1 Development Policy Review, 2011, 29 (s1): s5-s26 Aid, Institutions and Governance: What Have We Learned? David Booth Understanding of the relationship between institutions and economic progress made major headway in the 1990s. Since 2000 the pace of intellectual advance has been stepped up, with fresh thinking especially on governance systems and the role of aid. This article illustrates these propositions with special reference to relevant contributions to DPR over the period. Three particular topics are addressed: the rise and transformation of institutional theory; aid, governance and country ownership ; and the terms on which donors can contribute to the facilitation of institutional change. The article introduces a theme issue which celebrates the 50th anniversary of the Overseas Development Institute by reprinting nine of the highlighted DPR articles. Key words: Institutions, governance, policy, aid, Africa 1 Introduction It is sometimes said that it is impossible to have an original idea about development policy. Concerns and priorities go in cycles, and more often than not what looks like a new idea seems strangely familiar to those who are old enough to remember the debates of ten or fifteen years earlier. The most obvious examples involve the swings of the pendulum between the maximisation of economic growth and deliberate attention to income and wealth distribution, or more generically between market virtues and market failures. It seems very possible that when the history comes to be written, the main story about development-policy thinking since 1960 will involve cycles rather than genuine learning a somewhat depressing thought. Within the main story, however, there are definitely fields and sub-fields in which knowledge and understanding have had a more cumulative character. New evidence or sharper intellectual tools have led to fresh policy ideas (sometimes inspired by precursors from centuries rather than decades earlier) and these have become entrenched. When this happens, it should be celebrated. This DPR supplement makes the case that the field of institutions and governance exemplifies this more uplifting aspect of the evolution of development thinking. It does so by revisiting some of the best content that DPR itself has published in this field over the last decade. ODI began operations fifty years ago in 1960, and DPR has been publishing for a little over half of that period. The Editors were nevertheless persuaded that a celebration Research Fellow, Politics and Governance, and Director, Africa Power and Politics Programme, Overseas Development Institute, 111 Westminster Bridge Road, London SE1 7JD Published by Blackwell Publishing, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
2 s6 David Booth of this sustained contribution might reasonably take the form of a relatively focused retrospective dealing with a particular set of issues within the wide scope of the Institute s and the journal s concerns. As a likely guest editor, I leapt at the chance to focus on a field with which I have been involved and in relation to which the large gaps in my knowledge and skills as a social scientist would be least likely to be exposed. I would maintain, and I believe the articles referenced here demonstrate, that real progress has been made over the last decade in thinking about institutions, governance and development, including the possible contributions of development assistance or aid. I do not think it can be shown that there have been equivalent advances in policies and practices, which is one of the reasons why it seems important to write and publish more on the subject in a journal devoted to development policy. However, reshaping the intellectual agenda is a necessary first step, and not one that can be taken for granted. Looking forward, top priority needs to be given to how uptake of the ideas into practice is most likely to be achieved, and what the barriers are. The argument of the following pages is that we now understand better than we did as recently as ten years ago: how and with what significant qualifications institutions rule in the promotion of successful economic, political and social development; the limitations of the instruments for overcoming countries institutional constraints that lie within the easy reach of development-assistance organisations; and the idea of context-sensitive facilitation of institutional change and the challenges it poses to development policy and practice. The next three sections of this overview article are structured around these points, while the final section sums up and looks forward. Three of the DPR articles referenced in each section are reproduced as originally published, but with new pagination, in Parts 1, 2 and 3 of the supplement. Many others are available to subscribers via the electronic links embedded in the list of references. 2 The rise and transformation of institutional theory To the extent that it makes sense to treat the history of thought in terms of decades, it is the 1990s that has the best claim to be the decade of institutions. The flowering of various streams of institutionalist thinking in economics and political science that took place during those years acquired particular salience from the major world events of the years after the break-up of the Soviet bloc in However, it was in the following decade that the institutional turn convinced the sceptics, and critical scrutiny of its implications and limits could begin. In many respects, this is the more exciting story. 2.1 Origins and next steps In the 1990s, the world got institutions. Several analytical perspectives of older vintage were famously revived and brought into the social science mainstream during those years, including the transaction-cost theory of economic institutions (North, 1990),
3 Aid, Institutions and Governance: What Have We Learned? s7 rational-choice and game-theoretic perspectives on public policies and collective action (Ostrom, 1990; Olson, 2000) and the comparative history of state formation (Tilly, 1990; Evans, 1995), among others. But what really made the 1990s the decade of institutions was what was revealed for all to see by the course of events in two parts of the world: the former Soviet sphere of influence in Europe and Asia, and sub-saharan Africa. The divergent paths taken by the different member states of the former COMECON after liberalisation, and the sluggish response to the first generation of economic reforms in Africa, made the critical role of institutional factors increasingly hard to deny. The ferment of institutional thinking that was a feature of the 1990s created the conditions for the later advances which are the main focus of this section. It laid a foundation, but it also left many questions unanswered. I would argue that we have learned at least four things since the turn of the millennium which make an important difference to the way institutions figure in both academic and policy debates: Institutions rule, in the particular sense that cross-country regression analysis and comparative studies of country experience seem to converge in support of the proposition that the quality of a country s institutions is a more important factor in explaining per capita income growth than either of the variables emphasised in earlier econometrics: the geographic location of the country or the extent of its integration into world trade (Rodrik, 2003; Rodrik et al., 2004). The contribution of institutional quality also appears far more substantial than, and not closely related to, the volume of development assistance received (Doucouliagos and Paldam, 2009). While institutions are critical in the sense that certain essential functions need to be performed if market-based economic growth is to be sustained, the form in which this need is satisfied is to some extent undefined (Rodrik, 2007: Ch 6, Getting Institutions Right ). The history of the now highly industrialised countries does not provide a reliable template in this respect. Indeed, prescribing for poor countries the institutions which now prevail in those countries, at the end of their process of development, is equivalent to kicking away the ladder with which countries climbed up in the past (Chang, 2002). To an important degree, then, we do not know what the right institutions are. There is a strong case for distinguishing between the long- and the mediumterm needs of poor developing countries. Cross-country regressions covering the whole world and using standard measures of good governance indicate a clear correlation between those variables and economic outcomes, with some indication of causality. But a similar exercise covering only poor and middleincome countries is far more ambiguous (Khan, 2006). As confirmed by recent developmental successes in East and South-East Asia, the institutions that are good for obtaining growth and reducing poverty in the poorest countries may be quite different from the best practice arrangements that have proven their worth in moving forward from middle-income starting-points (Future State, 2010; Levy, 2010). None of the required change is self-generating, and relatively little of it has been the result of conscious design. As with the creation of the preconditions for accelerated development in the now most developed countries (Greif, 2006; North et al., 2009), institutional improvement in the remaining poor countries
4 s8 David Booth of the world will happen, if it happens, as a result of social changes and political conflicts and transactions which usher in new patterns of behaviour and social norms. Institutional change is a function of politics, a fact which is not sufficiently appreciated by providers of development assistance (IPPG, 2010). The DPR article on Politics and Growth by Williams et al. [2009, this issue] sets out boldly several of the main steps in this emerging thinking. The article begins with the ways economic and political institutions can inhibit private investment and economic growth, and goes on to explore the relationship between institutional change and political incentives at the country level. It ends with an engaging discussion of how development agencies can help to shift politicians incentives in a pro-growth direction. The Williams et al. article is comprehensive. However, it does not touch upon two issues on which others, including DPR authors, have had much to say. One is the extent to which growth rates and other development outcomes are influenced by policy choices as well as, and perhaps independently of, institutions and politics. The other is the extent to which aid that is delivered without reference to the institutional features of the recipient country can do harm as well as good. 2.2 Institutions versus policy? The wide acceptance that institutional theory now commands, whether in the North or the Khan variant, is very welcome. But it entails some dangers, the most important of which is the apparent implication that the content of the policies pursued by governments is a matter of little significance. I would argue strongly that this is not one of the implications of taking institutions seriously. An unusually compelling body of evidence in support of this assertion has been assembled by the Tracking Development research project at the University of Leiden (van Donge et al., 2010). The project has been investigating development processes and outcomes in matched pairs of African and South-East Asian countries over the fifty years since The main finding is that all of the selected Asian countries but none of the African countries adopted policies which combined (i) macro-economic balance, (ii) rural-biased public investment and (iii) economic freedom for smallholders. Close analysis of the country processes confirms that most of the very substantial divergence in growth and poverty-reduction performance between the two regions over the period is explained by these differences in the policies pursued. The Leiden study provides telling confirmation of the importance attached to policy in other work, notably the study by members of the African Economic Research Consortium of growth patterns within Africa up to 2000 (Ndulu et al., 2008). The inclusion of the South-East Asian cases in a comparative study is particularly helpful in that it allows rejection of the simple notion that Africa s inferior development performance is due to the fact that Africa s institutions are neopatrimonial that is, a blend of impersonal, law-based and personal, clientelistic forms of rule or that corruption is particularly widespread. The South-East Asian countries succeeded despite sharing these features in full measure and, indeed, despite the fact that several of them
5 Aid, Institutions and Governance: What Have We Learned? s9 began with institutions considered at the time to be a highly unpropitious basis for modern economic growth. Policy really does seem to matter, then. And if one is concerned with low-income Africa, policies towards agriculture matter more than any others. John Mellor (1976) and Michael Lipton (1977) were not wrong about this, which illustrates my opening point about cycles in development-policy thinking. In this sense, the arguments developed in the pages of DPR by Colin Poulton and his associates (Dorward et al., 2004; Poulton et al., 2006; Tschirley et al., 2010) justify careful reading and the fullest possible dissemination among those who will be the next generation of policy-makers in Africa. These articles have institutions at their centre. They are concerned with what kinds of institutions at the sector and sub-sector levels are likely to assist technological upgrading by cash-strapped, risk-averse smallholders, or the maintenance of production levels and output quality standards in major export sectors such as cotton. There is nevertheless a battle of policy ideas still to be fought about this. On the one hand, the bulk of the African intelligentsia remains wedded to the concept that agriculture is inherently backward and a poor basis for a national development strategy. On the other hand, the Washington Consensus is not yet entirely dead, and even the limited form of state interventionism that Poulton et al. are advocating continues to provoke nervousness in some international agencies. It will make a difference whether leading policy actors change their minds on this issue or not. The same applies to the debates about industrial policy options which are being aired in the pages of DPR (for example, Lin and Chang, 2009; Lin and Monga, forthcoming). Whether or not the new structural economics in industrial policy proposed by Lin (2010) is too radical or too timid in its stretching of the principle of comparative advantage, is a matter of some significance, at least for countries that already have the basic preconditions for industrial take-off. 2.3 Policy in institutional context Insisting that an institutional focus should not displace a concern about policy choice does not, however, take us back to square one. In two important ways, this appreciation of the centrality of policy choice is entirely consistent with the notion that institutions, and their political underpinnings, do rule. First, it is the real policy that counts, not the nominal policy. Policy is what policy does. A good deal of the nervousness about the remedies proposed by Poulton et al. relates not to what the policy would imply if well implemented, but to what it would mean in practice in the most likely kinds of political context. Bates (1981) account of the political economy of agricultural policy in 1970s Africa still haunts these debates, and for good reason. Brian Cooksey s article on Tanzanian policy [2003, this issue] shows that these concerns continue to be relevant. It documents how a widely commended policy option on agricultural marketing and input supply was quietly but effectively abandoned by the Government of Tanzania, to be replaced by a de facto policy with quite different implications. Although the article does not get into detail about the political drivers of the real policy towards agriculture in Tanzania, the implication is clear, that policy choice is seldom just a battle of ideas.
6 s10 David Booth Second, there are grounds for seeing a country s governance institutions and the elite incentives generated by them as a predisposing factor in the adoption of wise or unwise policies by its government. It is, of course, important to recognise the role of ideology and the intellectual climate of the time. Even the de facto policies of countries like Tanzania are strongly influenced by international opinion on what counts as sound development policy, and not only or mainly as a result of donor conditionality. But that having been said, it seems clear that some types of regime are more inclined than others to adopt and consistently implement wise policies. Implementation (or de facto policy) seems particularly susceptible to explanation in terms of the deep structures of a country s institutions, with the same weaknesses sometimes reappearing under a succession of different nominal policies. It is instructive in this sense to place Cooksey s analysis of Tanzania alongside Tony Killick s now updated study of policy in Ghana (1978; 2010). Killick was concerned to show, among other things, that Nkrumah s big push policy vision was not inconsistent with the mainstream development economics of the day, and he succeeded in doing this. However, his country analysis reveals not only that Nkrumah had his own reasons for acting as he did, but also that de facto policy was rendered fatally incoherent by socio-political factors which continued to operate under subsequent regimes, and in important respects remain at work up to this day. In short, recognising the importance of policy choice does not remove the imperative to understand the institutional make-up of a country before reaching conclusions about development prospects. It just provides a different entry point to consideration of the same issues. This is not to imply that the contrasting development performances of sub-saharan Africa and South-East Asia over the past 50 years are after all due to institutional differences, rather than the policy differences stressed by the Tracking Development project. Nor is it to revive the notion that Africa has suffered from forms of neopatrimonial politics or corruption from which South-East Asian countries have been free. That continues to be completely false. On the other hand, African economies have failed in different degrees and ways, as to a more limited extent have Asian economies. It may be that concepts as broad as corruption and neopatrimonialism fail to discriminate sufficiently. Perhaps there are forms of neopatrimonialism that are relatively good for development in both South-East Asia and sub-saharan Africa, in part because other things being equal they are favourable to the adoption and consistent implementation of sound policies. This is an avenue of enquiry which is being followed by a research stream of the Africa Power and Politics Programme (APPP), the consortium research programme which I direct. Pursuing a comparative study of post-colonial trajectories in Africa, with Asian experience and the Tracking Development findings in the background, this research argues that there are grounds for distinguishing among more and less developmental forms of neopatrimonial regime. It suggests that the main discriminating factor is the extent to which the collection and utilisation of economic rents are centralised by a political elite operating with a long time horizon (Kelsall et al., 2010). The argument has potentially large implications for what it means to get institutions right in the African context. It deepens the critique of the best practice approach to
7 Aid, Institutions and Governance: What Have We Learned? s11 institution building by pointing to the very different developmental dynamics to be found among regimes belonging to the broad family of neopatrimonial types. 2.4 Aid, institutions and development outcomes As Mary Shirley puts it, new institutional economics is not good news for foreign aid (2008: 76). And most of what we know about aid confirms this pessimistic view. History suggests that institutions change slowly, and mostly endogenously. Whether aid has any positive impact on economic growth continues to be contested territory, and the evidence of a contribution of aid through institutions to growth seems particularly weak. Indeed, there appear to be good theoretical and empirical reasons for suspecting that aid contributes perversely to elite incentives. In particular, it weakens the incentive to tackle difficult collective-action problems underlying the institutional barriers to faster and more equitable growth. It other words, it may do actual harm (van de Walle, 2001; Bräutigam and Knack, 2004; Moss et al., 2008). DPR articles suggest that the likelihood of actual harm being done can be high when aid is managed without taking proper account of the institutional features of the recipient country. For example, Borrman and Busse (2007) make a strong case for worrying about the Economic Partnership Agreements (EPAs) that are the European Union s favoured instrument for linking aid and trade in relations with the Africa, Caribbean and Pacific group of countries. According to this article, the trade-liberalisation components of EPAs benefit countries so long as they have a good framework of regulatory institutions which enable them to minimise the adjustment costs and take advantage of new trade opportunities abroad. However, many African countries have excessive regulation and reforming these arrangements under the tight deadlines implied by the EPA negotiations poses impossible political challenges. The countries with the weaker institutions are likely to incur losses to their trade if they follow the stronger members of their regional groupings into an agreement. If they opt out, regional integration processes will be badly affected. New flows of EU aid are supposed to support the needed reforms, but there are doubts about the appropriateness of the kinds of reform strategies that aid supports. So the promise of additional aid may simply serve to induce governments to accept unfavourable agreements. Other DPR contributions point to particular mechanisms which may help to explain the lack of evidence of positive aid impacts on growth. Although the debate continues, the econometric evidence that aid has a significant positive effect in aggregate does not seem to be strong. This, at least, is the conclusion of a recent metaanalysis of the large and contradictory literature on aid effectiveness (Doucouliagos and Paldam, 2009). The authors of this study and other economists who reach similar conclusions (for example, Rajan and Subramanian, 2009) commonly speculate that the reason has something to do with Dutch Disease effects, in which aid inflows work in the same way as natural-resource earnings to force up exchange rates, harming the profitability of the export sector. This is considered serious because dynamic export sectors are often needed to drive overall growth. Economists who advise donors on the potentially antiexport effects of aid (for example, McKinley, 2005; Barder, 2006) generally stress the
8 s12 David Booth offsetting effects on export dynamism of the cost-reducing investments that aid can (and was originally designed to) finance. However, empirical studies showing which of these tendencies prevails in practice are relatively few and far between, and there is almost no reflection on the institutional factors likely to influence the net result. The DPR article by Killick and Foster [2007, this issue] is a rare example of research which answers the question whether a surge of additional aid to poor African countries should be expected to have Dutch Disease effects, and under what conditions these would be offset by successful measures to stimulate the supply side of the economy. The authors did not find actual evidence of Dutch Disease, although in at least two countries (Tanzania and Uganda) they did find the related phenomenon of a crowding-out of private investment by aid-supported public spending. However, none of the seven African countries considered fully absorbed the surge in aid (preferring to accumulate the foreign exchange as reserves, in part out of fear of Dutch Disease). Therefore, the authors caution against the assumption that any large increases in aid to African countries will not cause this type of macroeconomic difficulty. The article proceeds to discuss the possibility that aid-financed investments would offset any exchange-rate effects by reducing the costs of enterprises producing tradeable goods and services. It makes the case that a compensating effect is more likely if the investments are targeted to the major components of exporters costs (for example, transport and power infrastructure) rather than the social sectors which have been emphasised so strongly by donors in recent years. It then comes to the issue of institutions. Whether extra aid is actually well deployed will depend on the strength of local public-expenditure management and the capacity of the civil service to channel resources into projects which are designed and implemented with reasonable efficiency. This is a bottleneck, and not one that aid itself has been good at addressing. As Killick has written elsewhere about Ghana, throughout the period when aid fluctuated around 10 per cent of GDP the capacity of the public service to put these resources to productive use was quite limited (2010: 452). The implication is clear. The lack of evidence of a strong positive link between aid and development outcomes such as economic growth should not surprise us if we have taken on board the general argument about the centrality of institutions. Obviously, no link should be expected if central banks do not allow the absorption of aid funds. But if they do, institutional bottlenecks are very likely to obstruct the use of the funds in the ways that, according to economic theories and models, could compensate for the antigrowth effects of stagnation in export sectors. These studies, of course, rest on descriptions of what has happened in the past. It is arguable that the past is a good enough guide to the future in matters of this sort. In this case, donors should probably pack up and go home. After all, shortage of foreign exchange as such is hardly a problem these days anywhere in Africa for Ghana, remittances alone have been of the order of 40% of GDP in some recent years, according to Killick (2010: 455). On the other hand, as Paul Collier (2006) puts it, aid is not oil; it is a particular sort of resource transfer that has at least the potential to impact positively on the institutional constraints which limit its own effectiveness as a contribution to development. Unless there are indications that the institutional problems of poor developing countries are likely to be self-correcting in the medium term, there seems to be an argument for asking the question whether the aid business could not do
9 Aid, Institutions and Governance: What Have We Learned? s13 rather better than it has in the past, in general and in respect of its impact on institutions in particular. The remaining two sections of this overview pick up this challenge. 3 Aid, governance and ownership During the twenty years or so in which academic research has come to terms with the centrality of institutional change in development, the policy world has been preoccupied with the limits of aid conditionality and the challenge of getting country ownership of aid-funded development efforts. The two trends have not been entirely disconnected, but they have remained distinct. 3.1 PRSPs: what went wrong? Donor and concessional lending agencies have a limited range of instruments at their disposal. They can be fully conscious of the depth and complexity of the deficiencies in development management at country level but still be faced with few options when it comes to engaging with this intractable reality. For this reason, the style of academic research which limits itself to debunking the latest World Bank or bilateral donor initiative without serious consideration of what it was attempting to do or what the alternative lines of action might be is seriously unsatisfying. An obvious case in point is the initiative of 1999 which gave the world Poverty Reduction Strategy Papers (PRSPs). This was a highly imperfect attempt to address a real problem, namely, the revealed inability of policy conditionality to secure commitment by the governments of aid-recipient countries to poverty-reduction goals and the steps required to achieve them. As Andrew Rogerson has written in DPR (2005: 550-1): the aid industry remains completely schizophrenic about conditionality. [Donors have] a deep-seated need to have multiple lock-in devices that either give us the power to rescind [long-term aid partnerships] at any time, or allow us to believe that we have it. It does not really matter whether these are framed as disbursement conditions or eligibility criteria. Most donor agencies use a belt-and-braces mixture of both. None of this changed in 1999/2000, except that the World Bank opted quite firmly for the eligibility criteria variant, otherwise known as ex-post conditionality, with its new instrument, the Poverty Reduction Support Credit. The Wolfensohn initiative at the Bank which led to the Comprehensive Development Framework and then PRSPs was, however, a genuine innovation. Like the subsequent multilateral decision to require Heavily Indebted Poor Country debt-relief candidates and later all recipients of the most concessional Bank and IMF credits to prepare PRSPs, Wolfensohn s idea was a thinking response to the failure of all previous forms of conditionality and technical assistance to bring about real commitment to policy-making for poverty reduction. The idea was misconceived in