IMF programs and economic growth: A meta-analysis

We examine 994 estimates of the effects of IMF programs on economic growth as reported by 36 studies. The mean reported effect is positive, but the estimates vary widely. We use meta-regression analysis to disentangle sources of this variation, addressing model uncertainty with Bayesian Model Averaging and LASSO. We find that estimates vary systematically depending on data and methods employed by the researchers. Reported effects of IMF programs tend to be more positive for samples that include countries with high levels of institutional and economic development, when measured on longer horizons or obtained with the propensity score matching technique. Estimates appear to depend on the types of IMF programs being considered, as general resource programs tend to result in less favorable growth outcomes compared to programs that lend from concessional resources. Estimates tend to be lower when the underlying data includes more programs from the 1980s. Authors with IMF affiliation tend to report estimates that are somewhat higher than those of outside researchers.