UC Berkeley

The use of technology by firms is changing the way insurance and lending markets function. I study the financial technology, or "fin-tech'', industry, which is characterized by a growing number of online lenders who use data on educational, employment, and financial outcomes to quickly assess the risk of prospective borrowers and offer individualized loan terms. In many ways, their financial "innovations'' can be thought of as movements towards more personalized products: interest rates that better reflect individuals' risk, payment plans that are tailored to individuals' monthly income and expenditures, and user-friendly interfaces that make financial decisions more intuitive and uncomplicated. On an individual level, as firms expand and customize product offerings, there is the potential for large efficiency gains. These innovations could also have wider implications for market structure; for example, if more accurate risk-based pricing creates clear winners and losers, it will change the distribution of consumer surplus.

Advances in data-driven underwriting have both efficiency and equity implications for consumer lending markets where private and public credit options coexist. In the $1 trillion student loan market, private lenders now offer a growing distribution of risk-based interest rates, while the federally-run loan program sets a break-even, uniform interest rate. In my first chapter, I measure the overall gains in consumer surplus from such risk-based pricing and quantify the redistributional consequences of low-risk types refinancing out of the government pool into the private market. The empirical analysis is based on a unique applicant-level dataset from an online refinancing firm that contains information on loan terms, household balance sheets, and risk-based interest rates. I first leverage a series of firm-conducted interest rate experiments to estimate the sensitivity of borrowers' maturity and refinancing choices to interest rates. Using the maturity response, I then estimate a structural model of borrowers' repayment preferences. Using the estimated model, I show that comprehensive risk-based pricing generates large absolute gains in welfare of $480 per borrower relative to a break-even uniform price, and $400 relative to a coarser method of FICO-based pricing. If the federal pool conducts breakeven pricing, these efficiency gains come at a direct equity cost -- low risk surplus will increase on average by $2,300, while high risk surplus will fall by $2,100. In order to maintain access to the current uniform rate, the government would have to transition from break-even pricing to an average net subsidy of $2,080 per borrower.

In the second chapter, I empirically analyze the fixed and variable rate decisions of borrowers who are financing large personal loans, and are given the option to switch rate types at any point. Many online lending firms now offer financial products that are more flexible and personalized than traditional loans; however, little is known about how consumers will interact with these more complete, but also more complex, contracts. Over my sample time period, the market index interest rate for the fixed and variable rate loans changed considerably. I first present reduced form evidence on the determinants of borrowers' initial rate decisions and the presence of switching costs, and then estimate a structural model that maps these findings to the coefficient of absolute risk aversion and a switching cost parameter. I compare the active and inactive rate choices of borrowers in different interest rate environments to separately identify switching costs from risk preferences.I show that while initial rate choices are very responsive to the prevailing interest rate environment, very few borrowers ever take advantage of the option to switch rate types even when interest rates increase. Specifically, I estimate a risk aversion parameter of .0564, which implies that borrowers are very risk averse, and lower and upper bounds on switching costs from $166 to $1,185. I also show that both the initial probability of choosing a variable rate loan and the probability of never switching are positively correlated with borrower liquidity constraints, which suggests that these borrowers are more focused on current monthly payments than future interest rate risk.