Essays on subprime lending, present bias, and risk salience

In chapter 1, I examine the consequences of a policy change in Rhode Island that lowered the cap on payday loan fees (interest rates) from 15% of the principal to 10%. I use a difference-in-difference framework and a unique proprietary dataset of payday loan transactions to estimate the impact on market outcomes. I find that the lenders always charge the prevailing cap, creating a strong first stage. I also find that demand for payday loans increases both at the extensive and intensive margins. I show that debt cycles become longer and more likely to end with default. Moreover, I find that no lenders exit the market after the policy change, implying that they had substantial market power. The increase in affordability of the loans increases consumer surplus by about 44%. Many consumer rights advocates believe that subprime consumers tend to be time-inconsistent. With this assumption, welfare implications of a fee cap are not straightforward, because the gain from higher affordability can be dominated by the loss from amplified time-inconsistent behavior. To address this issue, in chapter 2 I develop a dynamic model of payday loan usage with naïve hyperbolic discounting. I calibrate the model in such a way that the simulated means are as close as possible to empirical means for Rhode Island under both regulation regimes (10% and 15% fees). Using simulations of the model, I show that a tighter fee cap is welfare-improving for all consumers, regardless of their degree of time-inconsistency. Furthermore, I find that a ban is more beneficial than a fee cap to highly time-inconsistent consumers but harms time-consistent consumers. In chapter 3, I examine whether earthquake risk salience increases in an area in response to the news of earthquakes in other parts of the world. Using 20 years of housing and earthquake data, I show that disastrous earthquakes happening in other parts of the world decrease home prices in high-risk zip codes relative to low-risk zip codes. Moreover, I find that higher casualties are associated with higher price effects. I also show that the price effects decay after one month.