provides perhaps the best-known exposition of the approval culture (and how it has affected project implementation), but the theme makes frequent other appearances, as the brief literature review in Box 4 indicates. This literature does not have much to say about why this culture persists. The approval culture could conceivably derive from the belief that poor countries remain poor because they lack financial resources. This belief may help explain the origins of the approval culture 17A major study on the incentives faced by the Swedish donor agency, SIDA, is soon to be published. See Gibson, Ostrom, and Shivakumar (2001) 19 (for instance Faught Management Group Limited, 2000), but that view of the development process is now so unfashionable that other explanations must be sought. In the case of the World Bank, at least, financial and organizational features help explain the resilience of the approval culture. · Lending plays an important role, through the servicing of debts to the Bank, in covering the running costs of the Bank. The Bank's privileged status in financial markets weakens the normal link between loan profitability and loan repayment: the Bank's shareholders guarantee the loans by which it raises money on international capital markets and it in turn requires a government guarantee for the loans it makes to recipient countries. Moreover, the Bank's easy access to capital markets allow it, within reason, to ensure a country's repayment of loans by maintaining the volume of disbursements higher than debt service obligations. · The Bank's governance arrangements ­ the way in which shareholding governments exert control over Bank management and staff ­ give it substantial managerial autonomy.18 Box 2. Structural Adjustment Lending ­ a Defective Contract? An article by Killick (1997) reports on a research project on World Bank and IMF structural adjustment lending. It seeks to explain the frequent failings of conditionality through the contractual incentives that determine whether the objectives of donors (the principals seeking to "buy" the reforms) and recipient governments (the "agents" responsible for implementing them) are aligned and whether incentives (rewards and punishments) are adequate to overcome any misalignment of objectives. Donors know a lot about good policies, yet adjustment programs do not, according to the large literature surveyed, get strong results. These programs have limited ability to achieve their objectives, in large part because program implementation is often weak: programs are often abandoned or interrupted, their impact on key policy variables, and especially institutions, is weak. The research found that when donor and government objectives were misaligned, notably where conditionality implied political costs, governments typically did not "own" programs, and hence their implementation of conditionality was poor. Nor were incentives strong enough to "buy" ownership and implementation (or punish non- implementation): political, bureaucratic, and financial pressures encouraged donors to lend; multiple, conflicting, or unclear program objectives reduced effective program implementation; non-compliance was often not effectively punished by reduced access to new credits. In these circumstances of weak incentives on both sides for program implementation, it is hardly surprising that the adjustment programs typically did not lend credibility to the government's reform program. 18This point is made in Naim, 1994, a paper which used an analysis of organizational incentives and arrangements to understand the challenges that the Bank faced on its fiftieth anniversary. 20 Box 3. Structural Adjustment Lending and Donor-Recipient Bargaining: Conflicting Interests on Conditionality, Common Interests in Disbursing A multi-country research project of World Bank structural adjustment lending in the 1980s was carried out by Mosley, Harrigan, and Toye (1995). It treated structural adjustment lending in a contractual framework, characterizing it (in Chapter 3) as a bargaining process "in a game in which the donor (the Bank) seeks to influence economic policy in the manner desired by it, whereas the recipient resists all such attempts at influence which do not harmonise with its own political priorities." Thus, there was an inherent conflict, and the outcome of the process was not pre-determined, but depended on the players' strategies and on external circumstance. During the negotiation stage of this game, the recipient sought a maximum financial inflow, but at a minimum political cost. The donor wanted as much reform as possible, but there was little to make the Bank undertake a reality check on the reforms agreed since, if these were not implemented, it would be relatively easy to find subsequent excuses for non-implementation (external circumstance, political change, etc.) that were technically sustainable. Bank officers were under pressure to lend, not only because of the perceived developmental impact of policy-based, quick-disbursing lending, but also because the Bank needed to maintain its profitability in a period (the 1980s) of economic crisis and potential default. There were countervailing screening devices in the Bank (the technical loan appraisal process, the oversight of OED, the internal auditor, senior staff and Executive Board oversight), but these were typically weaker for program lending than for project lending. as a result organizational change and lack of evaluation methodology: Incentives in recipient governments It is clear that recipient governments are extremely keen to receive loans and grants from donors. The assumption is that these countries are "resource-poor and need to gain access to donor funds" (van de Walle and Johnston, 1996, page 55). There also appears to be a remarkable consensus that a lack of recipient-government "ownership" is a key problem in aid effectiveness. In the words of Dollar and Pritchett (1998, page 99): "Recipients often see projects as being generated and implemented by aid agencies alone, with little local ownership." 21 Box 4. Incentives in Donor Agencies: What the Literature Says In 1992 the Wapenhans Report on the World Bank found: "a pervasive preoccupation with new lending"; "an approval culture" in which Bank staff rush to approve new projects, without taking adequate account of the risks in implementing the project; and "a systematic and growing bias towards excessively optimistic rate of return expectations" during the appraisal process. In spite of the waves of managerial changes since this report was published, the tenor of the anecdotes ­ some cover the World Bank, others cover donors more generally ­ has not changed much since.19 These anecdotes stress the following: · Career incentives reward a large volume of lending. · There are similar budget incentives to move money. "Internal budget resources are awarded where loan volumes are high, not where the number of worthwhile projects is highest or where technical assistance and knowledge transfer are favored over funding." (Meltzer Commission Report, 2000). "Country program budget allocations are relatively invariant with respect to country policy and project performance, and are not always fully aligned with the Country Assistance Strategy." (World Bank, 2001, page xvii). · The countervailing incentives ­ the checks and balances applied in the project cycle ­ are too weak (Meltzer Commission Report, 2000; Mosley, Harrigan, and Toye, pages 72-73). · In the face of the incentives to move money, the incentive is not to monitor results too closely. "Since monitoring policy reform requires some subjective judgment, donors will likely find that governments are making a good effort ­ whether they are or not ­ and disburse the funds." (Dollar & Pritchett, 1998, page 50, on adjustment lending) · Accountability is further limited by staff turnover and the long lives of projects, while it is subsequently easy to blame extraneous factors for failure. · Donors prefer specific projects to larger programs because the former are easier to monitor and give the donor a better sense of control. By the same token, coordination among donors leads to loss of control and power (van de Walle and Johnston, 1996, pages 79-80). · "Host government guarantees, required on all loans, separate project failure from risk of loss to the Bank." (Meltzer Commission Report, 2000) In some cases the ownership problem may simply reflect a unilateralist style of project or program preparation on the part of the donor (and rectifiable by a more participatory style), but more often it reflects a deeper problem: the very mismatch of objectives that 19See: Wade (1989), quotes in Mosley, Harrigan, and Toye (1995), pages 80-81; Cassen (1994), page237; Mosley, Harrigan, and Toye (1995); van de Walle and Johnston (1996), pages 55-58 and 77-80; Edgren (1996), quoted in Dollar and Pritchett (1998), page 117; Killick (1997); Dollar & Pritchett (1998), pages 50 and 99; Faught Management Group Limited (2000); Meltzer Commission Report (2000), World Bank (2001), pages xvii and xx. 22 was found in the two research studies reviewed above.20 For this mismatch to occur, recipient governments must value loan disbursements in different ways from donors. Van de Walle and Johnston provide an insight with the recognition that clientelistic politics, stemming from weak or authoritarian governments, discourage beneficiaries from participating and asserting "ownership" (page 55) and that "Government ownership of development projects and effective management of aid resources are less likely in political systems that do not reward development success and in which the primary commitment is to political ends rather than to economic growth and poverty alleviation." (page 88). Put in other terms, recipient governments have their own objectives, more complex than the profit objective of private firms. Democratically-accountable governments may have "revealed" objectives to provide public services, but other governments may have substantially "unrevealed" objectives to serve private interests.21 Unrevealed political objectives help explain why recipients assiduously seek aid funding yet resist using it for the purposes contracted.. Donor-recipient contracting: an overview The discussion of this section, if it can be generalized, suggests that the obstacles to efficient contracting between donors and recipient governments are somewhat different from those where governments are principals (section III and Table 2, which summarily contrasts the different cases). However, these findings pertain more closely to the multinational development banks than to bilateral donors (where I am less aware of an analytical literature). The donor agencies can commit to providing financial assistance because their principals, OECD governments in the main, have a policy interest in maintaining this assistance. Moreover, the multinational development banks (which lend to their own shareholders) have a financial profitability objective which requires continued disbursement. But the agencies are less able to commit credibly on conditionality (understood in the broad sense of insisting on the implementation conditions of both adjustment and investment lending). This could be for several reasons. Project implementation is inherently difficult to monitor. Indeed, as the donors have moved into more institutional areas, external forces have helped impose an ever larger agenda on projects and programs, and loan administration has become more demanding, the growing complexity of projects may militate against effective oversight of implementation. Moreover, bureaucratic politics (i.e. the bureaucratic interest in the agencies) may favor this model of credible financial commitment and non-credible implementation commitment. 20In fact, there may be a confusion about who is principal and who is agent in the contract. In the formal loan contract the donor appears to act as principal and the recipient government as agent. Yet, in theory at least, the donor is simply helping the recipient government appears achieve its own development objectives. 21At the more general level of public administration ­ i.e. outside the narrow area of project administration ­ we are beginning to see a keener analysis of how political systems and the extent of democratic accountability they produce affect governments' interests in providing public goods for the public or private goods for private interests (see Chapters 2 and 7 of Burki and Perry, 1998). 23 IV. Donor Contracting: the Implications for OBA The contracting framework that has been discussed in this paper permits some commentary on how and where OBA improves the aid contract, where it does not, and what risks may remain to be handled.22 Table 4 makes a schematic presentation of these points. Information. OBA (as Brook and Smith, 2001, indicates) will not work with all services. It is necessarily limited to those services (including policies) which can be adequately monitored.23 OBA also benefits from the fact that private entities have (or should have if they operate in contestable markets) "revealed" objectives (as opposed to the unrevealed, often political, objectives of governments in IBA). If corruption is pervasive in a country, then private entities too may have unrevealed objectives, and that will weaken OBA.24 Incentives. OBA "solves" some but not all, of the contracting problems of the IBA model. · What OBA does do. First, payment by results and revealed objectives (i.e the principal knows the agent's objectives, which coincide with the principal's) certainly increase the likelihood of implementation. Second, the use of competitive markets increases the likelihood of efficient outcomes. · What OBA does not do. Contracting with a new agent does not improve the donor agency's incentive to ensure implementation, on two grounds. First, the incentive for volume lending, at the expense of ensuring implementation ­ shared by principal and agent alike ­ is not altered. Second, the implicit or explicit guarantee that donors will be preferred creditors (or, worse, that they will lend into default) continues to interrupt the link between the donor's interest in loan repayment and its interest in project implementation. · What OBA may do: moving towards results-based contracting will probably involve a higher managerial burden on the principal (i.e. higher transactions costs) that will partly offset the efficiency gains (see England, 2000, part 6). Commitment. OBA does not, of itself, solve the problem of low commitment of the public-sector principal which World Bank, 1995, emphasized. That OBA does not strengthen the commitment of either donor or government to ensuring implementation is a central challenge if the scheme is to work. However, given that OBA necessarily involves provision of monitorable services, it is an excellent candidate to benefit from third-party oversight. This oversight could be in the form of an independent public 22This review is more concerned with overall institutional arrangements than with the technical details of competitive mode, design of subsidy and payment modes covered in Brook and Smith (2001). 23 Keefer (1998) also suggests that "The potential benefits of contracting out are greater when there are substantial opportunities for cost savings, particularly those that require significant effort to discover." 24Keefer (1998) makes a similar point. 24 regulatory body to monitor service quality and to audit performance.25 Alternatively or additionally, beneficiaries could play an informal role (for instance through complaints) or formal role (for instance, PTAs) in quality control.26 Outcomes and risks. In areas where it can work, the OBA contract thus provides a strong impulse to improving the likelihood and efficiency of project implementation, even if there is no improvement in the incentives donors face to ensure this. But the role of the government in a donor-financed OBA project remains an issue. The temptation will remain for donors to create administrative "islands" to ensure project implementation. Indeed, OBA will facilitate that option, while giving the government a substantial role in a donor-financed OBA project will, by adding an extra agent-cum-principal to the contract, certainly complicate it and increase costs. But this path, easy to tread in the short term, will create longer-term difficulties. First, any scheme that delivers services with a public-interest content will require some form of informed and consistent government policy. The study of public-sector contracting in World Bank (1995) indicated how government pricing and labor-market policies could affect contract outcomes. Second, an "islands" policy sows its own seeds: experience indicates that islands of efficiency are often not sustained over the longer term, while a policy of thus fragmenting the public sector extracts a price in terms of the loss of coherence in public policy-making, expenditure control, and management. 25The Meltzer Commission Report (2000) indicates the use of independent audit to oversee implementation of the performance-based grants it proposes. 26Keefer (1998) makes the point that the gains from contracting out are greater if beneficiaries can provide feedback on the quality and quantity of service provision. 25 Table 4: IBA and OBA: Comparing Contract Attributes and Problems IBA: hypotheses OBA: implications and proposals Contract · Donor contracts recipient government · Private agent, chosen on a competitive characteristics to produce specified outputs in return basis, and donor have a contract for prior financing of inputs. specifying use of user fees, supplemented by subsidy payment from donor against delivery of outputs. Information · Product characteristics (complexity, · OBA is limited to products whose long gestation) make it difficult to characteristics permit monitoring. monitor implementation. · Government has unrevealed objectives · Private entities have more revealed (political) that conflict with donor's. objectives than governments (but not necessarily if there is corruption) Incentives · Donor and government share strong · Donor-agency incentives are not incentive for donor disbursement. changed by OBA. · Government has weak incentive to · Private entity has strong incentive to implement: unrevealed objectives. implement: payment by results and revealed objectives. · Loan guarantee separates donor's risk · OBA does not solve problem of of non-repayment from project donor's weak incentive to monitor on implementation. risk-of-non-repayment grounds. · Government has weak incentives for · Private entities have strong incentives efficiency in implementation. for efficiency in implementation (the more so if reputation is at stake). · Government management capacity is · OBA may place greater management weak. burden on government (unless OBA by-passes it). Commitment · Donor and recipient governments are · Introducing a third-party arbitration or sovereign, do not have an arms-length oversight mechanism (e.g. relationship, and do not have a third- independent regulator or beneficiaries- party enforcement mechanism. as-watchdog) could reinforce commitment of donor and government. Contract risks · Disbursement proceeds, but with low · To the extent government is by-passed level of attainment of donor and public-service "islands" created objectives. (to minimize donor and private-entity · Alternatively, donor creates "island" transactions costs), OBA service- to ensure implementation. delivery may prove unsustainable and public management will suffer from fragmentation. · Changing the model from input- financing of governments to output- financing of private entities changes service financing from a public- expenditure to a financial-market problem. 26 References Brook, Penelope J., and Suzanne M. Smith (2001), Contracting for Public Services: Output-Based Aid and its Applications, Washington, D.C.: the World Bank (http://rru.worldbank.org/obabook.asp) Burki, Shahid Javed, and Guillermo Perry (1998), Beyond the Washington Consensus: Institutions Matter, World Bank Latin American and Caribbean Studies, Viewpoints, Washington, D.C.: the World Bank Cassen, Robert (1994), Does Aid Work?, Oxford: Clarendon (2nd edition) Devarajan, Shantayanan, David Dollar, and Torgny Holmgren (2001), Aid and Reform in Africa: Lessons from Ten Case Studies, Washington, D.C.: The World Bank Dollar, David, and Lant Pritchett (1998), Assessing Aid--What Works, What Doesn't, and Why, published for the World Bank by Oxford University Press Edgren, Gus (1996), "A Challenge to the Aid Relationship", in Swedish International Development Authority (SIDA) (1996), Aid Dependency, Stockholm England, Roger, Contracting and Performance Management in the Health Sector: Some Pointers on How to Do It, Draft 2: March 2000, IHSD (Institute for Health Sector Development) Faught Management Group Limited (2000), "Management, Incentives and Administration at the World Bank", in Expert Papers, Washington: Government Printing Office for the International Financial Institution Advisory Committee, March Gibson, Clark, Elinor Ostrom, and Sujai Shivakumar (2001), Institutions, Incentives, and Aid Sustainability: a Workshop Study for SIDA, Workshop in Political Theory and Policy Analysis, Indiana University, Bloomington, Indiana. (see ) Gilbert, Chrostopher L., and David Vines, eds (2000), The World Bank: Policies and Structure, Cambridge: Cambridge University Press Horn, Murray J. (1995), The Political Economy of Public Administration: Institutional Choice in the Public Sector, Cambridge; New York, NY, Cambridge University Press Kapur, Devesh, John Prior Lewis, Richard Charles Webb (1997), The World Bank: Its First Half Century, Brookings Institution Keefer, Philip (1998), Contracting Out: An Opportunity For Public Sector Reform And Private Sector Development In Transition Economies, mimeo, The World Bank Killick, Tony (1997), "Principals, Agents and the Failings of Conditionality", Journal of International Development, Vol. 9, No. 4, 483-495 27 Meltzer Commission Report (2000), Report on the International Financial Institution Advisory Committee Moe, Terry (1990), "The Politics of Structural Choice: Toward a Theory of Public Bureaucracy", in Oliver Williamson, ed., Organization Theory: from Chester Barnard to the Present and Beyond, New York: Oxford University Press Mosley, Paul (1999), "Recent Changes in Aid Technology: is the White Paper an Adequate Response", Public Administration and Development, 19, 19-29 Mosley, Paul, Jane Harrigan, and John Toye (1995), Aid and Power: the World Bank and Policy Based Lending, Vol. 1, New York: Routledge Naim, Moises (1994), "The World Bank: Its Role, Governance and Organizational Culture", in Bretton Woods Commission, Bretton Woods: Looking to the Future, Background Papers, pp. C-273-C-287 OECD (1999), Performance Contracting: Lessons From Performance Contracting Case Studies; A Framework For Public Sector Performance Contracting, PUMA/PAC(99)2 Preker, Alexander S., April Harding, and Navin Girishankar (1999), The Economics of Private Participation in Health Care: New Insights from Institutional Economics, World Bank, Submitted to the International Social Security Association, August 1, Pugh, Derek S., and David J. Hickson (1996), Writers on Organization, Fifth Edition, London: Penguin Books Schick, Allen (1998), "Why Most Developing Countries Should Not Try New Zealand's Reforms", The World Bank Research Observer, 13:1, pp. 123-131 van de Walle, Nicholas, and Timothy Johnston (1996), Improving Aid to Africa, Washington, D.C.: Overseas Development Council Wade, Robert (1989), Unpacking the World Bank: Lending versus Leverage, unpublished paper, Washington, D.C., March Wilson, James Q. (1989), Bureaucracy: What Government Agencies Do and Why They Do It, Basic Books World Bank (1992), Effective Implementation: Key to Development Impact (July 24, 1992), also known as the Wapenhans Report World Bank (1995), Bureaucrats in Business: the Economics and Politics of Government Ownership, published for the World Bank by Oxford University Press World Bank (2001), 2000 Annual Review of Development Effectiveness: From Strategy to Results, Operations Evaluation Department, Washington, D.C. 28 World Bank, Operations Policy and Country Services (2001), Adjustment Lending Retrospective: Final Report, June 15 World Bank (2002), The Role and Effectiveness of Development Assistance: Lessons from World Bank Experience, Washington, D.C. (available on http://econ.worldbank.org/) 29 30