Cultural diversity and borrowers' behavior: evidence from peer-to-peer lending

We study cultural diversity and borrowers' behavior using data from peer-to-peer lending platform Renrendai. We proxy cultural diversity with the Linguistic Diversity Index, measured by the population-weighted number of dialects spoken in a region, and we show that it has a negative (positive) effect on the loan amount (default rate) of the borrowers. We address endogeneity using two novel instruments, the river length and land slope of Chinese cities, a Heckman two-stage model, and an IV-Heckit model. We also study areas where financial institutions' loan balances are higher (lower) than average. In areas with low (high) loan balances, the amount borrowed (the default rate) is affected more (less). We argue that lenders' behavior is a reason that borrowers in diverse cultures apply for smaller loans. Our results pass a number of robustness tests. Finally, we offer suggestions for improving risk management and inclusive financial development.