Journal of International Money and Finance 21 (2002) 565–587 www.elsevier.com/locate/econbase
Effectiveness of IMF-supported stabilization programs in developing countries Ays¸e Y. Evrensel Department of Economics, Portland State University, Portland, OR 97207-0751, USA
Abstract This paper examines the effectiveness of Fund-supported stabilization programs by investigating whether the IMF achieves its own objectives in such programs. Even though the Fund’s conditionality prescribes fiscal and monetary discipline in program countries, the results of the empirical analysis show that the IMF cannot impose its conditionality even during program years. Furthermore, when successive interprogram periods are considered, program countries enter a new program in a worse macroeconomic condition than they entered the previous program. These results and the fact that stabilization programs have a revolving nature are inconsistent with the effectiveness of IMF-supported stabilization programs and may signal the existence of moral hazard. 2002 Elsevier Science Ltd. All rights reserved. JEL classification: E63; F33; F40 Keywords: IMF; Stabilization programs; Conditionality; Moral hazard
1. Introduction In its 58 years of existence, the IMF has been criticized because of its institutional structure and lending practices. Some argue that the IMF is a bureaucratic and nontransparent institution with no accountability for its actions. It has also been suggested that Fund-supported stabilization programs are ineffective and may create moral hazard. The motivation to provide another study on the IMF is based on three points. First, although there are a large number of publications about the IMF and its pro-
E-mail address:
[email protected] (A.Y. Evrensel). 0261-5606/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 2 6 1 - 5 6 0 6 ( 0 2 ) 0 0 0 1 0 - 4
566
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
grams, most publications express opinions about the Fund without providing quantitative evidence. There are several quantitative evaluations of Fund programs, most of which have been provided by the IMF (Reichmann and Stillson, 1978; Donovan, 1982; Loxley, 1984; Goldstein and Montiel, 1986; Khan, 1990; Joyce, 1992; Doroodian, 1993; Conway, 1994; Killick, 1995; Santaella, 1996; Knight and Santaella, 1997).1 Second, the recent efforts to widen the IMF’s responsibilities make a quantitative study on the effectiveness of Fund-supported adjustment programs both timely and appropriate. In 1997, the IMF introduced the Supplemental Reserve Facility that provides large, short-term loans to countries in financial crises, as in the cases of Korea, Russia, and Brazil. Third, motivated by the idea that Fund resources provide ex-post assistance in a crisis, but realizing that they do not reduce the frequency and intensity of financial crises, it has recently been suggested that the IMF should assume the role of an international lender of last resort (Fischer, 1999). However, before defining new responsibilities for the IMF, one should be concerned with the performance of the Fund in its traditional roles. This paper focuses on the effectiveness of Fund-supported stabilization programs for developing countries and has four characteristics. First, this study does not question the IMF’s existence, its rationale, its programs, and the content of conditionality associated with these programs.2 This paper’s approach to program evaluation is to use the IMF’s criteria to see whether the IMF achieves its own goals in these programs. Second, this study uses a broader data set than previous program evaluations in terms of the types of balance of payments programs (four types), the number of program countries (91), and the length of the period under investigation (1971–97). Third, it provides a discussion of alternative evaluation methods and their weaknesses before the selection of the evaluation method. Fourth, the method of program evaluation is based on the observation of relevant variables during pre-program, program, and post-program years. Additionally, this paper attempts to relate program evaluation to moral hazard associated with the Fund’s lending. The organization of the paper is as follows. Section 2 describes this study’s approach to program evaluation, and provides a critique of previous evaluations. Section 3 constitutes the empirical part in which Fund-supported stabilization programs are evaluated using the data on 91 developing countries for the period 1971– 97. Finally, Section 4 summarizes the results of the empirical analysis and discusses the effectiveness of stabilization programs. 2. Approach to program evaluation 2.1. Alternative approaches to program evaluations There are three alternative approaches to the evaluation of adjustment programs. First, the outcome vs. alternative outcome approach compares the actual outcome in 1 2
Ul Haque et al. (1998) provide a review of previous program evaluations. See Vaubel (1991) and Willett (2000) on the political economy of international financial institutions.
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
567
an adjustment program with the outcome in an alternative program that would have achieved a similar degree of adjustment. This approach has been suggested as the closest approach to the ideal evaluation (Edwards, 1989; Krueger, 1998). However, in addition to the difficulty associated with estimating a robust alternative model, it is problematic to provide a meaningful definition of the term “similar degree of adjustment” (Edwards, 1989). Therefore, this approach has not been used in previous evaluations. Second, the outcome vs. counterfactual approach describes the program effect as the difference between the actual performance observed in a program and the performance that would have taken place in the absence of a program. Some evaluation studies use this approach by considering the developing members of the IMF that have not been involved in any IMF program as the control group (Goldstein and Montiel, 1986; Khan, 1990; Santaella, 1996). The third way of evaluating adjustment programs is the outcome vs. target approach that determines whether the program objectives have been achieved. The outcome vs. target approach is difficult to implement, because the IMF does not make the content of individual adjustment programs public.3 However, a generalized version of this approach, the outcome vs. purpose approach, has been used in almost all previous evaluations (Reichmann and Stillson, 1978; Donovan, 1982; Loxley, 1984; Goldstein and Montiel, 1986; Khan, 1990; Santaella, 1996; Joyce, 1992; Doroodian, 1993; Conway, 1994; Knight and Santaella, 1997). The outcome vs. purpose approach is based on the fact that the Fund’s purpose in adjustment programs is to reduce or eliminate balance of payments problems. The Monetary Approach to Balance of Payments (MBOP) that underlines the Fund’s conditionality suggests that the balance of payments problems are caused by incompatible exchange rate, monetary, and fiscal policies in program countries. Therefore, the main purpose of adjustment programs is to induce program countries to control the size of the public sector and to exercise monetary discipline under pegged exchange rate regimes to prevent the depletion of international reserves. 2.2. Critique of previous program evaluations The outcome vs. counterfactual approach (Goldstein and Montiel, 1986; Khan, 1990; Santaella, 1996) has been applied to the program country vs. nonprogram country comparison, where nonprogram countries are believed to represent the control group. The estimation of the counterfactual through the control group approach may be misleading. Viewing the control group as the counterfactual implies that the macroeconomic performance of nonprogram IMF-members is indicative for the 3 This means that the selection of target variables, their values as suggested by conditionality, and any changes in conditionality due to Article IV consultations are not made public. These consultations refer to the periodic meetings between the Fund’s staff and the authorities of member countries that take place in the member country to collect and analyze economic data (IMF, Annual Report, 1992). Since May 1997, the results of Article IV consultations are made public with the permission of the member country (IMF, Annual Report, 1998).
568
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
macroeconomic performance of program countries in the absence of an IMF program. However, being a program or a nonprogram country is a self-selected attribute. If anything, the difference between program and nonprogram countries in their macroeconomic performance will determine which country will become a program country. Therefore, the outcome vs. counterfactual approach will not be used in this study.4 Previous evaluations that use the outcome vs. purpose approach employ different methods to capture the Fund-effect: pooled regressions with a program year dummy, logit or probit models to discriminate between program and nonprogram years, and the before–after method. Pooled regressions with a program year dummy are used to explain the determination of the subaccounts of the balance of payments where the sign and the significance of the program dummy is believed to explain the program effect on balance of payments (Goldstein and Montiel, 1986; Khan, 1990; Doroodian, 1993). There are two problems associated with this approach. First, the fundamental problem is the lack of theoretical background. Dependent and independent variables are interchanged without any consideration of the direction of causality. Second, the meaning of the program year dummy in pooled regressions should be reconsidered. The program year dummy implies that adjustment programs affect the outcome in target variables in a way that is not captured by other independent variables in the model. For example, if being in a program year leads to significantly higher reserves, then this could be interpreted as the catalytic effect of adjustment programs. Such effects imply that program countries may have an easier access to private capital markets, if lenders view the IMF’s involvement in a country as a signal of stability. Therefore, the program dummy in pooled regressions does not reflect whether conditionality is imposed successfully. Third, there is an implied assumption regarding the counterfactual in pooled regressions with a program dummy. It is assumed that nonprogram years represent the counterfactual for program years. However, they do not, because the information regarding the availability of Fund programs is already imbedded in the variables for nonprogram years. Countries may follow riskier macroeconomic policies that will lead to a balance of payments crisis, knowing that Fund support will be available. Also, there may be a feedback from IMF programs on variables in question. Therefore, the sign and significance of the program year dummy in pooled regressions cannot be generalized as the program effect. The discrimination analysis (logit or probit) is also employed to distinguish between program and nonprogram years (Joyce, 1992; Conway, 1994; Knight and
4
Although nonmembers of the IMF may be a proxy for the counterfactual, there are various problems associated with using them as such. Monaco, Liechtenstein, Andorra, and Vatican City do not qualify to be in the sample not only because these countries do not have any control over their monetary policies, but also they are considered developed countries. No data are available for Nauru, Palau, and Tuvalu. Zaire became a nonmember country in 1997, which marks the end of the sample period. Taiwan is left as the only nonmember country that could be used as the counterfactual sample, which is not enough to draw any conclusions.
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
569
Santaella, 1997). The problem with this technique is that it is difficult to find explanatory variables that are not simultaneously determined. That is, it is very difficult to come up with truly exogenous variables with which one can distinguish between program and nonprogram years.5 Additionally, the problem with the counterfactual still exists. In this setting, the program year dummy is the dependent variable; however, the right-hand side variables may not appropriately distinguish between program and nonprogram years because of the above mentioned effects of expectations and feedback. The before–after method of program evaluation has also been used where the difference in the mean values of evaluation variables between program and nonprogram years is interpreted as the program effect (Reichmann and Stillson, 1978; Donovan, 1982; Loxley, 1984; Khan, 1990). As in other types of evaluations, the problem with the counterfactual also exists in the before–after method. Ul Haque et al. (1998) note that the before–after approach implies a constant counterfactual with respect to the policies and the external environment of the program country. It is obvious that estimating the counterfactual represents the main problem in program evaluations. In an attempt to calculate the “true” effects of Fund-supported programs, some studies employ an estimator of the Fund effect that distinguishes between policy and target variables (Goldstein and Montiel, 1986; Khan, 1990).6 The target variables are determined by the vector of policy variables, exogenous shocks, and the program dummy. Similar to pooled regressions, the program dummy is assumed to reflect the Fund effect. These studies recognize that policy variables are not directly observable in program countries. The counterfactual for policy variables is constructed by using the difference between desired and actual (lagged) values of the target variables. This estimation of the Fund effect suffers from the practice of interchanging the dependent and independent variables, because target variables are affected by policy variables and vice versa.7 The discussion of evaluation methods indicates that all types of program evaluations are problematic, and a perfect solution to the problems of program evaluation does not exist. However, the recognition of these problems is important with regard to the selection of the evaluation method and the interpretation of evaluation results. 2.3. This study’s approach to program evaluation Although there are many interesting questions regarding the effectiveness of Fund supported programs, the question to which there is a methodologically correct answer 5
The lack of exogeneity represents a problem also in the simulation study of Khan and Knight (1981) in which a system of equations determines output, prices, reserves, money, and fiscal policy. 6 In Goldstein and Montiel (1986), this is called the modified estimator of the Fund effect. 7 Theoretically, there is a superior way of estimating the counterfactual. Because an IMF member is expected to include the availability of Fund support in her macroeconomic decisions, the data based on membership years should not be used to estimate the counterfactual. Table 1 indicates that some countries became IMF members during the early or mid-1980s. Using the data from nonmember years, the counterfactual may be estimated for the member years. Unfortunately, the pre-membership data have many missing observations regarding late members so that forecasting the counterfactual was not possible.
SB
SB
SB
SB
EFF M
M
SB
SB
SB
SB/ EFF
SB
EFF
SB
SB/ SAF
EFF
SB/ SAF
SB/ SAF
SB/ SAF
SB
1987
1989
SB
SAF
SAF
SB
SB
1991
SB
1992
SAF
SAF
SAF
SB/ EFF SAF/ SAF/ SAF/ ESAF ESAF ESAF SB
SB
M
SB
1990
EFF
SB
1994
SAF
EFF
SB/ EFF
1995
SB/ EFF SAF
EFF
1996
SAF/ SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF
SAF/ SAF ESAF SB
EFF
1993
SB
EFF
1997
EFF
SB
SB/ SAF
SB
EFF/ SB
SAF
SAF
SB
SAF
SB
SB
SAF
SAF
SB
SAF
SB
SAF
SAF
SB
SAF
SAF
SB
SB SAF
SAF
SAF
SB SAF/ ESAF SAF/ SAF/ SAF/ ESAF ESAF ESAF SAF
SAF/ SB
SB/ SAF
SAF/ ESAF SAF/ ESAF ESAF SB
SAF
(Continued overleaf)
SAF/ SAF/ ESAF SB/ ESAF ESAF
SB/ SAF
ESAF ESAF ESAF SB SB
SAF/ SAF/ ESAF ESAF SAF SAF
SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF ESAF ESAF ESAF ESAF ESAF ESAF
SAF
SB
1988
EFF
SB
SB
EFF
SB/ SAF
SB
SB
SB
1986
M
SB
EFF
SB
SB
SB
1985
M
SB
M
SB SB
SB
1984
China
M
SB
EFF/ SB SB
SB
1983
Chile
SB
SB
SB M M
EFF
M
1982
SAF
NB
SB
SB
EFF
1981
SAF
SB
SB
SB
SB/ EFF
1980
SB/ SAF
SB
SB
SB
SB
1979
SB/ SAF
M
M M
M
M
M
SB
1978
Cambodia M Cameroon M Cape Verde Central M African Rep. Chad M
Botswana Brazil Burkino Faso Burundi
Bhutan Bolivia
Barbados Belize Benin
Bangladesh
SB
SB
1977
SB
M
SB
1976
Angola Antigua and Barbuda Argentina
SB
1975
SB
SB
1973 1974
Afghanistan M Algeria M
1971 1972
Table 1 IMF-supported adjustment programs in developing countries, 1971–97a
570 A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
Grenada
Gambia, The Ghana
M
M
El Salvador M Equatorial M Guinea Eritrea Ethiopia M Fiji M Gabon M
SB
SB
SB
SB
SB
SB
M
M SB SB
SB
SB
SB SB
1977
SB/ EFF
1978
SB
1979
SB
SB
SB
SB
SB
SB
EFF
SB
SB
1976
Dominican M Rep. Ecuador M Egypt M
SB
1975
M M
M
M
SB
1973 1974
Cote d’Ivoire Cyprus Djibouti Dominica
Colombia M Comoros Congo M Costa Rica M
1971 1972
Table 1 (Continued)
SB
SB
SB
EFF
SB SB
EFF
SB
SB SB
1980
SB
SB
SB
EFF
SB
SB EFF
SB
EFF
SB/ EFF EFF
1982
SB SB
EFF
EFF
SB
SB/ EFF EFF
1981
EFF
SB
SB
SB
SB
EFF
EFF
SB/ EFF EFF
1983
EFF
SB
SB
SB
EFF/ SB EFF
EFF
EFF
1984
EFF
SB
SB
SB
EFF/ SB SB
SB
SB
SB
1985
EFF
SB/ SAF SB
SB
SB
EFF/ SB SB
SAF
SB
SB SB
1986
SB/ SAF SB/ EFF/ SAF
SB
SB SB
SAF
SB
SB SB
1987
SAF/ ESAF EFF/ SAF/ ESAF
SB
SAF
SB SB
SAF
SB
SB SB
1988
SAF/ ESAF EFF/ SAF/ ESAF
SB
SAF
SB
SAF
SB
SB
1989
SAF/ ESAF EFF/ SAF/ ESAF
SB
SB SAF
SB
SAF
SB
SB SB
1990
SB
SAF
SB SAF
SB SB
SB
SAF
SB
SAF SB SB
1992
SB
SAF
SB/ EFF SB SAF/ ESAF
SB
SAF
SB
SAF
1993
SAF/ SAF SAF ESAF SAF/ SAF/ SAF ESAF ESAF
SB
SB SAF
SB SB
SB
SAF
SB
SAF SB SB
1991
SAF SB SB
1995
SB
SAF
1996
SB
1997
SAF
SAF
SB
SB EFF
SB SAF
SAF
EFF
SAF EFF
SB
SAF/ SAF/ ESAF ESAF ESAF
EFF/ SB SAF
SAF
SB SB SAF/ SAF/ ESAF ESAF
SB EFF
SAF
(Continued on next page)
SB SAF/ ESAF M SAF
SB EFF
SAF
ESAF ESAF ESAF ESAF
SAF SB SB
1994
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587 571
M M
M M
M
Kenya
Kiribati Korea Lao People’s Dem. Rep. Lebanon Lesotho
M
Jordan
SB
SB
M M M M M
India Indonesia Iran Iraq Jamaica
SB
SB
SB
SB
SB
SB
SB
SB
SB
SB
M SB SB
M SB SB
Haiti
SB
SB
1973 1974
M SB SB
SB
Honduras
GuineaBissau Guyana
Guatemala M Guinea M
1971 1972
Table 1 (Continued)
SB
EFF
SB
SB
1975
SB
EFF
SB
SB
1976
EFF
SB
SB
SB
M
1977
EFF/ SB
SB/ EFF
SB/ EFF
SB
1978
SB
SB/ EFF
EFF
SB/ EFF EFF
1979
SB SB
SB
EFF
EFF
EFF
EFF
1980
SB SB
SB
EFF
EFF
EFF
EFF
EFF
SB
1981
SB
SB
EFF
EFF/ SB EFF
SB
EFF
SB SB
1982
SB
SB
EFF
EFF
SB
SB
EFF
SB SB
1983
SB
SB
EFF/ SB
EFF
SB
SB
1984
SB
SB
SB
SB
1985
M SB
SB
SB
SAF
SB
1986
SB
SB
SAF
SB/ SAF SAF
1987
SAF
SB/ SAF
SB
SAF
SB SB/ SAF SAF
1988
SAF
SAF
SB/ SAF/ ESAF
SB
SB
SB/ SAF
SAF
SB SAF
1989
SAF
SAF
1994
1995
1997
SB/ EFF SB
SAF
EFF
EFF
EFF SAF/ ESAF ESAF
EFF
EFF
SAF/ SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF
SB
EFF
SAF SAF SB/ SB/ SAF SAF SB/ ESAF ESAF ESAF ESAF ESAF ESAF SB SB
SAF/ ESAF SAF/ ESAF ESAF ESAF ESAF
1996
SAF/ SAF/ SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF/ SB SB (Continued overleaf)
SAF
SB
SB
SB
SB
SB/ ESAF SAF SAF
SB SAF/ ESAF SAF
1993
SB SAF/ SAF/ ESAF ESAF SAF SAF/ ESAF ESAF ESAF ESAF ESAF
1992
SB SAF/ SAF/ ESAF ESAF SAF SAF
1991
SB/ EFF EFF SAF/ SAF/ SAF/ SAF/ SAF/ SAF ESAF ESAF ESAF ESAF ESAF
SB
SB
SB/ ESAF SB/ SAF SB
SAF
SB SAF
1990
572 A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
M
Malaysia Maldives Mali
SB
SB
SB
SB
SB
Panama M Papua New Guinea
M M
SB
SB
SB
SB
SB
SB M
SB
SB
SB
SB
EFF
SB
EFF
SB
SB
SB
SB SB
1981
EFF
SB
EFF/ SB
SB
SB
SB
SB
SB SB
1982
EFF/ SB
SB EFF
SB
SB/ EFF
SB SB
1983
SB
SB
SB
EFF
SB
SB
SB
EFF
SB
SB
EFF
SB
SB
SB
SB SB
1980
Nigeria Pakistan
SB
SB
SB EFF
SB
SB
1979
SB
SB
SB
SB
SB EFF
SB
M
SB
1978
Nicaragua M Niger M
M
M
SB
EFF
SB
SB SB
1977
SB
SB
SB
SB EFF
SB
EFF
SB SB
1984
SB
SB EFF
SB
SB
EFF
SB SB
1985
SB
SB
SB
SB
SB
1976
Myanmar Namibia Nepal
SB
SB
1975
M
SB
SB
1973 1974
Mozambique
Mauritius M Mexico M Micronesia Mongolia Morocco M
Marshall Islands Mauritania M
M SB SB
M
Malawi
Liberia M Madagascar M
1971 1972
Table 1 (Continued)
SB
SB/ SAF
SB
SB
SB/ SAF SB SB
SB
EFF
SB SB
1986
SB
SB/ SAF SB
SB/ SAF
SAF
SB
SB
SB/ SAF
SB
SB/ SAF
1987
SAF/ ESAF SB SB/ SAF
SAF
SAF
SB
SB
SB/ SAF
SB/ SAF
SB/ SAF SB/ ESAF
1988
1990
1991
1992
1993
1994
1995
1996
1997
SB/ SAF
SAF
SAF/ SAF/ SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF ESAF ESAF M
SAF/ ESAF SB SB/ SAF
SAF
SAF
SB
EFF
EFF M SB SB SB SB
EFF
SB SB M ESAF ESAF ESAF ESAF SB
EFF
SB
SAF/ ESAF SB SB/ SAF
SAF
SAF/ ESAF SB SAF
SB
SB SB
SB SAF/ ESAF SB SB SAF SAF
M SAF
SB
SAF/ SB
SAF/ ESAF SB SAF
SB SB
SB SB
SB SB
SB/ SB/ SB/ EFF/ EFF/ EFF/ ESAF ESAF ESAF
SAF/ SAF ESAF ESAF ESAF ESAF SB/ SAF SAF
(Continued on next page)
SAF/ SB/ EFF/ ESAF SB
SAF/ ESAF ESAF SB/ SAF
SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF ESAF ESAF ESAF ESAF ESAF ESAF
SB
EFF
SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF ESAF ESAF ESAF ESAF ESAF
SB/ SAF
SB/SAF/SAF/ SAF/ SAF/ SAF SAF SAF SAF ESAF ESAF ESAF ESAF SB/ ESAF ESAF ESAF ESAF SB/ SB/ ESAF ESAF ESAF ESAF ESAF
1989
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587 573
M M
SB
EFF
SB M
SB
M
EFF
EFF
SB
SB
EFF/ SB SB
EFF
SB
SB/ EFF
M M
SB
SB
SB
1980
SB
M
Suriname Swaziland M
SB
M SB
M
EFF
SB
1979
Sudan
SB/ EFF
SB
1978
EFF
SB
SB
SB
1977
SB
SB
1976
South M Africa Sri Lanka M
SB
SB
1975
SB
SB
SB
1973 1974
Seychelles Sierra M Leone Solomon Islands Somalia M
Rwanda M St Kitts and Nevis St Lucia St Vincent and the Grenadines Sao Tome and Principe Senegal M
Philippines M
Paraguay Peru
1971 1972
Table 1 (Continued)
EFF
EFF
SB
SB
EFF
SB/ EFF
SB
1981
EFF/ SB
EFF
SB
SB
EFF
SB/ EFF
EFF
1982
SB
SB
SB
SB
EFF
SB/ EFF
SB
EFF
1983
SB
SB
SB
EFF/ SB SB
SB
EFF/ SB SB
1984
SB
SB
SB
SB
M
EFF/ SB SB
1985
SB
SB/ SAF
SB/ SAF
SB
1986
SB/ SAF
SB/ SAF
SB/ SAF
SB
1987
EFF
1989
EFF
1990
SB/ EFF SAF
1991
SB/ EFF SAF
1992
SAF
SB
EFF
1993
SAF
SB/ SAF
SAF
SAF
SB/ SAF
SAF
SAF
SAF
SAF
SAF
SAF
SAF
SAF
SAF
EFF
EFF
1996
EFF
1997
SAF SAF SB/ SAF/ ESAF SAF/ ESAF ESAF
SAF
EFF
EFF
1995
SAF
SAF
SAF
SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF
SAF SB/ SAF/ ESAF
SAF
EFF
EFF
1994
(Continued overleaf)
SAF/ SAF/ SAF/ SAF/ SAF/ SAF ESAF ESAF ESAF ESAF ESAF
SAF
SAF
SAF SAF SAF SAF SAF SB/ SAF/ SAF/ SAF/ SAF/ SAF SAF/ ESAF ESAF ESAF ESAF ESAF
SB
1988
574 A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
M M
M
M M
M
Western Samoa Yemen Zaire
Zambia
SB
SB
SB
1973 1974
SB
SB
SB
1975
SB
SB
SB
SB
SB
1976
SB
SB
SB
SB
1977
SB
SB
SB
SB
SB
SB
1978
SB
SB
SB
SB
SB
SB SB
1979
SB
SB
SB
SB
SB SB
SB
SB
1980
EFF
EFF
SB M
SB SB
SB SB
SB
1982
M SB SB
SB/ EFF EFF
SB
SB SB
SB SB
SB
1981
EFF/ SB EFF/ SB SB
SB
SB
SB SB
SB SB
1983
EFF/ SB EFF/ SB SB
SB
SB
SB SB
SB
1984
SB
SB
SB
SB
SB
SB SB
1985
SB
SB
SB
SB
M
SB SB
SB
1986
SB/ SAF SB
SB
SAF
SB
SB/ SAF SB SB
1987
SB/ SAF SB
SAF
SB/ EFF
SB/ SAF
SB/ SAF
1988
SAF
1990
1992
1993
1994
1995
SAF/ SAF/ SAF/ SAF/ SAF ESAF ESAF ESAF ESAF
1991
SAF
1996
1997
EFF
SB EFF
SB EFF
SB/ SAF
EFF
SB/ SAF
EFF
SAF
EFF
SAF
EFF SB
SAF
SAF
SB SAF
SB NM
SB/ ESAF ESAF ESAF ESAF
SB SAF/ ESAF ESAF SB SB
SAF/ SAF/ ESAF ESAF ESAF EFF/ EFF/ EFF/ EFF/ ESAF ESAF ESAF ESAF
SAF
EFF
SB SB SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ ESAF ESAF ESAF ESAF ESAF ESAF ESAF SB SB SB SB
EFF
SB
SB/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ SAF/ ESAF SAF/ ESAF ESAF ESAF ESAF ESAF ESAF ESAF ESAF
SAF
1989
EFF, Extended Fund Facility; ESAF, Extended Structural Adjustment Facility; SAF, Structural Adjustment Facility; SB, Standby. a This table is constructed based on the information provided in various issues of the Annual Report published by the IMF. M indicates the year when a country became an IMF member. NM indicates non-member, and it is used for Zaire which discontinued its IMF membership in 1997.
Zimbabwe
SB
M M SB SB
Uruguay M Vanuatu Venezuela M Vietnam M
Turkey Uganda
Tonga Trinidad M and Tobago Tunisia M
Thailand Togo
Syrian M Arab Rep. Tanzania M
1971 1972
Table 1 (Continued)
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587 575
576
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
is the one regarding the conditional effect of a program. For example, the Fund views an unsustainable increase in domestic credit (unsustainable at the given exchange rate) as the main cause of balance of payments problems. Therefore, conditionality associated with adjustment programs prescribes a reduction in domestic credit creation. One can ask whether domestic credit creation differs significantly between the pre-program, program and post-program periods. In this paper, the premise of program evaluation is what the Fund expects program countries to do and whether these objectives are achieved. The Fund expects program countries to reduce their domestic credit creation, budget deficit, domestic borrowing, inflation rate, current account and capital account deficit. The relevant question is whether we observe significant improvement in these variables under an IMF program. At this point, the following trio of critique appears immediately. First, the IMF does not have any authority over sovereign countries, which means that program countries may not follow the IMF’s advice. Second, even if program countries try to follow the IMF’s advice, they may face exogenous shocks that prevent countries from improving upon evaluation variables. Third, program countries may have scored worse without an IMF program. With regard to the first point, there is a valid reason for conducting program evaluations to measure the IMF’s success, and not program countries’ willingness to improve their macroeconomic performance. Since the IMF provides program countries with subsidized loans justified by the existence of conditionality, the Fund is expected to demonstrate its ability to impose the content of conditionality. Second, if program countries were struck by exogenous, adverse shocks in a systematic fashion during a period of, say, 30 years, program countries would be considered as victims of economic disasters. Then the IMF would become the financial Red Cross and provide disaster relief without imposing conditionality. Third, the fact that program countries may have been worse off without the IMF’s support is not the only possible outcome. Suppose the inflation rate in a country before and during an IMF program was 40 and 60%, respectively. Although one may interpret this as a sign of IMF’s ineffectiveness in reducing the inflation rate, it is possible that the inflation rate could have been 80% in the absence of an IMF program. However, it is also possible that, in the absence of the IMF support, countries may have decided to follow sounder macroeconomic policies. Again, the point is that the IMF’s conditionality is supposed to increase the shadow price of the IMF support to program countries. The question is whether it does so. This does not mean that the counterfactual is irrelevant. However, the actual data should not be dismissed too quickly, because it shows the extent to which conditionality is enforced, everything else remaining the same. In this paper, Fund-supported programs are evaluated based on the outcome vs. purpose approach using the before–after method. As opposed to previous before– after evaluations that consider one-year lags before and after a program, this study uses lags of up to three years to observe changes in the evaluation variables from three years before the start of a program to three years after the end of a program. This method demonstrates how evaluation variables gradually change toward a pro-
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
577
gram and after a program. To support the results of the before–after analysis, the temporal interprogram analysis is used to illustrate the possibility of moral hazard associated with Fund programs.
3. Empirical analysis 3.1. Participation in IMF programs, 1971–978 In 1997, 181 out of 191 independent countries in the world were IMF-members.9 Table 1 contains 118 developing members of the IMF during the period 1971–97, and the types of adjustment programs they have received.10 During the sample period, 91 countries received any of the four structural adjustment programs: standby, Extended Fund Facility (EFF), Structural Adjustment Facility (SAF), and Enhanced Structural Adjustment Facility (ESAF).11 Table 1 indicates an increase in the frequency of IMF-supported stabilization programs since the mid-1980s, which can be explained by the introduction of SAF and ESAF in 1986 and 1987, respectively. Prior to the mid-1980s, only standby and EFF agreements were available. With the introduction of SAF and ESAF, most low-income developing countries have received different types of programs simultaneously.12 A summary of Table 1 in the form of chi-squared tables indicates that, during the period 1971–97, the probability of any developing IMF member receiving a structural
8 The choice of the period rests on the availability of data. Data on macroeconomic variables are obtained from the IMF’s International Financial Statistics of March 1997 on CD-ROM. Information regarding the type of programs comes from Annual Reports of the IMF for the years 1971 through 1997. 9 Nonmembers of the IMF include micro-states in Europe (Andorra, Liechtenstein, Vatican City, and Monaco), the Pacific (Tuvalu, Palau, and Nauru), Zaire, Cuba, and Taiwan. The remaining 63 nondeveloping members of the IMF include the former Soviet Union and the countries of Europe, Eastern Europe, North America, the Pacific, and the Middle East (oil exporters). Twenty-seven developing members of the IMF have never received stabilization programs and constitute the so-called nonprogram IMF members. 10 In Table 1, the duration of each program is not marked to keep the table simple, which means that tabulated standby, EFF, SAF, and ESAF arrangements correspond to different arrangements that last two to four years. Typically, there is a relatively short period between successive programs that extends from a couple of weeks to a couple of months. 11 While standby arrangements provide balance of payments support to middle- and high-income developing countries, EFF, SAF, and ESAF are designed for low-income developing countries. Standbys are provided for a year with a possible extension up to three years. Other programs last longer and imply longer periods of repayment (5–10 years). See Johnson (1993), Guitian (1995), and Schadler et al. (1995) for more information on these programs. 12 During the pre-1986 period, some countries received standby and EFF simultaneously. During the period 1986–97, the combination was changed to SAF and ESAF, and was primarily given to lowincome countries.
578
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
adjustment program is 0.3242.13 If 1986 is taken as a benchmark to divide the period in two sections, it becomes apparent that the probability of being in a program is higher during the period 1986–97. While the probability of receiving any program is 0.2192 for the pre-1986 period, it increases to 0.5054 during the period 1986–97. For low-income countries, the likelihood of being in a program increased from 0.216 to 0.6235 between the two periods. Similarly, middle-income countries were twice as likely to be in a program during the period 1986–97 (0.432) than during the pre1986 period (0.2177). Although the Fund’s Articles of Agreement state the temporary nature of the Fund’s financial support, the data on program status indicate that many program countries have been under the IMF’s care almost continuously. This raises the question as to why most of the program countries are repeat offenders, i.e., they keep getting into balance of payments problems and receiving financial support from the IMF. Provided that conditionality contains the correct description of and solution to balance of payments problems of program countries, and that conditionality is fully implemented during the program period and sustained by program countries’ governments during the post-program periods, it is plausible that continuing IMF support should be rare or nonexistent. In the following, using the before–after and temporal interprogram analysis, the revolving nature of program participation is explained. 3.2. Before–after analysis The almost continuous nature of Fund-supported programs creates a problem for the comparison of pre-program, program, and post-program years in terms of relevant variables. For the majority of low-income countries in the sample, as one program ends, another one starts in the same year. Additionally, since the mid-1980s most program countries have been involved in more than one program in a given year, which makes pre-program, program, and post-program comparisons noisy. If only one type of program is considered, the before–after comparison may be affected by another type of IMF-program. If all programs are considered, a possible difference among programs may be overlooked. To reduce the noise, mean values of the evaluation variables in different periods are compared not only for all programs but also for different program types (standby, EFF, SAF, and ESAF). Since program types and income levels of program countries are closely related, by considering the behavior of evaluation variables for all programs and under different programs, some of the noise associated with the aggregation of the Fund-effect under different (and sometimes overlapping) programs may be reduced.
13 Chi-squared tables are constructed using program types and income levels of countries. Income categories are defined based on the following levels of 1995 GDP per capita (Y): low-income country if YⱕUS$1500; middle-income country if YⱕUS$6000; high-income country if Y⬎US$6000. The Pearson c2 test on the independence of rows and columns indicates that program types and income levels are not independent, which is consistent with the IMF’s attempt to provide different programs to countries with different income levels.
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
579
Table 2 shows the results of the t-test that provides a comparison of evaluation variables in successive periods. Pre-program periods extend from three years to one year before a stabilization program starts (t⫺3, t⫺2, and t⫺1). Post-program periods extend from one year to three years after a program ends (t+1, t+2, and t+3). Using lags of three years allows the observation of changes in evaluation variables when a country is nearing a program and when it completes a program. Table 2 indicates that when all programs are considered, program countries’ performance worsens significantly toward a program in current account and overall balance of payment. Significantly smaller reserves and slower real growth are also observed. During program years, significant improvements are achieved in the areas of current account and reserves. While current account deficit is lower, reserves are higher during the program years. With respect to reserves, however, the definition of reserves makes a difference. If reserves are defined net of IMF-credit, there is no significant improvement in reserves due to a Fund program. Significantly smaller domestic borrowing and larger foreign debt are also observed. During post-program years, significant improvements occur in the immediate period after the end of a program (t+1) in financial account and overall balance of payments. However, in the latest post-program year (t+3), money supply increases significantly. When only standby arrangements are considered, a significant worsening in the current account, greater depreciation in the exchange rate, a lower real growth are observed during pre-program years. During a standby program, significant improvements in current account, financial account, and reserves are accomplished.14 Immediately after the end of a standby (t+1), significant improvements are observed in overall balance of payments. Considering only EFF arrangements, none of the mean differences among various periods in evaluation variables is significant during pre-program, program, and postprogram periods. When only SAF and ESAF are considered, mean values of evaluation variables do not differ significantly among pre-program periods. Program years are associated with significant improvements in financial account, reserves, and budget deficit. Although program years imply a significant increase in domestic credit, immediately after a SAF or ESAF (t+1), domestic credit becomes significantly smaller. In the later period (t+2), reserves decline significantly. The results of the before–after analysis can be generalized as follows. As countries approach a stabilization program, current account deficit and reserves deteriorate. This situation represents the very reason why a country asks for a Fund-stabilization program in the first place. The evidence is such that especially standby arrangements provide a balance of payments relief during the program period. The results summarized in Table 2 raise questions regarding the implementation of conditionality during a program. The Fund believes that the lack of monetary discipline under a pegged exchange rate regime is the source of the balance of pay-
14
This is the most common result of before–after evaluations. Previous before–after evaluations include Reichmann and Stillson (1978), Donovan (1982), Loxley (1984) and Khan (1990). Of course, these studies differ in terms of sample periods, program types, countries, and evaluation variables.
X¯ t⫺2⫺X¯ t⫺3b
X¯ progy⫺X¯ t⫺1c
Larger surplus Larger surplus Smaller surplus Larger deficit∗∗ Larger deficit∗∗ Larger deficit Smaller deficit Smaller∗∗
Larger surplus
Smaller surplus Smaller surplus
Smaller deficit Smaller deficit Larger deficit Larger deficit
Smaller
Smaller
Larger
SB
EFF SAF/ ESAF Overall BOP All programs SB EFF SAF/ ESAF Reservesf All programs
SB
EFF
Larger
Smaller
Larger surplus
Larger surplus∗∗ Smaller surplus
Smaller deficit Larger deficit Larger deficit
Smaller deficit
X¯ t ⫹ 1⫺X¯ progy
Larger∗∗∗ (larger) Larger∗∗ (larger) Larger (larger)
Smaller deficit Larger deficit Larger deficit Smaller deficit
Larger deficit Smaller deficit Larger deficit Smaller deficit
Deficit Larger surplus
Smaller surplus
Smaller surplus
Smaller deficit Smaller deficit Smaller deficit
Smaller deficit
X¯ t ⫹ 2⫺X¯ t ⫹ 1
Smaller Larger (smaller) Smaller
Larger (larger)
Smaller (smaller) Smaller
Smaller deficit∗ Smaller deficit∗ Smaller deficit Larger deficit∗
Smaller surplus∗∗ Smaller surplus Larger surplus Larger surplus∗ Smaller surplus
Smaller surplus
Larger deficit∗∗ Smaller deficit∗∗ Larger deficit∗ Smaller deficit∗ Larger deficit Smaller deficit Smaller deficit Larger deficit∗
X¯ t⫺1⫺X¯ t⫺2
SB Larger deficit EFF Smaller deficit SAF/ ESAFe Smaller deficit Financial account All programs Larger surplus
Current account All programs Smaller deficit
Evaluation variablesa
Smaller
Larger
Larger
0.0186
0.0859
0.0398 0.011 0.7598 0.9258
0.1002 0.7235
0.2
0.0753
0.3806 0.6414 0.7836
0.1801
Prob⬎Fd
0.9359 (Continued overleaf)
Smaller deficit Larger deficit Smaller deficit Smaller deficit
Surplus Larger surplus
Smaller surplus
Larger surplus
Larger deficit Smaller deficit Larger deficit
Smaller deficit
X¯ t ⫹ 3⫺X¯ t ⫹ 2
Table 2 Summary results regarding mean differences of evaluation variables in pre-program, program, and post-program yearsi
580 A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
Larger∗ Larger∗∗ Larger Larger Larger Larger Smaller Smaller Larger Larger Smaller
Higher Higher Lower Lower Smaller
Larger Larger Larger Larger
Smaller Smaller Smaller Smaller
Lower Lower Lower Lower
Larger
Larger
X¯ t⫺1⫺X¯ t⫺2
Smaller Smaller Smaller Larger
Smaller
SAF/ ESAF
Exchange rateg All programs SB EFF SAF/ ESAF Domestic credit All programs SB EFF SAF/ ESAF Money supply All programs SB EFF SAF/ ESAF Inflation rate All programs SB EFF SAF/ ESAF Budget deficit All programs
X¯ t⫺2⫺X¯ t⫺3b
Evaluation variablesa
Table 2 (Continued)
Smaller
Lower Lower Lower Higher
Smaller Smaller Larger Smaller
Larger∗∗∗ Larger Larger Larger∗∗∗
Larger Larger Smaller Smaller
Larger∗∗∗ (larger)
X¯ progy⫺X¯ t⫺1c
X¯ t ⫹ 2⫺X¯ t ⫹ 1
Smaller
Lower∗∗∗ Lower∗∗ Lower Lower
Larger Larger Smaller Smaller
Smaller∗∗∗ Larger Smaller Smaller∗∗∗
Larger Smaller Larger Larger
Smaller
Lower Lower Higher Lower
Smaller Smaller Larger Smaller
Larger Smaller Larger Smaller
Smaller Larger Larger Smaller
Smaller (smaller) Smaller∗
X¯ t ⫹ 1⫺X¯ progy
Larger
Higher Higher Higher Lower
0.002 0.5284 0.2013 0.0609
0.0052 0.4427 0.2795 0.2749
0.3777 0.4441 0.8629 0.4325
0.6685 0.8695 0.2545 0.1444
0.0015
Prob⬎Fd
0.9211 (Continued on next page)
Larger∗ Larger Larger Larger
Larger Smaller Smaller Larger
Larger Larger Smaller Larger
Larger
X¯ t ⫹ 3⫺X¯ t ⫹ 2
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587 581
X¯ t⫺2⫺X¯ t⫺3b
SB Larger EFF Smaller SF/ ESAF Smaller Net domestic borrowing All programs Smaller SB Smaller EFF Smaller SAF/ ESAf Smaller Net foreign borrowing All programs Smaller SB Smaller EFF Smaller SAF/ ESAF Larger Domestic debt All programs Smaller SB Larger EFF Larger SAF/ ESAF Smaller Foreign debt All programs Larger SB Smaller EFF Smaller SAF/ ESAF Larger Unemployment All programs Lower SB Lower EFF Lower SAF/ ESAF Lower
Evaluation variablesa
Table 2 (Continued)
Smaller Larger Smaller∗ Smaller∗ Smaller∗ Larger Smaller∗∗ Larger Larger∗∗ Larger Larger∗∗∗ Larger∗ Smaller Larger Smaller Larger∗ Larger∗ Larger Larger Lower Higher Higher Higher
Smaller Larger Smaller Smaller Smaller Smaller Smaller Smaller Smaller Smaller Smaller Smaller Larger Larger Smaller Larger Lower Higher Lower Lower
X¯ progy⫺X¯ t⫺1c
Smaller Smaller Smaller
X¯ t⫺1⫺X¯ t⫺2
Lower Lower Higher –h
Smaller∗ Larger Larger Smaller
Larger Larger Larger Smaller∗∗
Smaller∗ Smaller Larger Smaller
Larger Larger Larger Larger
Smaller Larger Smaller
X¯ t ⫹ 1⫺X¯ progy
Lower Lower Higher –h
Larger Smaller Smaller Smaller
Larger Smaller Larger Larger
Larger Smaller Smaller Smaller
Smaller Smaller Smaller Larger
Smaller Smaller Smaller
X¯ t ⫹ 2⫺X¯ t ⫹ 1
0.0575 0.2044 0.9771 0.4752
0.9841 0.9661 0.9724 0.0940
0.129 0.8018 0.8644 0.0490
0.2615 0.4468 0.5525 0.0132
0.1556 0.8581 0.0047
Prob⬎Fd
Lower 0.4059 Lowerv 0.9483 Higher 0.7143 –h 0.5906 (Continued on next page)
Smaller Smaller Smaller Smaller
Smaller Larger Smaller Larger
Smaller Smaller Larger Smaller
Larger Larger Larger Larger
Larger Larger Larger
X¯ t ⫹ 3⫺X¯ t ⫹ 2
582 A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
Lower∗ Lower∗ Lower Lower
X¯ t⫺2⫺X¯ t⫺3b
Lower Lower∗ Higher Lower
X¯ t⫺1⫺X¯ t⫺2
Higher Higher Lower Higher∗∗∗
X¯ progy⫺X¯ t⫺1c
Higher Higher∗ Higher Decline
X¯ t ⫹ 1⫺X¯ progy
Lower Lower Lower Higher
X¯ t ⫹ 2⫺X¯ t ⫹ 1
Higher Higher Higher Higher
X¯ t ⫹ 3⫺X¯ t ⫹ 2
0.295 0.0395 0.3324 0.0610
Prob⬎Fd
b
Evaluation variables are expressed as the proportion of GDP except for the exchange rate, money supply, inflation rate, and real growth. X refers to the variable under consideration. The t-test for the equality of the two population means is conducted for unpaired samples with unequal variances. c Xprogy indicates the program year average of the variable in question. d F statistic is based on the ANOVA where the equality of the mean differences among all periods is simultaneously tested. e Since SAF and ESAF are practically the same program, they are put in the same category. f Entries in parentheses indicate the changes in reserves net-of-IMF credit. g Adjectives indicate the extent of depreciation of the exchange rate. h No test results are available due to insufficient observations. As the table indicates, pre-program and post-program periods cannot be defined for some of the low-income countries because they have been in SAF or ESAF or both almost continuously since the mid-1980s. Additionally, the unemployment rate is calculated quite differently among countries. i ∗, ∗∗, ∗∗∗ denote 10, 5, and 1% level of significance, respectively.
a
Real growth All programs SB EFF SAF/ ESAF
Evaluation variablesa
Table 2 (Continued)
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587 583
584
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
ments problems. Consequently, the conditionality attached to stabilization programs advises a reduction in the size of the public sector and money creation along with the depreciation of the currency. At least during the program years, one would expect a significantly lower budget deficit, domestic creation, money supply, domestic borrowing, and inflation rate. However, these primary targets of stabilization programs are not significantly affected during the program period. When looking at the postprogram performance of program countries, the improvements in balance of payments and reserves achieved during an IMF program disappear. This raises questions with respect to the sustainability of the balance of payments improvement in the post-program period. The results of the before–after analysis should be considered in connection with Table 1, which indicates an almost continuous IMF-support in program countries. The revolving nature of the IMF support may be due to the short-term nature of its effects. The results indicate that the Fund provides program countries with hard currency, and eliminates the balance of payments crisis during the program period. However, in the absence of long-term incentives, the short-term improvement in balance of payments does not last, and it is even reversed once the program is over.15 3.3. Temporal interprogram analysis The issue of moral hazard regarding Fund-supported programs implies the possibility that the governments of program countries may adopt unsustainable macroeconomic policies due to the availability of the Fund credit. If stabilization programs created moral hazard, this would be inconsistent with the effectiveness of stabilization programs. Moreover, one would expect that the interprogram periods would be associated with increasingly unsustainable macroeconomic policies as the number of programs a country receives increases.16 If a country has had Fund support before, the cost of macroeconomic policies that lead to the depletion of international reserves may be lower to the country. The identification of interprogram periods is based on Table 1. The identification process is quite problematic, because the continuous nature of the Fund makes it difficult to identify two interprogram periods. It turns out that there are only 42 countries for which two interprogram periods can be identified during the period 1971–97.17 Table 3 shows the results of the temporal interprogram analysis. The results suggest that the second interprogram period is associated with significantly worsening macroeconomic performance than the first interprogram period. Compared
15 This point that the short-run nature of program effects is responsible for the ineffectiveness of IMF programs is also suggested by Edwards (1989). 16 Interprogram periods are defined as periods that are not associated with any of the four IMF programs and are located between two program periods. 17 It is important to realize that the temporal interprogram analysis requires the actual data. As discussed earlier, the use of the actual data in program evaluations is problematic, because the actual data incorporate program countries’ expectations regarding the availability of Fund programs. However, the actual data are exactly what is needed here.
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
585
Table 3 Temporal interprogram analysis Policy variablesa
Second vs. first inter-program period Difference
Reserves Domestic credit Inflation rate Budget deficit Net domestic borrowing Net foreign borrowing Net domestic debt Net foreign debt a b
⫺0.0116 0.1639 0.1192 0.0126 0.0449 0.0078 0.1922 0.2757
t-value
Change with respect to the earlier periodb
⫺2.7245 4.1875 3.5838 1.7148 2.5896 2.8405 2.683 3.3762
Smaller∗∗∗ Larger∗∗∗ Higher∗∗∗ Larger∗ Larger∗∗ Smaller∗∗∗ Larger∗∗∗ Larger∗∗∗
Policy variables are expressed as percentage of GDP except for the inflation rate. ∗, ∗∗, ∗∗∗ denote 10, 5, and 1% level of significance, respectively.
to the first interprogram period, the second interprogram period is associated with significantly smaller reserves, higher domestic credit creation, inflation rate, budget deficit, domestic borrowing, domestic and foreign debt. This means that, as a country goes in and out of IMF programs, on average, the country enters a new IMF program with worse macroeconomic conditions than when it entered the earlier program. The results of the temporal interprogram analysis should be interpreted in a strict all-else-equal sense. As in the before–after evaluation, this analysis also assumes away any external shocks that may have created the same results of program countries entering new programs in increasingly worse macroeconomic conditions. Moreover, the counterfactual remains to be a problem. However, as suggested by Tables 1 and 2, the Fund support tends to be revolving and of short-term nature with respect to its balance of payments relief. The fact that stabilization programs on average do not change the macroeconomic fundamentals of program countries and that the availability of these programs may imply moral hazard by motivating countries to follow riskier macroeconomic policies raises questions regarding the effectiveness of Fund-supported stabilization programs.
4. Conclusion During the period 1971–97, developing members of the IMF received a stabilization program for more than one-third of the time. There has been an increase in the frequency of IMF-supported stabilization programs since the introduction of SAF and ESAF in 1986 and 1987, respectively. In the post-1986 period, countries received support from the IMF almost half of the time. The results of the before–after analysis indicate that stabilization programs significantly improve current account and reserves during the program years. Although
586
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
stabilization programs seem to provide short-term balance of payments relief, these improvements are not sustained during the post-program period. Additionally, conditionality may not be effectively imposed even during the program years. Although conditionality prescribes a reduction in the size of the public sector, variables such as domestic credit creation and budget deficit are not significantly affected by conditionality during the program years. The results of the temporal interprogram analysis suggest that, on average, program countries enter a new program in a worse macroeconomic situation than before. Considering the revolving nature of the Fund support, this result is inconsistent with the effectiveness of stabilization programs and may be interpreted as a signal of moral hazard. The currency crises of the last few years have intensified discussions regarding the role of the IMF in the international financial system where the emphasis has been on the future role of the Fund. More recently, some would like to give the IMF the responsibility of becoming the international lender of the last resort. Suggestions regarding the Fund’s future will be misguided if its past performance is ignored.
Acknowledgements I am grateful to Gerald P. Dwyer Jr for his comments on the paper. Comments received from S. Nuri Erbas¸, David B. Gordon, Erdog˘ an Kumcu, James R. Lothian, Gary Santoni, Myles Wallace, John T. Warner, Thomas D. Willett, and two anonymous referees are also greatly appreciated. All errors are mine.
References Conway, P., 1994. IMF lending programs: participation and impact. Journal of Development Economics 45, 365–391. Donovan, D., 1982. Macroeconomic performance and adjustment under Fund-supported programs: the experience of the seventies. IMF Staff Papers 29 (June), 171–203. Doroodian, K., 1993. Macroeconomic performance and adjustment under policies commonly supported by the IMF. Economic Development and Cultural Change 41 (4), 849–864. Edwards, S., 1989. The International Monetary Fund and the developing countries: a critical evaluation. In: Brunner, K., Meltzer, A.H. (Eds.), IMF Policy Advice, Market Volatility, Commodity Price Rules, and Other Essays. Carnegie-Rochester Conference Series on Public Policy 31, pp. 7–68. Fischer, S., 1999. On the need for an international lender of last resort. Presented at the joint luncheon of the American Economic Association and the American Finance Association in New York on 3 January 1999. IMF website, speeches. Goldstein, M., Montiel, P., 1986. Evaluating Fund stabilization programs with multicountry data: some methodological pitfalls. IMF Staff Papers 33 (June), 304–344. Guitian, M., 1995. Conditionality: past, present, future. IMF Staff Papers 42 (December), 792–835. IMF, various years. Annual Report. The International Monetary Fund, Washington, DC. Johnson, M.E., 1993. The International Monetary Fund, 1944-1989. A Research Guide. Garland Publishing Inc. Joyce, J.P., 1992. The economic characteristics of IMF program countries. Economics Letters 38, 237– 242.
A.Y. Evrensel / Journal of International Money and Finance 21 (2002) 565–587
587
Khan, M.S., Knight, M., 1981. Stabilization programs in developing countries: a formal framework. IMF Staff Papers 28 (March), 1–53. Khan, M.S., 1990. The macroeconomic effects of Fund-supported adjustment programs. IMF Staff Papers 37 (June), 195–231. Killick, T., 1995. IMF Programmes in Developing Countries: Design and Impact. Routledge, London. Knight, M., Santaella, J.A., 1997. Economic determinants of IMF financial arrangements. Journal of Development Economics 54, 405–436. Krueger, A.O., 1998. Whither the World Bank and the IMF? Journal of Economic Literature 36 (4), 1983–2020. Loxley, J., 1984. The IMF and the Poorest Countries: The Performance of the Least Developed Countries under IMF Stand-By Arrangements. The North–South Institute, Ottawa. Reichmann, T.M., Stillson, R.T., 1978. Experience with programs of balance of payments adjustment: stand-by arrangements in the higher tranches, 1963–72. IMF Staff Papers 25 (June), 293–309. Santaella, J.A., 1996. Stylized facts before IMF-supported macroeconomic adjustment. IMF Staff Papers 43 (September), 502–544. Schadler, S., Bennett, A., Carkovic, M., Dicks-Mireaux, L., Mecagni, M., Morsink, J.H.J., Savastano, M.A., 1995. IMF conditionality: experience under stand-by and extended arrangements. Occasional paper (128). The International Monetary Fund, Washington, DC. Ul Haque, N., Khan, M.S., 1998. Do IMF-supported programs work? A survey of the cross-country empirical evidence. Working paper (169). The International Monetary Fund, Washington, DC. Vaubel, R., 1991. The political economy of the International Monetary Fund: A public choice analysis. In: Vaubel, R., Willett, T.D. (Eds.), The Political Economy of International Organizations. A Public Choice Approach. Westview Press, Boulder, pp. 204–243. Willett, T.D., 2000. Upping the ante for political economy of the international institutions. Working paper (58). Claremont Graduate University.
