The catalyzing role of policy-based lending by the IMF and the World Bank: fact or fiction?

THE CATALYZING ROLE OFPOLICY-BASED LENDING BY

THE IMF AND THE WORLD BANK:FACT OR FICTION?

GRAHAM BIRD1* and DANE ROWLANDS2

1Surrey Centre for International Economic Studies, University of Surrey, Guildford, UK2Norman Paterson School of International A�airs, Carleton University, Ottawa, Canada

Abstract:Does the involvement of the IMF and World Bank in developing countries and

countries in transition help them to attract capital from other sources? Do the

multilateral institutions exert a catalytic e�ect? While there is a strong body of opinion

that claims that they do, the catalytic e�ect has been under-researched.This paper brie¯y

reports the results from a research project ®nanced by the UK Department for

International Development. It examines the theory behind catalysis and also investigates

the empirical evidence relating to it, drawing on qualitative research, regression analysis

based on a large sample of developing countries, and systematic case studies. The

analysis shows how the catalytic e�ect is nuanced and complex but that overall it is

weak. The catalyzing role of the IMF and the World Bank remains potentially

important, however, and the paper assesses ways in which it may be strengthened.

Reforms are discussed in the context of the debate over a new international ®nancial

architecture. Copyright # 2000 John Wiley & Sons, Ltd.

1 INTRODUCTION

Does the involvement of the IMF and the World Bank in developing countries andcountries in transition (CITs) help them to attract capital from other sources, either inthe form of private ¯ows or bilateral aid? Do these multilateral institutions thereforeexert a catalytic e�ect on other ®nancial ¯ows? There is a strong body of opinion thatthe basic answer to this question is `yes'. Thus, in a recent UK Treasury Committeeinquiry into the IMF (Treasury Committee, 1997), the minutes of evidence report one

Journal of International DevelopmentJ. Int. Dev. 12, 951±973 (2000)

Copyright # 2000 John Wiley & Sons, Ltd.

* Correspondence to: Prof. Graham Bird, Surrey Centre for International Economic Studies, University ofSurrey, Guildford, UK. E-mail: [email protected]

member as saying that, `the drift of the evidence' received by the inquiry from `peoplewho claim to know' was `a lot about catalytic e�ects and a lot about public leverage ofprivate money . . . it is just a very commonly held view,' p. 30.

In a recent review of the World Bank and the IMF, Anne Krueger (1998) claimsthat an `argument may be made that the presence of a Fund programme providesimportant signals for private creditors' (p. 2010) also suggesting that `privatecreditors are often unwilling to extend credit lines unless the Bank and Fund have ®rstsignalled their acceptance of economic policies,' (p. 2007). Dani Rodrik (1996) alsoclaims that `multilateral ¯ows should act as a catalyst for private ¯ows,' (p. 182); ane�ect which he claims should work through the information and signals contained inconditionality.

The IMF itself has advocated a `three pronged' approach to policy-based lendingunder which the third prong is `the securing of external ®nance to supportprogrammes,' (Schadler et al., 1995), and in the IMF Survey in March 1997 the Fundclaims that by signalling policy credibility to markets IMF conditionality encouragesadditional private capital ¯ows, (see Dhonte, 1997).

At the same time, however, there are contra-indications. Fund involvement in EastAsian economies in 1997 was associated with accelerating capital out¯ows such thatcatalysis appeared to be negative. Here, Fund and Bank programmes seemed to besignalling severe economic distress. Moreover, a broad examination of aggregate datasuggests that lending by the Fund tends to increase at times when private lending isfalling, implying that lending by the multilaterals substitutes for private capital ¯owsrather than complements them as the catalytic e�ect would predict.

But does it matter what the sign and strength of the catalytic e�ect is? There are anumber of reasons why it does. First, in dealing with balance of payments de®citsthere will, in principle, be an optimal combination of adjustment and ®nancing. Thus,a country experiencing a structural balance of payments problem may do better tophase adjustment over the medium to long term. This policy approach will imply arelatively strong demand for external ®nancing in the short term. Can this demand bemet? If not, the country will be forced to adopt an economic programme which aimsto eliminate the payments de®cit more rapidly, and this is likely to focus oncompressing domestic aggregate demand. Such a strategy may have adverse supply-side consequences, with a negative impact on economic growth. Without adequateexternal ®nancing developing countries and CITs will therefore be pushed towardssub-optimal adjustment strategies, with the availability of external ®nance constrain-ing economic development.1

A second implication is that if the supply of external ®nance constrains the designof adjustment programmes, it is also likely to limit their success. Sub-optimaladjustment strategies will be less successful than optimal ones. At a time when themajority of Fund programmes remain uncompleted it is therefore relevant to askwhether this failure is connected with ®nancial starvation.2

However the lending capacities of the IMF and the World Bank are themselvesconstrained by political factors, with the major shareholders often being reluctant toexpand their resource base. In these circumstances the needs of developing countries for

1This point is examined in more detail in Bird (1997).2 Evidence on the completion rates of Fund programmes is presented in Killick (1995) with Bird andRowlands (1999a) providing up-dated information. Killick ®nds evidence to suggest that the `success' ofFund programmes is positively linked to the amount of ®nance that is provided.

952 G. Bird and D. Rowlands

Copyright # 2000 John Wiley & Sons, Ltd. J. Int. Dev. 12, 951±973 (2000)

external ®nance will not be met by the multilateral institutions alone. There willthereforebea®nancing gapwhich either has tobe®lledby capital fromother sources, oreliminated by policies which reduce economic growth. The catalytic e�ect is about theextent to which the multilateral institutions can entice others to lend and help close theexternal ®nancing gap. The more e�ective the multilaterals are in catalyzing other®nancial ¯ows the less the need for them to lend themselves. By the same token a weakcatalytic e�ect implies that the multilaterals will need to take on a larger share of the®nancing role or accept lower economic welfare in developing countries and CITs.

Third, therefore, the catalytic e�ect is important in the context of the debate aboutthe international ®nancial architecture and the roles and resource requirements of themultilateral agencies. Should the IMF and the World Bank concentrate on givingpolicy advice and rely on others to provide ®nance which they can catalyze viaconditionality, or should they be lending institutions, and if so, how large should theirlending role be? Catalysis is fundamental to answering these questions.

However, althoughwidely discussed, catalysis by themultilaterals has also tended tobe discussed rather loosely, and there have been relatively few studies that have set out toexamine it in any depth. Perhaps the best known and frequently cited analysis ofmultilateral lending is that by Rodrik (1996), although other studies which have abearing on the catalytic e�ect are reviewed in Bird and Rowlands (1997), which alsoprovides additional new evidence. However, even Rodrik's research exhibits someshortcomings. First, it does not explore in any detail the theory of catalysis and simplyassumes that, via signalling, IMFandWorldBank conditionality shouldhave apositivecatalytic e�ect. Second, the econometric results reported rely on grouping observationsinto six year periods for which there is only limited justi®cation. Do other econometricmethodologies and other time periods give di�erent results? Third, Rodrik does notattempt to control for the economic characteristics of each country. Finally theeconometrics is not augmented by case studies. Nor did Rodrik talk to fundmanagers,credit-rating agencies, and bilateral aid donors to gauge how important they perceivethe multilaterals to be in in¯uencing their investment and lending decisions.

The research reported in the paper attempts to overcome some of these de®ciencies.The paper presents, in broad terms, the ®ndings of a research project ®nanced by theDepartment for International Development (DfID) in the United Kingdom.However, since in essence the paper is a summary of the project, a considerableamount of signi®cant detail has been omitted to ensure that its length does notbecome unmanageable. For those who want to pursue some of the issues raised herein more detail this paper should therefore be read in conjunction with the otherpapers that have arisen from the project (Bird and Rowlands, 1997; 1999a; 1999b;Bird, 1998; 1999; Bird et al., 1999b).3

The layout of the paper is as follows. Section 2 discusses the theory of catalysis byexploring the principal analytical issues to which it gives rise. Section 3 brie¯ysummarizes the results of previous research that has a bearing on catalysis. Section 4discusses the objective functions of the principal players and reports what privatelenders and aid donors say about the importance of the Fund and the Bank in thecontext of catalysis. Section 5 presents a summary of our new econometric evidenceincorporating various levels of disaggregation in terms of types of capital ¯ow and

3These papers also contain more extensive references, which have been kept to a minimum here, and areonly used where directly relevant to a point that is being made in the text.

Catalyzing Role of Policy-based Lending 953


multilateral lending. Section 6 provides a brief review of existing case study evidenceand also reports some of our own case study ®ndings. Section 7 extracts the mainpolicy implications from the foregoing discussion, and examines how the catalytice�ect could be strengthened in future. Section 8 o�ers a few concluding remarkswhich, in particular, attempt to place our research on the catalytic e�ect in the contextof the debate about the new international ®nancial architecture.

2 ANALYTICAL ISSUES

What theory lies behind the `very commonly held view' that the multilaterals exert acatalytic e�ect on other ®nancial ¯ows? In principle, there are a number of modalitiesthrough which catalysis may occur. Route 1 is via conditionality. The idea here is thatby agreeing to an IMF or World Bank programme, governments can signal to privatecapital markets or to aid donors that they are serious about, and committed to,economic reform. By involving the multilaterals, governments can overcome the timeinconsistency problem and, e�ectively, import the superior reputation of the Fund orthe Bank for economic management. The multilaterals will thus add value in terms ofthe design of the economic programmes and the monitoring and enforcement ofimplementation to make sure that they are carried through to completion.

According to this approach, private markets and bilateral aid donors lack theinformation and, individually, the clout to carry out these functions, and they aretherefore content to delegate them to the multilaterals. In a sense the multilateralsthen act as the agents of private lenders and bilateral aid donors. Route 1 hasconditionality enhancing credibility which in turn generates catalysis.

Of these two links it is reasonable to assume that greater credibility will have apositive catalytic e�ect. However, the link between conditionality and credibilityneeds to be looked at more critically. Can it be safely assumed that the IMF and theWorld Bank will design appropriate policies, and furthermore does conditionalityo�er an e�ective means of ensuring that these policies will be implemented? This isnot the place to get into a detailed discussion of the design of IMF and World Bankconditionality, since such discussions exist elsewhere.4

However, it can be noted that there are plenty of claims in the literature to suggestthat the design of IMF and World Bank conditionality is inappropriate, or at leastbuilt on economic ideas about which there is legitimate debate.5 Most recently in thecontext of the East Asian crisis the argument has been forcefully made by critics thatthe IMF endeavoured to deal with `new' problems using `old' policies which weremisplaced and counterproductive (see, for example, Feldstein, 1998). Meanwhile, theWorld Bank has been accused of showing over-zealous support for a market-basedapproach to economic development and economic growth, while in practicepremature economic liberalisation often leads to di�culties, (Fanelli et al., 1994).

4 Indeed discussion of IMF and to a lesser extent World Bank conditionality has been something of agrowth industry. Early contributions to the literature included Williamson (1983) and Killick et al. (1984).Other important and frequently cited papers include Edwards (1989) and Khan (1990). More recent studiesinclude Conway (1994), Killick (1995) and ul Haque and Khan (1998). More generally IMF conditionalityhas often been linked to discussions of IMF lending (Bird, 1995). Mosley et al. (1991) still represents themost thorough investigation into the World Bank's policy-based lending.5 Bird (1999) explores this issue in the context of discussing the meaning and relevance of `sound' economicpolicy. Similar arguments are put forward by Rodrik (1999).



Stalwart components of conditionality in the form of raising interest rates andcurrency devaluation may, for example, have negative e�ects on capital ¯ows just aseasily as positive ones if they are believed to engender recession, domestic bankingcollapse, and the risk of further currency depreciation (Radelet and Sachs, 1998).

So doubts exist about the design of conditionality. On top of this, however, thereare doubts about whether conditionality enhances implementation. Again somecommentators have argued that there are built-in inconsistencies in conditionality interms of the functions it is seeking to perform, and that implementation relies ongovernment commitment which is, in many ways, at odds with conditionality.6

The acid test is to see whether conditionality as used by the multilaterals has beensuccessful. But what are the criteria for success? Does conditionality ensure that agreedpolicy changes aremade? Is conditionality fully implemented?Does conditionality leadto an improvement in ultimate indicators of economic performance?Again there is verylarge literature dealing with each of these questions, but in summary conditionalityexhibits a less than spectacular track record. Thus the majority of Fund programmesremain uncompleted. Neither Fund programmes nor structural adjustment pro-grammes supported by the World Bank have strongly signi®cant positive results onsubsequent economic performance, though this remains a controversial issue.7

Moreover, recent evidence on World Bank conditionality suggests that even wherespeci®c policies do lead to bene®cial outcomes, their enactment has little to do withconditionality (Dollar and Svensson, 1998).

Even the enforcement aspect of conditionality lacks teeth for as long as non-compliance is not heavily penalized by the multilaterals. Recidivism is a feature ofboth IMF and World Bank lending and, in this respect, a current programme witheither multilateral may be seen as a lead indicator of future economic di�culties.8

Therefore while private markets and aid donors will certainly be looking forgovernmental commitment to sound economic policies, there must be some doubtabout whether the multilaterals will identify the best policies, and considerable doubtabout whether conditionality will ensure that these policies are pursued. Indeed,conditionality may be inimicable to the commitment upon which implementationrelies. If catalysis relies on conditionality it may be fundamentally ¯awed.

But conditionality may not be the only modality through which catalysis works.Route 2 is via liquidity. The multilaterals not only provide an adjustment input viaconditionality but they also provide ®nancial support. Financing by the multilateralsmay ful®l a number of functions. First, it may improve the quality of their adjustmentadvice since theyare nowputting theirmoneywhere theirmouths are.Even if it does notimprove the quality of conditionality, private markets may believe that it does and thiswill lead to positive catalysis. Second, where the problems encountered by countriesturning to themultilaterals for assistance are of a liquidity type, the provision of ®nancemay be seen as central in helping to resolve them. Thus private markets may be moreprepared to lend to countries where they perceive the multilaterals as ®lling a large

6The incompatibilities between the various functions that conditionality is seeking to achieve are exploredin detail in Collier et al. (1997). The question of implementation and its links with conditionality areexplored in Killick (1998) and Bird (1998).7Reference can again be made to the works cited in footnote 5. ul Haque and Khan (1998) provide what isprobably the most sanguine view of the e�ects of IMF programmes.8Many studies have identi®ed these recidivist tendencies, for example, Bird (1995), Killick (1995), andConway (1994). There has, however, generally been little analysis of recidivism. For an attempt to providejust such an analysis, see Bird et al. (1999a).



proportion of the ®nancing gap. Of course the irony here is that the catalytic e�ect willthen be stronger in precisely those cases where it is less needed.

Where the multilaterals provide large amounts of ®nance, it may also be the casethat better designed adjustment strategies will be feasible which work more on thesupply side of the economy and on the structural causes of balance of paymentsproblems. To the extent that these allow adjustment with growth, private capitalmarkets may be more attracted than they would be where adjustment relied on thecompression of aggregate demand, which, especially where it hits investment, is likelyto have a negative e�ect on economic growth at least in the short term.

How important this point is depends on what it is that private markets and bilateralaid donors are looking for. What are their objective functions, and how closely dothese match those of the multilaterals? Analytically this raises a number ofpossibilities. If the IMF is more concerned about short-run stabilization thanabout long-run economic growth, perhaps IMF conditionality will have a strongerimpact on short-term capital ¯ows than on long-term capital ¯ows. If the World Bankis more growth and development oriented, its conditionality may be relatively moree�ective in catalyzing long-term capital ¯ows.

In the context of low income countries, multilateral lending is likely a priori to havea stronger in¯uence over bilateral aid ¯ows than private capital ¯ows because aiddonors are not looking for a commercial rate of return and will have been involved inthe design of the multilaterals' programmes. All this suggests that catalysis will onlybe properly understood at the appropriate level of disaggregation.

Finally, the moral hazard and bailout dimensions of multilateral lending call intoquestion whether the theory of catalysis as conventionally presented has the correctsequencing. Instead of multilateral lending taking the lead, with private lendingfollowing it, the moral hazard hypothesis suggests that it is the perceived availabilityof multilateral lending, should the need arise, which encourages private markets tolend, and indeed overlend. Overlending by private capital markets then contributes tocreating economic crises which in turn lead countries to turn to the IMF and WorldBank. A catalytic e�ect is at work here, but it is an ex ante e�ect where private lendingprecedes and in a sense catalyzes lending by the multilaterals.9

What emerges from this discussion of the analytics of catalysis? The basic messageis that there is no clear a priori causal connection between multilateral lending in thecontext of IMF and World Bank programmes and ®nancial ¯ows from other sources.The catalytic e�ect is essentially an empirical issue. So it is to the empirical evidencethat we now turn.

3 EXISTING EMPIRICAL EVIDENCE

Although it is di�cult to identify exactly when the catalytic e�ect was ®rst described,its widespread currency was clearly facilitated by numerous anecdotal accounts of(primarily) IMF agreements being followed by fresh capital in¯ows, (Kenen, 1986;McCauley, 1986; de Vries, 1986; Buira, 1987; Pool and Stamos, 1987). The

9At the same moment as the multilaterals are becoming involved private capital may be attempting to exit,implying negative contemporary catalysis. According to the moral hazard argument, however, it has beenthe expectation of future multilateral involvement that has positively catalyzed previous period private¯ows. Rodrik (1996) reports some evidence to support this bail-out story.



`gatekeeper' role of the IMF was subsequently formalized in institutional settingssuch as the Paris Club (Milivojevic, 1985). The catalytic e�ect, however, cannot beproved de®nitively on the basis of these anecdotal observations since there was no wayof testing the counterfactual. More systematic evidence is required.

Most of the existing systematic empirical evidence on the catalytic e�ect has beencoincidental to work that has had a di�erent focus. Attempts to estimate the demandfunction for IMF credit and to explain the economic characteristics of user and non-user countries have, for example, often included a variable designed to capture thesigni®cance of alternative (usually private) sources of ®nance, (Bird and Orme, 1981;Cornelius, 1987; Joyce, 1992; Bird, 1994; 1995). The results of such studies havegenerally suggested the absence of any signi®cant relationship, or at best only a weakrelationship which is as likely to be negative as positive. A negative coe�cient impliesthat IMF lending and private lending are substitutes, rather than the complementsthat the catalytic e�ect predicts.

With a focus on aspects of external debt, other research has also raised doubtsabout the complementarity hypothesis. Hajivassiliou (1987) examines the pattern ofdebt servicing problems in LDCs between 1970 and 1982. His results reveal asigni®cant negative relationship between IMF involvement and the supply of newloans, and with the level of arrears accepted by existing creditors. Unfortunately themodel is not designed speci®cally to test the catalytic e�ect, and the results may bebiased by the use of IMF agreements as part of a proxy for debt servicing problems.The absence of complementarily between IMF involvement and other forms of newlending, however, provides no support for the catalytic e�ect.

In a study of the impact of structural adjustment under the sponsorship of the IMFand the World Bank over the period 1982±86, Faini et al. (1991) also ®nd a signi®cantnegative correlation between multilateral lending and net private credit. A similarresult is discovered by Killick (1995), who ®nds that IMF agreements are generallyassociated with a net deterioration in the capital account of the balance of paymentsof the countries involved, with loans from the Fund in e�ect being used to pay o�previous commercial loans rather than encourage new ones.10

The need to distinguish between the e�ects on private and public ¯ows ishighlighted by Rowlands (1994) who, on the basis of aggregate econometricinvestigation, discovers a catalytic e�ect of Fund lending on other public ¯ows butnot on private ¯ows. Rowlands also shows that the catalytic e�ect, to the extent that itexists, is time variant.

Further evidence against the idea of a catalytic e�ect of IMF lending on privatecapital ¯ows comes from Ozler (1993) who sets out to explain the determinants ofinterest rate spreads on commercial loans to developing countries over 1968 to 1981.He ®nds that countries that had had IMF stand-by agreements seem to have beencharged higher interest rates than those that had not, suggesting that IMF

10Of course care needs to be exercised in interpreting these results. Conventional balance of paymentsaccounting tells us that, with no changes in reserves, an `improvement' in the current account of thebalance of payments will be matched by an equivalent `deterioration' in the capital account. Surely thenIMF programmes that are successful in strengthening the current account must as a matter of accountingweaken the capital account? Of course in the midst of a current account balance of payments crisis,countries will almost certainly have experienced reserve losses which they will be anxious to make good,and in these circumstances current and capital account disequilibria will not sum to zero. Moreovergovernments will be anxious to make themselves more creditworthy as a consequence of agreeing to anIMF or World Bank programme, and in this context total capital in¯ows are not the relevant variable.



involvement is taken to indicate higher risks and providing no support for a positiveimpact on credibility.

Most of the studies cited have concentrated on the IMF, where the claims for acatalytic e�ect have been strongest. Research that has examined the e�ects of World

Bank structural adjustment lending has also discovered a generally neutral e�ect on

private ¯ows (Mosley et al., 1991), or a negative correlation (Faini et al., 1991). Thusthe overall assessment must be that neither of the key multilaterals have been shown

to have a particularly strong catalytic e�ect on lending.

As noted in the Introduction, the two most focused published studies on the

catalytic e�ect to date are Rodrik (1996), and Bird and Rowlands (1997). Rodrik ®ndsthat multilateral lending is negatively but not signi®cantly associated with future

period private capital ¯ows. He also notes a `sharp dichotomy' between the IMF and

other multilaterals. Thus while other multilaterals have a negative e�ect onsubsequent private ¯ows, particularly bank lending, the IMF's catalytic e�ect

appears to be positive. Bird and Rowlands (1997) search for a catalytic e�ect

disaggregating the data in various ways. Although they ®nd limited evidence for IMFcatalysis on public ¯ows, they conclude that their `estimations do not generally

support the notion of a positive catalytic e�ect'.

However, the limited scope of these studies counsels caution over their results. In

what follows, a more rounded view of the empirical evidence is presented not only byrunning some new regressions but also by systematically examining the case study

evidence and by seeking qualitatively the views of those involved in moving money

around the world.

However there is another way in which existing empirical evidence may be used.

With the surge of private capital ¯ows to Latin America and Asia in the early part ofthe 1990s a number of studies were undertaken with a view to better understanding

why these capital movements occurred. Again this is not the place to o�er a detailed

review of this research, but one aspect of it is important in the context of the catalytice�ect. The research conventionally distinguishes between pull and push factors. Pull

factors include aspects of domestic economic policy and performance while push

factors include changes in world interest rates and economic activity. Although thedetails of the results vary across studies a fairly consistent theme is the signi®cance of

push factors that lie outside the control of capital importing developing countries. Inmany cases changes in US interest rates were found to be the single most dominant

factor in causing capital movements.11

Where does IMF and World Bank conditionality ®t in? In principle, conditionality

could in¯uence pull factors. But in practice, and for reasons discussed in the previous

section, conditionality is unlikely to have exerted a signi®cant impact on them. Theimportant factor here is the government's commitment to a programme of solid

macroeconomic and microeconomic policies. Over the years the multilaterals have

almost certainly had an impact over the choice of policy via their continuing dialoguewith countries, but this is not captured by the conditionality contained in speci®c

agreements. The conclusion follows that IMF and World Bank conditionalityprobably plays an insigni®cant role in in¯uencing pull factors that in any case may be

11 This research is summarized in Bird (1999). Illustrative examples include Calvo et al. (1993), Chuhan etal. (1993), Claessens et al. (1995), Fernandez-Arias (1996), Fernandez-Arias and Montiel (1996), andTaylor and Sarno (1997).



dominated by push factors that lie outside the in¯uence of the multilaterals. The mereobservation that there have been periods of feast and famine in terms of privatecapital ¯ows to Latin America while the Fund's presence has been much moreenduring suggests that capital ¯ows are largely determined by other factors. Thisimplies at best a de minimus role for the catalytic e�ect. But what about thequalitative evidence?

4 QUALITATIVE EVIDENCE

What do money managers say about the relevance of the IMF and the World Bank tothem in making their decisions? On the basis of a questionnaire we interviewedbankers, fund managers (including hedge funds, and pension and mutual funds),credit-rating agencies and aid agencies.12 First, and without exception, they claimedthat their decisions are `in¯uenced' by the involvement of the IMF and the WorldBank in individual countries. For the most part the in¯uence is linked to adjustmentprogrammes negotiated with the multilaterals, although other aspects of involvementin terms of policy dialogue or meeting standards of data dissemination may also beseen as relevant. Although some respondents claimed to distinguish between thenature of IMF and World Bank lending, this does not appear to involvediscriminating between di�erent IMF lending windows or even between the variousarms of the World Bank. Closer investigation suggests that where a distinction ismade, it relates to the sectors or projects in which the World Bank is involved.

With little exception, signing an agreement with the Fund or the Bank is claimed tomake a country more attractive than it would otherwise have been, which providessome evidence against negative catalysis. IMF agreements are without exception seenas more important than World Bank structural adjustment lending. Interestingly,however, it is the level of ®nancing involved that is seen as more signi®cant than theprecise nature of conditionality. This calls into question the often made assumptionthat catalysis works primarily through the modality of conditionality. However it didalso come over in interviews that the importance of conditionality relative to liquiditydepends on the nature of the loan or investment. For short-term ®nancial ¯ows theliquidity aspect of the agreement dominates more than it does in the case of longer-term capital movements. In no case, however, was the conditionality component seenas more important than the liquidity component. A strong message was that moneymanagers form their own views about a country's policies, performance, andprospects, and that this might not always tally with that of the Fund or the Bank, inwhich case managers tend to back their own judgement.

At the same time, it was always maintained that the implementation ofconditionality is monitored, albeit usually by press reports, and that the degree ofimplementation is interpreted as being important. If this is the case, the lowcompletion rate of programmes serves to undermine the Fund's potential catalyzingrole.

There were mixed views as to whether Fund and Bank involvement had beenperceived as increasingly important. Where the claim was made that it had, this wasusually explained in the context of the incidence of speci®c crises rather than any

12Copies of the questionnaire can be provided by the authors upon request. In total about ®fteeninterviews were conducted.



systemic change. The involvement of the multilaterals is seen as more important in a

bear market.

In no case, however, did the involvement of either the Fund or the Bank feature

directly as one of the principal ®ve determinants of lending and investment

decisions. What factors are important depend on whether you talk to hedge fund

managers or pension and mutual fund managers. But amongst the factors listed

were the prospects for economic growth, the degree of industrial diversi®cation, the

quality of infrastructure, the size and quality of domestic ®nancial markets, and the

degree of political stability. Credit-rating agencies appear to prioritize the short-run

liquidity position and then move onto solvency indicators such as indebtedness

before moving on to the quality of macroeconomic policy, the structural reform

agenda and political stability. A broad observation is that in undertaking country

risk analysis the involvement of the Fund and the Bank is not generally taken into

account in any formal way. Individual countries are treated sui generis, although

the importance put on Fund and Bank involvement depends on whether a country

is in a crisis phase. Certainly IMF and World Bank programmes do not confer an

automatic seal of approval as the most extreme presentations of the catalytic e�ect

claim.

But do things di�er for bilateral aid donors? Fund agreements are seen as providing

certain `assurances' about macroeconomic policy and are claimed to be more

important than Bank agreements in the context of short-term aid ¯ows, although

donors still make their own independent assessments. The crucial di�erence here is

that donors will be involved in designing IMF and World Bank programmes both `on

the ground' and via their Executive Directors in Washington. The in¯uence of

individual aid donors in this respect has a regional dimension with, for example, the

UK having a stronger in¯uence over policy design in Africa than in Latin America or

Asia. However, the general point is that bilateral aid and assistance from the

multilaterals comes as a package, with IMF programmes only `standing up' if

accompanied by bilateral aid.

Aid donors claim to have received the `big message' from recent research on aid

e�ectiveness that aid works best when accompanied by the `right' policy environment,

and in this respect IMF endorsement is important, even though aid donors will look

beyond this to issues of governance and social sector policies. There appears to be less

recognition that the same research is rather ambiguous on the role of conditionality in

bringing this environment about, with some researchers claiming that it has ful®lled a

more important function than others.13 Aid donors do observe, however, that the

Fund can provide assurances outside formal agreements and that these can in¯uence

their decisions. Moreover, bilateral aid does still go to countries with no current IMF

agreement.

Neither the IMF nor the World Bank is taken as providing any guarantee on

political stability and donors form their own views about this, although in some cases

individual countries will be given the `bene®t of the doubt' on governance issues once

a Fund agreement has been signed.

13 Thus while Burnside and Dollar (1997) attribute at best a very modest role to the conditionality of themultilaterals, Mosley and Hudson (1997) claim that increasing aid e�ectiveness is linked to improveddialogue with the World Bank as well as conditionality.



IMF agreements certainly appear to be seen as more signi®cant than the WorldBank's policy-based lending, which plays only a supporting role. Where the WorldBank does become more important is in terms of its project lending, where itin¯uences the sectoral composition of longer-term bilateral aid ¯ows.

As in the case of private ¯ows, there is no simple and universal model of catalysiswhich links IMF and World Bank policy-based lending to bilateral aid ¯ows. Onbalance the multilaterals appear to be more important in the case of bilateral aid ¯owsthan in the case of private capital ¯ows since the objective functions of bilateral aiddonors may be closer to the objective functions of the multilaterals. However, the veryfact that donor governments choose to direct some of their foreign aid throughbilateral channels con®rms that there will be other, probably country-speci®c,political factors in¯uencing bilateral aid. Bilateral aid should not therefore beexpected to perfectly match multilateral lending.

Nevertheless, a fundamental fact is that the modality through which catalysisoccurs di�ers signi®cantly as between public and private capital ¯ows, since aiddonors, unlike private creditors, will be directly involved in putting together IMF andWorld Bank programmes. Apart from anything else, this means that to the extent thatIMF and World Bank lending is political, similar politics may also in¯uence bilateralaid ¯ows. It may be anticipated therefore that there will be a closer associationbetween lending by the multilaterals and other public capital ¯ows than there is withprivate capital ¯ows.

5 QUANTITATIVE EVIDENCE

As noted earlier, much of the evidence on catalysis has emerged from anecdotalaccounts of individual cases, as an indirect implication of studies designed for otherpurposes, or from larger sample studies that do not control for the economiccircumstances of countries belonging to the Bretton Woods institutions. In order toaddress these de®ciencies, our research has placed considerable emphasis onconstructing a su�ciently extensive data set to allow for a large-sample analysis ofthe catalytic e�ect. While several models have been examined for both the IMF andthe World Bank, the results reported here use an unbalanced panel of approximately96 developing countries from the early 1980s to 1995. The sample size was restrictedlargely by the availability of data.

The activities of the Fund and the Bank were examined in terms of their e�ects ondi�erent types of capital ¯ow. The di�erent categories of ¯ows included three privatechannelsÐ foreign direct investment, portfolio ¯ows, and government or governmentguaranteed debt (commercial bank, bond or other)Ðand bilateral o�cial ¯ows inthe form of either grants or debt. These ¯ows are presumably driven by di�erentmotivations, with private ¯ows being assumed to be more concerned with commercialrates of return than bilateral ¯ows. Furthermore, foreign direct investment may beexpected to be more closely related to long-term economic performance than are, say,portfolio ¯ows. By exploiting these di�erences, it was hoped that further insight intothe catalytic e�ect could be gained.

The activities of the Fund and Bank were also di�erentiated to isolate the potentialsources of catalysis. Thus lending instruments were di�erentiated on the basis of thedegree of conditionality, and estimations were carried out using indicators of activity



Table

1.Resultsof®xed-e�ects

regressionsusingIM

Fprogrammes.

VARIA

BLE

Dependentvariable�P

ortfolio

Flows/GDP

Dependentvariable�N

etFDI/GDP


etPrivate

Debt/GDP


etO�cialFlows/GDP

Explanatory

variables

SBA

tÿ1

ÿ0.00039(ÿ

0.91)

ÿ0.00032(ÿ

0.28)

ÿ0.00462��(ÿ

3.27)

ÿ0.01204(ÿ

0.92)

EFFtÿ

10.00068(0.42)

0.00029(0.09)

ÿ0.00714�(ÿ

1.87)

ÿ0.01286(ÿ

0.72)

ESAFtÿ

1ÿ0

.00068(ÿ

1.01)

0.0036(1.21)

ÿ0.00443�(ÿ

1.83)

0.00825(0.17)

SAFtÿ

1ÿ0

.00037(ÿ

0.82)

0.00011(0.02)

0.00051(0.033)

0.01698(0.74)

Observations

1008

1008

1008

1008

R2

0.182

0.489

0.171

0.802

Adjusted

R2

0.079

0.425

0.068

0.777

SBA,EFF,ESAF

andSAF

referrespectively

todummyindicators

forthepresence

ofStand-byarrangem

ents,Extended

FundFacility,

EnhancedStructuralAdjust-

mentFacility,

andStructuralAdjustmentFacility,

alllagged

byoneyear.

Thet-statistics

appearin

parentheses.Coe�

cients

identi®ed

with�� ,��,and�are

signi®cant

atthe1%

,5%

and10%

two-tailed

test

levelofsigni®cance

respectively.Theequationsare

estimated

usinga®xed-e�ects

model

withheteroscedastic-consistentesti-

mates

andasimple

correctionfor®rst-order

autocorrelation.Only

theresultsfortheIM

Fvariablesare

presented:thefullestimatingequationis

similarto

that

pro-

vided

inBirdandRowlands(1999a).



Table


regressionsusingWorldBankindicators.

VARIA

BLE

Dependent

variable�P

ortfolioFlows/

GDP

Dependent

variable�N

etFDI/

GDP


etPrivateDebt/GDP


etO�cialFlows/GDP

Explanatory

variables

SALtÿ

1ÿ1

120.0

(ÿ1.26)

ÿ917.8

(ÿ0.84)

ÿ2427.7�(ÿ

1.82)

2921.1

(0.14)

SALtÿ

287.35(0.18)

1133.7

(0.93)

ÿ1326.0

(ÿ1.04)

8645.9

(0.46)

SECALtÿ

1ÿ1

932.0

(ÿ1.49)

3965.7

(0.29)

1831.1

(1.48)

15648

(1.07)

SECALtÿ

2248.9

(0.51)

ÿ3314.6��

(ÿ2.32)

ÿ2017.2

(ÿ1.53)

ÿ4322.6

(ÿ0.35)

WBPROJECTtÿ

1ÿ2

90.1

(ÿ0.66)

ÿ1497.1

(ÿ0.66)

5398.1��

(2.28)

6878.0

(0.18)

Observations

935

935

935

935

R2

0.19

0.54

0.18

0.76

Adjusted

R2

0.09

0.48

0.07

0.73

SAL,SECAL

andWBPROJE

CT

referrespectively

todummyindicators

forthepresence

ofWorldBankStructuralAdjustmentLoans,

SectoralAdjustmentLoans,

andIB

RD

orID

Aproject

lending.Lagsare

indicated

inthesubscripts.Thet-statistics

appears

inparentheses.Coe�

cients

identi®ed

with�� ,��,and�are

signi®cant

atthe1%

,5%

and10%

two-tailed

test

levelofsigni®cance

respectively.Theequationsare

estimated

usinga®xed-e�ects

model

withheteroscedastic-consistentesti-

mates

andasimple

correctionfor®rst-order

autocorrelation.Only

theresultsfortheBankvariablesare

presented:thefullestimatingequationis

similarto

that

pro-

vided

inBirdandRowlands(1999b).



Table


estimationsusingWorldBank®nancinglevels.

VARIA

BLE

Dependent

variable�P

ortfolioFlows/

GDP

Dependent

variable�N

etFDI/

GDP


etPrivateDebt/GDP


etO�cialFlows/GDP

Explanatory

variables

SALtÿ

1ÿ0

.0220(ÿ

1.04)

ÿ0.0273(ÿ

0.46)

ÿ0.0729�(ÿ

1.73)

0.499(0.37)

SALtÿ

2ÿ0

.0142(ÿ

1.02)

0.0125(0.31)

ÿ0.1038��(ÿ

3.43)

0.708(0.91)

SECALtÿ

1ÿ0

.0347(ÿ

1.12)

0.0711(0.72)

0.0181(0.42)

ÿ0.192(ÿ

0.22)

SECALtÿ

2ÿ0

.00371(ÿ

0.36)

ÿ0.0214(ÿ

1.56)

ÿ0.0654(ÿ

1.22)

ÿ1.239(ÿ

1.58)

WBPROJECTtÿ

10.0331(1.09)

ÿ0.0303(ÿ

0.24)

0.00551(0.09)

3.404��(2.63)

Observations

935

935

935

935

R2

0.19

0.54

0.17

0.80

Adjusted

R2

0.08

0.49

0.07

0.77

SAL

and

SECAL

referrespectively

totheamounts

of®nancing

agreed

toin

World

Bank

StructuralAdjustmentLoansand

SectoralAdjustmentLoans,

while

WBPROJE

CTrefers

totheamountofdisbursem

ents

ofIB

RD

orID

Aproject

lending.See

Table

2above

forother

relevantnotes.



as well as levels of associated ®nancing. In this way liquidity e�ects were separated

from the in¯uence of conditionality.

The estimation models performed quite well in terms of conforming to prior

expectations about the e�ects of di�erent economic characteristics.14 On the one

hand, commercially motivated capital ¯ows did indeed seem to be a�ected positively

by good economic events, and vice versa. The behaviour of o�cial bilateral ¯ows, on

the other hand, was consistent with more altruistic motivations. While the details of

these estimations are presented in Bird and Rowlands (1997; 1999a; 1999b), the basic

conclusion is that it is important to control for economic characteristics in any

examination of the catalytic e�ect.

In terms of the catalytic e�ect itself, a nuanced story emerges. Speci®c results are

provided for IMF programmes in Table 1, and for World Bank programmes in Tables

2 and 3. As a general conclusion, however, it seems that the empirical estimations cast

considerable doubt on the proposition that IMF or World Bank conditionality

transmits a positive signal to private capital markets. Private sources tend to react

negatively, if at all, to these arrangements, and the negative reaction tends to be more

pronounced for the higher conditionality agreements. Finally, there is no evidence to

suggest that these private sources have begun to react more favourably in recent years,

at least to IMF arrangements.

More speci®cally, while portfolio and FDI ¯ows seem una�ected by any IMF

activity, there is reasonably strong evidence to suggest that private debt ¯ows react

adversely to high conditionality IMF programmes. The reaction of private creditors

to World Bank activity is also subdued. Conditional programmes have either no

e�ect or a negative e�ect on these ¯ows. The only positive reaction appears to

accompany Bank project loans. Thus, viewed as a whole, the evidence from the

large sample investigation casts considerable doubt on the proposition that

conditionality is an important modality of positive catalysis. What emerges instead

are reasonable grounds for presuming that Fund and Bank conditional

programmes are not perceived by markets as enhancing the commercial prospects

for recipient countries.

The only evidence of positive catalysis that emerges from these estimations is in the

case of o�cial bilateral sources of ®nance. Even in this case, however, the e�ect is

generally weak and the signi®cance of the results is not robust. In some formulations

of the estimating equation o�cial ¯ows are associated positively with Fund

agreements, especially SAF programmes, but longer-term EFF and ESAF arrange-

ments as well. When corrected for cross-country heterogeneity and simple

autocorrelationÐboth problems being evident from diagnostic testsÐ the statistical

association becomes insigni®cant. A similar characterization seems appropriate for

World Bank catalysis, though in this case the positive association between project

lending and o�cial bilateral ¯ows remains robust in the face of corrections to the

estimating procedure.

14 The results presented in Tables 1±3 are minor variants of the estimating equations in Bird and Rowlands(1999a; 1999b). These papers have a more comprehensive discussion of the estimations, and report thecomplete results. It should be noted that while the results reported here are for equations using only laggedversions of the IMF or Bank variables, these are generally consistent with equations that used both laggedand unlagged variables. The latter are excluded to avoid contemporaneity and the resulting complicationsfor interpreting causality.



In conclusion, the large sample empirical evidence suggests that the catalytic e�ectof the IMF and World Bank is generally insigni®cant or negative for commercial¯ows, and that conditionality does not have the expected positive role suggested bysome of the earlier literature. If anything, high conditionality programmes have amore adverse e�ect on commercial ®nance than other Bank and Fund activity. OnlyBank project ®nancing appears to have any positive in¯uence on commercial ¯ows,though the e�ect is neither strong nor robust. The strongest evidence of positivecatalysis is with o�cial bilateral ¯ows, though again this e�ect appears inverselyrelated to the degree of conditionality.

6 CASE STUDY EVIDENCE

It may be that cross-country regression analysis obscures a catalytic e�ect that wouldbe revealed by closer examination of individual case studies and episodes. To allowfor this we selected 18 countries to look at in some detail. These covered variousregional locations and per capita income levels. Our approach was to build on andextend existing case study work which has focused more broadly on the e�ects of IMFand World Bank programmes, (Killick, 1995; 1998). To do this we undertook anextensive and systematic electronic literature search in order to double check that theconclusions drawn in these earlier studies were sound and that important pieces ofevidence had not been omitted. Initially we did this for the period covered by thestudies themselves, but then updated our search as far as possible. We also collectedour own data to see whether these were consistent with the conclusions that came outof the systematic review.15

In fact our case study evidence strongly con®rms the ®ndings of the qualitative andeconometric research reported earlier. What general messages emerge from the casestudies? These may be summarized in terms of the following working hypotheses:

(i) involvement by the multilaterals will not guarantee an in¯ow of capital fromother sources;

(ii) what really matters is the perceived commitment by a government to a policyagenda that is seen as sound and internally consistent;

(iii) catalysis associated with the multilaterals is likely to be stronger and morepositive in the case of bilateral aid ¯ows; and

(iv) the nature and extent of catalysis may di�er as between the Fund and the Bank.

Experiences drawn from almost every one of our case studies con®rmed hypotheses(i) and (ii). It was only perhaps in Russia that an IMF-backed programme in 1996seemed to induce positive catalysis in spite of a perceived lack of commitment by thegovernment. Here the signal seemed to be that Russia would not be allowed to fail. Inother countries the involvement of the multilaterals seemed to be largely irrelevant indetermining capital ¯ows. Argentina established a reputation for commitment topolicy reform under its Convertibility Plan in conjunction with the IMF, while Chileand Malaysia (and for a time other countries in East Asia) achieved a strongreputation largely outside the auspices of the multilaterals, by in fact pursuing capitalaccount policies of which the multilaterals did not approve. The apparent failure of

15 The case study evidence upon which this section draws is much more fully reported and discussed in Birdet al. (1999b).



the early IMF-based programmes in the context of the East Asian crisis had much todo with the perceived lack of commitment by the governments in Thailand, SouthKorea and Indonesia to the design of the programmes at that time. Similarly, themarkets perceived the Fund's analysis of the Brazilian economy in the aftermath ofthe Real Crisis as being over-optimistic, with the government's failure to implementall the agreed ®scal measures calling into question the credibility of the adjustmentprogramme. Previously it had been the government's perceived commitment to thePlan, and not involvement by the multilaterals that impressed capital markets. Similarexamples may be drawn from Mexico, Turkey, Bulgaria, Ghana, Kenya, Zambia andTanzania.

Experience in Africa is particularly relevant in the context of hypothesis (iii), wherethe case study evidence lends some support. Bilateral and multilateral aid arefrequently interconnected, with aid donors also providing input into the design ofprogrammes. However, the relationship is not always straightforward and where itdoes exist it is more in the nature of co-ordination than catalysis. In Ghana bilateralaid seems to have re¯ected a desire by donors to provide support, and it continued toincrease during times when support from the multilaterals was being interrupted. InZambia there was the opposite scenario with bilateral donors reducing aid at the sametime as the multilaterals continued to lend since Zambia was perceived to be meetingconditionality requirements. In Kenya the relationship between bilateral aid andmultilateral lending was closer although it was not a catalytic one but one ofmutuality. While Tanzania provides some support for a catalytic e�ect again whatreally seems to matter is the government's perceived commitment to reform.

The case study evidence is less clear with respect to hypothesis (iv), although it doesappear that to the extent that there was a negative catalytic e�ect in the case of theEast Asian crisis in 1997±98 this may be attributed to the design of the related IMFprogrammes.

It transpires that our case study evidence is, in large measure, consistent with bothour qualitative and econometric evidence, as well as the case study research that haspreviously been undertaken, where little, if any, evidence for a catalytic e�ect isdiscovered, (Killick, 1995; 1998; Mosley et al., 1991). In general, markets, and to alesser extent bilateral aid donors, will want to form their own views about the qualityof and commitment to economic policy reform. On its own IMF or World Bankconditionality is insu�cient (or even unnecessary) to generate a positive catalytice�ect.

7 STRENGTHENING THE CATALYZING ROLE OF THE IMF AND THEWORLD BANK

The ®rst question is whether we should be seeking to strengthen the catalytic e�ect atall. Should the multilaterals always be encouraging private capital markets and aiddonors to lend more? The East Asian crisis demonstrates how problems may becaused by excessive lending as well as by de®cient lending. The multilaterals shouldtherefore form a view at to the appropriate combination of economic adjustment andexternal ®nancing and endeavour to ensure that this optimum quantity of ®nancing issupplied (but not surpassed). How can they do this?



There are four channels through which the multilaterals may seek to exert acatalytic e�ect; leaving to one side the project lending activities of the Bank which lieoutside the scope of this paper. The ®rst channel is via the conditionality associatedwith policy-based lending. As noted earlier this has often been super®cially taken tobe the route through which catalysis already occurs. However deeper analysiscombined with empirical evidence of various sorts suggests otherwise. But to say thatconditionality has not had a catalytic e�ect in the past is not to say that it could not inthe future. There is scope for catalysis via conditionality. The trick is to makeconditionality more e�ective by working to improve both its design and itsimplementation. This implies taking a serious look at the range of proposals thatthere are for reforming conditionality rather than assuming that catalysis will bestrengthened simply by more of the same sort of conditionality that there has been inthe past. The key here seems to be in enhancing conditionality as a modality forencouraging economic reform by increasing the degree of ownership that govern-ments feel. There is a good deal of evidence to support the idea that conditionality hasworked better when it is combined with strong country ownership. If private marketsand aid donors are able to sense the degree of government commitment toconditionality this also helps to explain why conditionality may appear to have anuneven catalytic e�ect. Implementation could also be strengthened by applyingstricter penalties to non-completion via impaired access to future multilateral lending.In these circumstances private markets would have greater con®dence in both thedesign of conditionality and its implementation and would be more prepared tointerpret it as a lead indicator of future economic improvement.16

Yet it may also be appropriate that countries be allowed within the context ofconditionality to manage capital in¯ows via various forms of taxation. In the long runthe better management of capital in¯ows could enhance a country's creditworthinessand the durability of its access to private capital markets.17

The design and implementation of conditionality also depends on the amount of®nance provided by the multilaterals; there is some evidence to suggest that thesuccess of IMF programmes is an increasing function of the amount of ®nance that isprovided (Killick, 1995). A message which comes strongly out of the qualitativeresearch reported in this paper is that liquidity is an important factor in catalysis, andit is liquidity that represents the second channel through which the multilaterals canhave a catalytic e�ect.18 While it is widely accepted that the multilaterals themselvesneed to lend in order to encourage others to do the same, it may be that in the past thelending capacity of the IMF and the Bank has been too low to inspire the con®dencein country performance upon which catalysis relies. In this sense there may be acomplementarity between lending by the multilaterals and lending by privatemarkets. To some extent this has been recognized by the multilaterals since during theAsian crisis and its aftermath various devices were introduced, such as the

16 For a much fuller discussion of reform proposals along these lines see Killick (1995; 1998), Bird (1995,1996), Collier et al. (1997).17As the IMF has sought to extend conditionality to include the capital account of the balance ofpayments a great deal of interest has been shown in Chile which has set out to manage capital ¯ows. For anessentially positive assessment of Chilean policies see for example Eichengreen (1999). There is perhaps agrowing consensus that developing countries should not rush to liberalise their capital accounts, certainlynot in advance of domestic ®nancial liberalization and the introduction of an adequate prudential andregulatory framework.18 The quantitative support for this qualitative view is, however, not strong.



Supplemental Reserve Facility, to allow countries to signi®cantly exceed theirconventional quota limits on borrowing from the IMF.

At the same time some early attempts to quantify the resource needs of the IMFsuggested that even with on-going quota reviews the Fund could often requiresubstantial additional resources, (Williamson, 1983; Bird, 1987). And although therehave been periods when the Fund has had signi®cant spare lending capacity therehave also been times, particularly during the mid 1990s when its liquidity ratio hasfallen to very low levels.

One way of providing these additional resources would be for the Fund to borrowdirectly from private capital markets in order to supplement the subscriptions itreceives.19 In this way the Fund would become more of a pure international ®nancialintermediary. This could represent a more e�ective and e�cient means of channellingprivate capital to developing countries. Certainly it would by-pass the need forcatalysis. To the extent that the Fund believed that private markets would underlendin the aftermath of a Fund-backed agreement, and further believed that this wouldendanger the success of the related adjustment programme, direct borrowing fromprivate capital markets would be a way of circumventing such problems. However thisproposal is likely to be controversial.

The third channel of catalysis is less controversial. One problem associated withlending by private markets to developing countries and CITs relates to inadequaciesof information. The multilaterals have a comparative advantage in data collection anddissemination. Superior information should, other things being equal, lead tosuperior lending decisions. In terms of private capital ¯ows this may, of course, meansmaller capital ¯ows in some cases where markets would otherwise have beenexcessively exuberant, as well as larger ¯ows in other cases where markets would havebeen excessively pessimistic. Markets already take into account whether countriescomply with the data dissemination standards of the IMF, and this has an impact onperceived country risk and therefore capital ¯ows. Widening the range of relevantdata disseminated, and understanding better what it means for future economicperformance will have implications for catalysis.

Finally, co-ordination of creditors in conditions of crisis will have consequences forcatalysis. Evidence suggests that the Fund's catalytic e�ect on private capital ¯owswas perhaps at its strongest following the Third World debt crisis in 1982. This wasbecause the Fund coerced commercial banks to lend and made its own involvementconditional on the banks continuing to lend. For various reasons, not least the widerarray of creditors involved, coercion does not represent a long-term option. Howeverin the immediate aftermath of currency crises the Fund could play a more signi®cantrole in co-ordinating creditors and in discouraging them from withdrawing capital

19 This idea has been around for sometime but has received closer attention recently in the light of theFund's liquidity shortage, (Lerrick, 1999). The standard ripostes to this proposal are that ®rst it isinconsistent with the basic nature of the Fund as a club of members, second, it is not something that centralbanks do, and third, it might crowd out other borrowers such as the World Bank and the developingcountries themselves, (see IMF Survey, June 21, 1999). Less often voiced is the argument that privateborrowing could loosen the short leash by which the Fund's major shareholders may seek to constrain itsactivities. These criticisms are themselves open to criticism. Clubs borrow from time to time, The Fund isnot a central bank and cannot easily create its own resources. There is little reason to believe that Fundborrowing would crowd out other borrowers, Indeed expanded Fund lending ®nanced by borrowing frommarkets would crowd in additional ¯ows if it resulted in superior programmes. Moreover if the Fund wereonly to borrow when the demand for its resources was high this would almost certainly be at a time whenprivate capital markets were less prepared to lend to developing countries directly.



simply in an attempt to get out ahead of the pack. The details of how such co-ordination might work (and the related problems) have been discussed in some detailby Eichengreen (1999). Via the coordination of creditors the Fund could seek tominimize the chances of its involvement having a negative catalytic e�ect in thecontext of currency crises.

Apart from these broad channels through which the multilaterals could strengthentheir catalyzing role, another feature of the research reported here is that catalysisneeds to be analysed at a relatively high level of disaggregation. While, for example,IMF conditionality may be more important in catalyzing short-term capital ¯ows, theWorld Bank may have a more important catalyzing role in the context of foreigndirect investment. In order to maximise the e�ect of the multilaterals on other capital¯ows it is necessary to understand the details of their respective comparativeadvantages. However, to do this, it is also necessary to gain a better understanding ofwhy capital moves in the way that it does. At present there is no clear cut theory ofcapital movements which allows us to see where the involvement of the Fund and theBank ®ts in.

8 CONCLUDING REMARKS

The catalytic e�ect of lending by the IMF andWorld Bank is important in the contextof the debate surrounding the new international ®nancial architecture. One modelmight envisage almost perfect catalysis. This would present the Fund as an exclusivelyadjustment-oriented agency not making its own loans. Here, the endorsement of theFund would be su�cient to encourage private markets to lend. The Fund would be agenuine gatekeeper and the Fund's seal of approval would have a very high marketvalue. Although this model is not so far removed from the `commonly held view'reported in the introduction, it is a long way from the reality revealed by our research.The adjustment input of the Fund via the modality of conditionality just does notwork in this way. Nor does World Bank conditionality perform any better.

At the other extreme a second model would have the multilaterals providing theirown direct lending to ®ll the ®nancing gaps encountered by developing countries andcountries in transition. Here the multilaterals would not rely on there being anycatalytic e�ect. Indeed if there were to be such an e�ect there would be a danger ofoverlending and of sub-optimal adjustment which would make economic crises of onesort or another more likely. Two things are wrong with this model. First, both thequantitative and qualitative evidence shows that the IMF and the World Bank areÐalbeit to a limited extentÐ relevant in the decisions made by private markets and aiddonors. Second, and more importantly political economy considerations rule out theexpansion in IMF and World Bank resources via conventional means that would benecessary for them to take on this ®nancing role.

A third model is the most relevant. This recognizes the potential for the IMF andthe World Bank to in¯uence other capital ¯ows. However, it also recognizes thatcatalysis is complex and nuanced and that the potential for catalysis has remainedlargely untapped. At the very least a clear distinction needs to be made betweenprivate and bilateral public ¯ows. In order to extract the full potential for catalysis themultilaterals will need to undergo fairly fundamental reform, covering the design andgeneral modus operandi of their conditionality and the size of their resource base. The



former needs to focus on improving the success rate of the multilaterals'conditionality, which in turn implies measures to strengthen ownership. With regardsto the latter, there is scope for the Fund to take on a larger role as a ®nancialintermediary, borrowing directly from private capital markets rather than relying onthe vagaries of catalysis. There also needs to be closer co-ordination between themultilaterals and the private markets, particularly in the immediate aftermath ofcrises.

While only one part of the architecture debate, the issue of how best themultilaterals can mobilize other sources of external ®nance is important. The dangeris that commonly held views that are in fact at odds with reality will result ininappropriate reform.

ACKNOWLEDGEMENTS

Financial support for the research upon which this article is based was provided bythe UK Department for International Development. While gratefully acknowledgingthis support, the views and opinions expressed are those of the authors alone.Research assistance has been provided by Connie Tulus, Helgi Maki, and AntonellaMori.

REFERENCES

Bird G. 1987. International Financial Policy and Economic Development. Macmillan: London.

Bird G. 1994. The myths and realities of IMF lending. The World Economy 17(5): 759±778.

Bird G. 1995. IMF Lending to Developing Countries: Issues and Evidence. Routledge: London.

Bird G. 1996. The International Monetary Fund and developing countries: a review of the

evidence and policy options. International Organisation 50(3): 477±511.

Bird G. 1997. External ®nancing and balance of payments adjustment in developing countries:

getting a better policy mix. World Development 25(9): 1409±1420.

Bird G. 1998. IMF conditionality and the political economy of policy reform: is it simply a

matter of political will? Journal of Policy Reform 1(1): 89±113.

Bird G. 1999. How important is sound domestic macroeconomics in attracting capital in¯ows

to developing countries. Journal of International Development 11(1): 1±26.

Bird G, Orme T. 1981. An analysis of drawings on the International Monetary Fund by

developing countries. World Development 9(6): 563±568.

Bird G, Rowlands D. 1997. The catalytic e�ect of lending by the international ®nancial

institutions. The World Economy 20(7): 967±991.

Bird G, Hussain M, Joyce J. 1999a. Many happy returns: recidivism and the IMF.

(mimeograph).

Bird G, Mori A, Rowlands D. 1999b. Do the multilaterals catalyse other capital ¯ows: what

does the case study evidence show? Surrey Centre for International Economic Studies;

Working Paper Series 99/2.

Bird G, Rowlands D. 1999a. Does IMF conditionality signal policy credibility to markets.

Surrey Centre for International Economic Studies; Working Paper Series 99/5.

Bird G, Rowlands D. 1999b. Does World Bank policy based lending catalyse other ®nancial



¯ows? Surrey Centre for International Economic Studies; Working Paper Series 99/4.

Buira A. 1987. Adjustment with growth and the role of IMF in the 1980s. In The International

Monetary System and its Reform: Papers prepared for the Group of Twenty-Four, Dell S (ed.).

Elsevier Science Publishers: The Netherlands.

Burnside C, Dollar DR. 1997. Aid, policies and growth. World Bank Policy Research Working

Paper. World Bank: Washington, DC No. 17777.

Calvo GA, Leiderman L, Reinhart CM. 1993. Capital in¯ows and real exchange rate

appreciation in Latin America: the role of external factors. IMF Sta� papers 40(1): 108±151.

Chuhan P, Claessens S, Mamingi N. 1993. Equity and Bond Flows to Asia and Latin America.

Policy Research Working Papers. The World Bank: Washington, DC No. 1150.

Claessens S, Dooley M, Warner A. 1995. Portfolio capital ¯ows: hot or cold? The World Bank

Economic Review 9(1): 153±174.

Collier P, Guillaumont P, Guillaumont S, Gunning JW. 1997. Redesigning conditionality.

World Development 25(9): 1399±1407.

Conway P. 1994. IMF lending programs: participation and impact. Journal of Development

Economics 45(2): 365±391.

Cornelius P. 1987. The demand for IMF credits by sub-Saharan African countries. Economics

Letters 23: 99±102.

de Vries M. 1986. The role of the International Monetary Fund in the world debt problem. In

World Debt Crisis: International Lending on Trial, Claudon M (ed.). Ballinger Publishing

Company: Cambridge, MA; 111±122.

Dhonte P. 1997.Conditionality as an instrument of borrower credibility. IMF Paper on Policy

Analysis and Assessment PPAA/97/2; IMF: Washington, DC.

Dollar D, Svensson J. 1998. What explains the success or failure of structural adjustment

programs? World Bank Policy Research Working Paper. World Bank: Washington, DC No.

1998.

Edwards S. 1989. The International Monetary Fund and the developing countries: a critical

evaluation. Carnegie Rochester Conference Series on Public Policy 31: 7±68.

Eichengreen B. 1999. Toward a New International Financial Architecture. Institute for

International Economics: Washington, DC.

Faini R, deMelo J, Senhadji-Semlali A, Stanton J. 1991. Macro performance under adjustment

lending. In Restructuring Economies in Distress. Policy Reform and the World Bank, Thomas

V, Chhibber A, Dailami M, deMelo J (eds). The World Bank: Washington, DC; 222±242.

Fanelli JM, Frenkel R, Taylor L. 1994. Is the Market-Friendly Approach Friendly to

Development: A Critical Review. In Latin America's Economic Future, Bird G, Helwege A

(eds). Academic Press: London and San Diego.

Feldstein M. 1998. Refocusing the IMF. Foreign A�airs 77(2): 20±33.

Fernandez-Arias E. 1996. The new wave of capital in¯ows: push or pull? Journal of

Development Economics 48(2): 389±418.

Fernandez-Arias E, Monteil P. 1996. The surge of capital in¯ows to developing countries: an

analytical overview. The World Bank Economic Review 10(1): 51±77.

Hajivassiliou V. 1987. The external debt repayment problems of LDCs: an econometric model

based on panel data. Journal of Econometrics 36(1/2): 205±230.

IMF. 1999. IMF Survey. June 21, 1999; IMF: Washington, DC.

Joyce J. 1992. The economic characteristics of IMF program countries. Economics Letters

38(2): 237±242.

Kenen P. 1986. Financing, Adjustment and the International Monetary Fund. The Brookings

Institution: Washington, DC.



Khan MS. 1990. The macroeconomic e�ects of fund-supported adjustment programs. IMF

Sta� Papers 37(2): 195±231.

Killick T. 1995. IMF Programmes in Developing Countries: Design and Impact. Routledge/ODI:

London.

Killick T. 1998. Aid and the Political Economy of Policy Change. Routledge: London.

Killick T, Bird G, Sharpley J, Sutton M. 1984. The Quest for Economic Stabilisation: The IMF

and the Third World. Overseas Development Institute: London.

Krueger A. 1998. Whither the World Bank and the IMF? Journal of Economic Literature

XXXVI(4): 1983±2020.

Lerrick A. 1999. Private Sector Financing for the IMF: Now Part of an Optimal Funding Mix.

Bretton Woods Committee: Washington, DC April.

McCauley R. 1986. IMF: managed lending. In World Debt Crisis: International Lending on

Trial, Claudon M (ed.). Ballinger Publishing Company: Cambridge, MA; 123±146.

Milivojevic M. 1985. The Debt Rescheduling Process. St. Martin's Press: New York.

Mosley P, Harrigan J, Toye J. 1991. Aid and Power: The World Bank and Policy-Based Lending,

2 vols. Routledge: London.

Mosley P, Hudson J. 1997. Has aid e�ectiveness increased? (mimeographed).

Ozler U. 1993. Have commercial banks ignored history? American Economic Review 83(3): 608±

620.

Pool JC, Stamos S. 1987. The ABCs of International Finance. Lexington Books: Lexington,

MA.

Radelet S, Sachs J. 1998. The East Asian ®nancial crisis: diagnosis, remedies, prospects.

Brookings Papers on Economic Activity 1: 1±75.

Rodrik D. 1996. Why is there multilateral lending? In Annual World Bank Conference on

Development Economics 1995, Bruno M, Bleskovic B (eds). World Bank: Washington, DC;

167±193.

Rodrik D. 1999. Governing the global economy: does one architectural style ®t all? Paper

prepared for the Brookings Institution Trade Policy Forum conference on Governing in a

Global Economy, 15±16 April.

Rowlands D. 1994. The response of new lending to the IMF. The Norman Paterson School of

International A�airs Development Studies Working Paper 7: 1994.

Schadler S, Bennett A, Carkovic M, Dicks-Mireaux L, Mecagni M, Morsink JHJ, Savastano

MA. 1995. IMF Conditionality: Experience Under Stand-By and Extended Arrangements:

Part I: Key Issues and Findings, and Part II: Background Papers. Occasional Papers, Nos. 128

and 129, IMF: Washington.

Taylor MP, Sarno L. 1997. Capital ¯ows to developing countries: long and short term

determinants. World Bank Economic Review 11(3): 451±470.

Treasury Committee. 1997. International Monetary Fund, Fourth Report. London Stationary

O�ce: House of Commons, London (together with Proceedings, Minutes of Evidence and

Appendices).

ul Haque N, KhanM. 1998. Do IMF-supported programs work? A survey of the cross-country

empirical evidence. IMF Working Paper. IMF: Washington, DC WP/98/169.

Williamson J. 1983. IMF Conditionality. Institute for International Economics: Washington,

DC.



The catalyzing role of policy-based lending by the IMF and the World Bank: fact or fiction?

Documents

Transcript of The catalyzing role of policy-based lending by the IMF and the World Bank: fact or fiction?