IMF conditionality, implementation and the new political economy of ownership.

Bird, Graham ; Willett, Thomas D.

INTRODUCTION

For a long time, the conditionality of the International Monetary Fund has been criticised by those on the left of the political spectrum for being too harsh and invasive. More recently, it has also come under attack from some on the right who claim that conditions are seldom enforced. Thus, rather than imposing excessive adjustment, IMF programmes are seen by them as postponing adjustment. (1) Academic commentators pointed to the escalation in IMF conditionality during the 1980s and early 1990s and suggested that this may have discouraged governments from implementing programmes (Bird, 2001b). Certainly, a number of studies, including ones conducted by the IMF itself, have found that the track record of implementation has been poor. (2) This in turn has raised questions about the value of the IMF's seal of approval and its value as a catalytic agent to influence private capital flows.

Some critics, such as the majority of a commission appointed by the US Congress and chaired by Allen Meltzer (International Financial Institution Advisory Commission (IFIAC), 2000) have gone so far as to propose that IMF conditionality should be entirely abandoned and replaced with ex ante rules for eligibly to borrow. Others have proposed a wider range of reforms relating to conditionality.

Faced with these criticisms, the Fund undertook a review in which it broadly accepted that conditionality had been expanded excessively and expressed a desire to return to a focus on its core responsibilities for macroeconomic and exchange rate policies. Conditionality was to be 'streamlined'. (3)

There has also been a second strand of recent reform. This has placed increasing emphasis on the notion of 'ownership'. The basic idea here is that governments are more likely to implement programmes that they 'own', rather than those that they feel have been imposed on them. The link to the first strand is that the feeling of ownership is likely to decline as the number of conditions rises. Hence, streamlining is presented by the Fund as contributing to ownership. The Fund has also sought to involve more elements of society in the discussions that culminate in an agreed programme.

The thrust of recent IMF policy towards conditionality is therefore reasonably straightforward. The Fund retains the view that conditionality is useful. However, as a modality for encouraging economic reform its effectiveness has been impaired by poor implementation. This, in turn, has been associated with a lack of ownership. By streamlining conditionality, and in addition, by involving more elements of society in the discussions that result in IMF-approved programmes, ownership may be strengthened. As a result, implementation may be improved and effectiveness enhanced. But is this an appropriate strategy and will it work? The basic purpose of this paper is to examine this question.

As the IMF has recognised, ownership is a fuzzy concept; it means rather different things to different people. At one extreme its meaning can be so broad as to be amorphous; at the other it can be turned into a non-operational tautology that equates ownership and implementation.

We review this literature and suggest that, for the IMF's purposes, a narrow definition of ownership that is directly related to the prospects for programme implementation is the most appropriate. (4) Broad concepts related to good governance and the roles of civil society may well be worth discussing, but, in the context of IMF-backed programmes, could be unhelpful. They could run counter to the efforts to streamline IMF conditionality.

Focusing on implementation highlights the need for the IMF to consider political economy as well as technical economic aspects of conditionality. From this perspective, a technically less desirable programme but one that has a substantially higher probability of implementation should be preferred to a technically superior programme that, however, has little chance of being accepted by a national government and of being successfully implemented.

The operational focus should be on implementation. The Fund needs to consider ways in which it can analyse and encourage this. In our view, programme ownership should be viewed only as a means to an end, not an end in itself.

The organisation of the paper is as follows. The following section briefly examines the various purposes behind conditionality. The subsequent section goes on to evaluate the effectiveness of conditionality and in particular the record on its implementation. The section thereafter analyses the concept of ownership and examines the extent to which it is compatible with conditionality and connected to implementation. In the light of this, the next section critically evaluates recent policy reforms and goes on to put forward a number of complementary or alternative proposals. The last section offers concluding remarks.

THE PURPOSES OF CONDITIONALITY

In principle, conditionality could serve a number of purposes. These have been discussed at some length in the literature. (5) Most simply, the Fund needs some mechanism for ensuring that countries pursue policies that will enable them to repay the resources they have borrowed. Khan and Sharma (2001) draw on finance and agency theory to argue that 'IMF conditionality can be perceived as a complex covenant written into the loan agreement ... (it) therefore can be thought of as a substitute for collateral' (p. 6). This is of course a sensible rationale, but levels of IMF conditionality often exceed what is needed on that score (while enforcement sometimes falls below what is needed).

Going beyond this, conditionality may also be a method of policing moral hazard; governments could otherwise misuse the resources borrowed, in the sense of pursuing 'inappropriate' policies and seeking to avoid economic adjustment. Or they could pursue policies that would have deleterious effects on other countries--so-called 'beggar thy neighbour' policies. Conditionality is similar to tied aid where the donor is concerned with improving the lot of the recipient, but judges improvement in terms of its own values or preferences or the ones which it believes the recipient should have, rather than those the recipient actually has. (6)

Extending this idea, conditionality may be viewed as a way of helping countries to help themselves and thereby making financial support more effective. Important in this context is conditionality as a commitment device. Many activities ranging from dieting to macroeconomic stabilisation have asymmetric time properties of costs and benefits. Many humans have short time horizons and find it difficult to be happy with delayed gratification.

With dieting, the costs of giving up eating delicious food are immediate while the benefits are delayed. The same is true of most macroeconomic stabilisation plans. The costs in terms of high unemployment and slow growth may only be transitory, but they show up quickly while the benefits of lower inflation often come only with a substantial lag. By the same token, overly expansionary monetary and fiscal policies have their costs delayed while the transitory benefits show up quickly. In the face of such asymmetrical cost-benefit streams, we may need to adopt constraints over some types of behaviour in order to act according to long-run rather than short-term costs and benefits.

A second problem is that it is now well understood in public choice and macroeconomic theory that the combination of rational (or irrational) ignorance and short time horizons can, in some circumstances, create situations where popular participation is likely to lead to bad outcomes; for example, generating political business cycles and inflationary biases. (7) Thus, in many cases governments create independent regulatory agencies and/or constitutional rules to limit their freedom to do the 'wrong thing'.

When we turn from individual to collective action issues, it is not inherently antidemocratic to adopt measures to limit the effects of our bad habits. Independent central banks and regulatory agencies and limitations on the sizes of budget deficits are frequent examples. Many IMF policies serve as such commitment devices which in effect lengthen the time horizon of the political process.

In addition to the above, some outside commentators have suggested that conditionality may fulfil functions to which the Fund would not subscribe. Conditionality has been presented as the 'price' of IMF financial support; a price that varies according to the demand for and supply of IMF resources. Here, conditionality is, in effect, a rationing device. As the demand for IMF resources increases, because of a heavier incidence of balance of payments problems or because private capital is less available or more expensive, so conditionality becomes stricter. On the other hand, after a quota review and an increase in the IMF's lending capacity conditionality is relaxed. This explanation of IMF conditionality presents it as a tool that is manipulated by the IMF's staff and management in order to maximise its own objective function as a bureaucracy (Dreher and Vaubel, 2003).

Other observers have suggested that conditionality is the means by which politically powerful countries advance their own global geopolitical interests or those of special interests to which they are beholden. Here the Fund is presented as an agent for such countries or interests. (8)

While the Fund would deny being an agent for private markets, it does claim that conditionality has an effect on private capital flows as well as on bi-lateral and other aid flows. In this context, conditionality is presented by the Fund as a signalling or commitment device that has a beneficial influence on market confidence; encouraging others to lend in circumstances where, without conditionality, they would have been reluctant to do so. Commitment also runs the other way. Via conditionality the Fund is committing itself to supply a specific amount of financial support to countries at certain stages, on the proviso that pre-defined conditions are met. This reduces the uncertainty that governments would otherwise face about the availability of external finance. Thus, conditionality is seen as a mechanism through which the Fund exerts a catalytic effect on other financial flows. (9) As will be discussed below, in this context conditionality need not just be an externally imposed constraint; it can be a commitment device desired by the national government.

Many different scenarios are possible. Assume, for example, that the government concerned is a unified actor--there is little or no disagreement within the government about the design of economic policy. However, the government's preferred policies may differ from those preferred by the IMF. Conditionality may then be a way of tying the government into the IME's preferred policies. It causes the government to alter its policies.

More likely, domestic policy-makers will not be unified. In such circumstances, the relationship between conditionality and ownership becomes particularly interesting. There may be significant disagreements within governments about the design of appropriate economic policy. The preferences of the finance ministry may, for example, differ sharply from those of spending ministries. Or, more broadly, there may be disagreements between those in favor of economic reform and those who are keen to retain the rents from the status quo. These disagreements may result in political and economic instability as well as uncertainty about the future. Conditionality may then be a mechanism for reducing this uncertainty, inasmuch as the executive branch of the government commits to an agreement with the IMF. The arrangement with the Fund may 'tip the balance' in favour of economic reform or may allow reformers to use the IMF as a 'scapegoat'. (10) The gains will, of course, depend on the relative merits of the alternative policy strategies--is the one favoured by the Fund in some sense superior to the others that are being considered?

It could be that the conditionality embedded in an agreement with the Fund allows a commitment to be made to compensate those who perceive themselves as potential losers from the micro- and macro-economic policies being followed by the government and this may reduce the hostility of opposition groups. In this way, the impediments placed in the path of reform by veto players or by groups who could disrupt the process of reform may be minimised. (11) Where such compensation takes the form of the adoption of economically inefficient policies, it runs counter to the IMF's technocratic tradition. Some IMF senior staff and management are quite politically savvy and there is little doubt that at times the Fund has willingly acquiesced to such policies. We shall argue that while typically distasteful to pure economists, such accommodation is likely to be a key part of more effective policy implementation.

The general message emerging from the above discussion is that conditionality should reduce uncertainty about the future conduct of economic policy. To the extent that it raises the probability that policies supported by private capital markets will be followed and not abandoned, conditionality may be expected to exert a positive influence over capital flows to countries under IMF programmes.

But all this hinges on programmes being implemented. Without implementation, the credibility of conditionality will be undermined. Conditionality cannot be an effective commitment device if governments do not comply with the agreed conditions. (12) So what is the record on implementation? Do governments carry through IMF programmes to completion?

THE IMPLEMENTATION OF CONDITIONALITY

Measuring implementation

Measuring the implementation of IMF conditionality is far from straightforward. The most convenient and widely used measure is the proportion of committed resources that are disbursed, or the rate of completion (Killick, 1995; Mussa and Savastano, 2000; Bird, 2002b). An advantage of this approach is that it provides continuous data. However, there are problems with this measure. Resources may not always be drawn on in spite of the fact that economic policy reform is undertaken. Indeed, a 'failure' to complete a programme in terms of the disbursement of resources may reflect economic 'success' in the sense that finance from the IMF is no longer required. Some programmes agreed with the Fund will be precautionary or will turn precautionary and in these cases there is no intention to draw resources; it would therefore be inappropriate to evaluate them in terms of disbursements relative to commitments. In contrast, a programme may be completed in the sense of using all the agreed resources in spite of a government failing to fully implement all the conditions originally laid down. The IMF may feel that substantial progress has been made and may allow modest deviations from targets to be accommodated through the use of waivers or modifications to the initial programme.

Since the beginning of the 1990s, another measure of implementation has become feasible as the IMF has collected data in the form of its MONA database (Monitoring Fund Arrangements) on the extent to which both the macroeconomic and structural conditions stipulated within programmes are implemented. This allows an index of implementation to be constructed. However, there are again problems with this measure; it only covers programmes that come up for review by the Executive Board and therefore excludes those that are cancelled permanently or interrupted. Since it may be assumed that it will be in these programmes that implementation is likely to be at its weakest, the MONA data will have an upward bias towards implementation. Moreover, there is only a limited amount of data as yet.

Other measures of implementation focus on whether programmes are interrupted either reversibly or irreversibly. Reversible interruption is where a review of a programme is delayed, but the programme is subsequently revived. Irreversible interruption is where scheduled reviews are not completed or, even if they are, the instalments of the arrangement are not approved. This measure of implementation does not, however, tell us whether an interrupted programme is replaced by another one. Moreover, it does not provide continuous data and only allows programmes to be classified as uninterrupted, reversibly interrupted or irreversibly interrupted. Even so, it does usefully complement the implementation index based on the MONA database by capturing cases where programmes are interrupted and, as a result, not subject to further review by the Fund's Executive Board.

Although these four measures of implementation have been found to be significantly mutually correlated with one another (Ivanova et al., 2001) the correlation coefficients are not very high, suggesting that the measures are picking up different dimensions of implementation. This needs to be borne in mind when comparing studies that use different measures.

What is the record on implementation?

Table 1 presents data showing the completion rate of IMF programmes or the ratio of disbursements to commitments. A number of features stand out. First, noncompletion is quite widespread. Indeed, over the full period 1973-1997 covered by Table 1 only about 35 per cent of arrangements were fully disbursed. Second, whereas the completion rate was fairly stable over the period 1973-1987 at between 40 and 45 per cent, it fell in the subsequent time period to as little as 20 per cent. Third, the completion rate is rather higher for stand-bys than for structural adjustment lending and much higher than for extended loans.

Using the interruption measure, Ivanova et al. (2001) report that in the period between 1992 and 1998, 44 per cent of all programmes experienced an irreversible interruption, while 70 per cent experienced either a major or a minor interruption. They also report that the average implementation index for programmes for which information was available was 76 per cent, with the macro implementation index at 80 per cent being significantly higher than the structural implementation index at 67 per cent. But, as noted above, these indices overstate implementation since irreversibly interrupted programmes are not captured by the MONA database on which the index is founded. In any event, and according to all the measures, the failure to fully implement programmes is quite widespread.

This record on the implementation of conditionality creates a conundrum. If implementation is far from complete, this may suggest a low degree of commitment. One of the basic purposes of conditionality as articulated earlier is therefore challenged. A natural response is to ask why implementation is poor. The IMF's answer is that it is associated with a lack of ownership. Without ownership there will be little commitment. IMF conditionality will therefore fail to carry credibility, and without credibility it will fail to generate a catalytic effect. But just how useful is the idea of ownership when considering the implementation of IMF conditionality? And just what is ownership? We now turn to that question.

OWNERSHIP AND IMPLEMENTATION

As a concept, 'ownership' definitely has its attractions. It is widely used in a number of contexts and contains sufficient ambiguity to accommodate a range of interpretations. In part, it is the ambiguity that contributes to the wide usage of the term. Moreover, there is a commonsensical appeal to the basic notion that people will be more committed to ideas that they perceive as their own.

At the same time, there are problems with the concept. The very ambiguity that is part of its appeal also implies that it is difficult to use operationally. What does it really mean and how can it be measured and monitored? Is there a sharp distinction between ownership and lack of ownership, or are there degrees of ownership? Do governments need to feel ownership of a programme in order to implement it and, if so, what degree of ownership do they need to feel? Does ownership guarantee implementation; can implementation occur without ownership? Is ownership compatible with conditionality or does conditionality, by its very nature, undermine ownership? Is 'national' ownership the same thing as ownership by the government, and which is more relevant in the case of implementing IMF conditionality? What is the connection between ownership and governance, and how far should the IMF become involved in ownership and governance issues? For example, would it be better for the Fund to deal with an authoritarian regime that is committed to pursue IMF-approved conditionality in spite of strong but contained opposition, or with a democratic regime where there is vocal and effective opposition to conditionality both within the various components of government and across elements of civil society, sufficient to threaten implementation? Is ownership in some sense exogenous and beyond the control of the Fund, or can it be influenced by means of persuasion? Is it legitimate or even appropriate for the Fund to try and influence public opinion?

In this section we investigate some of these issues. (13) To begin, it is important to establish some view about the IMF's terms of reference and its areas of responsibility. What is the Fund's role? There is probably no 'right' answer to this question. Some may claim that the Fund should take a position on the acceptability of different political regimes and on what constitutes good governance, and should then take decisions that discriminate against unacceptable political regimes and bad governance. Clearly, the Fund should not be indifferent about corruption or about the violation of human and civil rights. But there are constraints on just how involved the Fund should become. For example, does the Fund have the right to 'impose' social policies on democratically elected governments? Where should the line be drawn between what is legitimate and illegitimate for the Fund to seek to influence.

These are all arguable issues. In our view, however, there is a strong case for having the IMF take a fairly narrow view of its responsibilities. This is based on two major considerations. One is that attempting to achieve broader objectives seems likely to reduce the IMF's ability to achieve its core responsibilities. Secondly, the IMF is not an appropriate body to decide what is right with respect to such broader issues. The desire of various groups to use IMF leverage to help achieve their goals is quite understandable, but in our judgement this can greatly damage the IMF's ability to achieve its core objectives.

The discussion that follows is oriented towards raising the degree to which IMF conditionality is implemented, subject to certain political constraints. We assume that it is the role of legitimate governments to govern, and not that of unelected bureaucrats from the IMF. We also implicitly accept, however, the idea that it is reasonable for the Fund to make its loans conditional in an attempt to improve the economic performance of countries where it must have been poor economic performance that contributed to their referral to the IMF in the first place. (14) In this context, ownership is seen as a means to an end, rather than an end in itself. The end is to encourage the implementation of appropriate economic reform. The 'end' contains two elements. The first relates to the design of conditionality and the appropriateness of the policies it incorporates. The second relates to the implementation of these policies. It seems reasonable to assume that governments are more likely to implement policies that they think are appropriate.

What does ownership mean; is it an operational concept?

Like a lot of similar terms, many people would feel comfortable that they understood in general what is meant by 'ownership'. However, they would have some difficulty in coming up with a tight definition of it. The IMF's staff (International Monetary Fund (IMF), 2001b) makes the following suggestion. 'National ownership refers to a willing assumption of responsibility for a programme of policies, by country officials who have the responsibility to formulate and carry out these policies, based on the understanding that the programme is achievable and is in the country's best interests'. They go on to argue that 'conditionality, if well designed and established through a mutually acceptable collaborative process, can promote and strengthen ownership, in particular by demonstrating the authorities' commitment to a course of action'.

Khan and Sharma (2001) develop these ideas further by noting that '.... ownership is intricately connected to questions of trust in domestic institutions, the effectiveness of political structures, and whether the government negotiating on behalf of its citizens has sufficient support to speak for a fair majority' (p. 15). Boughton and Mourmouras (2002), in a similar vein, acknowledge that 'judging the breadth and depth of ownership is a delicate proposition that requires knowledge and understanding of the country's political economy and sensitivity to the limits of the Fund's own role. Are the economic officials fully committed? Do they have support throughout the country? Are there groups in parliament or elsewhere with the power and the incentive to block implementation?' (p. 90).

These are all sensible statements about the meaning of ownership, but they illustrate just how vague the term is. The first definition equates 'national' ownership with the willing assumption of responsibility by a subgroup of the government. Furthermore, it could be seen as implying that ownership depends on implementability (rather than the other way around). The quote from Khan and Sharma offers a broader definition of national ownership by requiring that the government has 'sufficient' support to speak for a 'fair majority'. But what is 'sufficient' and what is 'a fair majority'? The quotation from Boughton and Moumouras suggests that ownership requires the support, or at least the indifference, of non-governmental groups in parliament and also possibly interest groups outside parliament, provided they have the power to block (or postpone?) implementation. This implies a much broader and more difficult definition of ownership. Not only is there the question of identifying these groups, but there is also the question of the extent to which opposition groups in parliament, or interest groups in society as a whole, should be given a say by the IMF in the design of economic policy. For example, if the Fund engages in direct discussion with some of these groups, this could be regarded as anti-democratic.

Moreover, broader participation in the discussions that culminate in the signing of IMF agreements does not guarantee ownership. Special interest groups may be consulted merely to discover that their advice is then ignored. This seems hardly likely to ensure that they feel as though they own the resulting programme. Indeed they may feel even more antagonistic towards it.

Participation is not the same thing as ownership. It may or may not contribute to ownership depending on how it works. Do special interest groups feel that they have been genuinely listened to? Do they feel that a sincere attempt has been made to accommodate their views? As a result of discussions, do they better understand and appreciate why the government may be unable to accommodate their views? Or do they feel that the participatory process has been a purely cosmetic exercise which allows the government and the Fund to express the rhetoric of ownership without ever intending to allow it to influence the reality of policy?

Ownership, conditionality and implementation

Even if ownership could be made operational, would it be consistent with conditionality? Imagine two situations. In the first, the government (and society) is strongly committed to the reforms and policies advocated by the Fund. The programme is nationally owned. But if strong ownership maximises the chance of implementation why is conditionality required in these circumstances? Indeed, it could impair implementation by giving the impression that policy has been imposed from outside, thereby reducing national sovereignty (and ownership).

In the second case, assume that there is weak national ownership. The government is not committed to the programme. But if ownership is necessary to ensure implementation, conditionality in these circumstances will not work. The conclusion seems to follow that if ownership is indeed required in order to deliver implementation, conditionality is either redundant or ineffective. (15)

This would be too strong a conclusion. For example, conditionality may usefully endorse a domestically owned programme. Conditionality may add value via the IMF's seal of approval and this may improve a country's access to external finance. Moreover, there is the possibility that conditionality helps to lock in future governments. It may help guard against a subsequent loss of ownership. (16) The above example disaggregates the unified actor by adding a time dimension. Also important is recognising that, even at one point in time, the unified actor model will seldom be appropriate for considering the relationship between the IMF and its client countries. (17)

With a pluralist society, conditionality generally helps those groups favouring reform and makes implementation more likely? (18) The problem here is that where there are powerful and opposed factions in government and in society it is difficult to claim that there is strong national ownership, and, without this, conditionality may be ineffective. This raises the crucial issue of partial ownership to which we return below.

Does implementation depend on ownership?

Ownership will surely, in principle, assist implementation. However, the required degree of ownership may only relate to those enacting the agreed policies and their ability to stay in office. There may be an ownership threshold. Below the threshold ownership will be too weak to sustain the programme of reform agreed with the IMF. Above the threshold ownership will be strong enough to motivate implementation, and stronger ownership will confer little benefit in terms of implementation.

Disaggregation of programmes may be needed. Certain aspects of a programme can be more strongly owned than others and therefore the probability that these aspects are implemented will be higher. A special interest group may be opposed to a particular policy instrument that erodes the rents they enjoy. They may then be able to prevent the implementation of this particular policy without necessarily preventing the implementation of other parts of the programme to which they are not opposed. (19)

And while ownership is helpful, it is clearly neither a necessary nor sufficient condition for implementation. Even where there is strong national ownership of an entire programme, there may be other factors that disrupt implementation. For example, there may be external shocks that drive a programme off course, or the targets set may have been over-ambitious. On the other hand, implementation may be feasible even in the absence of ownership. Bird (2003) presents a model of implementation based on marginal benefits and marginal costs. While the incentive to implement depends on whether the marginal benefits exceed the marginal costs, ownership is related to the costs of implementation which, in part, depend on the difference between the policy preferences of the government and those of the Fund. The bigger this difference, the weaker is ownership. However, if the benefits of implementation are increased by, for example, the Fund providing larger amounts of finance, it is possible that implementation could improve at the same time as ownership declines. Ownership helps, but it is not a necessary condition for implementation.

What is really needed is a better understanding of the determinants of implementation that will then enable us to examine how factors related to ownership fit in. This is made difficult because, as established earlier, it is hard to measure ownership, or even to define what is really meant by the term. Thus, existing empirical studies use a range of political variables that are assumed to capture elements of ownership in terms, for example, of the strength of opposition groups within parliament. However, the current contribution of this research is somewhat limited. Not only are there doubts as to whether the chosen variables are actually capturing ownership, but also the empirical evidence is mixed. Some studies find that the political economy variables that are intended to reflect ownership are indeed statistically significant in explaining implementation, but others fail to confirm these results. Empirically, therefore, the jury is still out in terms of quantifiable relationships between ownership and implementation.

To date perhaps the most comprehensive single study of the implementation of IMF programmes has been conducted by Ivanova et al. (2001). They analyse the implementation of 170 programmes approved between 1992 and 1998, using multiple measures of implementation in the form of reversible interruptions, irreversible interruptions, an overall index of implementation derived from the MONA database, and the ratio of disbursements to commitments. They test econometrically the effects on implementation of political conditions in the borrowing country that could be interpreted as proxies for ownership, IMF effort, conditionality and initial and external conditions.

They summarise their findings as follows: 'on the one hand, the implementation of IMF-supported programs is strongly influenced by recipient countries' domestic political economy. Strong special interests, lack of political cohesion, inefficient bureaucracies, and ethno-linguistic divisions are strongly associated with weak program implementation. The strong association between program implementation and political economy variables is robust across different econometric specifications. On the other hand, initial economic conditions, IMF effort and the breadth and depth of conditionality do not seem to materially influence program prospects when they are properly instrumented for.' (Ivanova et al., 2001, p. 4).

In stressing the overall significance of political economy variables, this research confirms earlier work that examined the success of World Bank programmes. Dollar and Svensson (2000) find that these programmes are more successful, according to the Bank's Operations Evaluation Department, where the political environment is stable, there are few ethnic divisions, and governments are democratic and have not been in power for long. In a similar vein and based on a study of major interruptions in the context of 36 ESAF programmes with the IMF, Mecagni (1999) discovers that they often depend on 'political disruptions serious enough to call into question the continuing authority of the government ... the nature of political upheavals and the intensity of political and ethnic turmoil varied, but all cases were characterised by a severe reduction of the authorities' ability to commit credibly to and implement adjustment policies" (p. 9). Mecagni also finds some statistical evidence to support the suggestion that the poor implementation of programmes may be linked to external shocks such as export shortfalls or shortfalls in external financing; something also found by Killick (1995).

While all of these studies share the common theme that political variables are important when seeking to explain implementation, there are important differences between them in terms of the precise nature of the relationships found. For example, Dollar and Svensson do not include a variable to capture the importance of special interests within parliament, whereas, building on recent theoretical work, Ivanova et al. (2001) find this to be highly significant statistically. The only currently available proxy for special interests--contained in the World Bank's political data base--looks only at the presence of special interest parties in legislatures, and hence has only a very loose connection to the types of special interest influence emphasised in public choice analysis. The studies also differ over the impact of a government's length of tenure on implementation as well as on whether a democratic orientation makes any difference. Indeed, whereas Dollar and Svensson find that democracy aids implementation and Ivanova et al. find no statistical link, in another study of IMF programme interruptions Thomas (2002) discovers that autocratic regimes have a better record of implementation.

In an as yet unpublished paper, Joyce (2003) finds that democracy helps and that politically more open regimes have a superior record of implementation. Using various measures of special interests within government, he finds no statistically significant connection between them and implementation, nor does he find a link between the cohesion of the executive and legislative branches of government, and implementation. He finds that regimes that have been in power for longer are less likely to complete programmes and that recently elected governments are more likely to complete them. His results also suggest that more open economies are more likely to complete programmes, which could suggest that proximity to the Fund's underlying economic paradigm is relevant.

There are some resonances between the findings reported by Joyce who examined 77 programmes over the period 1975-1999, and those discovered by Dreher (2003) who examines programme completion across 104 countries over the period 1975-1998. Dreher finds 'no robustly significant coefficients' when he tests for political explanations in terms of government fractionalisation, the political leaning of the chief executive's party, the existence of autonomous regions, the political power of the leader, the degree of political cohesion and various other political variables. He does, however, find some not completely robust evidence that IMF programmes are more likely to be interrupted prior to elections, and that while democratic regimes are generally associated with less compliance, the increase in the probability of interruption at election times is less severe in democracies. He also finds that initial economic conditions in the form of government consumption relative to GDP, short-term debt relative to GDP and GDP per capita exert a statistically significant effect on implementation. Interruptions appear to vary positively with the first two of these variables and negatively with the third.

Ivanova et al. (2001) also find some evidence based on bi-variate correlations that implementation is affected by the severity of some initial conditions but, as noted earlier, this relationship loses statistical significance once political variables are included. Earlier research by Killick (1995) suggested that the degree to which programmes are completed is positively related to the amount of finance provided by the Fund in relation to the size of the initial current account deficit, although Ivanova et al. do not find a similar relationship when the size of the loan is expressed in relation to the borrowing country's quota.

While the recent quantitative literature helps offer some guidance to the IMF about the conditions to look for in judging the likelihood of programmes being effectively implemented, it is far from sufficiently robust to be used as a strong guide for the IMF's decision making. For the foreseeable future, there is no alternative to the careful formulation of highly fallible judgements about the prospects for implementation. Modern political economy analysis can be extremely helpful in highlighting political considerations to which the IMF should pay attention, but it does not provide many easy answers. It would not be hard to become daunted by the difficulties of making good forecasts, but the situation becomes brighter if we apply a standard of whether such political economy analysis could have enough validity to lead to a substantial improvement in the IMF's track record. (20) Perhaps more difficult than the political forecasting is creating conditions where the IMF feels able--and is given the freedom by its major shareholders--to say no to programmes whose prospects for effective implementation are widely recognised to be poor.

Just how good the prospects for implementation need to be to justify loans is, of course, an issue of delicate judgement. There will be no one correct scientific answer, but it is an issue to which the IFIs should give serious attention. In doing so, the prospect of doing good in any one case will need to be balanced against the costs to the credibility of the institution if things turn out wrong. This suggests that because of the importance of the credibility for the catalytic effect on private capital flows, the IMF should set a higher threshold of probability of success on its crisis-related loans than should the World Bank on its project loans. (21) There is mounting support within the IMF for being more careful in agreeing to programmes. For example, the Staff Statement of Principles underlying the Guidelines on Conditionality stresses that 'the guidelines also note that the need for ownership implies selectivity'. Such recognition is not a sufficient step, but it is certainly a necessary one.

Increased selectivity will require a sharp tilt in the way the IMF negotiates. Understandably, the IMF likes to view itself as a technocratic institution that puts its emphasis on deciding what policies are needed to solve the problem at hand. If the government in question agrees to such policy packages, then a deal is made. What the new view requires is that the IMF must also make a judgment about the ability and/or willingness of the government to implement its promises.

It is already the duty of the Managing Director of the IMF in effect to certify that he believes that not only is a proposed programme a good one, but that the prospects for its implementation are also good. The latter has typically not been a point of public emphasis.

In the new view, the Fund must countenance telling governments that their promise to implement an agreed programme is not believed. This is by its nature a much more sensitive topic than negotiating the policy content of conditionality. Furthermore, while the Fund is full of good economists, and some of these have developed excellent political judgments, few at the Fund have expertise in political science. Some of Fund's management and staff are beginning to recognise the need for such political economy expertise. This runs against the traditional culture of the Fund and will face considerable opposition, but its importance is inherent in the recognition of the need to worry about ownership and the prospects for programme implementation.

It is important to emphasise that the need for the Fund to develop a capacity for political economy analysis does not imply greater 'politicisation' of the Fund. The latter commonly refers to political influence on the Fund to follow policies for political rather than technical reasons and it is quite understandable and appropriate that the Fund's management and staff should oppose such politicisation. Of course, the appropriate dividing line between undesirable politicisation and desirable accountability is the subject of conflicting opinions and it is possible that explicit recognition of the need for the Fund to make political economy as well as technical judgments could subject it greater outside manifestations. On the other hand, it could provide the Fund with more ammunition to counter such pressures.

It should also be recognised that concerns with ownership and awareness of this ever does not imply that the IMF would become more intrusive in the political processes of borrowing countries. To be aware.

POLICY IMPLICATIONS

The analysis in this paper points in a number of policy directions and carries a number of policy messages. First, attempting to impose conditionality that is significantly at odds with a government's preferences is unlikely to be successful unless the Fund supplies substantial amounts of financial support. (22) Second, while the Fund will need to retain the right to decide whether the basic macroeconomic framework is sufficient to get the job done, it should allow governments as much discretion as possible to design their own programmes of economic reform. This would maximise the ownership of those who have the responsibility of carrying through the programmes. The Fund should draw the line at negotiating outside and beyond the government. But it would also be wise for the Fund to inform itself about the possible domestic political constraints to implementing conditionality, and to make a discussion of these constraints and ways in which governments intend to overcome them, part of the negotiations. (23) Governments would thereby be encouraged to consult widely and to canvas support for some policy. An implication of this is that it would be better for the Fund to support some programmes that it perceives are less good technically in order to secure a better chance of implementation. This is illustrated in Table 2 where column 2 gives a rating of the contents of conditionality involved in a range of programmes based on 'technical' criteria and using a rising scale of 1-10.24 Column 3 gives the percentage probability of the programmes being implemented based on an evaluation of political factors. Column 4 gives the overall score for each programme calculated by multiplying columns 2 and 3 together. It therefore marries the technical and the political economy dimension of various forms of conditionality. According to the illustrative numbers used in Table 2, the Fund should be indifferent between programmes 6 and 7. It should, however, favour either of these programmes over programmes 1-5, in spite of the fact that each of these may be viewed as technically superior.

From the Fund's perspective, conditionality would become an exercise in the theory of the second best. The first best solution would be to have the government and society fully committed to programme 1. This would give an overall score of a perfect 10. However, in the imperfect real world, all the other programmes (apart from programme 10) are likely to be more effective at improving economic performance.

As a consequence of the above, conditionality might become substantially self-designed by governments, but the IMF would still provide a mechanism for monitoring progress in implementation. The macroeconomic framework that could constitute mandatory conditionality from the IMF would impose constraints on the other elements of programmes. It would, in effect, impose a discipline on other policies. The logic here is that the balance of payments is a constraint. Countries often end up borrowing from the Fund because they have paid insufficient attention to this constraint. IMF conditionality in the form of mandating a macroeconomic framework reestablishes the constraint. But within this constraint governments would be granted freedom to design their own programmes, which would then be monitored by the Fund.

There are two rather different messages for the current thrust of IMF policy towards conditionality. The first message is broadly supportive of streamlining as a way of reducing the scope of the conditionality that is, in some sense, designed in Washington DC. An external benefit of this is that greater ownership may follow.

The second message is that there may be little to be gained and some prospect for loss from attempting to operationalise a broad concept of ownership by having the Fund engage a wider array of society in negotiating conditionality. This should remain the responsibility of legitimate governments. However, this is not to argue that the Fund should ignore the political economy dimension of implementing conditionality. Indeed, this dimension should be discussed much more fully and 'scientifically' than it has in the past as part of the negotiations over conditionality. Moreover, the Fund might be called upon by governments to help explain to a wider audience the logic behind the programmes that are supported, and by implication why it is that alternative programmes have been rejected.

CONCLUDING REMARKS

The effectiveness of IMF conditionality depends upon both its design and its implementation. The record shows that many IMF-backed programmes are not fully implemented. For some years, the Fund has sought to explain this in terms of a lack of commitment on behalf of the governments concerned or a lack of political will. More recently, these explanations have been repackaged using the concept of 'ownership'.

A literature on ownership has begun to emerge and, alongside this, policy changes have been introduced with the intention of fostering it. The thrust of the policy has been to streamline conditionality, reducing its scope and involving a wider array of constituencies in the discussions that culminate in the signing of agreements. The thinking seems to be that narrower conditionality and broader participation will lead to greater ownership and therefore to better implementation.

This paper raises a series of questions about some aspects of this approach. Ownership is an imprecise term and this limits its operational value. In particular, broader participation does not necessarily lead to stronger ownership. Attempts to strengthen ownership by means of broadening the range of people with whom conditionality is discussed risks involving the IMF too heavily in domestic politics.

It is, however, appropriate that the Fund should pay close attention to implementation. The policy objective should be to improve the implementation rate of Fund-backed programmes. To achieve this, it is useful to consider how the benefits and costs of implementation evolve. In this context, many of the political economy variables that are conventionally linked to ownership may indeed be significant. The Fund needs to be aware of the political economy dimension of implementation and to make this a part of its discussions with the governments with which it negotiates conditionality. In this way, the Fund can form a view, albeit imprecisely, of the probability that alternative programmes will be implemented. This, in turn, may mean that the Fund supports a programme that in purely economic terms it views as being inferior. Again, the objective should remain that of maximising the beneficial economic impact of programmes subject to political constraints. Thus we advocate a narrow concept of ownership that corresponds to the expected probability of sufficient political support for effective implementation.

The desirability of developing analytical capacity at the IMF to engage in such political economy analysis is a major implication of the new emphasis on ownership. Even with the best analysis possible, there will typically still be a great deal of uncertainty concerning the probabilities of implementation. But the IMF's performance could be improved substantially just by giving more attention to cases where there is widespread agreement among independent observers that prospects for implementation look poor. More difficult than this will be fostering the IMF's ability to say no when the chances of implementation are low. This in turn will require the willingness of its principal stakeholders to let it say no, even when their geopolitical interests dictate otherwise. This will require an appeal to their enlightened self-interest, emphasising that the lack of selectivity in past IMF programmes may have seriously eroded the effectiveness of the institution.

Our analysis has important implications for the design of conditionality. One element of conditionality should seek to impose an external discipline on domestic economic policy. This would relate to restoring macroeconomic equilibrium and alleviating balance of payments constraints. Much as consumers tend to accept that they face a budget constraint, governments may also be expected to accept this aspect of IMF conditionality; particularly as it may have been the attempt to ignore the constraint that has led to the Fund's involvement. During a crisis, governments may, in any case, place a relatively high priority on restoring macroeconomic stability. By definition, the situation without the Fund has been adjudged to be less desirable.

The second element of conditionality relates to structural adjustment and involves issues of microeconomic efficiency rather than macroeconomic stability. This needs to be designed as far as possible by those who have the responsibility for carrying through the associated policies. The Fund's role in designing this element of conditionality should be more re-active than proactive. It should be more advisory and less mandatory. At the same time, the Fund should retain the role of monitoring the implementation of the resulting programmes and should be able to reject those that it regards as ill thought out and completely inappropriate. As part of the exercise, the Fund would need to assess the probability of implementation by taking into account the domestic political realities faced by governments and governments' plans for overcoming political impediments. Governments would therefore be encouraged to consider issues pertaining to broader national 'ownership' in the context of persuading the Fund that programmes would be implemented.

These reforms would in effect seek to reallocate the locus of ownership of programmes between governments and the Fund. They would be directed towards making more realistic the Fund's claim that programmes are 'country programmes' backed by the IMF and not 'IMF programmes'. They would, however, play down the concept of ownership in the sense of broad participation and play up the implementation of conditionality and the range of factors, including political economy ones, that influence implementation. (25)

The development of effective national ownership even in the narrow sense advocated here will often be a time-consuming process. However, in today's world of high capital mobility, crises can develop quickly and failure to deal with them promptly can be disastrous. As Khan and Sharma (2001) argue '... one drawback with the options strategy is that the discussion of alternative options is bound to stretch out the negotiation process. In crisis situations, there may simply not be the time to examine in detail the merits of different policy packages ...' (p. 19).

In the middle of a crisis, there may be little opportunity for time-consuming discussions. This suggests the need to consider developing a new IMF facility for lending short term in order to allow some breathing room for the negotiation of owned policies. Functionally, this would be a quasi-lender of last resort type crisis lending facility, but, because of strong opposition to giving the IMF such power, it might be better to present it as a bridging facility. (26)

Preconditions for access to such bridging loans would need to be developed. This will be no easy task, as reflected by the failure of the Fund's Contingent Credit Lines which have recently been abandoned. The case for such a new, rapidly disbursing IMF lending facility is one of the least recognised implications of the new political economy of ownership.

Table 1: Fraction of IMF loan actually disbursed under each
arrangement, distribution by quartiles (x=fraction of total IMF
loan disbursed under each arrangement) (a)

 0.50 [less
 0.25 [less than or
 than or equal to]
 equal to] x <0.75
 x<0.25 x <0.50 (in per cent)
All arrangements (b)
 1973-1977 36.5 7.1 5.9
 1978-1982 19.4 16.1 10.5
 1983-1987 12.9 15.8 19.4
 1988-1992 17.5 15.1 20.6
 1993-1997 (c) 27.0 19.1 26.2

Full period 21.6 15.3 17.6
(1973-1997) (c)
of which:
 Stand-by (c) 23.1 13.4 15.0
 EFF (c) 33.3 22.2 19.0
 SAF/ESAF (c) 9.0 18.9 27.0

 0.75 [less
 than or Fully
 equal to] disbursed Number of
 x <1.0 (x=1.0) arrangements
All arrangements (b)
 1973-1977 5.9 44.7 85
 1978-1982 12.9 41.1 124
 1983-1987 7.9 43.9 139
 1988-1992 14.3 32.5 126
 1993-1997 (c) 11.3 16.3 141

Full period 10.7 34.8 615
(1973-1997) (c)
of which:
 Stand-by (c) 9.5 39.0 441
 EFF (c) 15.9 9.5 63
 SAF/ESAF (c) 12.6 32.4 111

Source: Mussa and Savastano (2000)

(a) Calculated as the ratio of the total purchases made to the full
amount of IMF resources committed under each arrangement.

(b) Includes stand-by arrangements, EFF arrangements, and arrangements
under the SAF and ESAF. Excludes STF arrangements, and drawings under
the first credit tranche and the CCFF.

(c) The distribution of the ratio x for the 1993-1997 period is biased
(downward) by the inclusion of arrangements with expiration date
posterior to 1997. This bias is also present in the distributions
reported for the full period.

Table 2: The ranking of IMF-supported programmes

1 2 3 4

Programme Technical rating Probability of Overall score
 (based on economics and implementation (based on
 scored 1-10) (based on domestic political economy,
 politics in columns 2 x 3)
 percentages)

 1 10 10 1.00
 2 9 15 1.35
 3 8 25 2.00
 4 7 35 2.45
 5 6 45 2.70
 6 5 60 3.00
 7 4 75 3.00
 8 3 85 2.55
 9 2 95 1.90
10 1 100 1.00

(1) For a review of such arguments from the left and right, see Willett (2001).

(2) For a review of the literature dealing with the effectiveness of IMF programmes, see ul Haque and Khan (1998). Bird (2001a) offers a broader assessment of the extent to which IMF programmes work. The IMF's most recent review of conditionality is reported at length in IMF (2001a). Also, see IMF (2001b). Dreher and Vaubel (2003) present a succinct summary of developments in IMF conditionality, showing how the breadth of conditionality increased during the 1980s and 1990s as greater emphasis was placed on structural adjustment and supply-side conditions.

(3) In this view, supply-side policies should be considered only where they are directly relevant for macroeconomic and exchange rate stability. Financial sector problems have played an important role in a number of recent crises and are now widely considered as a core issue for the IMF.

(4) This paper focuses on IMF conditionality rather than conditionality in general. We do not explore the question of World Bank conditionality or the ways in which the IMF and World Bank differ in terms of their attitudes towards conditionality. It is, however, probably reasonable to claim that the World Bank is less sanguine about the extent to which conditionality is compatible with country ownership. The significance of ownership was also probably identified rather earlier by the World Bank than by the IMF. For an excellent discussion of the concept of ownership and a review of the early work on it as it applies to the international financial institutions, see Killick (1998).

(5 One of the earliest reviews of conditionality may be found in Williamson (1983). Guitian (1995) provides a summary of the Fund's general approach to conditionality. For a briefer summary of the purposes behind conditionality see Bird (2001b). The IMF's web site (www.imf.org) also provides much useful information about conditionality.

(6) Of course, our ethical judgments about such 'exchanges' will depend on the range of options open to the actors. If there is no practical choice but to accept, then the 'exchange' is clearly coercive. The bandit's 'your money or your life' would be such a 'transaction'.

(7) See, for example, Drazen (2002) and Willett (1988).

(8) This is part of a broader literature that examines the extent to which various aspects of the Fund's operations are influenced by political factors. This research initially focused on IMF lending and the extent to which powerful shareholding countries exerted a discernible effect on which countries received financial support from the IMF (Thacker, 1999; Bird and Rowlands. 2001a; Barro and Lee, 2001). More recently research has also examined political influences over the size of loans (Oatley and Yackee, 2004) and various dimensions of conditionality (Dreher and Jensen, 2003). Gould (2003) argues that conditionality has been modified to better serve the interests of private international creditors. Feldstein (1998) has also argued that during the East Asian crisis in 1997 and 1998 the IMF was used as a conduit to push policies that favoured industrial countries, private capital markets and multinational enterprises.

(9) Bird and Rowlands (2002) evaluate the logic behind the catalytic effect, They also review the existing empirical evidence and conclude that there is little support for the hypothesis. A somewhat more favourable conclusion is drawn by Mody and Saravia (2003).

(10) See, for example, Krueger (1998) and Vreeland (2003).

(11) As Rodrik (1996) explains, economic reform may be prevented by opposition groups believing that they will lose even when they will not. Thus, policies that actually favour a majority may still be rejected. In principle, conditionality could be used as a device for circumventing this problem.

(12) For a more detailed discussion of the credibility of IMF programmes and how this influences the usefulness of IMF conditionality as a signalling device, see Bird (2002b). One of us has written a couple of papers that analyse implementation and completion in some detail (Bird, 2002b, 2003) and, in places, what follows draws on these papers.

(13) As explained in the Introduction, 'ownership' in the context of programmes supported by the international financial institutions is a relatively new concept. At the same time, it is a concept that has spawned a number of recent papers. From amongst these, particularly useful are those by Boughton and Mourmouras (2002), Drazen (2001), Khan and Sharma (2001), and Mayer and Mourmouras (2002).

(14) We judge it as being largely beyond the role of the IMF to define the legitimacy of governments. This needs to be determined elsewhere. For a useful discussion of governance issues involving the IMF, see Woods (2000, 2001).

(15) For a clear and concise presentation of the view that conditionality is fundamentally inconsistent with ownership, see Killick (1997).

(16) The degree of ownership may change over time; there may be stronger ownership of some parts of a programme than others. Where the programme is phased over time the degree of ownership may therefore depend on the contemporary preponderance of preferred policies. At the outset of a programme, the government may recognise the need to strengthen the balance of payments since this constraint has, almost by definition, become an effective one. There may therefore be a relatively high degree of national ownership. However, as time passes, the balance of payments constraint may relax and other policy objectives may take priority for the government. Its focus may shift to regenerating economic growth and increasing consumption. It may therefore feel less ownership of policies that continue to emphasise fiscal and monetary rectitude. Bird (2003) examines these issues in much more detail in the context of explaining implementation and why it is that a government may initially sign an agreement with the IMF that it does not then implement.

(17) See Drazen (2001) and Willett (2002).

(18) It is via this sort of argument that Drazen (2001) argues that conditionality, which is inconsistent with ownership in the context of a government opposed to reform and unnecessary where the government already owns the programme, may still fulfil a function in cases where government (or society) is not unified. In these circumstances, conditionality is being used as a way of overcoming an impasse that is associated with disunity. It does this by increasing the leverage exerted by one group relative to the rest; a prime concern will be to gain access to the Fund's financial support. Conditionality would only increase the degree of national ownership in these circumstances where it is used, to some extent, to tie these resources to uses that buy out opposition groups.

(19) This overlooks the point that programmes may be holistic. If a particular reduction in government expenditure is removed--for example the elimination of a subsidy or a particular tax hike rejected, this will mean that it becomes more difficult to attain broader fiscal targets.

(20) A significant amount of political forecasting using agent-based models has been recently applied in the multilateral donor community. See Kugler and Feng (1997) for an overview, Feder (1987) for the US Government's independent evaluation and Abdollahian and Root (1999) for a specific application of creating politically sustainable structural reforms in the wake of the Korean financial crisis.

(21) In cases of particularly high uncertainty, it may be desirable to require actual passage of some measures as preconditions for programmes. For discussions of the appropriate mix of pre- and post-loan conditions, see the analysis in Willett (2003). As Khan and Sharma (2001) note 'ex ante penalties have limited credibility, since they are unlikely to be enforced ex post ... this is one of the reasons why there have been suggestions that the IMF should lend only to pre-qualified countries with good policy environments' (p. 9).

(22) And even in these circumstances there may be some theoretical inconsistencies. Larger amounts of financing may be seen as a means of reducing the emphasis on short-term adjustment based on reducing aggregate demand. This argument therefore appears to be that governments will only accept conditionality that emphasises consumption sacrifices if finance is provided that allows these sacrifices to be avoided in the short term.

(23) Bird (2003) discusses various ways in which this might be done. These differ slightly from the ideas expressed by Willett (2002).

(24) We do not explore how this rating would be calculated and what the 'technical criteria' would be. But in a sense this is unimportant to the argument. The rating simply shows the IMF's ranking of alternative programmes independent of considerations of political feasibility and implementability. For clarity of exposition, we assume that there is an inverse relationship between the Fund's ranking and the probability of implementation but this need not necessarily be the case. Our principal argument is simply that the Fund needs to take into account both dimensions of programmes in order to calculate the probable welfare gains.

(25) Although we have not discussed it in this paper, ownership could also be considered from the viewpoint of the IMF. If a government does not own a programme does it follow that the IMF does? Is the IMF a unified actor, or does the programme favoured by the Fund represent compromises and coalitions within the IMF? There is some evidence that the Fund tends to be overambitious in setting targets (Baqir et al., 2003) with some Fund staff realising that although the targets may not be achieved, they are necessary in order to get the programme approved by the Executive Board. It might then be possible to envisage programmes that neither the IMF nor the government owns, weakening implementation still further. This in spite of the fact that the Fund's Managing Director implicitly certifies the prospects for implementation as being good when approving programmes.

(26) For more on these issues, see the analysis and references in Willett (2003).

REFERENCES

Abdollahian, M and Root, H. 1999: Economic crisis & the future of oligarchy. In: Institutional Reform & Democratic Consolidation. Hoover Press: Stanford, CA.

Baqir, R, Ramcharan, R and Sahay, R. 2003: IMF program design and growth: What is the link? Mimeograph.

Barro, RJ and Lee, J-W. 2001: IMF programs: Who is chosen and what are the effects? Paper presented at Second Research Conference. IMF: Washington, DC, USA.

Bhagwati, J. 1998: The capital myth: The difference between trade in wiggets and dollars. Foreign Affairs 77(3).

Bird, G. 2001a: IMF programs: Do they work? Can they be made to work better? World Development 29(11): 1849-1865.

Bird, G. 2001b: IMF programs: Is there a conditionality Laffer curve? World Economics 2(2): 29-49.

Bird, G. 2002a: The credibility and signalling effect of IMF programs. Journal of Policy Modelling 24: 799-811.

Bird, G. 2002b: The completion rate of IMF programs: What we know, don't know and need to know. The World Economy 25(6): 833-847.

Bird, G. 2003: The implementation of IMF programs and how to improve it. Working paper, October, Surrey Centre for International Economic Studies: Surrey.

Bird, G and Rowlands, D. 2001a: IMF lending: How is it affected by economic, political and institutional factors? Journal of Policy Reform 4(3): 243-270.

Bird, G and Rowlands, D. 2002: Do IMF programs have a catalytic effect on other international capital flows? Oxford Development Studies 20(3): 229-249.

Bonghton, J and Mourmouras, A. 2002: Is policy ownership an operational concept? International Monetary Fund Working paper no. 02/72. In: Gilbert, C and Vines, D. (eds). The IMF and the International Financial Architecture. Cambridge University Press: Cambridge pp. 225-253.

Dollar, D and Svensson, J. 2000: What explains the success or failure of bank-supported adjustment programmes? Economic Journal 110(446): 894-917.

Drazen, A. 2001: Conditionality and ownership in IMF lending: A political economy approach. Paper presented at Second IMF Research Conference, 29-30 November 2001.

Drazen, A. 2002: Political economy in macroeconomics. Princeton University Press: Princeton, NJ.

Dreher, A. 2003: The influence of elections on IMF program interruptions. Mimeograph, Journal of Development Studies 39 (6): 101-120.

Dreher, A and Jensen, N. 2003: Independent actor or agent? an empirical analysis of the impact of US interests on IMF conditions. Mimeograph.

Dreher, A and Vaubel, R. 2003: The causes and consequences of IMF conditionality. Mimeograph.

Feder, S. 1987: Factions and policon: New ways to analyze politics. Studies in Intelligence, US Government, declassified 1994.

Feldstein, M. 1998: Refocusing the IMK Foreign Affairs 77: 20-33.

Gould, E. 2005: Money talks: Supplementary financiers and International Monetary Fund conditionality. Paper presented to a conference on The Impact of Globalization on the Nation State from Above: The IMF and the World Bank, Yale University, April.

Guitian, M. 1995: Conditionality: Past, present and future. IMF Staff Papers 42(4): 792-835.

Haque, NUI and Khan, MS. 1998: Do IMF supported programs work? A survey of the cross-country empirical evidence. IMF Working paper no. 98/169.

International Financial Institution Advisory Commission. 2000: Report of the IFIAC (the Meltzer Report). US Government Printing Office: Washington, DC.

International Monetary Fund. 2001a: Conditionality in fund supported programs: External consultations, 17 July, http://www.imf.org/external/np/pdr/cond/2001/eng/collab/07170l.pdf.

International Monetary Fund. 2001b: Strengthening country ownership of fund-supported programs, 5 December, http://www.imf.org/external/np/pdr/2001/eng/strengt/120501.htm.

Ivanova, A, Mayer, W, Mourmouras, A and Anayiotos, G. 2001: What determines the success or failure of fund-supported programs, Paper presented at the Second Annual IMF Research Conference, 29-30 November 2001.

Joyce, JP. 2003: Promises made, promises broken: a model of IMF program implementation. Wellesley College Department of Economics Working paper no. 2003-03.

Khan, MS and Sharma, S. 2001: IMF conditionality and country ownership of programs. IMF Working paper no. 01/142.

Killick, T. 1995: IMF programs in developing countries. Routledge: London.

Killick, T. 1997: Principals, agents and the failings of conditionality. Journal of International Development 9(4): 483-495.

Killick, T. 1998: Aid and the political economy of policy change. Routledge: ODI.

Krueger, A. 1998: Whither the World Bank and the IMF? Journal of Economic Literature 36(4): 1983-2020.

Kugler, J and Feng, Y (eds) 1997: The expected utility approach to policy decision making. International Interactions 23.

Mayer, W and Mourmouras, A. 2002: Vested interests in a political theory of IFI conditionality. IMF Working paper no. 02/73.

Mecagni, M. (1999): The causes of program interruptions. In: Bredenkamp, H and Schadler, S. (eds). Economic Adjustment and Reform in Low-Income Countries. IMF: Washington, DC pp. 215-276.

Mody, A and Saravia, D. 2003: Catalyzing capital flows: Do IMF programs work as commitment devices? Presented at the conference on 'The Impact of Globalization on the Nation-State from Above: The International Monetary Fund and The World Bank' 25-27 April 2003. Yale University: New Haven, CT.

Mussa, M and Savastano, M. 2000: The IMF approach to economic stabilization, in: Bernanke, BS and Rotemberg, JJ. (eds). NBER Macroeconomics Annual 1999. MIT Press: Cambridge, MA pp. 79-122.

Oatley, T and Yackee, J. 2004: American interests and IMP lending. International Politics 41(3): 415-429.

Rodrik, D. 1996: Understanding economic policy reform. Journal of Economic Literature 34(1): 9-41.

Thacker, S. 1999: The high politics of IMP lending. World Politics 52(1): 38-75.

Thomas, A. 2002: Do prior actions achieve their objectives? An evaluation based on active programs over the 1992-98 period. IMF Working paper, IMF: Washington, DC February.

Vreeland, J. 2003: The IMF and economic development. Cambridge University Press: Cambridge.

Willett, T.D. 1988: Political Business Cycles. Dukes University Press: Durham, NC.

Willett, T. 2001: Understanding the IMF Debate. The Independent Review: A Journal of Political Economy, Spring 2001, pp. 593-610.

Willett, T. 2002: Towards a broader public choice analysis of the International Monetary Fund. In: Andrews, D, Henning, R and Pauly, L. (eds). Organizing the World's Money, a Festschrift in honor of Benjamin J. Cohen. Cornell University Press: Ithaca, NY, 2002, pp. 60-77.

Willett, T. 2003: The IMF and Capital Account Crises: The Case for Separate Lender of Last Resort and Conditionality Functions. Presented at the conference on 'The Impact of Globalization on the Nation State from Above: The International Monetary Fund and The World Bank' 25-27 April 2003. Yale University: New Haven, CT.

Williamson, J. (ed) 1983: IMF conditionality. Institute for International Economics: Washington, DC.

Woods, N. 2000: The challenge of good governance for the IMF and the World Bank Themselves. World Development 28(5): 823-841.

Woods, N. 2001: Making the IMF and the World Bank more accountable. International Affairs 77: 83-100.

GRAHAM BIRD (1) AND THOMAS D WILLETT (2)

(1) Surrey Centre for International Economic Studies, University of Surrey, Guildford, Surrey GU2 7XH, UK. E-mail: g.bird@surrey.ac.uk

(2) The Claremont Colleges, 160 East 10th Street, Claremont, CA 91711, USA. E-mail: Thomas.Willett@cgu.edu