WPS7069 Policy Research Working Paper 7069 The Second Wave of Independence Shopping for Solutions Marie M. Gaarder Ulrich Bartsch Independent Evaluation Group Public Sector Evaluation Department October 2014 Policy Research Working Paper 7069 Abstract In the 21st century, many developing countries will become the right outcomes, but also take away incentives to learn emerging markets and will no longer be in need of the about outcomes. A fundamental rethink of responsi- carrot-and-stick approach to development assistance most bilities and accountabilities in the development business prevalent today: development financing made available would make sure that development outcomes are traded conditional on certain policies and interventions. This in the development marketplace. It would put recipient paper suggests that interactions between development agen- countries in charge of contracting development agencies cies and recipient governments are mostly about inputs to provide these outcomes. Development agencies would deemed (but not known) to contribute to improvements commit to and be held financially accountable for out- in living standards in recipient countries, rather than out- comes, that is, real improvements in welfare indicators. comes. The paper argues that the development marketplace The paper describes the role of the evaluation function is beset by market imperfections because of externalities, in aligning incentives with the ultimate goal of improv- principal-agent problems, and decision making under ing lives and provides examples of emerging solutions. uncertainty, which not only make it difficult to achieve This paper is a product of the Public Sector Evaluation Department, Independent Evaluation Group. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at mgaarder@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 The Second Wave of Independence: Shopping for Solutions Marie M. Gaarder and Ulrich Bartsch 1 Keywords: market failure; principal–agent; development outcomes; evaluation function; externalities JEL Classification: D80, D86, D43, D62, F34, F35, O22 1 Gaarder: Independent Evaluation Group (IEG), World Bank, mgaarder@worldbank.org; Bartsch: GMFDR - Macro & Fiscal Mgmt, World Bank, ubartsch@worldbank.org. Acknowledgements: The authors thank Howard White, Vibecke Dixon, Cheryl Gray, Emanuel Jimenez, and Soniya Carvalho for comments on previous drafts of this paper. 1. Salient Features of the Development Business In the sunset days of the British Empire and at the end of the myth of European superiority following the calamity of World War II, a new branch of the service sector was born: the development business. Colonial administrations and missionaries bringing ‘enlightenment’ gave way to independent governments in erstwhile colonies and a new army of bankers and social scientists arrived. They believed that progress in living standards could be accelerated with money and knowledge from the ‘North’, because market failures would otherwise slow down growth. The ‘demand’ came from developing country governments who, realizing that the private sector within their countries underprovided the desired outcomes, tried to make up for it by intervening and, because they lacked financing, then sought development assistance. Development agencies were created with the stated objective to improve the welfare of populations in developing countries, as measured by development outcomes. 2 Thus, they promised, say, improving the quality of education, road safety, or agricultural incomes among poor farmers. Yet, 70 years hence, not all is well with the development business. The fundamental problem with the development business, this paper suggests, is that the actual interactions between development agencies and their counterparts in developing countries relate to negotiations around the inputs and outputs (e.g. schools built; teachers trained), not outcomes (e.g. an improved literacy rate; a decrease in the repetition rate). The parties involved in the exchange are most often not directly benefiting from the provision of the outcome good, which is intended to help 'the poor'. While much money and brain power has been expended on inputs, such as bricks-and-mortar investments and smart laws and regulations, all too often the outcomes fall far short of expectations. Thus, schools have been built that see neither students nor teachers, roads have been built that lead to nowhere and there is often a wide gap between de jure and de facto rules. Even worse, notwithstanding decades of experience, there is a disconcerting lack of knowledge about what works in development, and what does not. 2 With development agencies we mean single-government institutions such as the Department for International Development in the UK, Norad in Norway, etc., and multilateral development banks, including the World Bank and the Inter-American Development Bank. 2 In addition to the stated objectives of improving living standards in recipient countries, there may be a myriad additional objectives, mostly unstated, driving the activities of development agencies, such as being seen as doing good (the 'warm glow' benefit), promoting national interests (political and economic), and staying in business (the 'development' business). Thus, giving and being seen as doing so sometimes seem to be at least as important as achieving improved welfare of the intended recipients of assistance, which means ‘the shoppers’ in the development market place care more about being seen shopping than about what ends up in the cart. ‘After decades in which development agencies have disbursed billions of dollars for social programs, and developing country governments and nongovernmental organizations (NGOs) have spent hundreds of billions more, it is deeply disappointing to recognize that we know relatively little about the net impact of most of these social programs.’ (CGD, 2006, p.1) 3 At fault is not our ability to acquire knowledge, but a lack of trying to do so. Appropriate monitoring and evaluation systems and experimental designs would give us the knowledge to link development interventions with outcomes, and evaluate the outcomes after the project finished. However, these systems and designs are missing with surprising regularity. A recent report commissioned by the Evaluation Department in Norad (2014) concluded that ‘implementation of a results-focus fails to ensure evaluability, partly because there is little clarity about minimum standards, but also pressures of time on staff, low priority by senior managers and a lack of incentives to prioritize results’. Most development agencies produce reports about implementation, completion and results (ICR) at the time of project closure. However, these may not be sufficient to reliably link interventions to outcomes. Impact evaluations with robust methodologies can help to establish such links, but they are not routinely provided. For example, the World Bank’s Independent Evaluation Group (IEG) recently found that only 17% percent of recently-evaluated operations cited impact evaluations in the ICR (IEG, 2014). In many cases where no comparison or control groups exist, alternative factors outside of the intervention that may have affected the outcomes are not discussed or considered. So for example, a road infrastructure program may report on decreased travel-time on a certain stretch of road after its rehabilitation, without looking into what 3 As a direct result of this report, the International Initiative for Impact Evaluation, 3ie, was created devoted to ‘enhancing development effectiveness through supporting the production and use of evidence from rigorous impact studies’ (Gaarder and White, 2009, p.2). 3 happened to fuel prices and traffic density over the same period. 4 The ICR’s ability to report satisfactorily on outcomes is to a large extent dependent on the quality of the monitoring and evaluation (M&E) activities built into the projects. The evaluation of learning in World Bank operations found that only 23% of a sample of recently closed projects had established good M&E systems. 5 This seeming lack of attention to M&E should not be taken to indicate a lack of commitment on the part of the staff of development agencies to pursue their stated goals. Most of them are deeply committed to improving lives in developing countries. Another explanation therefore needs to be found why we know so little about the goods sold and bought in the ‘development market place’. This paper suggests that the salient features of the development business itself make it susceptible to market imperfections and misaligned incentives. • First, the actual interactions between development agencies and their counterparts in developing countries relate to negotiations around the inputs and outputs (e.g. schools built; teachers trained), not outcomes (e.g. an improved literacy rate; a decrease in the repetition rate). The parties involved in the exchange are most often not directly benefiting from the provision of the outcome good, which is intended to help 'the poor'. In other words, the 'outcome good' that is the ultimate purpose of the engagement is not part of the transaction. If it ensues, it can be treated as a positive externality, and transactions can and will go ahead entirely oblivious of this part of their ultimate outcomes, as economic theory tells us. • Second, the interests, incentives and actions of the development agency and the recipient country, respectively, are not perfectly known and observable to the other party. Several additional layers exist in this principal-agent problem: the recipient country’s government may or may not act in the best interest of ‘the poor’, who have incomplete information about the extent to which their representatives do what they are supposed to. Likewise, principal-agent relationships exist in development agencies between their boards of directors, management, and staff. Principal-agent 4 If fuel prices go up some people may decide to use public transport, thereby leading to a decrease in congestion and travel time even without any road improvements. 5 World Bank, 2014. 4 problems lead to misalignment of incentives, incomplete contracts, and suboptimal outcomes. • Third, the extent to which inputs and outputs will deliver the outcome good is marred in uncertainty as the theory of change, or the ‘production functions’ through which inputs and outputs result in outcomes, are uncertain, and evidence on effectiveness of particular interventions is scarce and context specific where it exists. In addition, it usually takes years before it can be known whether the interventions actually yielded the intended outcomes, and only if someone has made the effort early-on to put in place an effective system to monitor and evaluate them. Decision making under uncertainty, with unclear assignation of accountability, will lead to inefficient provision of goods, as the literature tells us. 2. Market Imperfection Theories and Development Outcomes What does the economics literature say about market imperfections? Market failure in economic theory refers to a situation in which the allocation of goods and services by a free market is not efficient, implying that a better overall allocation could exist. This type of situation is often associated with public goods, externalities, principal-agent problems, information asymmetries, and uncertainty. 6 The existence of market imperfections is often the reason for government interventions to improve outcomes. And although the dealings between development agencies and governments currently do not display many of the features we are accustomed to seeing in markets, we will in the next sections discuss three market failure concepts which help us in understanding and possibly overcoming the shortcomings highlighted above: externalities, agency theory, and uncertainty. 2.1 Externalities An externality could be defined, for example, as in Rothengatter (1994): ‘an externality is a relevant cost or benefit that individuals fail to consider when making rational decisions’ (p. 361). 6 Pigou, A. 1920; Samuelson, P. 1954; Akerlof, G. 1970; Spence, A. 1973; Stiglitz, J. 1976, 1987. 5 Thus, in addition to the direct and willing parties to transactions, there are also unwilling or unknowing parties. While we are more familiar with negative externalities such as air pollution from cars affecting people's health or carbon emissions from productive activities affecting the global climate, an externality can just as well be of a positive nature such as the effect of a well- educated labor force on a company’s productivity, or the protection against infectious disease provided to someone living amongst people with up to date vaccinations. Development outcomes have a number of features that could lead us to viewing them as externalities. First, the parties involved in the exchange are most often not directly benefiting from the provision of the outcome good, which are intended to benefit 'the poor'. In the case of a loan from a development bank or official development aid (ODA) from a bilateral agency, the typical transfer of funds is conditional on a negotiated set of activities (project components) and implementation arrangements, including procurement and fiduciary rules, and lines of responsibility in implementation, with the objective of producing outcomes, such as ‘increasing the productivity of irrigated agriculture’, ‘reducing road transport costs’, or ‘raising the incomes of communities affected by resource degradation’. Whether or not these outcomes actually ensue will usually not influence in any way the contract terms, interest and principal payments (in the case of loans) or indeed future lending or aid negotiations with the same country and sector. Loans will have to be repaid, whether or not agriculture is more productive, and the next set of activities will be planned with a bilateral donor whether or not incomes of previously targeted communities are now higher than before or not. In other words, outcomes are not part of the transaction in the development business. While the term ’externality’ is often used for unintended effects, we argue here that the development business actually treats ostensibly intended outcomes as unintended because they often remain outside of the transaction and unmeasured, and because achieving these outcomes or not has little implications for the parties involved in the transaction. How do we internalize the externalities? By creating a market for said ‘good’ or ‘bad’ with appropriate regulation. Key to effective regulation is reliable measurement of the externality, putting an economic value on it, and establishing an enforcement mechanism. For example, a well-known response to the negative externality of CO2 emissions, which lead to climate change, is the creation of a market in which carbon emission permits can be traded. The market 6 for emissions works because they are measurable (tons of CO2 produced from burning fossil fuel), and governments have created tradable permits and an enforcement mechanism. The market scheme puts a price on emissions, and the externality created by these emissions is thus internalized in production and consumption decisions. 7 In a similar fashion, we could imagine internalizing development outcomes, such as a decrease in under-five mortality by 10% in a certain region in, say, Tanzania by a certain date. A donor agency would commit to achieving the outcome, and could tender out the delivery of the outcome to implementing agencies to achieve it at least cost. The role of evaluation clearly would be crucial: it would measure and verify the achievement of the outcome target indicator, and payment would be effected upon achievement of results. Some variants of such market arrangements already exist and will be discussed further in Section 3. 2.2 Principal-Agent Problems The principal-agent relationship exists when one entity (an 'agent') makes decisions on behalf of another entity (the 'principal'). In many situations, conflicting objectives and decentralized information are the two basic ingredients of the principal-agent problem. 8 In this case, the principal cannot directly observe that the agent is always acting in her best interests. This is especially problematic when activities that are useful to the principal are costly to the agent, and observing the agent’s activities is costly for the principal. If the agent has different preferences than the principal but his/her efforts are perfectly observable and monitorable, the principal will propose a contract which perfectly controls the agent to act in the principal’s best interest. The incentive problem disappears. While applications of agency theory have typically assumed outcomes to be observable with specific preference representations for the principal and agent, there are many situations in which outcomes are not easily observable. In the same way that the shareholders of a firm typically do not know the impact on their future returns of a manager’s action, so too is the impact of development professional’s actions obfuscated by many other intervening factors. These factors 7 Bartsch, 2000. 8 Spence & Zeckhauser 1971, Holmstrom 1979, Shavell 1979, Stiglitz 1974, Mirrlees 1975, Holmstrom 1979, Shavell 1979, Rees 1985a,b. 7 can include the uncertainty around the way the actions will convert into development outcomes, also known as the theory of change, 9 as well as the difficulties in measuring outcomes. This makes it next to impossible to write the optimal contract as a function of outcomes and contracts are instead written on inputs and surrogate measures of outcomes. To make matters worse, the principal-agent problem is actually multi-layered in the development business. It occurs in the relationship between the development agency and the recipient (or partner) country, and in the relationship between a recipient government and the people it is purported to represent. Then there are the board of directors, management, and team leaders in the development agency, which have principal-agent relations as well. On the partner country side, there is the government, the implementing agencies, and the ultimate ‘beneficiaries’ which again can create principal-agent problems at various levels. 10 Evaluation is again crucial in establishing the empirical basis for contracts that overcome principal-agent problems. Some variants of contracts intended to incentivize behaviors and processes that are believed to be conducive to outcomes (surrogate measures) will be further discussed in Section 3. 2.3 Decision Making under Uncertainty of Outcomes While known factors are already reflected in efficient market prices, the main sources of market instability are unknown factors - also known as market uncertainty. If development agencies’ survival depended on the developmental return on their investments rather than on the repayment of loans and on capital replenishments, they would shun projects (read ‘assets’) with a higher degree of uncertainty of outcomes. The agencies would have to treat exposure to these assets with a higher degree of caution, which in turn would limit the growth of these assets. In financial markets, institutions with higher exposure to market uncertainty face a higher cost of credit and other penalties, which limit their ability to become overexposed to these assets. Development agencies currently are not exposed to the good or bad outcomes of their interventions, because recipient countries have to repay loans even when development projects 9 For further insights about theory of change and theory-based evaluations, refer to White (2009). 10 Martens et al., 2001. 8 fail to produce the desired outcomes, or politicians replenish grant funds disregarding outcomes, because of the ‘warm glow’ of doing so. Again, outcomes are externalities, which are not taken into account adequately when designing interventions. Instead of development agencies, recipient countries are exposed to the risks of interventions. They have to repay loans, or face criticism over the use of ODA when outcomes do not materialize, or, possibly, make matters worse, actually, rather than improving the lives of people. Such a lop-sided allocation of exposure to risks leads to inefficient decisions about development interventions. 3. Internalizing Development Outcomes Possible solutions to the three-pronged problem discussed above – development outcomes are externalities, the development business is beset by principal-agent problems, and decisions have to be made under a high degree of uncertainty – involve creating markets for outcomes, or creating the right incentives for behaviors and processes that are believed to be conducive to outcomes. 3.1 Developing Solutions: Bringing Outcomes into the Contracts in the Development Business The solution for better taking development outcomes into account is to make sure that development outcomes form an integral part of the contracts in the development business and make development agencies financially accountable for these outcomes. This would turn the usual donor-recipient relationship on its head: the country would contract the development agency to produce development outcomes. This would obviously only work in countries with governments that have the ambition and capacity to design strategies to promote improvements in living conditions of the poorer parts of their societies. In the case of development loans, the country would pay back the loan and interest in full only if the outcomes were achieved. In the case of development grants rather than loans, let them be counted as official development aid (ODA) only if promised outcomes were achieved, and no ‘warm glow’ shine otherwise. Arguably, if aid did not ‘aid’ it should not be counted as aid. So for example, the funds intended for reductions in mother and child morbidity and mortality in Norway’s support to a flagship health program in India, the Norway India Partnership Initiative (NIPI), should be counted as ODA only if the reductions actually ensued and were measured. The funds that went into the 9 program would be subtracted from Norway’s statistics of ODA if results were either falling short or were not measurable (incidentally, this was the conclusion in the recent evaluation of said program). 11 Norway’s claim to be one of the most generous donors among high-income countries, with its 1% of GDP annual budget allocation to ODA, would be in jeopardy. The Norwegian aid administration would then obviously come under pressure to care much more about outcomes, and internalize the externality. Once outcomes are internalized into the development business, market-based contracting becomes possible, just as emissions trading becomes possible with the creation of tradable permits and a constraint on their availability. Similar to emissions trading, which is expected to produce desired emission cuts at lower costs through spurring competition and innovation, market forces in the development business could be used to produce superior outcomes at lower cost. To open up the possibility for a market solution, let the ‘purchaser’ (the partner country in question) of the desired outcome open a tender for turn-key solutions (e.g. improve maternal health by x percent), and let implementing agencies or development institutions bid. The country chooses the best bid, and pays upon independent verification of results. Development agencies would have to develop credible profiles by showing their successes in similar programs in order to win bids. And while it may be counterintuitive at first – why should they bid for being contracted to finance development activities? – they would be compelled to bid, because, after all, it is their mandate to engage with recipient countries, and they want to give, even if only for the ‘warm glow’. When development outcomes are internalized in the development business, risks will also be taken into account more explicitly. Risk-reward relationships are most explicitly modeled in contracts between oil and gas companies and governments in resource-rich countries. The companies invest large sums before knowing the extent to which hydrocarbons found underground can be monetized. Governments, on the other hand, want to make sure that the resources they suspect are within their jurisdiction are developed in the most efficient way possible. Contracts therefore specify the pace of exploration, and have detailed formulae by which the returns on the investment are shared between developers and host country. Sliding scale royalties, for example, allow investors a higher share of rewards early-on to repay their 11 Norad, 2013. 10 investment outlays, but give host countries a larger share of the resource wealth once the initial investment has paid off. Similarly, varying production shares increase the return for investors when resource finds are small, but limit the upside when they are large. 12 In the development business, risks are overwhelmingly left with recipient countries. In the worst case, recipient countries are suffering from unintended negative effects of development projects, while their governments pay loan interest and principal for a decade or more. Even when investment projects are delayed for years because of unforeseen complications, country governments pay a commitment fee, because the development agency has tied its funds to a particular project and does not have them available for other projects anymore, although they are still in the agency’s accounts. The risks of intended or unintended outcomes materializing lie with recipient countries. Turning the principal-agent relationship around, and making development agencies accountable for outcomes changes the assignation of risks. They will henceforth devote greater efforts to understand and mitigate those risks, and invest in better M&E and learning systems. However, ways will have to be found to share risks: leaving them entirely with development agencies could result in overly risk-averse development efforts, and high-risk, high-return activities that could really change lives could remain unexplored. Furthermore, some have argued that the off-setting incentives on the recipient country – on the one hand the positive incentive of getting the desired results (and the political goodwill that comes with it), on the other hand a rebate if it does not happen (this would only be in the case of loans) – could lead to a sub-optimal investment on their part in the actual M&E efforts on the ground, over which they are ultimately in control. 13 As in the oil and gas business, contracts have to be designed with risk-reward sharing in mind. 3.2 Existing Approaches A plethora of recent initiatives bring outcomes directly into the transaction between development partners (Cash On Delivery, COD; Program for Results, P4R, etc.). They allow setting up 12 Bartsch, 1998, 2005. 13 We do not find the argument very persuasive, as the loan was taken up in the first place to address a development issue. The fact that the loan becomes ‘cheaper’ does not mean that it carries zero costs, and hence the desired development outcome remains the purpose of incurring the cost. 11 contracts that explicitly spell out the desired outcomes against which payments will be made— such as increased primary school completion rates, vaccination coverage, or access to clean water. Nancy Birdsall, President of the Center for Global Development (CGD), explains that this approach provides incentives for governments to identify problems and design and implement locally appropriate solutions. 14 However, these are recent efforts to circumvent the informational and observational challenges, but they keep the usual principal-agent relationship between development agencies and recipient countries untouched: the buyer of the outcome (the principal) is the development institution and the agent providing it is the country. The agent gets paid against achievements of outcomes, which is an innovation over the current situation, where aid or loans are given against deployment of inputs. On the one hand, this may appear empowering the country to find the most appropriate solution to a development problem, because the aid/loan transaction does not anymore specify which inputs are to be used. On the other hand, it still uses conditionality to achieve desired results. The conditioned cash implies, rather paternalistically according to some, that the development agencies value outcomes more than their government counterparts in the client countries 15, but it is an acknowledgement that after almost 70 years of development assistance the development agencies have little in terms of knowledge of what works and under what circumstances, and therefore leave it to the recipients to decide the best manner in which to reach the outcomes. Whether the recipient country becomes the agent responsible for providing specified outcomes, or the development agency, it is clear that incentives align between the country and the development agency to put in place an M&E system to verify the achievement of the desired outcomes. It also needs to be able to give clear insights into how much it should cost to obtain certain development outcomes based on previous evaluations in similar circumstances, and hence be able to calibrate ex ante the payments for the expected outcomes. 4. The Role of the Evaluation Function While the development agencies’ mandate is to improve development outcomes, the evaluation functions in these organizations have been established to better demonstrate the effects of aid and 14 CGD, 2010. 15 Gaarder, 2012. 12 to report on them. It is worth distinguishing between the internal and the independent evaluation functions: the former focuses mainly on strengthening the agencies’ internal ability to demonstrate the effects of aid by building evaluation capacity and evaluation frameworks into individual projects. The latter focuses mainly on reporting objectively and ex post on the achievement of development outcomes and the agencies’ ability to produce these efficiently (what is known as the accountability-focus). In the traditional development business model, formal evaluation serves the function of remedying the ‘broken feedback loop’ that exists between the beneficiaries of the interventions and the donors. Peter Murrell (as quoted in Martens et al. 2001) suggests this is necessary to overcome moral hazard on behalf of aid suppliers who are seeking business opportunities rather than (or in addition to) development objectives. Nevertheless, all too often the lack of focus on outcomes and on the evaluability of projects during design and implementation leaves the independent evaluation functions with little data to analyze and report on in their ex post evaluations. In the model where outcomes form an explicit part of the contract, the evaluation function should play a key role to reliably verify outcomes. The internal evaluation function would be responsible for ensuring that sufficient capacity is in place, or otherwise for building capacity in monitoring and evaluation. The independent evaluation function could be an important arbitrator of the M&E system and the validity of what it produces. The need for independent verification of results has been strongly emphasized, for example by CGD. 16 Truly independent verification would require certification by outside actors; see as an example the way auditing firms certify company balance sheets. Nevertheless, in order to fully achieve the internalization of development outcomes in the development business contracts, the feedback loop needs not only to bridge the spacial separation between beneficiaries and donors by focusing on outcomes but also the temporal separation between causes (activities) and their effects (outcomes). For this to happen, incentives need to be in place for ensuring evaluability of all development interventions. In a similar fashion that the internal evaluation function is in charge of overseeing and ensuring that project 16 Birdsall and Savedoff, 2010. 13 reviews take place upon project completion in a number of development agencies, and the independent evaluation function is there to assure the objectivity and soundness of the assessments, so too can they divide up the roles of ensuring evaluability at entry. This ex ante review has two aspects: first, it overcomes the time-inconsistency of incentives, when relevant staff is not held accountable for the outcomes several years down the line; and second, it increases the chances of knowing the development outcome of the project at the end of it. Both the Inter-American Development Bank and the UK Department for International Development have recently introduced evaluability as a criterion for project approval. Whether it also increases the likelihood of the projects achieving their development outcomes is a question that remains to be further evaluated. 17 5. Conclusions In the 21st century, developing countries became known as Emerging Markets, and increasingly rise out of their dependency on knowledge from the ‘North’ and its charity. This second wave of independence changes the power structure and will lead to a rethink of interactions in the development marketplace: interactions between development agencies and recipient country governments will no longer be negotiations around the inputs and outputs for development, but instead will focus on the ultimate purpose of the engagement, the improvement of living standards. The paper suggests improvements in development outcomes and our knowledge about the links between inputs and outcomes, will be achieved by making outcomes integral parts of the contracts in the development business, and making development agencies (more) accountable for these outcomes. The proposal is to turn the usual donor-recipient relationship on its head. In the case of development loans, the country would pay back the loan and interest in full only if the outcomes were achieved, and in the case of development grants, the official development aid (ODA) would only be counted as such if promised outcomes were achieved. Once outcomes are internalized into the development business in this way, we suggest that market-based contracting becomes possible, which could produce superior outcomes at lower 17 Gaarder and Bartsch, 2014 (forthcoming); Gray, 2014 (forthcoming); OVE, 2010, 2013; IDB, 2010; DFID, 2011, 2014 14 cost. The ‘purchaser’ (the partner country in question) of the desired outcome would open a tender for turn-key solutions (e.g. improve maternal health by x percent), and let development agencies bid. They would have to develop credible profiles by showing their successes in similar programs in order to win bids, and thereby would have a strong incentive to build strong M&E systems to be able to do so. When development outcomes are internalized in the development business and development agencies are made more accountable for results, risks will also be taken into account more explicitly and shared more appropriately. In the traditional development business, risks are overwhelmingly left with recipient countries. In the worst case, recipient countries are suffering from unintended negative effects of development projects, while their governments pay loan interest and principal for a decade or more. With development agencies also accountable for outcomes they will devote greater efforts to understand and mitigate those risks, and invest in better M&E systems. What role can the evaluation functions play in aligning incentives and helping the process of internalizing development outcomes into the development market? In the traditional development business model where outcomes are treated as externalities, formal evaluation serves the function of remedying the ‘broken feedback loop’ that exists between the beneficiaries of the interventions and the donors. However, the evaluation function can also play a key role in supporting the creation of a market for development outcomes. The internal evaluation function would be responsible for ensuring that sufficient capacity is in place, or otherwise for building capacity in monitoring and evaluation, while the independent evaluation function would be an important arbitrator of the M&E system and the validity of what it produces and in helping identify and provide evidence on promising solutions and their costs. In the future, governments in Emerging Markets will have a choice between several development agencies, which proffer evaluations of their previous efforts as evidence that they can provide solutions the governments want. They will no longer be able to be the shoppers who care more about being seen shopping than about what ends up in their cart. 15 References Akerlof, G. A. 1970. “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism.” The Quarterly Journal of Economics 84 (3): 488–500. doi:10.2307/1879431. Bartsch, U. 1998. Financial Risks and Rewards in LNG Projects: Qatar, Oman, and Yemen. Oxford, UK: Oxford Institute for Energy Studies, NG3. Bartsch, U. and B. Müller. 2000. Fossil Fuels in a Changing Climate. Oxford Institute for Energy Studies, Oxford, UK: Oxford University Press. Birdsall, N., and W. Savedoff. 2010. Cash on Delivery: A New Approach to Foreign Aid. Washington, DC: Center for Global Development. CGD. 2006. When Will We Ever Learn? Improving Lives through Impact Evaluation. Report of theEvaluation Gap Working Group. Washington, DC: Center for Global Development. CGD. 2010. Nancy Birsall on Cash on Delivery Aid. Accessed September 9, 2014. http://international.cgdev.org/blog/nancy-birdsall-cashdelivery-aid DFID. 2011. “Writing a Business Case, How To Note.” A DFID practice paper. London, UK: Department for International Development. DFID. 2014. Review of Embedding Evaluation in the Department for International Development; Final Report. London, UK: Department for International Development. Gaarder, M. and U. Bartsch. Forthcoming 2014. “Who Cares About Development Outcomes? Market Failures and the Role of the Evaluation Function”. Journal of Development Effectiveness 6 (4). Gaarder, M. 2012. “Conditional versus unconditional cash: a commentary.” Journal of Development Effectiveness 4 (1), 130-133. Gaarder, M. M. and B. Briceño. 2010. “Institutionalisation of Government Evaluation: Balancing Trade- Offs.” Journal of Development Effectiveness 2 (3). doi:10.1080/19439342.2010.505027. Gaarder, M. M. and H. White. 2009. “The International Initiative for Impact Evaluation (3ie): An Introduction.” Journal of Development Effectiveness 1 (3). doi:10.1080/19439340903147685. Gray, C. Forthcoming 2014. “Finding Out What Works: Tracking Results in the Inter-American Development Bank”. Journal of Development Effectiveness 6 (4). Holmstrom, B. 1979. “Moral Hazard and Observability.” The Bell Journal of Economics 10: 74–91. doi:10.2307/3003320. 16 IDB. 2010. Report on the Ninth General Increase in the Resources of the Inter-American Development Bank. Washington, DC: Inter-American Development Bank. IEG. 2014. Learning and Results in World Bank Operations. The World Bank Group. Martens, B., U. Mummert, P. Murrell, and P. Seabright. 2001. The Institutional Economics of Foreign Aid. Cambridge: Cambridge University Press. Mirrlees, J. 1999. “The Theory of Moral Hazard and Unobservable Behaviour: Part I.” The Review of Economic Studies 66 (1): 3–21. Special Issue: Contracts (Jan.). doi:10.1111/1467-937X.00075. Norad. 2013. Evaluation of the Norway India Partnership Initiative for Maternal and Child Health. Evaluation Report 2013/3. Oslo, Norway: Norwegian Agency for Development Cooperation. Norad. 2014. Can We Demonstrate the Difference that Norwegian Aid Makes? Evaluation Report 2014/1. Oslo, Norway: Norwegian Agency for Development Cooperation. OVE. 2010. Ten Years of Evaluability at the IDB, Presentation at the OECD. Washington, DC: Inter- American Development Bank. OVE. 2013. “Mid-Term Evaluation of IDB-9 Commitments.” The Development EffectivenessFramework and the Development Effectiveness Overview. Background paper, Inter-American Development Bank, March. Pigou, A. C. 1920. The Economics of Welfare. London: Macmillan. 755 Rees, R. 1985a. “The Theory of Principal and Agent—Part I.” Bulletin of Economic Research 37(1): 3– 26. doi:10.1111/j.1467-8586.1985.tb00179.x. Rees, R. 1985b. “The Theory of Principal and Agent –Part II.” Bulletin of Economic Research 37(2): 75– 97. doi:10.1111/j.1467-8586.1985.tb00185.x. Rothengatter, W. 1994. “Do External Benefits Compensate for External Costs of Transport.” Transportation Research 28 (4): 321–328. Samuelson, P. 1954. “The Pure Theory of Public Expenditure.” The Review of Economics and Statistics 36 (4): 387–389. doi:10.2307/1925895. Shavell, S. 1979. “Risk Sharing and Incentives in the Principal and Agent Relationship.” The Bell Journal of Economics 10: 55–73. doi:10.2307/3003319. Slovik, P., 2010, Market uncertainty and market instability, IFC Bulletin No 34. Basel, Switzerland: Bank for International Settlement. 17 Spence, A. M. 1973. “Job Market Signaling.” The Quarterly Journal of Economics 87 (3): 355–374. doi:10.2307/1882010. Spence, A. M., and R. Zeckhauser. 1971. “Insurance, Information, and Individual Action, Am.” Economics Reviews 61: 380–387. Stiglitz, J. 1974. “Incentives and Risk Sharing in Sharecropping.” The Review of Economic Studies 41: 219–255. doi:10.2307/2296714. Stiglitz, J. 1976. “The Efficiency Wage Hypothesis, Surplus Labour, and the Distribution of Income in L.D.C.s.” Oxford Economic Papers, New Series 28 (2), 185–207. Stiglitz, J. 1987. “Principal and agent.” The New Palgrave: A Dictionary of Economics 3: 966–971. White, H. 2009. “Theory-Based Impact Evaluation: Principles and Practice.” Journal of Development Effectiveness 1 (3). doi:10.1080/19439340903114628. 18