Research Topic 1

Journal of Economic Perspectives—Volume 30, Number 1—Winter 2016—Pages 77–94

P oor people find it harder to save, so poor countries can find it difficult to finance needed investments from domestic savings alone. In an ideal world, this would not be a problem; capital would flow from high-income capital-rich countries to low-income capital-poor countries, because the marginal return should be higher in countries where capital is relatively scarce. But that was not what people saw happening in the world 70 years ago. In the years just after World War II, global capital markets were thin and not trusted as a source of finance. It seemed that new institutions were needed.

In response, delegates from 44 countries met in 1944 at a hotel in Bretton Woods, New Hampshire, and agreed to create the International Monetary Fund (IMF) and the International Bank for Reconstruction and Development (IBRD). The latter is a core component of what came to be known as the World Bank Group, or more often the World Bank. The IMF was charged with managing imbalances of payments to avoid destabilizing currency devaluations, while the World Bank was to be the channel for longer-term development finance.

Much has changed since then. There have been prominent calls for radically reforming the World Bank, or even closing it. Two main concerns have been raised by the Bank’s critics. The first is that the Bank’s efforts are largely wasted because poor countries face nonfinancial constraints that limit their development. The second is

The World Bank: Why It Is Still Needed and Why It Still Disappoints

■ Martin Ravallion holds the inaugural Edmond D. Villani Chair of Economics, Georgetown University, Washington, DC. Prior to joining Georgetown in December 2012, he worked for the World Bank for 24 years, serving most recently as the director of its research department. His email address is [email protected]. † For supplementary materials such as appendices, datasets, and author disclosure statements, see the article page at http://dx.doi.org/10.1257/jep.30.1.77 doi=10.1257/jep.30.1.77

Martin Ravallion

j_ravallion_301.indd 77 1/20/16 6:54 AM

78 Journal of Economic Perspectives

that global financial markets are no longer thin and can now serve the Bank’s original role. In 1945, the global stock of international investments (measured by asset values) represented 5 percent of world GDP, while 50 years later it had risen to 62 percent (Obstfeld and Taylor 2004). Today, developing countries turn often to the private sector to finance investment; World Bank lending in 2012 represented only about 5 percent of the aggregate private capital flows to developing countries.

Does the World Bank still have an important role to play? How might it fulfill that role? The paper begins with a brief account of how the Bank works. It then argues that, while the Bank is no longer the primary conduit for capital from high-income to low-income countries, it still has an important role in supplying the public good of development knowledge—a role that is no less pressing today than ever. This argument is not a new one. In 1996, the Bank’s President at the time, James D. Wolfensohn (1996), laid out a vision for the “knowledge bank,” an implicit counterpoint to what can be called the “lending bank.” A knowledge bank might serve a number of functions. It can be a broker that taps into existing knowledge and redirects it to needy clients (which was the role emphasized by Wolfensohn). But this vision is rather limited. There is also the task of identifying pressing knowl- edge gaps—our key areas of ignorance constraining development—and filling those gaps.

The paper argues that the past rhetoric of the “knowledge bank” has not matched the reality. An institution such as the World Bank—explicitly committed to global poverty reduction—should be more heavily invested in knowing what is needed in its client countries as well as in international coordination. It should be consistently arguing for well-informed pro-poor policies in its member countries, tailored to the needs of each country, even when such policies are unpopular with the powers-that-be. It should also be using its financial weight, combined with its analytic and convening powers, to support global public goods. In all this, there is a continuing role for lending, but it must be driven by knowledge—both in terms of what gets done and how it is geared to learning. The paper argues that the Bank disappoints in these tasks but that it could perform better.

How the World Bank Functions

The World Bank currently has 188 member countries, and it employs over 12,000 staff working from 120 offices globally. The Bank is divided into five groups, which together disbursed $44 billion in 2014. Two of the groups are focused on lending and aid. The International Bank for Reconstruction and Develop- ment (IBRD) is the original World Bank institution. It primarily makes loans to middle-income countries and made $19 billion in loans in 2014. The Interna- tional Development Association (IDA) provides grants and loans at favorable terms targeted to low-income countries; it disbursed $13 billion in 2014. The Bank’s cumulative lending (IBRD + IDA) between 1945 and 2011 was $788 billion, spread over about 180 countries of which the largest share went to India (11.3 percent),

j_ravallion_301.indd 78 1/20/16 6:54 AM

Martin Ravallion 79

followed by Mexico (6.5 percent), Brazil (6.3 percent), China (6.3 percent), and Indonesia (5.5 percent).

The other three members of the World Bank Group are more focused on directly encouraging private-sector activity. The International Finance Corporation (IFC) lends to private institutions and disbursed $9 billion in 2014. (IFC profits also support IDA.) The Multilateral Investment Guarantee Agency (MIGA) focuses on insurance and credit guarantees for private investors. The International Center for the Settlement of Investment Disputes (ICSID) is a forum for disputes between investors and governments; use of the ICSID process is written into many interna- tional investment treaties, domestic investment laws, and specific contracts.

The World Bank raises money in various ways. Most of its income comes from lending the Bank’s own capital, which includes both funds accumulated over time and funds paid in by the member countries. The Bank can sell AAA-rated bonds in the global financial markets, thanks to its conservative lending policies relative to its capital. It can then re-lend these funds at higher interest rates through the IBRD. For the low-interest loans and grants made through IDA, 40 donor coun- tries contribute funds triennially. The Bank also receives some funds from donors for administering their aid and from client countries for reimbursable services. Finally, there are short-term trust funds, which the Bank manages on behalf of other nonprofit agencies.

Formally, the World Bank is run by a Board of Governors, with representatives from all the member countries, meeting annually. The Executive Board (here- after “the Board”) meets regularly and comprises representatives appointed by the six largest shareholders—currently China, France, Germany, Japan, the United Kingdom, and the United States—plus 19 members each representing groups of countries. Membership entails a minimum weight in voting, which then rises according to ownership of the Bank’s capital stock. The Bank President presides over the World Bank Group and chairs the Board. Reforms in 2010 increased the voting rights of borrowing countries, notably (but not only) China, which is now the third-ranked in IBRD votes after the United States and Japan.

The Bank’s lending operations have long been organized around country teams, each led by a country manager/director. The countries are assigned to six regional groupings, each with its Vice President. This country-based model is backed up by some cross-cutting central units. For example, there are sectoral support units now called Global Practices, which provide specialized expertise and project lending in agriculture, education, energy, health, the environment, transportation, and other areas. The Development Economics Vice Presidency (DEC) is the chief research arm of the World Bank, led by a Chief Economist who reports directly to the President. There is also an Independent Evaluation Group (IEG), which has the task of evaluating World Bank lending and projects in both the public and private sectors, and reports to the Board.

The World Bank is not the only international development bank. The three largest regional development banks—the Inter-American Development Bank, the Asian Development Bank, and the African Development Bank—have expanded

j_ravallion_301.indd 79 1/20/16 6:54 AM

80 Journal of Economic Perspectives

operations since the 1960s. They are collectively still smaller than the World Bank, but there is clearly a degree of competition (as discussed in Kanbur 2003). Recently, China has taken the lead in establishing a new Asian Infrastructure Investment Bank. The New Development Bank has also been created by the BRICSs: Brazil, Russia, India, China, and South Africa. It appears that expanding private finance is not displacing public development finance globally, but instead both are expanding.

Why the World Bank is Still Needed

While developing countries have greatly improved their access to global capital markets, private capital flows have tended to be selective, not reaching all coun- tries and sectors. The Bank has a role in facilitating private finance when needed. But underdevelopment is not only due to a lack of external finance; it has deeper causes in poor policy-making and governance in developing countries—in short, their political economy.

In making the case for the World Bank today, one cannot simply point to unfunded projects. One must explain how the Bank’s lending or aid addresses the reasons why such projects are not already funded. That requires the Bank to be a credible knowledge leader.

Why a Development Bank? Capital markets encounter persistent problems of uninsured risk (including

from asymmetric information), externalities, and contract enforcement. Private sector lending to low-income countries can be risky. While private capital flows have increased substantially, the flows are still quite volatile, with potentially destabilizing macroeconomic effects. The Bank can address these problems in several ways: by making loans directly; by giving the private sector a positive signal through its decision to make loans; and by providing trusted sources of information that give the private sector the ability to assess risk and to make loans. The Bank’s ability to develop and disseminate knowledge underpins its ability to fulfill these roles.

Development knowledge has properties of a public good. Agents in the private sector have little obvious incentive for publicly documenting what they have learned about development, so that it can be available for the benefit of others. Scale economies in knowledge production can also entail large costs at the outset. If the supply of development knowledge depends on voluntary contribu- tions by individuals or countries, then there will be too little supply. In principle, an institution such as the Bank is well suited to resolving the deficiencies of decen- tralized knowledge provision.

Development challenges spillover across country borders in the form of pandemics, wars, refugee migrations, and environmental disasters. A global financing institution can play a role in helping to address these regional and global public bads. For example, during the recent Ebola outbreak in West Africa, the Bank deployed $400 million for improving health systems in the affected countries. In 2015, the Bank

j_ravallion_301.indd 80 1/20/16 6:54 AM

The World Bank: Why It Is Still Needed and Why It Still Disappoints 81

created a Pandemic Financing Facility to provide health workers, equipment, and drugs in response to future pandemics. The discussion will return to these issues.

Why a World Bank? The arguments in favor of multilateral development lending and aid reflect

concerns over how national governments politicize aid in either bilateral or regional settings. Too often, country preferences over who receives lending or aid reflect foreign-policy considerations and historical ties rather than genuine need or efficacy. Simulations by Collier and Dollar (2002) suggest that an allocation of aid that mini- mized aggregate poverty would differ greatly from the allocation that existed in the 1990s. According to their calculations, the poverty-minimizing allocation would have almost doubled poverty reduction relative to the actual allocation. Furthermore, bilateral aid has often been tied to recipient countries buying goods and services produced by the donor, a practice that has reduced the real value of aid (Temple 2010). Evidence also points to bilateral aid being used to buy support in major world forums, such as the UN Security Council (Kuziemko and Werker 2006).

The “pet projects” of national development ministries (possibly serving the interests of a local lobby in the donor country) need not make a lot of sense in the context of a sound strategy for poverty reduction. In contrast, a well-functioning global institution can generate economies of scale in knowledge and lending that are out-of-reach for a bilateral agency or even a regional institution. A global institution can also encourage broader participation by high-income countries, thus reducing what otherwise could be a severe free-rider problem. A multilateral institution can also serve a coordination function, embracing both bilateral and regional develop- ment lending and aid programs.

Escaping Traps and Overcoming Constraints A source of market failure that has been prominent in arguments for develop-

ment assistance concerns the scope for poverty traps. Rosenstein-Rodan (1943), who went on to be a prominent economist at the World Bank in 1947–53, pointed to complementarities between the investments made by different firms in an underde- veloped economy. If all firms invested, then they would all do well, but no individual firm has the incentive to invest when others do not. Development stalls in the inferior equilibrium—a poverty trap. The idea of coordination failures prompted Rosenstein-Rodan and others to advocate what came to be known as a “big push”—a large injection of aid for low-income countries. More recently, Sachs (2005) invoked the poverty trap idea to argue for an increase in development aid. Better public information can also help address coordination failures stemming from comple- mentarities in the investment decisions of firms (Englmaier and Reisinger 2008).

While the idea of a poverty trap as a low-level attractor has been influential, and there can be little doubt that such inferior equilibria exist, their empirical relevance in normal times is less evident (see, for example, Kraay and McKenzie 2014). Models with multiple equilibria are not easily identified empirically and a slow adjustment processes over time can be mistaken for a trap. For the purposes of the present

j_ravallion_301.indd 81 1/20/16 6:54 AM

82 Journal of Economic Perspectives

argument, however, it is not essential to resolve the question as to whether constraints on development are best viewed as “traps” or as substantial hindrances. Instead, one can postulate the existence of constraints on development that the private sector or bilateral agencies cannot address on their own. Hausmann, Rodrik, and Velasco (2008) provided an influential formulation of the development problem in terms of binding constraints specific to each country. The policy idea here is not necessarily of the “big push” variety, although that may well be valid in some cases. Instead, it is to assess for each country what is constraining poverty reduction and to target policy reforms accordingly. Identifying the relevant constraints is not easy and requires considerable country-level expertise. Relaxing constraints may require complemen- tary public inputs such as spreading technical knowledge, supporting more capable public administrations, and helping to supply public goods.

The Continuing Case for Bundling Lending with Knowledge There has been an ongoing controversy over the extent to which development

assistance has benefited the recipient countries. Some observers have argued that badly governed people in a poor country will be worse off with aid as it will reward and support the regime (for example, Deaton 2013, chap. 7). While the attribution problems are severe, given that aid is endogenous, my own review of the evidence from many studies suggests a credible case that past development aid has helped (Ravallion 2016, chap. 9).

My purpose here is not to revisit this debate, but to point out that the experi- ence with development assistance has lessons to teach about what works and what doesn’t. Learning these lessons requires the development institution to be centrally focused on generating and disseminating relevant knowledge at country, regional, and global levels. The gains from bundling knowledge with lending provide the key rationale for the Bank’s existence in a world of more developed capital markets (as argued by Gilbert, Powell, and Vines 1999). It also points to a key difference between the World Bank and dedicated research institutions, including academia.

The rest of this paper will argue that a valid case for World Bank lending opera- tions remains, but knowledge must drive that lending—both informing the nature of the lending and learning from it—rather than simply serving lending when it happens to be called upon. From this perspective, the Bank is falling short of its potential.

Why the World Bank Still Disappoints

Sound evaluations both before and after its operations are clearly crucial to a knowledge bank, so this topic is a good place to start. The discussion then turns to other things to be expected of a knowledge bank, and how the World Bank performs.

Evaluations of Lending Operations The first question we would surely ask of a knowledge bank is whether it

establishes a sound prior case for its own interventions and systematically assesses

j_ravallion_301.indd 82 1/20/16 6:54 AM

Martin Ravallion 83

whether that case turned out to be valid. The World Bank has not, however, lived up to this ideal. Evaluation is generally weak and unbalanced, both before and after implementation. This reflects a lack of focus on the welfare outcomes of projects and policies. Instead of studying the effect on its stated goal of poverty reduction, the focus tends to be on monitoring inputs—for example, schools built rather than education attainments (Gaarder and Bartsch 2015).

A true knowledge bank will address questions like: Why is the proposed project needed? How does it relate to overall development goals? What are the market, or governmental, failures it addresses? What are its distributional goals? What are the trade-offs? Social cost–benefit analysis provides the economic frame- work for addressing these questions (Devarajan, Squire, and Suthiwart-Narueput 1997). The Bank was once a leader in cost–benefit analysis, but this is no longer true. While the Bank’s operational directives call for cost–benefit analysis, it is not implemented for most Bank projects (World Bank 2010). The proportion of projects quoting an expected rate of return has fallen over time.

Cost–benefit analysis has clearly fallen out of favor among World Bank staff and managers. There have been justifiable concerns about the quality of the key inputs to the analysis, notably on the benefits side—both the magnitude of the benefits and their monetary values. Uncertainty about key parameters creates scope for manip- ulation by project staff keen to get their loan approved. But these are not good reasons for abandoning project appraisal. We still need to know what the case is for the project, given what we know and recognizing the uncertainties. The identified knowledge gaps should then be addressed in follow-up work to reduce the uncer- tainties for future appraisals. Checks can be done on the quality of the analysis.

The decline in cost–benefit analysis at the World Bank came with a welcome rise in the use of impact evaluations done after projects are completed. The Devel- opment Impact Evaluation initiative based in DEC has helped, and many new impact evaluations are underway. Nonetheless, there is still much to do. The vast majority of Bank lending operations still are not properly evaluated after they are completed. Recent assessments by the Independent Evaluation Group indicate that three-quarters or more of Bank lending operations do not have impact evaluations (World Bank 2014d), although that is still an improvement compared to 15 years ago (World Bank 2012a).

The concerns go beyond the number of evaluations. The subset that is evalu- ated cannot be considered to be representative of the whole, as World Bank (2010) shows in the case of before-the-project evaluation, and World Bank (2012a) shows for after-the-fact evaluation. The already limited after-the-fact evaluations have been skewed in the last 10 years or so toward projects, or aspects of projects, amenable to randomized control trials (Ravallion 2009; World Bank 2012a). As a result, we see fewer evaluations of other types of projects when a simple assignment of participants and nonparticipants does not exist or when such an analysis is severely contaminated by spillover effects. There are other concerns. Evaluations tend to be biased toward short-run impacts; there have been remarkably few impact evalu- ations that can claim to have tested for the long-term impacts of Bank operations.

j_ravallion_301.indd 83 1/20/16 6:54 AM

84 Journal of Economic Perspectives

Granted, long-term evaluation can be difficult, but it is still possible: for an example, see Chen, Mu, and Ravallion (2009), an evaluation of a Bank lending operation in China. The Independent Evaluation Group has also raised doubts about how much the limited impact evaluations that have been done (inside and outside the Bank) are being used in project preparation and reporting (World Bank 2015). Moreover, the evaluations that have been done have only rarely measured impacts on poverty, even though poverty reduction is the Bank’s overall goal (Goldstein 2014). It cannot be claimed that these shortcomings stem solely from technical problems or costs of doing evaluations; the problems lie elsewhere in how the Bank functions, a subject to which we will return.

From the 1990s on, it came to be understood that traditional project-specific evaluations need to be augmented by a broader assessment of public spending. There are two main reasons for taking a broader perspective. First, aid is to some degree fungible, so that aid ostensibly tied to a specific project is really just freeing up money to be spent in other areas. A rigorous cost–benefit analysis of a specific project may tell us very little about what the aid is actually funding. There is evidence of fungibility (Feyzioglu, Swaroop, and Zhu 1998), although it is less plausible in some relevant circumstances: for example, in heavily aid-dependent countries or for projects that require the external technical assistance that comes with the aid.

Second, portfolio effects arise when the multiple elements of the program package interact. The success of an education project (say) may depend crucially on whether infrastructure or public sector reform projects have worked. Evalu- ating each bit separately and adding up the results will not (in general) give us an unbiased estimate of the portfolio’s impact (Ravallion 2016, chap. 6).

More holistic country-level approaches to assessing aid effectiveness have emerged, aiming to put each specific project in a broader public finance context. Various World Bank analytic documents (called Poverty Assessments and Public Expenditure Reviews) have played a role. However, while these analyses are useful complementary elements to cost–benefit analysis, they should not be a substitute for it. We should still know what the economic rationale is for any public project and what was learnt about its impact—both the good and the bad news.

Development Data The World Bank has long been the one-stop shop for development data. Histor-

ically, much of this effort was through compilations of country-level data, initially in the Bank’s annual World Development Reports, but breaking off to form the World Development Indicators. Since 2010, the Bank has provided open access to these data.

The Bank’s data compilations are valuable, but one can also feel a degree of frustration in what has not been accomplished on the data production side. The sorry state of the national accounts in much of the developing world—for example, Jerven (2013) points to serious concerns about the quality of national accounts data for sub-Saharan Africa—cannot be entirely blamed on the World Bank and the IMF, but these organizations bear some responsibility. The Bank has not used its own power as much as it could to encourage governments to make public their own data,

j_ravallion_301.indd 84 1/20/16 6:54 AM

The World Bank: Why It Is Still Needed and Why It Still Disappoints 85

which can help improve its quality. I recall attending a meeting with then–Bank President Robert Zoellick in 2010 at which he expressed justified alarm on realizing that the Bank provides budget support to some countries that do not provide fully public budget documents. Zoellick pushed hard on this point and budget trans- parency improved. But an institution committed to poverty reduction should also insist, as a prerequisite for support, that countries provide open public access to the micro- and administrative data from their own statistical offices that are needed to monitor progress against poverty.

Starting in the 1980s, the Bank’s data efforts started to be more analyti- cally driven and policy relevant. Looking at the 1979 World Development Report, Bank President Robert McNamara was shocked to see that only 17 developing countries had data on poverty and inequality, even though estimates of macroeco- nomic aggregates from national accounts were available for virtually all countries. McNamara asked his research staff to collect the missing data. With the founding of the Living Standards Measurement Study in the 1980s—which involved detailed household-level surveys of a wide range of data—as well and subsequent initiatives like the Enterprise Surveys that collect firm-level data and the Quantitative Service Delivery Surveys that collect data on health and education facilities, the Bank soon emerged as a major source of microdatasets. Bank researchers have played an important role in public microdata production.

Complementary initiatives in software development to make microdata more accessible in developing countries have greatly expanded policy and analytic capa- bilities; a good example for the description and analysis of microdata is the ADePT software platform (available at the World Bank website1). Facilitating data collec- tion and access to relevant analytic tools should be central to the mandate of a knowledge bank.

The Bank itself also needs to be open about the data related to its own lending operations. Much data is collected in loan preparation, supervision, monitoring, and evaluation. These data can be valuable to other aid agencies and potential private financiers in facilitating learning from Bank operations—both the successes and failures. However, much of these data are not made public, and there is scope for selectivity in what is made public. Currently the incentives are weak to change this practice.

Research Research and analytic capability is crucial to the rationale for the World Bank

as a “knowledge bank.” A significant share of that capability needs to be in-house, given the difficulties of structuring incentives for outsiders to deliver what is needed (Squire 2000). The Bank’s research department aims to span all sectors of the Bank’s work. Research is also done in some of the sectoral/regional units. Bank

1 The ADePT software platform is available here: http://econ.worldbank.org/WBSITE/EXTERNAL/ EXTDEC/EXTRESEARCH/EXTPROGRAMS/EXTADEPT/0,,menuPK:7108381~pagePK:64168176~pi PK:64168140~theSitePK:7108360,00.html.

j_ravallion_301.indd 85 1/20/16 6:54 AM

86 Journal of Economic Perspectives

research takes many forms, ranging from project evaluations to analytic assessments of the constraints on development in specific settings.

To be a true knowledge bank, research needs adequate and secure funding. The Bank spends less on research as a share of its budget than comparable organiza- tions, and Bank spending devoted to research has declined in real terms over recent years (World Bank 2012b). The Bank’s knowledge activities (not just research) have also become more dependent on “soft money,” notably trust funds.

The research function at a knowledge bank poses an organizational balancing act that needs to be more explicitly acknowledged and managed. On one side, there is a risk of “ivory-tower” researchers becoming isolated from operations. On the other side, Bank research sometimes needs to be protected by its management from the efforts of the sectoral and regional empires to influence the themes and messages of that research. A research paper that identifies deficiencies in the policies of any prominent national borrower may have a hard time getting cleared—and clearance by the Bank’s country director for the country concerned is a requirement for publication. It is rare for a research paper to not be cleared, although edits are often called for. And of course, the need for this clearance is anticipated in choices made about what to research and how to present the results.

There must also be effective demand for knowledge in operations. The bulk of the Bank’s senior operational staff appears to value Bank research for their work, and come to know it well (based on a survey of senior Bank operational staff discussed in Ravallion 2013). But there is a marked unevenness. The staff members working on poverty, human development, and economic policy tend to value and use Bank research more than staff in the more traditional sectors of Bank lending—agricul- ture and rural development, energy and mining, transport, and urban development. The latter sectors account for 45 percent of lending, but of the Bank staff who report they are highly familiar with Bank research, only 15 percent are in these sectors (Ravallion 2013). Of course, there are two sides to this problem. Demand for Bank research is interrelated with supply, and stronger incentives for learning within the Bank must come with more relevant and accessible research products.

The Bank’s knowledge role should continue to include facilitating indepen- dent research outside the Bank, especially in developing countries. A good example is the Bank’s support of the Global Development Network, which since its inception in 1999 supports researchers from developing countries on a competitive basis, with both financial support and by connecting researchers globally.

Policy Advice For a knowledge bank to be credible, all parties must have confidence that the

institution is not under the undue influence of powerful shareholders. At one time or another, it is likely that all of the World Bank’s major shareholders and borrowers have attempted to influence Bank policies and processes. For example, some coun- tries have been known to lobby against a Bank index of performance (such as on governance) when it ranks that country low, although such lobbying rarely appears to succeed. However, the influence of the United States has been a longstanding

j_ravallion_301.indd 86 1/20/16 6:54 AM

Martin Ravallion 87

concern in some quarters. The United States does have considerable power at the Bank, including in selecting the Bank’s President, its weight in formal voting at the Board and in (more subtle) policy positions, and even in project implemen- tation (for example, Kilby 2013 looks at how politics affects project preparation times). Critics question American influence on the policies advocated by the Bank in developing countries.

The bulk of this critique has focused on the Bank (and IMF) advocacy of a set of “neoliberal” economic policies that came to be known as the “Washington Consensus” (Williamson 1990). The policies included fiscal discipline, cutting generalized subsidies, tax reforms, market interest rates, liberalizing trade and foreign direct investment, privatizing state-owned enterprises, de-regulation to encourage competition, and assuring legal security for property rights. From a marketing point of view, the label “Washington Consensus” could hardly have been more damaging. The label suggests a policy agenda formed amongst an elite group in one high-income country, making the policies an easy target for some critics (for example, Broad and Cavanagh 2009).

The critics were not always well-informed about the economic rationales for those policies. There were clearly specific contexts where the policies made sense. Nor did the critics always make clear what alternative policies they had in mind and what their welfare impacts would be. For example, research has often shown that inflation is costly to poor people (for a review, see Ravallion 2016, chap. 8), so the poor have an interest in macroeconomic stability. Exaggerated claims were heard about the adverse impacts of macroeconomic adjustment on poverty; careful analysis (also considering the costs to poor people of not adjusting) often painted a more nuanced picture (for example, World Bank 1994; Jayarajah, Branson, and Sen 1996; Sahn, Dorosh, and Younger 1997).

However, some of the criticisms were valid. Early World Bank (and IMF) programs for “structural adjustment” paid too little attention to the implications for poverty reduction and human development. A welcome change in thinking within the Bank was already underway by the late 1980s. Add-on programs to “compensate the losers from adjustment” were becoming common. There was also a mounting effort to use evidence to understand the social impacts of economy-wide and sectoral policies.

The Washington Consensus was too formulaic to be credible as a policy prescrip- tion. It listed a single set of policies, but governments of developing countries could see for themselves that there were multiple paths to development success. In partic- ular, the non-Washington Consensus route taken by China since 1980 stood out as an example for all to see. Development policy-making has become more open to what were once considered heterodox ideas, though it remains true that all policy advocates should justify their case. Theory and evidence remain no less relevant when one takes a more contextual and pragmatic approach.

An objective country-specific assessment of the binding constraints on poverty reduction should ideally guide all World Bank support. About one-quarter of total Bank lending involves what are now called Development-Policy Loans (formerly

j_ravallion_301.indd 87 1/20/16 6:54 AM

88 Journal of Economic Perspectives

structural adjustment loans), which are quick-disbursing loans to support a govern- ment’s policy reform plans. While Development-Policy Lending operations often draw on high-level expertise within and outside the Bank, it is not clear how much influence that expertise has or how well the operations are tailored to addressing the most important constraints on development in each country. This is even less clear for the Bank’s investment-lending portfolios. A series of innovations have tried to make the rationales clearer, such as in the “Country Assistance Strategy” papers. But too often, these appear to be little more than post hoc rationalizations for the lending program, rather than decisive independent analyses of what needs to be done to assure more rapid progress against poverty in the specific context. I am not the only observer to note the generally declining quality of these types of papers over time; they do not appear to be getting the attention that they once held.

Striking a balance between independent World Bank judgment and what its client countries wish to do is a continuing challenge. In 2014, the Bank introduced Systematic Country Diagnostics, in which the Bank’s country teams try to identify the main development problems the country faces (and which serve as an input to the Country Partnership Framework, developed with the government). In prin- ciple, the new diagnostic tool is not confined to issues identified by the government, acknowledging the desirability of the Bank’s independent view. However, the official guidelines for the new diagnostic tool say that it is to be done “in close consultation with national authorities” (World Bank 2014b, p. 1). It remains to be seen how inde- pendent the country diagnostics will be in practice, and whether politically sensitive analytics will surface in policy dialogues, especially in the large borrowing countries.

The compartmentalization of knowledge has also constrained policy advice. The Bank’s sectoral silos (now called Global Practices) have not been well-suited to identifying trade-offs across sectors. More attention to trade-offs among different methods of fighting poverty is needed, and this would also be welcome for many of the Bank’s clients who face hard allocative decisions.

Over the last 15 years or so, an increase in social protection spending by developing-country governments came with considerable financial support from the Bank (World Bank 2014c). This area is less attractive to the private sector (compared to infrastructure, say). But here too, the Bank’s policy stances seem to strive too much for universality. Social-protection policy advocacy turned “targeting” (avoiding leakage to the “non-poor”) into a fetish—oddly confusing the ends and means of social protection (Ravallion 2016, chap. 10). Lending and policy advice in this area has been dominated by a “flavor-of-the-month” approach. For a time, there was a rush to create “conditional cash transfer” schemes, providing transfer payments conditional on keeping children in school and attending to their health care. The popularity of these programs was to some degree informed by evaluations that had demonstrated impact. For example, well-documented research on the Progresa program in Mexico was very influential; on reviewing this and other evidence, a Bank research report by Fiszbein and Schady (2010) stimulated greater Bank support for conditional cash transfer schemes in numerous countries. However, conditional cash transfer advo- cates did not always pay proper attention to other research findings on the supply-side

j_ravallion_301.indd 88 1/20/16 6:54 AM

The World Bank: Why It Is Still Needed and Why It Still Disappoints 89

delivery problems in health and education. Conditional cash transfers work less well in settings where the problem does not obviously appear to be on the demand side, given the evident failings of public service provision—failings to which Bank research has often pointed (for example, World Bank 2003).

Enthusiasm amongst practitioners ran well ahead of evaluative research for some other social policies. As one example, the weakness of local states led to well-intentioned efforts to implement Community Driven Development, in which local communities would ostensibly drive the development process rather than the state. Many development agencies, along with the Bank, provided substan- tial funding for community-based projects. But evaluative work soon pointed to concerns, including project capture by local elites. A more nuanced view emerged amongst researchers, which acknowledged the potential benefits of citizen partici- pation but also warned that local states needed to be strong enough to assure that participation was effective and pro-poor (Mansuri and Rao 2013). Citizen participa- tion is not a substitute for local state capacity. There could be a trade-off between the local-level fairness of participatory implementation and a development proj- ect’s impact on poverty (Chen, Mu, and Ravallion 2009). Such trade-offs need to be taken more seriously in lending and aid, such as in poor-area development efforts.

Taking a Longer-Term Perspective on Development World Bank policy advice needs to take a longer-term perspective on a country’s

development. Countries are essentially locked out of support from the development banks and most bilateral donors if their institutional environment is deemed to be too poor; in the case of the World Bank this is measured by a very low score in the Bank’s Country Policy and Institutional Assessments. Once the quality of the institutional environment rises above a minimum threshold, lending and aid start to flow, with the aim (in part) of improving governance and the institutional envi- ronment more generally. This model is based on a belief that development lending and aid can improve governance (in contrast to the view of some aid critics that it promotes bad governance). External assistance eventually stabilizes when institu- tions are sufficiently well developed. Beyond some point, development assistance declines and eventually vanishes.

The parameters of this model are open to debate. The lack of justification for the Bank’s income thresholds has been a long-standing concern—in part because the Bank’s questionable criteria are widely used by other aid agencies. A more flexible approach based on relevant economic factors, such as creditwor- thiness and domestic capacity for redistribution, is long overdue.

But even taking the parameters as given, a feature of this model often not acknowledged properly by either aid critics or supporters is that such a model can readily yield multiple equilibria in institutional development (Ravallion 2016, Ch. 9). This has important implications for policy. For example, getting out of the low equilibrium of weak institutions—what I dub a “poor institutions trap” (PIT)— will often not be possible with only a small positive incentive for reform. As another example, fragile states could be destabilized enough to easily end up in a PIT.

j_ravallion_301.indd 89 1/20/16 6:54 AM

90 Journal of Economic Perspectives

This argument points to a role for the Bank in longer-term institutional devel- opment. If the World Bank were to anchor its engagement to a plan for addressing the relevant constraints in each country, its engagement would not be capricious— buffeted by short-term political shocks in its client countries or foreign-policy considerations amongst its major shareholders. To its credit, the Bank does take a longer-term perspective on development than most other aid agencies; this is evident in the attention that the Bank has given to institutional development (Birdsall and Kharas 2014) and its greater use of the recipient country’s own perfor- mance management system (Knack 2013).

International Public Goods While the World Bank is increasingly called upon to address development

problems that spillover across country borders—such as pandemics and climate change—it is far from clear that it is currently well equipped for such tasks. The Bank looks for opportunities to address international public bads and has responded at times, but its present country-lending model is not well-suited to such tasks. As Birdsall (2014) points out, the Bank’s $400 million Ebola response in 2014 was a fortuitous fit with the country model, rather than the systematic application of an adequately funded institutional mandate.

The Bank’s new Global Practices have the capability of significant sectoral knowledge transfer across borders. The Bank also has a convening power that can help in the cross-country coordination needed in addressing global commons issues. But the required level of demand for international public goods cannot be expected to come from individual nations on their own, given the externali- ties involved. For the Bank to play a larger role in this area, a stronger mandate is required from its shareholders and there must be dedicated funding for global commons tasks (Morris and Gleave 2015). Birdsall (2014) suggests a new arm of the Bank is needed, or even a new institution.

It is hard to see any of this happening soon; the Bank’s major shareholders have shown little enthusiasm for providing the extra capital required for new global initiatives, and many of the Bank’s borrowing countries are inclined to oppose any potential diversion of funds from traditional country-based lending.

Knowledge Dissemination There is little point in producing development knowledge that cannot be

shared. A knowledge bank will naturally produce a wider range of knowledge prod- ucts than a dedicated research center alone, or an academic institution. There is a role for the aforementioned “knowledge broker” function. More broadly, the task of “learning in lending” will require effort at careful documentation. Bank research should meet scholarly standards when relevant, but it should not be judged solely by narrow academic criteria. Instead, its aim must be to inform policy debates and to provide a constructively critical perspective on Bank operations. While acknowl- edging the differences from academic research outputs, there are a number of concerns about the Bank’s current knowledge products.

j_ravallion_301.indd 90 1/20/16 6:54 AM

Martin Ravallion 91

First, there are quality concerns. Publication processes entail peer review, which provides a degree of quality control; the research of DEC (the chief research arm of the World Bank) tends to be published and so is subject to peer review, generally external to the Bank. However, while unpublished knowledge products customized to client needs are important to the Bank’s impact on the ground, the quality of the internal review process and final output is in my experience uneven, and this should be a source of concern.

Second, the Bank’s more operationally oriented knowledge products (whether published or not) have often struck me as remarkably self-referential, with rather limited signs of new knowledge entering from outside the institution. If something has not already been tried within the Bank, then it is often treated as risky—even if there is outside experience that might help evaluate that risk more clearly. Established methodologies within the Bank have a persistence that often defies innovation, new knowledge, and sometimes even old knowledge.

Third, the Bank’s size and the pressures on each unit to stay big also foster knowledge products that are essentially “make-work” schemes that make little or no contribution to knowledge and so have attracted little attention. Using the very broad citation data that can be assembled from Google Scholar,2 in Ravallion and Wagstaff (2012), my coauthor and I find that it is hard to discern more than a negli- gible impact for many Bank publications, though certainly not all.

Must the Lending Bank Rule?

The World Bank is not a monolithic, technocratic, poverty-minimizing agent. While eliminating absolute poverty and sharing prosperity are espoused as its overarching goals, the objectives of its staff and managers are not as well aligned with those goals as they should be. Instead, more diverse and complex motives emerge out of the Bank’s governance and the multiple interests of its various stakeholders.

One important motive is to maintain and expand the institution itself. The profits from its lending have historically been an important source of revenue for Bank staffing, so it can be no surprise that the Bank’s “lending culture” rewards operational staff for the volume of their lending. However, as we have seen, weak evaluative practices entail weak connectivity between Bank lending and its goal of poverty reduction. The managers/directors of the country teams have an incentive to push a high volume of lending to satisfy their bosses and ensure a decent budget for their unit, without giving sufficient consideration to the quality of that lending and how it will benefit poor people, or how it will affect the transfer of knowledge. In the process, the lending bank also generates a gauntlet of procurement rules and

2 Google Scholar casts a broader net than other bibliographic databases, including citations by books, working papers, reports, conference proceedings, open-access journals, new, and less well-established journals. It is also more “global” in its reach, as it includes research outputs from everywhere in the world and all languages.

j_ravallion_301.indd 91 1/20/16 6:54 AM

92 Journal of Economic Perspectives

other administrative hurdles that absorb much staff time. Maintaining, let alone developing, the human capital of staff can be a challenge.

Concerns about the alignment of incentives in the Bank are not new. For example, this was a theme of a high-level Bank report nearly a quarter-century ago (Wapenhans 1992). Organizational changes in 1987, 1996, and 2014 sought to improve incentives for learning from lending. But with reference to the changes in 1987 and 1996, the Independent Evaluation Group concluded that: “These changes have not led to a significant change in learning from lending because they touched neither the culture nor the incentives” (World Bank 2014d, p. vii). While the new Global Practices are a promising step, all indications are that the lending culture thrives today, and still with generally weak accountability to the Bank’s overall goals. Bank insiders continue to debate how to better assure that managerial choices are consistent with the Bank’s overall goals (Over and Ravallion 2012; Gaarder and Bartsch 2015).

The idea of bundling knowledge with lending is still attractive to the Bank’s clients. The traditional country-based model remains relevant as a means of iden- tifying and solving pressing development problems. The complementarities with private finance point to a continuing relevance of the Bank’s projects and policy support. The challenge for the Bank today is to assure that knowledge drives lending and aid, rather than simply serving them when called upon. This requires a quite fundamental change in the Bank’s culture such that managerial and staff incentives are reoriented from lending to learning.

■ For helpful comments, the author thanks Ulrich Bartsch, Tim Besley, Nancy Birdsall, Laurent Bouton, Michael Clemens, Asli Demirguc-Kunt, Shanta Devarajan, Francisco Ferreira, Marie Gaarder, Alan Gelb, Garance Genicot, Manny Jimenez, Ravi Kanbur, Steve Knack, Aart Kraay, Branko Milanovic, Rinku Murgai, Mead Over, Giovanna Prennushi, Martin Rama, Biju Rao, John Rust, Luis Serven, Lyn Squire, Jon Strand, Vinod Thomas, Dominique van de Walle, Nicolas van de Walle, Adam Wagstaff, and Bob Zoellick. The author is also grateful to the journal’s managing editor, Timothy Taylor, and co-editors Enrico Morelli and Gordon Hansen for many useful comments.

References

Birdsall, Nancy. 2014. “My Two Big Worries about the World Bank.” Center for Global Devel- opment blog post, November 3. http://www.cgdev. org/blog/my-two-big-worries-about-world-bank.

Birdsall, Nancy, and Homi Kharas. 2014. The Quality of Official Development Assistance. Third Edition. Center for Global Development and Brookings Institution.

j_ravallion_301.indd 92 1/20/16 6:54 AM

The World Bank: Why It Is Still Needed and Why It Still Disappoints 93

Broad, Robin, and John Cavanagh. 2009. Development Redefined: How the Market Met Its Match. Boulder: Paradigm.

Chen, Shaohua, Ren Mu, and Martin Ravallion. 2009. “Are There Lasting Impacts of Aid to Poor Areas? Evidence from Rural China.” Journal of Public Economics 93(3–4): 512–28.

Collier, Paul, and David Dollar. 2002. “Aid Allo- cation and Poverty Reduction.” European Economic Review 46(8): 1475–1500.

Deaton, Angus. 2013. The Great Escape: Health, Wealth, and the Origins of Inequality. Princeton University Press.

Devarajan, Shantayanan, Lyn Squire, and Sethaput Suthiwart-Narueput. 1997. “Beyond Rate of Return: Reorienting Project Appraisal.” World Bank Research Observer 12(1): 35–46.

Englmaier, Florian, and Markus Reisinger. 2008. “Information Coordination, and the Industrializa- tion of Countries.” CESifo Economic Studies 54(3): 534–50.

Feyzioglu, Tarhan, Vinaya Swaroop, and Min Zhu. 1998. “A Panel Data Analysis of the Fungi- bility of Foreign Aid.” World Bank Economic Review 12(1): 29–58.

Fiszbein, Ariel, and Norbert Schady. 2010. Condi- tional Cash Transfers for Attacking Present and Future Poverty. With Francisco H. G. Ferreira, Margaret Grosh, Nial Kelleher, Pedro Olinto, and Emmanuel Skoufias. Washington, DC: World Bank.

Gaarder, Marie, and Ulrich Bartsch. 2015. “Creating a Market for Outcomes: Shopping for Solutions.” Journal of Development Effectiveness 7(3): 304–316.

Gilbert, Christopher, Andrew Powell, and David Vines. 1999. “Positioning the World Bank.” Economic Journal 109 (459): F598–F633.

Goldstein, Markus. 2014. “Do Impact Evalua- tions Tell Us Anything about Reducing Poverty?” Development Impact blog, World Bank. http:// b l o g s . w o r l d b a n k . o r g / i m p a c t e v a l u a t i o n s / do-impact-evaluations-tell-us-anything-about- reducing-poverty.

Hausmann, Ricardo, Dani Rodrik, and Andrés Velasco. 2008. “Growth Diagnostics.” Chap. 15 in The Washington Consensus Reconsidered: Towards a New Global Governance, edited by Narcis Serra and Joseph E. Stiglitz. New York: Oxford University Press.

Jayarajah, Carl, William Branson, and Binayak Sen. 1996. Social Dimensions of Adjustment: World Bank Experience 1980–93. Washington, DC: Opera- tions Evaluation Department, World Bank.

Jerven, Morten. 2013. Poor Numbers. How We Are Misled by African Development Statistics and What to Do about It. Ithaca, NY: Cornell University Press.

Kanbur, Ravi. 2003. “Regional Versus

International Financial Institutions.” In Regional Public Goods: From Theory to Practice, edited by A. Estevadeordal, B. Frantz, and T. R. Nguyen. Inter-American Development Bank and Asian Development Bank.

Kilby, Christopher. 2013. “The Political Economy of Project Preparation: An Empirical Analysis of World Bank Projects.” Journal of Develop- ment Economics 105: 211–25.

Knack, Stephen. 2013. “Aid and Donor Trust in Recipient Country Systems.” Journal of Development Economics 101: 316–29.

Kraay, Aart, and David McKenzie. 2014. “Do Poverty Traps Exist? Assessing the Evidence.” Journal of Economic Perspectives 28(3): 127–48.

Kuziemko, IIyana, and Eric Werker. 2006. “How Much is a Seat on the Security Council Worth? Foreign Aid and Bribery at the United Nations.” Journal of Political Economy 114(5): 905–930.

Mansuri, Ghazala, and Vijayendra Rao. 2013. Localizing Development: Does Participation Work? Washington, DC: World Bank.

Morris, Scott, and Madeleine Gleave. 2015. “The World Bank at 75.” CDG Policy Paper 58, Center for Global Development, Washington, DC.

Obstfeld, Maurice, and Alan M. Taylor. 2004. Global Capital Markets: Integration, Crisis, and Growth. Cambridge University Press.

Over, Mead, and Martin Ravallion. 2012. “Recognizing and Rewarding the Best Develop- ment Professionals.” Development Impact blog post, October 11. World Bank, Washington, DC.

Ravallion, Martin. 2009. “Should the Randomistas Rule?” Economists’ Voice 6(2): 1–5.

Ravallion, Martin. 2013. “Knowledgeable Bankers? The Demand for Research in World Bank Operations.” Journal of Development Effective- ness 5(1): 1–29.

Ravallion, Martin. 2016. The Economics of Poverty: History, Measurement, and Policy. New York: Oxford University Press.

Ravallion, Martin, and Adam Wagstaff. 2012. “The World Bank’s Publication Record.” Review of International Organizations 7(4): 343–68.

Rosenstein-Rodan, Paul. 1943. “Problems of Industrialization of Eastern and Southeastern Europe.” Economic Journal 53: 202–11.

Sachs, Jeffrey D. 2005. Investing in Development: A Practical Plan to Achieve the Millennium Develop- ment Goals. New York, NY: Millennium Project, United Nations.

Sahn, David E., Paul A. Dorosh, and Stephen D. Younger. 1997. Structural Adjustment Reconsidered: Economic Policy and Poverty in Africa. Cambridge University Press.

Squire, Lyn. 2000. “Why the World Bank Should Be Involved in Development Research.”

j_ravallion_301.indd 93 1/20/16 6:54 AM

94 Journal of Economic Perspectives

Chap. 4 in The World Bank: Policies and Structure, edited by Christopher L. Gilbert and David Vines. Cambridge University Press.

Temple, Jonathan R. W. 2010. “Aid and Condi- tionality.” Chap. 67 in Handbook of Development Economics, Vol. 5, edited by Dani Rodrik and Mark Rosenzweig. Amsterdam: North-Holland. Avail- able at: http://www.sciencedirect.com/science/ handbooks/15734471.

Wapenhans, Willi. 1992. Effective Implementa- tion: Key to Development Impact. Report of the World Bank’s Portfolio Management Task Force. World Bank.

Williamson, John. 1990. “What Washington Means by Policy Reform.” Chap. 2 in Latin Amer- ican Adjustment: How Much Has Happened? edited by John Williamson. Washington, DC: Institute for International Economics.

Wolfensohn, James D. 1996. “Annual Meetings Address: People and Development.” October 1. http://web.worldbank.org/WBSITE/ EXTERNAL/NEWS/0,,contentMDK:20025269~ menuPK:34474~pagePK:34370~piPK:34424~the SitePK:4607,00.html.

World Bank. 1994. Adjustment in Africa. New York: Oxford University Press.

World Bank. 2003. World Development Report 2004: Making Services Work for Poor People. New York: Oxford University Press.

World Bank. 2005. World Development

Report:  Equity and Development. New York: Oxford University Press.

World Bank. 2010. “Cost-Benefit Analysis in World Bank Projects.” Washington, DC: Indepen- dent Evaluation Group, World Bank.

World Bank. 2012a. World Bank Group Impact Evaluations: Relevance and Effectiveness. Washington, DC: Independent Evaluation Group, World Bank.

World Bank. 2012b. Research at Work: Assessing the Influence of World Bank Research. Report to World Bank’s Board.

World Bank. 2014a. Annual Report 2014. Washington, DC: World Bank.

World Bank. 2014b. “Interim Guidelines for Systematic Country Diagnostic (SCD).” Report 85905. Available at: http://documents.worldbank. org/curated/en/2014/02/19227271/interim- guidelines-systematic-country-diagnostic-scd.

World Bank. 2014c. The State of Social Safety Nets 2014. Washington, DC: World Bank. http:// documents.worldbank.org/curated/en/2014/ 05/19487568/state-social-safety-nets-2014.

World Bank. 2014d. Learning and Results in World Bank Operations: How the Bank Learns, Evalu- ation 1. Independent Evaluation Group, World Bank, Washington, DC.

World Bank. 2015. Learning and Results in World Bank Operations: Towards a New Learning Strategy, Evaluation 2. Independent Evaluation Group, World Bank, Washington, DC.

j_ravallion_301.indd 94 1/20/16 6:54 AM

Copyright of Journal of Economic Perspectives is the property of American Economic Association and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.