Evaluating the impact of IMF programs: A comparison of matching and instrumental-variable estimators

We examine the impact of IMF programs on economic performance in 95 developing countries over the period 1993–2002. Three macroeconomic measures of economic performance are considered: the real per capita economic growth rate, the ratio of the fiscal surplus to GDP, and the ratio of the current account surplus to GDP. Three estimation techniques are used: censored-sample, full-sample instrumental-variable, and matching. Substantively, we find little statistical support that IMF programs contemporaneously improve real economic growth in participating countries, but stronger evidence of an improvement in economic growth in years following a program. We find that both the fiscal ratio and the current-account ratio improve contemporaneously with IMF participation relative to the counterfactual, with effects in succeeding years differing little from the impact effects. We conclude that the program-effect estimates of matching and other estimators will differ largely because of the sample included in estimation. Matching by its nature excludes country episodes associated with extreme values of the propensity score, while the instrumental-variable estimator includes those. If there is heterogeneity of performance response in extreme vs. moderate cases, the estimates differ systematically between the two techniques. JEL codes F33 · F34 · C34