This paper replicates and extends Forbes and Warnock (2012). Using quarterly data for 60 countries over 1980–2024, the authors reconstruct the original dataset, reproduce the identification of surge, stop, flight, and retrenchment episodes, and re-estimate binary prediction models with a cloglog specification.
The replication broadly confirms the main findings of the original paper. Global factors, particularly global risk and financial contagion, remain the most important predictors of capital-flow episodes, while domestic factors generally play a limited role.
The analysis is extended in three directions. First, the sample is updated through 2024, covering major events such as the European sovereign debt crisis and the COVID-19 pandemic. Second, heterogeneity across income groups is examined, finding that the relative importance of global and domestic factors varies with the level of development. Third, local projections are estimated to assess the effects of episodes on economic activity.
Robustness checks based on alternative episode-identification thresholds show that the main conclusions remain largely unchanged.