On the Determinants and Outcomes of IMF Loans in Low- and Middle-Income Countries: Do Politics Matter?

ABSTRACT

The objective of this paper is to analyze the economic and political determinants of IMF loans in low- and middle-income countries and their impact on economic growth. Our contribution is threefold. First, we use the IMF Monitoring of Fund Agreements database along with international political economy factors to analyze IMF lending determinants through a Heckman two-stage selection procedure. Second, we use the predicted values of determinants of IMF lending to explain the consequences of this lending on growth. We also investigate how the domestic political regime of the recipient country would affect the outcomes of these loans. Third, we study the dynamic effects of IMF loans on economic growth using the local projection method. Our main findings show that economic and political proximity to the IMF major shareholders matter for the likelihood of obtaining an IMF non-concessional loan. Furthermore, most of the loans exert a negative effect on the trend component of GDP, confirming that such loans can stabilize the economies in the short term without improving the long-run steady growth. The analysis of the dynamic effects of loans also confirmed these findings. Finally, democratic regimes compared to autocratic ones improve the effects of these loans on economic growth.

KEYWORDS:

• IMF lending
• economic growth
• low- and middle-income countries

JEL:

• F33
• F34
• F55

Supplemental data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/1540496X.2023.2199116.

Acknowledgments

We thank the editor Paresh Kumar Narayan and two anonymous reviewers for their very constructive comments and suggestions. Comments from Kabbashi Suliman on an earlier draft are gratefully acknowledged. We also thank participants in the Economic Research Forum (ERF) 27^th Annual Conference for useful discussions, especially Ibrahim Elbadawi and Thibault Lemaire.

This paper has benefited from a financial support from the Economic Research Forum, Cairo, Egypt.

Disclosure Statement

No potential conflict of interest was reported by the authors.

Notes

1. We use a group of 156 countries including all low- and middle-income countries according to the World Bank classification and all countries which were enrolled in any IMF agreement over the period of our study. Annex 2 provides a list of these countries.

2. Some studies exclusively focus on two types of IMF programs (SBA and EFF) since they are the major programs offered by the IMF (Barro and Lee 2005; Dreher 2006). Other studies account for all IMF programs altogether. However, we distinguish between the programs given their different time span, extent of concessionality, and policies. For instance, some programs focus on managing aggregate demand or adjustment to shocks whereas others focus on strengthening the supply side. Furthermore, some programs focus on low-income countries in particular. We believe these differences are important to consider while studying the impact of IMF loans on economic growth (Bird and Rowlands 2017).

3. We refer the reader to Annex 3 that provides some descriptive statistics related to the size and different types of IMF lending by region.

4. Given that we have a panel of countries, we have to take into consideration the unobserved heterogeneity. Accordingly, as a baseline we undertake a static panel model. In order to choose between the fixed effects model and the random effects one, we run the Hausman test. Hausman test results (reported in results tables) mostly confirm that the fixed effects model is preferred compared to the random effects since the null hypothesis of the test is rejected in most of the specifications. The null hypothesis of the test is that the individual unobserved effects are uncorrelated with the included explanatory variables. If that hypothesis is rejected, the fixed effects is the preferred model since it is consistent whereas the random effects is not.

It is also worth noting that we run these regressions using IMF loans as a share to GDP and the results remain globally the same.

5. High-pass filters, such as the HP and the BW filters, allow for stochastic cycles meeting a minimum frequency level. Band filters, such as Christiano-Fitzgerald (CF), and Baxter-King (BK) filters, allow only stochastic cycles within a specified range of frequencies with any frequencies outside this range filtered out (Fedderke and Mengisteab 2017).

6. The regime type classification is based on Polity Scores. The Polity Score captures a regime authority spectrum on a 21-point scale ranging from −10 (hereditary monarchy) to + 10 (consolidated democracy). The Polity scores can also be converted into regime categories in a suggested three part categorization of “autocracies” (−10 to −6), “anocracies” (−5 to + 5 and three special values: −66, −77 and −88), and “democracies” (+6 to+10).