When Does the IMF Expect More? Explaining Legalistic Variation in Conditionality

• In Event: IMF and Its Effectiveness

Abstract

The IMF asks countries to undertake different types of conditionality reforms in exchange for loan packages. Some countries are asked to pass new laws and implement them, while others are asked to pursue mostly administrative reforms that do not require changes in the law. We explain this variation in the form of conditionality, arguing that the IMF is more likely to ask for legislative action when it expects implementation. We employ a novel dataset encompassing 35,000 IMF conditions contained in loan packages since 1980. We utilize GPT-3.5 Turbo to categorize short textual descriptions of these conditions, separating them into those conditions which require the creation or modification of laws by a legislative body, such as a parliament or congress, and those which require no legislative action. We find that partial reform programs, non-completion, and recidivisim all associate with administrative reforms rather than legislative reforms. We argue that the IMF injects legislative reforms into its loan conditionality packages when these reforms stand a reasonable chance of success. We pursue this argument with analysis of temporal variation in conditionality characteristics, both within countries and as the IMF has shifted conditionality priorities across countries. By analyzing conditionality text and categorizing sub-types of conditionality requirements, this paper brings a new perspective to debates about conditionality, which have so far tended to focus on counting the number of conditions. We also contribute to a new wave of empirical research on IMF conditionality, which focuses on key topics like gender, climate change, and corruption, among other topics.

Authors

• Patrick Egan, Tulane University

• Michael Breen, Dublin City University