A Bayesian analysis of payday loans and their regulation

Payday loans are small short-term loans that a borrower must repay or renew on his/her next payday. In states where payday lending is legal, many terms of these loans are regulated, ostensibly to protect the consumer from excessively burdensome lending practices. isting literature on payday loans has primarily focused on estimating causal effects of access to those loans, including work by Morse (2011), Skiba and Tobacman (2009) and Melzer (2011). Using individual-level administrative records on borrowers in 38 states from an online payday lender, this paper departs from past work by estimating how payday loan regulation affects borrower behavior, specifically how much they choose to borrow, how many times they choose to renew the loan, and whether or not they choose to default. State-level variation in maximum loan sizes and renewal caps are used as exclusion restrictions for identification purposes. We pay particular attention to the calculation of posterior predictive distributions that summarize the sensitivities of borrower behavior to various changes in state-level policies.