Essays on FinTech and Machine Learning in Finance

Abstract

My dissertation investigates the economic driving forces behind financial technologies adoption and examines how financial technologies could contribute to more efficient, equitable, and inclusive financial markets. In chapter one, I study the role of trust in incumbent lenders (banks) as an entry barrier to emerging FinTech lenders in credit markets. The empirical setting exploits the outbreak of the Wells Fargo scandal as a negative shock to borrowers' trust in banks. Using a difference-in-differences framework, I find that increased exposure to the Wells Fargo scandal leads to an increase in the probability of borrowers using FinTech as mortgage originators. Utilizing political affiliation to proxy for the magnitude of trust erosion in banks in a triple-differences specification, I find that, conditional on the same exposure to the scandal, a county experiencing a greater erosion of trust has a larger increase in FinTech share relative to a county experiencing less of an erosion of trust. Chapter two estimates treatment effect heterogeneity using generic machine learning inference to provide additional support of the trust channel and better understand the borrowers' heterogeneous responses to the Wells Fargo scandal. I exploit a generic machine learning inference approach proposed by Chernozhukov et al. (2020) (CDDF) to estimate treatment effect heterogeneity. This approach allows me to ex-ante stay agnostic about the characteristics of borrowers that will be more affected by the Wells Fargo scandal and let the machine learning algorithm choose those who will be more affected. Chapter three is co-authored with Murray Frank. We study the predictability of firm profits using Fama-MacBeth regressions and gradient boosting. Gradient boosting can use more relevant factors and it predicts better. Profits are more predictable at firms that are large, investment grade, low R\&D, low market-to-book, low cash flow volatility. Effects on financing decisions, and cross-section of stock returns are studied. During recessions profits are less predictable - particularly non-investment grade firms. Both algorithms produce estimates like those interpreted in the literature as evidence of excessive human optimism during booms and excessive pessimism during recessions.