D
N
B
W
o
r
k
iN
g
P
a
P
e
r
DNB Working Paper
imf-Supported Programs: Stimulating
Capital to Solvent Countries
Koen van der Veer and Eelke de Jong No. 244 / March 2010
Working Paper No. 244/2010 March 2010 De Nederlandsche Bank NV P.O. Box 98 1000 AB AMSTERDAM The Netherlands
IMF-Supported Programs: Stimulating Capital to Solvent
Countries
Koen van der Veer and Eelke de Jong
*
* Views expressed are those of the authors and do not necessarily reflect official
positions of De Nederlandsche Bank.
IMF-Supported Programs:
Stimulating Capital to Solvent Countries
Koen J.M. van der Veery
De Nederlandsche Bank
Eelke de Jongz
Radboud University Nijmegen
March 15, 2010
Abstract
Sovereign default is the switching state between successful and unsuccessful Fund catalysis. We …nd the IMF to be e¤ective in mobilising private capital ‡ows to middle-income countries that participate in a Fund program, but do not restructure their debt. A debt restructuring is a clear signal of very weak economic fundamentals, deterring creditors from resuming lending, even when the IMF intervenes. As long as default is avoided, IMF programs help a country signal its willingness to reform and repay debts, thereby catalysing private capital. This signalling role appears to be more important for Fund catalysis, than the size of IMF lending.
JEL codes: F32, F33, F34.
Keywords: IMF, Sovereign default, Private capital ‡ows, Catalytic e¤ect
We are grateful to Martin Admiraal and Henk van Kerkho¤ for collecting the data and to Marco Hoeberichts for excellent research advice. For comments, we thank Marloes Foudraine, Poonam Gupta, Pierre Lafourcade, Ashoka Mody, Marc Roovers and seminar participants at De Nederlandsche Bank. The views expressed in this paper are those of the authors and do not necessarily represent those of the institutions with which they are a¢ liated.
yCorresponding author: [email protected], Economics and Research Division, De Nederlandsche Bank, P.O. Box 98, 1000 AB, Amsterdam, The Netherlands, tel: +31 (20) 524 5836, fax: +31 (20) 524 2506.
zDepartment of Economics, Radboud University Nijmegen, P.O Box 9108, 6500 HK Nijmegen, The Netherlands, Email: [email protected]
1
Introduction
The International Monetary Fund (IMF or Fund) has an important role in triggering private capital ‡ows to countries that participate in an adjustment program. Fund-supported programs are intended to pave the way toward a return to balance of payments viability and sustainable growth. In order to accomplish these goals, a country’s funding gap needs to be covered. Private capital in‡ows are important in this regard. IMF loans tend to fall short of a country’s …nancing needs, and the international community usually avoids a full bailout by o¢ cial rescue loans in order to limit moral hazard. When the IMF fails to catalyse private capital ‡ows, a funding gap remains, and a breakdown of the Fund program becomes more likely.
Is the IMF successful in "catalysing" private capital ‡ows to developing countries? Surveying the empirical literature, Cottarelli and Giannini (2006) conclude that ‘the potential of catalytic o¢ cial …nance as a crisis management tool appears at best limited.’ Similarly, Bird (2007) observes that ‘the evidence certainly implies that it is misplaced to rely on the Fund catalysing private capital markets to lend.’ Indeed, it is often argued that Fund programs signal payment di¢ culties, increasing spreads and triggering private capital out‡ows ( ½Ozler, 1993; Bird and Rowlands, 2002, 2009; Jensen, 2004; Edwards, 2006). Thus, in general, the data seems to reject Fund catalysis.
Our …ndings, however, go counter to this empirical rejection of catalysis. We show that IMF programs do stimulate private capital ‡ows, but only to countries that do not default. Thus, our main contribution is to identify empirically this primary condition for a catalytic e¤ect of IMF programs.
Mody and Saravia (2006) and Eichengreen, Kletzer and Mody (2006) examine the theoretical insight that Fund catalysis is most likely to work when a country’s fundamentals are poor, but not hopelessly so (Morris and Shin, 2006; Corsetti, Guimarães and Roubini, 2006). These authors show that IMF programs lower bond spreads and increase bond issuance, but only in countries within a limited "intermediate" range of external debt or reserves. We build on their analyses but show that the scope for successful Fund catalysis is broader,and that Fund catalysis applies to private capital ‡ows as well. Basically, IMF programs turn out to catalyse private capital ‡ows when countries are solvent (i.e. do not restructure their debt), almost regardless of a country’s level of debt or reserves. Thus, our evidence does not support the idea of IMF programs signalling trouble in country’s with relatively sound fundamentals.
But how can the IMF catalyse private capital? The IMF describes the mechanism underlying catalysis as follows:
In most cases, IMF loans provide only a small portion of what a country needs to …nance its balance of payments. But, because IMF lending signals that a country’s eco-nomic policies are on the right track, it reassures investors and the o¢ cial community and helps generate additional …nancing. Thus, IMF …nancing can act as a catalyst for attracting funds from other sources.1
This "signalling role" of the IMF features in the theoretical models by Marchesi and Thomas (1999) and Tirole (2002), who see IMF policy conditionality as a way to signal a country’s adjustment e¤ort to stave o¤ default. Other models highlight the liquidity provision by the IMF as an explanation for catalysis (Morris and Shin, 2006; Corsetti, Guimarães and Roubini, 2006). Liquidity support lowers the likelihood of default by enlarging the range of economic fundamentals at which private creditors roll-over their credit to the country.
We remain, however, agnostic on these theories. Our results indicate Fund catalysis of private capital ‡ows in the …rst program year, which mirrors the predictions of models highlighting the role of liquidity provision in preventing default. Since IMF lending is conditional on policy adjustment, however, our …nding can also be due to the Fund program signalling a country’s reform e¤ort to avoid default.
We apply an instrumental variable approach à la Barro and Lee (2005) to account for the endo-geneity of IMF programs. Our model con…rms that failure to control for the non-random selection of IMF programs would introduce a bias against Fund catalysis of capital ‡ows. Contrary to previous studies, we di¤erentiate between Fund programs in countries that simultaneously do and do not restructure their debt. This way, we single out the e¤ect of IMF intervention in insolvent countries for which we –in line with theory –do not expect a catalytic e¤ect on capital ‡ows.
In what follows, we discuss the recent theoretical literature that focuses on the conditions for e¤ective Fund catalysis (Section 2) and review the latest empirical …ndings (Section 3). We describe our methodology and data in Section 4. Section 5 presents our benchmark results. In Section 6 we test the sensitivity of our benchmark results to a number of changes in the …rst and second stage
regression of our system, and show how our …ndings relate to existing evidence on Fund catalysis in countries with "intermediate" fundamentals. Section 7 concludes.
2
Catalysis and default: theoretical considerations
We draw on recent theoretical studies to identify how the IMF could trigger private capital ‡ows, and under what conditions this catalytic role of the IMF may be successful. While the models focus on either IMF lending or program conditionality as the driver of Fund catalysis, they are all build on the interaction between the debtor country, private creditors and the IMF. The decision of each player is endogenous to the actions of the other. Private creditors will resume lending if they believe the IMF can induce the debtor country to adopt policies conducive to repayment of private debt. Recognizing its pivotal role, the IMF will intervene. The debtor country, in turn, realizes that the private creditors and the IMF will only act accordingly if costly domestic adjustment policies are undertaken.
E¤ective catalysis of Fund programs, then, hinges on IMF lending being a complement (not a substitute) to private lending and on the ability of the IMF to induce the debtor country to implement the necessary adjustment policies. Crucially, IMF intervention must not weaken a government’s incentive to implement desirable but costly policies. The literature suggests that the extent to which such debtor moral hazard can be avoided depends on the country’s economic fundamentals.
First, consider the catalytic role of IMF lending. Morris and Shin (2006) and Corsetti, Guimar¼aes and Roubini (2006) argue that even in the absence of explicit conditionality, IMF lending can trigger private lending by alleviating the debtor country’s costs of implementing adjustment policies. Yet, the IMF can only e¤ectively counteract the risk of insolvency and restore con…dence when the economic fundamentals of the country are poor, but not hopelessly so. When fundamentals are very weak, an IMF program is unable to compensate for the economic di¢ culties faced by the country, at least not in the short or medium term. Similarly, Penalver (2004) relates the success of Fund catalysis to the ability to avoid default. The IMF lends funds below the prevailing market interest rate and it is this subsidy that induces the borrowing country to exert adjustment e¤ort to avoid default. By preventing default, future marginal rates of return on investments are kept high, thereby encouraging private capital ‡ows.
signalling that a country’s economic policies are on the right track. Marchesi and Thomas (1999) argue that by signing a Fund program, countries can signal their willingness and ability to reform and use new money for investment and debt repayments. As such, Fund policy conditionality can work as a "screening device", enabling private creditors to discriminate between debtor countries that are willing and unwilling to adjust policies. Likewise, Tirole (2002) argues that the IMF can monitor a country’s policies and thus serve as a ‘delegated monitor’for private creditors. The Fund’s role is to substitute for the missing contracts between the sovereign and individual foreign investors and thereby help the country to attract private capital.
Like the catalytic role of IMF lending, Mody and Saravia (2006) observe that catalysis via the signalling role of the Fund also depends on the country’s economic fundamentals. Country commit-ment through the Fund is likely to be e¤ective when countries are vulnerable but have not yet crossed thresholds that imply inability to service external debts even with Fund assistance. Consequently, an IMF program is unlikely to catalyse new capital when solvency is at stake.
3
Recent empirical studies
This Section brie‡y reviews the recent empirical literature on the catalytic e¤ect of IMF programs. These studies are di¤erent from the earlier attempts to identify Fund catalysis in that they allow for the impact of Fund programs to di¤er across country fundamentals and/or control for the nonrandom selection of countries with an IMF program.2
Mody and Saravia (2006) take into account country conditions when examining the catalytic e¤ectiveness of Fund programs. Their results imply that IMF programs are e¤ective in reducing bond spreads when a country’s reserves cover between 4 and 12 months of imports or the debt-to-GNP ratio is between 36 and 54%. Also, they …nd that Fund presence raises the frequency of bond issuance in a comparable ‘intermediate’range of debt or reserves. Thus, the important point Mody and Saravia demonstrate, is that Fund catalysis can be e¤ective, but that the catalytic e¤ect is limited to countries within an intermediate range of fundamentals.
Studies estimating the impact of Fund programs on private capital ‡ows have traditionally gen-erated evidence against Fund catalysis (see Cottarelli and Gianini, 2006). Edwards (2006), however, noticed a ‡aw within this empirical literature, namely that it overlooks the nonrandom nature of
2
IMF programs. Countries sign an IMF program when they face a balance of payments crisis. Due to their economic di¢ culties, these countries are less likely to attract private capital ‡ows. Failure to control for this sample selection when estimating the impact of Fund programs leads to a bias against Fund catalysis. Edwards controls for sample selection, but rea¢ rms that Fund programs lead to private capital out‡ows. Surprisingly, he …nds that correction for sample selection makes no di¤erence to his results. A likely reason, however, is his focus on portfolio equity ‡ows. Most low and middle-income countries have zero in‡ow of equity investment, possibly precluding identi…cation of sample selectivity.
Bird and Rowlands (2002, 2009), in particular, cast doubt on the catalytic e¤ect of IMF programs. In their most recent contribution, they estimate net total private capital ‡ows and account for sample selectivity, but …nd no grounds for catalytic conversion. Likewise, Jensen (2004) reports a negative e¤ect of Fund programs on foreign direct investment, without allowing for the e¤ect to vary across countries’economic fundamentals.
Our main concern with these studies, however, is the way they di¤erentiate between solvent and insolvent countries that sign an IMF program. The practice is to categorize countries according to their level of external debt or reserves, assuming that a certain level of debt or reserves implies the same state of economic fundamentals across countries. This assumption, however, is easily refuted. For example, sovereign default is a clear measure of very weak country fundamentals. Yet, the levels of external debt or reserves in countries that default diverge greatly.3 Consequently, the rejection of Fund catalysis in previous studies could be due to an insu¢ cient divide between solvent and insolvent countries. Indeed, Arteta and Hale (2008) show that when countries restructure their debt, they experience a decline in bond and bank lending that persists for over two years.
In this article, we improve on the way country fundamentals are accounted for when evaluating Fund catalysis. We examine seperately the reaction of private creditors to Fund intervention in countries that do and do not restructure their commercial or o¢ cial debt. This way, we focus on the IMF’s traditional role of addressing short-term balance of payment crises.
3A simple check of the raw data reveals that a third of the countries that default have a level of external debt to GDP below the mean value of the countries that do not default. Likewise, more than a third of the defaulting countries have a level of reserves higher than the mean level of the non-defaulting group.
4
Methodology and data
We aim to examine whether IMF programs trigger private lending. We focus on Stand-by Arrange-ments (SBAs) and Extended Fund Facilities (EFFs), which are the traditional Fund programs to address balance of payments crises in middle-income countries. SBAs are designed to help countries address short-term balance of payments problems, while EFFs assist member countries in overcoming balance of payments problems that stem largely from structural problems and require a longer period of adjustment.
Our dataset includes total private capital ‡ows to all middle-income countries in the period from 1984 to 2004. Due to missing observations our …nal sample includes 49 countries (see Appendix Table A.1 for the countries in our sample). We measure the e¤ect on private capital ‡ows one year after the IMF program is signed. Focusing on the …rst program year allows us to ignore the issue of program implementation, which becomes important when estimating the impact of Fund programs over a longer horizon (Edwards, 2005). Contrary to previous studies, we di¤erentiate between Fund programs in countries that do and do not default. This way, we single out the e¤ect of IMF intervention in insolvent countries for which wea priori do not expect Fund catalysis.
We apply an instrumental variable approach to account for the endogeneity of IMF programs and follow the strategy proposed by Barro and Lee (2005) (see also Eichengreen, Gupta and Mody, 2008). They argue that the IMF is a political organization, where the decision to approve a Fund program is in‡uenced by the Fund’s major shareholders, in particular the United States. As such, program participation is determined by a country’s economic situation and its political or economic proximity to the United States. The authors come up with various measures to capture the link between a country and the United States and use these variables as instruments for IMF programs. Accordingly, we estimate a "…rst stage" probit model for the probability a country signed an IMF program:
Bit= i+ t+Zit +"1it
where ‘B’ is a binary variable with value one when a country signed an IMF program but did not default, and are country and year …xed e¤ects respectively, and ‘Z’is a vector of variables predicting participation in a Fund program. Aside from country characteristics, ‘Z’includes variables re‡ecting a country’s economic and political proximity to the United States.
From the probit estimation we generate predicted values of Fund program participation and use these predicted values (or propensity scores) as an instrument for program participation in the "second stage" regression for private capital ‡ows:
Yit= i+ t+Xit +"2it
where ‘Y’are total private capital ‡ows in percent of GDP, and are country and year …xed e¤ects, and ‘X’is a vector of variables explaining capital ‡ows (including the propensity score) and"2
is a second error term. All explanatory variables are lagged one year to avert the issue of simultaneity or reverse causality.
4.1 Variables and data sources
We measure total private capital ‡ows as the aggregate of bond and bank loans, equity investment, short term ‡ows and foreign direct investment, all taken from the World Bank’s Global Development Finance. The World Bank’s data on capital ‡ows records gross ‡ows (except for foreign direct investment).4 Focusing on gross capital ‡ows assures that a positive change is driven by increased in‡ows instead of reduced out‡ows. If IMF programs are catalytic, they should be expected to increase gross capital ‡ows (Bordo, Mody and Oomes, 2004).
The second stage equation includes "pull" and "push" factors of private capital ‡ows. The pull factors are a common set of economic variables measuring a country’s solvency and liquidity position: real GDP growth, trade to GDP, debt servicing to exports, external debt, reserves in months of imports, short term debt to reserves, (change in) domestic credit to GDP. These conventional measures come from the IMF’s International Financial Statistics or World Economic Outlook; except for trade, external debt and debt service for which we used World Bank resources (see Appendix Table A.2 for the source and calculation of each variable). Aside from these more conventional measures, we include a measure of exchange rate volatility, calculated as the annual variance to the mean of a country’s monthly US dollar exchange rate. The volatility of a country’s exchange rate is an indicator of …nancial instability and exchange rate risk (Jeanneau and Micu, 2002). Finally, we
4The World Bank’s data on capital ‡ows are net of amortizations on account of principal repayment. Foreign direct investment (FDI) is measured as the balance of in- and out‡ows of the reporting country. Since FDI is an important source of capital for middle income countries we prefer to include FDI in our measure of capital ‡ows.
include year …xed e¤ects to capture global "push" factors of private capital ‡ows.5
The …rst stage equation examines the probability for a solvent country to participate in a Fund program. We de…ne a country that participates in an IMF program as insolvent when it restructures its commercial or o¢ cial debt in the same year. The data on commercial debt restructuring agree-ments are available from the World Bank’s Global Development Finance (2002). We used subsequent issues of the Global Development Finance for agreements concluded in the period from 2001 to 2004. The Paris Club website reports data on restructuring agreements of o¢ cial bilateral debt. The IMF calls on the Paris Club creditors – mainly OECD countries – when the success of a rescue package is linked to the ability of a debtor to restructure its existing claims (IMF, 2001). As such, a Paris Club agreement is a signal of very weak country fundamentals. Indeed, Arteta and Hale (2008) …nd that o¢ cial debt restructuring leads to a larger decline in credit than commercial agreements. As a possible explanation, the authors argue that commercial agreements contain no new information since o¢ cial creditors usually negotiate with sovereigns before commercial creditors.
Table 1 shows the number of IMF programs approved and commercial and o¢ cial debt restructur-ings concluded in middle income countries in the period from 1984 to 2004. Standby Arrangements –the short-term stabilization program –make up the bulk of the IMF programs (83%). A little over a third (37%) of all countries that signed a Fund program also restructured their commercial and/or o¢ cial debt in the same year. Thus, about two-third of all IMF programs were signed in what we de…ne as solvent countries.
Previous research has identi…ed various determinants of Fund program participation.6 Our
bench-mark model is similar to the model used by Eichengreen, Gupta and Mody (2008). To capture a country’s economic and political proximity to the United States, we include a variable measuring U.S. aid as a percentage of total foreign aid received by a country. The data on foreign aid are from the OECD DAC database. We also examine alternative political-economy variables, such as the fraction of votes that each country cast in the United Nations General Assembly along with the United States (Barro and Lee, 2005) and U.S. bank exposure (Oatley and Yackee, 2004).
In addition to U.S aid receipts, we include other country-speci…c variables. These macroeconomic controls are similar to the variables used in the capital ‡ows equation. Yet, instead of our measure
5Some authors propose using United States interest rates as a global ‘push’ factor for private capital ‡ows to developing countries (Calvo, Leiderman and Reinhart, 1993), although other studies failed to con…rm this relationship (World Bank, 1997). We did not …nd US interest rates to a¤ect a country’s capital ‡ows.
6
of exchange range volatility, we added dummy variables for pegged rates and regimes of limited ‡exibility using the Reinhart-Rogo¤ classi…cation of exchange rate regimes (Reinhart and Rogo¤, 2004).7 The summary statistics for all variables are in Table 2.
5
Benchmark results
5.1 Predicting IMF program participation in solvent countries
Table 3 presents the results for the …rst stage regression estimating the likelihood of an IMF program. The share of U.S. aid is our key instrumental variable. Estimated without other conditioning vari-ables, we …nd a positive and statistically signi…cant in‡uence of U.S. aid on program participation (Table 3, Column 1). Columns 2 and 3 show, however, that U.S aid is positively and signi…cantly related to Standby Arrangements (SBAs) but not to Extended Fund Facilities (EFFs). This result is consistent with the …nding by Stone (2008) that U.S in‡uence operates to constrain conditionality. Since the level of conditionality of EFFs is higher than for SBAs, we are not surprised to …nd a signi…cant in‡uence of U.S. aid on SBAs only.8 Nevertheless, as the signi…cant correlation between U.S aid and Fund program participation is crucial to our instrumental variable approach, we prefer to limit our analysis to SBAs. We show below that our second stage results regarding Fund catalysis are robust to including EFFs in the …rst stage regression.
The results of our benchmark model for program participation are presented in Table 3 column 4. Importantly, U.S. aid remains positive and statistically di¤erent from zero, even when additional country-speci…c variables are included in the model.9 The signs of the conditioning variables are plausible and generally consistent with other studies. Solvent countries are more likely to participate in a Standby Arrangement (SBA) when they have a higher level of debt service, a rising level of external debt or a pegged exchange rate. The latter …nding is consistent with the idea that ‡exible exchange rates allow countries to better accommodate external shocks (Edwards and Levy Yeyati,
7
Also, in the capital ‡ows equation we replaced the variable ‘change in debt/GDP’ for the level of external debt, which turned out to be a better determinant of capital ‡ows. The results do not change when debt/GDP and the change in debt/GDP are included in the model.
8Standby Arrangements test an average of …ve categories per month, whereas Extended Fund Facilities test an average of seven categories (Stone, 2008).
9
While these additional variables do not provide identi…cation in the second stage, because they do not plausibly satisfy the exclusion criterion for an instrumental variable, they reassure us that any signi…cance imputed to the political-economic determinants are not really attributable to these other characteristics (see Eichengreen, Gupta and Mody (2008) for similar reasoning).
2005). We also …nd a positive but marginally signi…cant e¤ect of a change in domestic credit. A possible explanation could be that a domestic credit-boom increases the chance of a sudden stop in capital ‡ows (IMF, 2004; Eichengreen, Gupta and Mody 2008), thereby increasing the likelihood of Fund intervention.10 Finally, solvent countries are less likely to sign a SBA the better their access
to capital markets (as measured by the ratio of short term debt to reserves). Economic growth, trade, reserves, and domestic credit, are also negatively related to SBAs, but none of these e¤ects is signi…cantly di¤erent from zero.
In Table 3 columns 5 to 7, we examine alternative measures to capture a country’s link with the United States. While both UN voting in line with the United States and U.S. bank exposure enter the regression positively, neither is statistically signi…cant. The estimate of U.S. aid, however, is unchanged and remains statistically distinguishable from zero at all reasonable signi…cance levels. Therefore, we use the speci…cation of the fourth column of Table 3 for instrumenting IMF programs in the second stage regression of capital ‡ows.
5.2 The catalytic e¤ect of IMF programs in solvent countries
Figure 1 shows the pattern of capital ‡ows in countries that participate in an IMF program. The di¤erence between solvent and insolvent countries is clear. Only solvent countries experience a strong recovery of capital ‡ows in the …rst program year. Countries that restructure their debt see a large decline in capital ‡ows in the year before signing an IMF program, and no recovery to the pre-program level in the …rst pre-program year. Thus, the raw data are consistent with the idea that Fund intervention can catalyse private lending but only to solvent countries.
The question is whether this result still holds when we condition on other variables explaining capital ‡ows and account for the endogeneity of IMF programs. The results for our benchmark model of capital ‡ows are presented in Table 4 column 1. Before we discuss the catalytic e¤ect of IMF programs, we brie‡y discuss the other determinants of capital ‡ows.
The benchmark regression results show that economic growth is associated with higher capital in‡ows. A higher level of external debt, on the contrary, is strongly related with lower capital in‡ows. Also, we …nd evidence that exchange rate volatility reduces capital in‡ows. Thus, the model seems to indicate that …nancially instable countries or countries with a higher risk of facing external payments
1 0
We do, however, not …nd evidence supporting this reasoning in our “second stage” regression, where we …nd the change in domestic credit to have a positive and statistically insigni…cant e¤ect on capital ‡ows.
problems receive less private capital. Finally, we do not …nd evidence for the idea that a domestic credit boom heightens the risk of a sudden stop of capital ‡ows (see i.e. Eichengreen, Gupta and Mody 2008). In our model, domestic credit growth is positively and insigni…cantly associated with capital in‡ows.
Turning to the estimate of greatest interest; Fund programs seem to catalyse capital in‡ows to solvent countries. The instrumental variable estimate for IMF programs is positive and statistically distinguishable from zero, but only when a country does not restructure its debt (Table 4, column 1). The estimate for IMF programs interacted with a debt restructuring dummy is negative and statistically insigni…cant. The second column of Table 4 presents the results when endogeneity bias is not accounted for. The estimated IMF program e¤ect is also positive but statistically insigni…cant, underlining the importance of dealing with endogeneity. As for the magnitude of the catalytic e¤ect in solvent countries, the point estimate of 9.140 implies that an increase in the probability of program participation by 0.135 (one standard deviation) increases capital in‡ows by 1.2%.
6
Sensitivity analysis
We start our sensitivity analysis with some robustness checks based on changes to the …rst stage regression. Then, we test the sensitivity of our main result to a number of changes in the second stage regression for capital ‡ows. Subsequently, we i) estimate a dynamic model, ii) account for unobserved heterogeneity in program participation, and examine the in‡uence of iii) past and iv) precautionary IMF programs, v) the size of IMF lending, vi) impact of exchange rate regime and vii) the sample period. We conclude by examining how our …ndings relate to previous studies indicating that Fund catalysis is most e¤ective when a country’s economic fundamentals are in an "intermediate" range.
First, we check the sensitivity of our main result to changes in the dependent variable of the …rst stage regression of program participation. We examine whether it matters to account for both commercial and o¢ cial debt restructurings in the dependent variable. Recall, that the …rst stage dependent variable is a dummy variable equal to one if a country signed an IMF program and did not restructure its commercial or o¢ cial debt in the same year. In the …rst and second column of Table 5, we changed the dependent variable by respectively accounting for commercial or o¢ cial debt restructurings only. The catalytic e¤ect is robust to both measures, but the estimate is less signi…cant
when o¢ cial debt restructurings are not taken into account. Subsequently, the third column of Table 5 shows that our basic result remains resilient when we include Extended Fund Facilities in the dependent variable.
Next, we conduct a battery of robustness checks based on reasonable changes to the second stage regression of capital ‡ows. First, we estimate a dynamic version of the model by including the lagged dependent variable in the regression of capital ‡ows. This way, we account for the fact that capital ‡ows are often correlated over time. Indeed, the lagged dependent variable enters the regression positively and is statistically signi…cant (Table 6, Column 1). Our main result, however, is unchanged.11 Including additional lags of the dependent variable did not a¤ect our …ndings either, moreover these lags were not signi…cantly correlated with the dependent variable.12
Subsequently, we test whether the results hold when we account for unobserved di¤erences be-tween program and non-program countries. Our benchmark model adjusts for the non-random selection of IMF programs by controlling for observed di¤erences only. To adjust for unobserved het-erogeneity, we use the results of the …rst stage regression to construct a selection bias control factor – lambda –equivalent to the Inverse Mill’s Ratio. Lambda is a summary measure re‡ecting unobserved variables correlated with Fund program participation. Our main result is robust to the inclusion of lambda in the speci…cation, although the estimate now is signi…cant at the 10 percent level (Table 6, Column 2). As to be expected, lambda is negative (but statistically insigni…cant), indicating that unobserved variables correlated with Fund program participation reduce capital in‡ows.
We continue the sensitivity analysis by examining whether the catalytic e¤ect is related to the size of the IMF loan. We measure the size of the IMF loan as a percentage of a country’s quota within the IMF, but …nd no signi…cant impact on capital ‡ows (Table 6, Column 3). Also, our main result is unchanged. These results suggest that countries can use a Fund program to signal adjustment e¤ort and the intention to repay debts. The size of the IMF loan does not seem to matter for e¤ective Fund catalysis.
A related test is to see whether our …nding is driven by so-called precautionary programs. When an IMF program is precautionary, the country agrees to meet speci…c conditions for use of IMF
1 1We decided not to include the lagged dependent variable in our benchmark model since the parameters estimates are known to be biased in models with …xed e¤ects and lagged dependent variables. The problem is that the di¤erenced residual is necessarily correlated with the lagged dependent variable, because both are a function of the lagged residual (see for further details Angris and Pischke, 2009).
resources, although it indicates to the Executive Board its intention not to make purchases. Previous studies …nd that the catalytic e¤ect of IMF programs appears salient in the context of precautionary programs (Mody and Rebucci, 2006). The fourth column of Table 6 shows that our …nding is robust to the inclusion of a dummy for precautionary programs, although the size and the statistical signi…cance of the catalytic e¤ect reduce slightly. Lastly, we add controls for current and previous IMF programs (Table 6, Column 5); our result remains resilient.
Finally, we check the sensitivity of our results by accounting for the type of exchange rate regime (Table 6, Column 6), estimating a full model including all the controls examined above (Column 7), and shortening the sample to the period from 1990 to 2004 (Column 8). Again, none of these changes destroy our …ndings. We conclude that our results are relatively robust to reasonable changes in the …rst and second stage regression. IMF programs seem to stimulate private capital ‡ows to solvent countries in the …rst program year.
6.1 The scope for e¤ective Fund catalysis
The …nal question we address is how our results relate to previous studies indicating that Fund catalysis is most e¤ective when a country’s economic fundamentals are in an intermediate range (see Mody and Saravia, 2006; Eichengreen, Kletzer and Mody, 2006). These authors categorize countries into those with good, intermediate or bad fundamentals according to their level of external debt or reserves, and …nd IMF programs to reduce bond spreads and raise bond issuance in a limited intermediate range of economic fundamentals.
To put our results into perspective, we examine whether our …nding of Fund catalysis of total private capital ‡ows is due to the subsample of countries with an intermediate level of debt or reserves. The raw data indicate that countries with an IMF program and no commercial or o¢ cial debt restructuring (our …rst stage dependent variable) are almost equally represented within the three categories of fundamentals as de…ned by the studies cited above. The intermediate category, for respectively debt or reserves, accounts for 30 or 32% of the 125 observations with a Standby Arrangement and no debt restructuring. Thus, our sample of IMF programs in solvent countries is not concentrated particularly in countries with an intermediate level of debt or reserves.
with an intermediate range of debt or reserves.13 The e¤ect of IMF programs on capital ‡ows is positive and statistically signi…cant even when we drop all observations in the intermediate range (Table 7, Column 2 and 5). Likewise, our result is also fairly robust to dropping all observations with good fundamentals (Table 7, Column 1 and 4). Finally, when we remove all observations with bad fundamentals, the results are ambiguous. The positive e¤ect of Fund programs disappears when we exclude countries with a level of debt above 60%. Yet, the estimate for Fund programs remains positive and statistically signi…cant when observations with low reserves are dropped. Either way, contrary to earlier studies, our evidence does not support the idea of IMF programs signalling payment di¢ culties, even when a country’s fundamentals are relatively sound.14
We conclude that our results are not due to some subset of our sample. IMF programs seem to catalyse private capital ‡ows to countries that do not restructure their debt, almost irrespective of a country’s level of debt or reserves. Thus, the scope for e¤ective Fund catalysis seems to be broader than suggested in previous studies.
7
Conclusions
By distinguishing between solvent and insolvent countries, we improve on the way country funda-mentals are accounted for when measuring the impact of IMF programs on private capital ‡ows. Previous studies either reject the idea of e¤ective Fund catalysis or …nd only a narrow window of e¤ectiveness. Our results indicate that IMF programs catalyse capital ‡ows to countries that do not restructure their commercial or o¢ cial debt. These …ndings are consistent with the idea that the success of Fund catalysis depends on the ability to avoid default. To better understand the factors driving catalysis, we also examined whether the size or use of IMF credit matters. Although the sig-nalling and lending function of IMF programs are di¢ cult to analyse separately, the evidence points out that countries can use IMF programs to signal their commitment to reform and willingness to repay debts to private creditors.
Preventing default, then, is crucial for the IMF to play its role as a catalyst. Whenever a country
1 3This result also goes when we measure fundamentals by a country´ s short term debt as percent of reserves (see for the respective categories of fundamentals, Bordo, Mody and Oomes, 2004).
1 4
Thus, while IMF programs in countries with good fundamentals, in theory, could send a negative signal to capital markets. In practice, such countries do not seem to participate in IMF programs. Also, Cottarelli and Giannini (2006, note 29) note that the argument of a negative signal of IMF programs seems not much convincing, as typically, markets know already that a country is in trouble when it reaches the point of starting program discussions with the IMF.
defaults, the ability of the IMF to catalyse private capital ‡ows wears o¤, at least in the short run. A policy implication, therefore, is that early IMF intervention is important.
References
[1] Angrist, J.D and J. Pischke (2009). "Mostly Harmless Econometrics. An Empiricist’s Compan-ion." Princeton University Press: Princeton.
[2] Arteta, C. and G. Hale (2008). "Sovereign Debt Crises and Credit to the Private Sector."Journal of International Economics, 74, pp. 53-69.
[3] Barro, R.J. and J.W. Lee (2005). "IMF Programs: Who Is Chosen and What Are The E¤ects?" Journal of Monetary Economics, 52, pp. 1245-1269.
[4] Bird, G. (2007). "The IMF: A Bird’s Eye View of Its Role and Operations."Journal of Economic Surveys, 21 (4), pp. 683-745.
[5] Bird, G. and D. Rowlands (2002). "Do IMF Programmes Have a Catalytic E¤ect on Other International Capital Flows?" Oxford Development Studies, 30 (3), pp. 229-249.
[6] Bird, G. and D. Rowlands (2009). "The IMF’s Role in Mobilizing Private Capital Flows: Are There Grounds for Catalytic Conversion?"Applied Economics Letters, 16(17), pp. 1705-1708. [7] Bordo, M.D., A. Mody and N. Oomes (2004). "Keeping Capital Flowing: The Role of the IMF."
International Finance, 7(3), pp. 421-450.
[8] Calvo, G., L. Leiderman and C. Reinhart (1993). "Capital In‡ows and Real Exchange Rate Appreciation in Latin America."IMF Sta¤ Papers, 40, pp. 108-151.
[9] Corsetti, G., B. Guimarães and N Roubini (2006). "International Lending of Last Resort and Moral Hazard Distortions: A Model of IMF’s Catalytic Finance.", Journal of Monetary Eco-nomics, 53, pp. 441-471.
[10] Cottarelli, C. and C. Giannini (2006). "Bedfellows, Hostages, or Perfect Strangers? Global Capital Markets and the Catalytic E¤ect of IMF Crisis Lending." In: IMF-Supported Programs. Recent Sta¤ Research, ed. Mody A. and A. Rebucci. Washington, D.C.: International Monetary Fund.
[11] Eichengreen, B., Kletzer, K. and A. Mody (2006). "The IMF in a World of Private Capital Markets."Journal of Banking and Finance, 30, pp. 1335-1357.
[12] Eichengreen, B., P. Gupta and A. Mody (2008). "Sudden Stops and IMF-Supported Programs." In: Financial Markets Volatility and Performance in Emerging Markets, ed. Gomes, M. and M.G.P. Garcia. Chicago: University of Chicago Press.
[13] Edwards, M. S. (2005). "Investor Responses to IMF Program Suspensions: Is Non-Compliance Costly?"Social Science Quarterly, 86 (4), pp. 854-873.
[14] Edwards, M. S. (2006). "Signalling Credibility? The IMF and Catalytic Finance." Journal of International Relations and Development, 9 (1), pp. 27-52.
[15] Edwards, S. and E. Levy Yeyati (2005). "Flexible Exchange Rates as Shock Absorbers." Euro-pean Economic Review, 49, pp. 2079-2105.
[16] International Monetary Fund (2001). "Involving the Private Sector in the Resolution of Financial Crises – the Treatment of the Claims of Private Sector and Paris Club Creditors Preliminary Considerations." Policy Development and Review Department, Washington D.C: International Monetary Fund.
[17] International Monetary Fund (2004). "Are Credit Booms in Emerging Markets a Concern?" In: World Economic Outlook (April), pp. 147-166.
[18] Jensen, N. M. (2004). "Crisis, Conditions, and Capital. The E¤ect of International Monetary Fund Agreements on Foreign Direct Investment Flows."Journal of Con‡ict Resolution, 48(2), pp. 194-210.
[19] Jeanneau, S. and M. Micu (2002). "Determinants of International Bank Lending to Emerging Market Countries." BIS Working Paper No 112.
[20] Marchesi, S. and J.P. Thomas (1999). "IMF Conditionality as a Screening Device." Economic Journal, 109 (454) (March), pp. 111-125.
[21] Mody, A. and A. Rebucci (Eds.) (2006). "IMF-Supported Programs. Recent Sta¤ Research." Washington, D.C.: International Monetary Fund.
[22] Mody, A. and D. Saravia (2006). "Catalysing Capital Flows: Do IMF-Supported Programs Work As Commitment Devices?"Economic Journal, 116 (513), pp. 843-867.
[23] Morris, S. and H.S. Shin (2006). "Catalytic Finance: When Does It Work?" Journal of Inter-national Economics, 70 (1), pp. 161-177.
[24] Oatley, T. and J. Yackee (2004). "American Interests and IMF Lending."International Politics, 41(3), pp. 415-429.
[25] ½Ozler, S. (1993). "Have Commercial Banks Ignored History?" American Economic Review, 83 (3) (June), pp. 608-620.
[26] Penalver, A. (2002). "How Can The IMF Catalyse Private Capital Flows? A Model." Bank of England Working Paper 215.
[27] Reinhart, C. and K. Rogo¤, K. (2004). "The Modern History of Exchange Rate Arrangements: A Reinterpretation." Quarterly Journal of Economics, 119, pp. 1-48.
[28] Stone, R.W. (2008). "The Scope of IMF Conditionality." International Organization, 62 (Fall), pp. 589-620.
[29] Steinwald, M.C. and R.W. Stone (2008). "The International Monetary Fund: A Review of the Recent Evidence."Review of International Organizations, 3, pp. 123-149.
[30] Tirole, J. (2002). "Financial Crises, Liquidity, and the International Monetary System." Prince-ton: Princeton University Press.
[31] World Bank (1997). "Private Capital Flows to Developing Countries: The Road to Financial Integration, Chapter 2, pp. 75-149, Oxford: Oxford University Press.
[32] World Bank (2002). "Global Development Finance. Financing the Poorest Countries." Wash-ington D.C: World Bank.
Figure 1: Private Capital Flows to Countries With an IMF Program, 1984-2004
0
1
2
3
4
5
6
t-3
t-2
t-1
IMF
t+1
Years
% GDP
No debt restructuring
Commercial and/or official debt restructuring
Table 1: IMF programs in Middle Income Countries, 1984-2004
Standby Arrangements Extended Fund Facilities Debt restructuring Debt restructuring All None Commercial O¢ cial All None Commercial O¢ cial
1984 9 5 1 4 0 0 0 0 1985 10 3 4 7 1 0 1 1 1986 7 3 2 2 0 0 0 0 1987 4 1 2 3 0 0 0 0 1988 8 4 1 4 1 1 0 0 1989 5 0 1 5 3 1 0 2 1990 9 3 2 6 0 0 0 0 1991 16 7 1 9 2 1 0 1 1992 12 8 1 4 3 2 0 1 1993 9 8 0 1 2 1 0 1 1994 12 6 3 5 4 2 0 2 1995 19 16 1 3 2 0 1 2 1996 10 10 0 0 6 4 1 2 1997 8 8 0 0 3 2 1 0 1998 6 5 0 1 4 1 3 1 1999 7 4 2 1 4 3 0 1 2000 8 5 2 2 2 0 0 2 2001 7 6 1 0 0 0 0 0 2002 9 8 0 1 0 0 0 0 2003 9 9 0 0 1 1 0 0 2004 6 6 0 0 0 0 0 0 1984 2004 190 125 24 58 38 19 7 16
Note that countries can (and often) have a commercial and a o¢ cial debt restructuring in the same year.
Table 2: Summary Statistics for Variables Used in Regression
Variable Mean Standard Deviation Min Max
Total private capital ‡ows/GDP 5:3 7:4 18:1 52:6
Real GDP growth 4:1 4:1 13:4 22:0
Trade/GDP 83:5 42:0 16:3 220:9
Debt servicing/Exports 18:5 13:8 0:3 117:9
Total external debt (US billions) 24:8 47:7 0:0 247:6
Reserves/Imports (Months) 4:0 3:6 0 27:1
Short term debt/Reserves 7:9 76:6 0 1483:2
Domestic credit/GDP 44:0 46:4 71:5 494:4
Exchange rate volatility 2:2 21:6 0 468:3
Exchange rate regime 0:3 0:5 0 1
IMF program 0:1 0:3 0 1
Precautionary program 0:0 0:2 0 1
IMF loan (percent of quota) 18:3 66:3 0 751:7
US aid/Total aid 18:7 19:6 0:1 84
UN voting in line with the US 0:2 0:1 0 0:4
US bank exposure (US billions) 3:3 8:1 0 77:2
Note: The sample consists of the 633 observations for the period 1984 to 2004 that are used in the regressions in Tables 4-7.
Table 3: Determinants of IMF Programs Without a Debt Restructuring
Dependent Variable SBA or EFF SBA EFF SBA
(1) (2) (3) (4) (5) (6) (7) US aid/Total aid 0:009 0:011 0:010 0:017 0:017 0:017 0:017 (0:005) (0:005) (0:017) (0:006) (0:006) (0:006) (0:006) Real GDP growth 0:004 0:007 0:004 0:001 (0:030) (0:029) (0:032) (0:030) Trade/GDP 0:007 0:010 0:010 0:014 (0:007) (0:008) (0:008) (0:009) Debt servicing/Exports 0:022 0:020 0:020 0:019 (0:008) (0:007) (0:007) (0:006) Change in Debt/GDP 0:023 0:021 0:022 0:021 (0:009) (0:009) (0:009) (0:010) Reserves/Imports 0:015 0:034 0:005 0:021 (0:071) (0:072) (0:082) (0:086)
Short term debt/Reserves 0:010 0:011 0:009 0:009
(0:003) (0:003) (0:003) (0:003)
Domestic credit/GDP 0:001 0:003 0:003 0:005
(0:005) (0:005) (0:005) (0:006)
Change Domestic credit/GDP 0:008 0:009 0:011 0:012 (0:004) (0:005) (0:005) (0:005)
Exchange Rate Regime: 0:786 0:752 0:724 0:691
Pegged (0:393) (0:406) (0:408) (0:424)
Exchange Rate Regime: 0:215 0:134 0:354 0:294
Limited Flexibility (0:380) (0:379) (0:350) (0:350)
UN voting 3:289 3:823
(2:045) (2:441)
US bank exposure 0:003 0:003
(0:019) (0:019)
Country …xed e¤ects Yes Yes Yes Yes Yes Yes Yes
Time …xed e¤ects Yes Yes Yes Yes Yes Yes Yes
Psuedo R-squared 0.13 0.14 0.09 0.35 0.35 0.33 0.33
Number of Observations 772 693 95 684 640 552 514
Note: all conditioning variables are lagged by one year. Standard errors are presented in parentheses. Signi…cance levels: ***: 1%, **: 5%, *: 10%.
Table 4: Catalytic E¤ect of Fund Programs With and Without a Debt Restructuring
Dependent Variable Total private capital ‡ows/GDP
Benchmark model IMF programs not instrumented
IMF program, Instrumented 9:140
(4:351)
IMF program, Instrumented * Debt Restructuring 6:755 (4:227)
IMF program 1:830
(1:491)
IMF program * Debt Restructuring 1:044
(0:795) Real GDP growth 0:161 0:167 (0:074) (0:074) Trade/GDP 0:016 0:018 (0:026) (0:027) Debt servicing/Exports 0:060 0:062 (0:036) (0:035)
Total external debt 0:090 0:081
(0:017) (0:015)
Reserves/Imports (Months) 0:136 0:139
(0:210) (0:211)
Exchange rate volatility 0:024 0:026
(0:005) (0:005)
Short term debt/Reserves 0:000 0:000
(0:001) (0:001)
Domestic credit/GDP 0:019 0:020
(0:016) (0:016)
Change Domestic credit/GDP 0:001 0:002
(0:012) (0:012)
Country …xed e¤ects Yes Yes
Time …xed e¤ects Yes Yes
Psuedo R-squared 0.17 0.16
Number of Observations 633 633
Note: all conditioning variables are lagged by one year. Standard errors are presented in parentheses. Signi…cance levels: ***: 1%, **: 5%, *: 10%.
Table 5: Sensitivity of the Catalytic E¤ect to Changes in the First Stage Regression
Dependent Variable Total private capital ‡ows/GDP
Account for commercial Account for o¢ cial Include Extended restructuring only restructuring only Fund Facilities
(1) (2) (3)
IMF program, Instrumented 8:247 8:323 7:474
(4:671) (3:927) (3:864)
IMF program, Instrumented * Debt Restructuring 5:812 4:931 6:560
(3:333) (5:191) (3:912) Real GDP growth 0:177 0:161 0:179 (0:073) (0:072) (0:075) Trade/GDP 0:013 0:015 0:016 (0:027) (0:025) (0:026) Debt servicing/Exports 0:050 0:049 0:064 (0:041) (0:033) (0:038)
Total external debt 0:087 0:095 0:088
(0:017) (0:017) (0:017)
Reserves/Imports (Months) 0:117 0:121 0:146
(0:212) (0:215) (0:208)
Exchange rate volatility 0:026 0:024 0:025
(0:005) (0:005) (0:005)
Short term debt/Reserves 0:001 0:001 0:000
(0:001) (0:001) (0:001)
Domestic credit/GDP 0:020 0:020 0:020
(0:016) (0:015) (0:016)
Change Domestic credit/GDP 0:001 0:000 0:001
(0:011) (0:012) (0:012)
Country …xed e¤ects Yes Yes Yes
Time …xed e¤ects Yes Yes Yes
Psuedo R-squared 0.17 0.17 0.17
Number of Observations 633 633 633
Note: all conditioning variables are lagged by one year. Standard errors are presented in parentheses. Signi…cance levels: ***: 1%, **: 5%, *: 10%.
Table 6: Sensitivity of the Catalytic E¤ect to Changes in the Second Stage Regression
Dependent Variable Total private capital ‡ows/GDP
Include lagged Include Account for Account for dependent lambda size of precautionary
variable IMF loan programs
(1) (2) (3) (4)
IMF program, Instrumented 8:655 8:915 9:462 7:618
(3:968) (4:542) (4:381) (3:890)
IMF program, Instrumented * Debt Restructuring 6:125 6:807 7:024 6:165 (3:445) (4:197) (4:173) (4:043) Real GDP growth 0:010 0:162 0:155 0:157 (0:059) (0:076) (0:073) (0:075) Trade/GDP 0:007 0:015 0:016 0:015 (0:021) (0:025) (0:026) (0:025) Debt servicing/Exports 0:050 0:057 0:062 0:057 (0:027) (0:039) (0:037) (0:035)
Total external debt 0:078 0:090 0:086 0:087
(0:015) (0:017) (0:017) (0:017)
Reserves/Imports (Months) 0:129 0:136 0:136 0:144
(0:153) (0:211) (0:209) (0:208)
Exchange rate volatility 0:017 0:025 0:023 0:025
(0:006) (0:005) (0:005) (0:005)
Short term debt/Reserves 0:000 0:001 0:001 0:000
(0:001) (0:002) (0:001) (0:001)
Domestic credit/GDP 0:009 0:018 0:019 0:019
(0:014) (0:016) (0:016) (0:016)
Change Domestic credit/GDP 0:002 0:002 0:001 0:001 (0:010) (0:013) (0:012) (0:012)
Lagged dependent variable 0:243
(0:107)
Lambda 0:181
(0:578)
IMF loan (percent of quota) 0:003
(0:002)
Precautionary program 2:911
(2:205)
Country …xed e¤ects Yes Yes Yes Yes
Time …xed e¤ects Yes Yes Yes Yes
Psuedo R-squared 0.22 0.17 0.17 0.18
Number of Observations 633 633 633 633
Note: all conditioning variables are lagged by one year. Standard errors are presented in parentheses. Signi…cance levels: ***: 1%, **: 5%, *: 10%.
Table 6 (continued): Sensitivity of the Catalytic E¤ect to Changes in the Second Stage Regression
Dependent Variable Total private capital ‡ows/GDP
Account for Account for Include 1990-2004 current and past exchange rate controls
IMF programs regime (1) to (6)
(5) (6) (7) (8)
IMF program, Instrumented 9:625 9:399 8:422 8:602
(4:085) (4:480) (3:791) (4:270)
IMF program, Instrumented * Debt Restructuring 7:434 6:562 6:114 2:704 (4:933) (4:351) (3:848) (4:991) Real GDP growth 0:160 0:176 0:106 0:105 (0:074) (0:068) (0:056) (0:094) Trade/GDP 0:018 0:016 0:008 0:009 (0:025) (0:026) (0:021) (0:042) Debt servicing/Exports 0:062 0:055 0:050 0:067 (0:035) (0:036) (0:032) (0:047)
Total external debt 0:089 0:097 0:077 0:110
(0:018) (0:021) (0:016) (0:018)
Reserves/Imports (Months) 0:157 0:161 0:180 0:212
(0:205) (0:207) (0:143) (0:282)
Exchange rate volatility 0:024 0:027 0:017 0:026
(0:001) (0:005) (0:006) (0:005)
Short term debt/Reserves 0:001 0:000 0:000 0:002
(0:001) (0:001) (0:002) (0:001)
Domestic credit/GDP 0:019 0:019 0:009 0:005
(0:015) (0:016) (0:015) (0:032)
Change Domestic credit/GDP 0:001 0:000 0:002 0:008
(0:011) (0:012) (0:011) (0:020)
Lagged dependent variable 0:242
(0:101)
Lambda 0:012
(0:484)
IMF loan (percent of quota) 0:004
(0:003) Precautionary program 3:127 (1:978) IMF program at t 1:406 1:637 (0:756) (0:689) IMF program at t-2 0:441 0:098 (0:874) (0:633)
Exchange Rate Regime: 0:595 0:928
Pegged (1:843) (1:518)
Exchange Rate Regime: 1:412 1:717
Limited Flexibility (0:965) (0:931)
Country …xed e¤ects Yes Yes Yes Yes
Time …xed e¤ects Yes Yes Yes Yes
Psuedo R-squared 0.18 0.18 0.24 0.13
Table 7: The Scope of Fund Catalysis in Solvent Countries
Dependent Variable Total private capital ‡ows/GDP
Debt/GDP Reserves/Imports (months)
Drop observations with.. <40% 40-60% >60% >3 1.25-3 <1.25
(Good) (Intermediate) (Bad) (Good) (Intermediate) (Bad)
(1) (2) (3) (4) (5) (6)
IMF program, Instrumented 15:983 13:146 0:110 11:934 7:767 9:326 (4:972) (5:658) (3:123) (6:401) (4:442) (4:440)
IMF program, Instrumented 8:045 7:741 5:887 10:324 2:583 6:622
* Debt Restruncturing (4:540) (4:667) (5:754) (6:488) (3:399) (5:113) Real GDP growth 0:121 0:135 0:112 0:198 0:138 0:166 (0:089) (0:089) (0:062) (0:109) (0:085) (0:090) Trade/GDP 0:055 0:014 0:008 0:022 0:007 0:008 (0:025) (0:033) (0:033) (0:025) (0:032) (0:028) Debt servicing/Exports 0:027 0:060 0:011 0:079 0:036 0:063 (0:033) (0:042) (0:047) (0:048) (0:046) (0:042)
Total external debt 0:123 0:122 0:018 0:049 0:085 0:087 (0:024) (0:021) (0:020) (0:017) (0:014) (0:020)
Reserves/Imports (Months) 0:282 0:129 0:288 0:538 0:009 0:203 (0:539) (0:203) (0:102) (0:423) (0:230) (0:201)
Exchange rate volatility 0:022 0:019 0:056 0:075 0:026 0:024 (0:006) (0:005) (0:051) (0:040) (0:006) (0:005)
Short term debt/Reserves 0:004 0:000 0:081 0:000 0:000 0:001 (0:002) (0:001) (0:078) (0:002) (0:002) (0:087)
Domestic credit/GDP 0:021 0:022 0:037 0:028 0:017 0:004 (0:017) (0:014) (0:021) (0:026) (0:018) (0:024)
Change Domestic credit/GDP 0:004 0:007 0:024 0:000 0:001 0:004 (0:012) (0:014) (0:015) (0:016) (0:014) (0:015)
Country …xed e¤ects Yes Yes Yes Yes Yes Yes
Time …xed e¤ects Yes Yes Yes Yes Yes Yes
Psuedo R-squared 0.27 0.18 0.12 0.26 0.21 0.16
Number of Observations 398 474 394 296 422 548
Note: all conditioning variables are lagged by one year. Standard errors are presented in parentheses. Signi…cance levels: ***: 1%, **: 5%, *: 10%.
Table A.1: Country List Albania Honduras Algeria Indonesia Angola Jamaica Argentina Kazachstan Armenia Lesotho Azerbaijan Macedonia Belize Malaysia Bolivia Mauritius Botswana Mexico Brazil Moldova Cameroon Nicaragua
Cape Verde Panama
Chile Papua New Guinea
China Paraguay
Colombia Seychelles
Costa Rica South Africa
Croatia St. Kitts and Nevis
Dominica St. Lucia
Dominican Republic St. Vincent and Grenadines
Egypt Swaziland El Salvador Thailand Georgia Tonga Grenada Turkey Guatemala Uruguay Guyana
Table A.2: Sources of Data
Variable Source
Total private capital ‡ows/GDP Calculated using data from Global Development Finance, WB
Real GDP growth World Economic Outlook, IMF
Trade/GDP World Development Indicators, WB
Total debt service/Exports Global Development Finance, WB Total external debt (US billions) Global Development Finance, WB Reserves/Imports (Months) International Financial Statistics, IMF Short term debt/Reserves Calculated using data from IFS and GDF
Domestic credit/GDP Calculated using data from IFS
Exchange rate volatility Calculated using data from IFS
Exchange rate regime From Reinhart and Rogo¤ (2004)
IMF program Data from the Policy Development and Review Department of the IMF
Precautionary program Data from the Policy Development and Review Department of the IMF
IMF loan (percent of quota) Data from the Policy Development and Review Department of the IMF
US aid/Total aid OECD DAC database
UN voting Provided by Axel Dreher
Previous DNB Working Papers in 2010
No. 242 Leo de Haan and Jan Kakes, Momentum or Contrarian Investment Strategies: Evidence from Dutch institutional investors
No. 243 Ron Berndsen, Toward a Uniform Functional Model of Payment and Securities Settlement Systems
DNB W
ORKING
P
APER
D N B W o r k i n g P a p e r
No. 35/April 2005Jan Kakes and Cees Ullersma
