Summary: | This paper considers the question of whether international banks learn from their previous crisis experiences and reduce their lending to developing countries in the event of a financial crisis. The analysis combines a bank-level dataset of bank activity and ownership with country-level data on the stock of historical crisis events between 1800 and 2005. To circumvent selection and endogeneity concerns, the paper exploits temporal variations in the relative recency of crises as instruments for crisis experience. The results indicate that foreign banks with greater crisis experience reduced their lending significantly more relative to other foreign banks, which can be interpreted as evidence in favor of a learning effect. The findings survive robustness checks that include alternative measures of crisis experience, additional controls, and decompositions into different types of crises. The question of learning is also examined from the perspective of other measures of bank performance
|