Growth Collapses

We study episodes where economic growth decelerates to negative rates. While the majority of these episodes are of short duration, a substantial fraction last for a longer period of time than can be explained as the result of businesscycle dynamics. The duration, depth and associated output loss of these episodes differs dramatically across regions. We investigate the factors associated with the entry of countries into these episodes as well as their duration. We find that while countries fall into crises for multiple reasons, including wars, export collapses, sudden stops and political transitions, most of these variables do not help predict the duration of crises episodes. In contrast, a measure of export flexibility given by the distance-weighted density of unexploited export products is significantly associated with lower crisis duration. We also find that unconditional and conditional hazard rates – the probability of exiting the crisis in a given year decrease over time, a fact that is consistent with either permanent shocks to fundamentals or with models of poverty traps. 1 We would like to thank Masami Imai, Cameron Shelton, Sanjay Reddy, Eduardo Zambrano and seminar participants at Harvard University for their valuable comments and suggestions. Alejandro Izquierdo and Bailey Klinger kindly provided data used in this analysis. The usual disclaimer applies.