Essays in Financial Economics

Abstract

This dissertation contains three essays in financial economics, with a focus on the effects of new technologies on traditional financial markets including venture capital market and mortgage market.

In a joint work with Fangzhou Lu, we document in Chapter 1 the heterogeneous effects of new technology on venture capital investment. Venture capital, an undoubtedly successful way of financing entrepreneurship, is believed to suffer from search costs and information frictions. New technologies have been invented to overcome these issues. Online platforms such as Product Hunt maintain user-generated profiles of and comments on a portfolio of start-ups. Venture capital investors has been increasingly relying on data from these platforms and artificial intelligence tools to guide investment decisions. However, information frictions could be intensified due to the presence of manipulation. Using data on venture capital deals and data from Product Hunt, we present supporting evidences that entrepreneurs manipulate investor perceptions by manufacturing comments that praise their products. Using COVID-19 as a positive shock to investor online presence, we examine the differences in online opinions for similar start-up products before and after the pandemic. We argue that the net gains from manipulating online opinions are highest for entrepreneurs who are new to the online community, for start-ups in early stage, and for start-ups facing fierce competition. We demonstrate that start-ups with a high incentive to manipulate have more positive but less useful comments post COVID-19 relative to prior. Furthermore, investors tend to relate their investment decisions to online sentiment, but the effect is heterogeneous. Only young and inexperienced investors are responsive to online sentiment.

In Chapter 2, I study Fintech mortgage lenders, who collect wide forms of borrower data entirely online and rely on big data to make credit decisions through the use of machine learning algorithms. Compared to traditional lenders, Fintech lenders are more likely to originate loans with high loan-to-value ratio (LTV) and particularly high debt-to-income ratio (DTI), possibly working through greater loan size instead of lower income. Conditional on predicted default rate using only observables, ex-post default rate is not significantly differed whether the loan is originated by a Fintech or a traditional lender. Fintech lenders also set interest rates that are more sensitive to LTV but less sensitive to DTI, and consequently, their interest rates have higher forecastability in prepayment but lower forecastability in delinquency and default, resulting from a premium charged to high LTV loans that get prepaid more often. Fintech lenders get cross-subsidies in the to-be-announced (TBA) mortgage-backed-securities (MBS) market since high prepayment rate loans are pooled together with low prepayment rate loans in the same forward contract. The findings suggest that new technology might be able to identify credit risks at the margin but may also be used to facilitate lenders in extracting rents.

In Chapter 3, I document five facts on labor productivity in the U.S. mortgage lending industry using a novel dataset that matches lenders, lender branch locations, loan officers, mortgage applications, originations and loan performance. First, labor productivity at non-depository lenders is on average two times higher than that at banks or credit unions. Second, labor productivity growth has been accelerating since 2014. The trend is driven by banks and credit unions, not non-depository lenders. Third, banks with larger assets, higher return on assets, but lower growth in deposits have higher labor productivity. Fourth, high labor productivity is associated with long lending distance. Fifth, high labor productivity is associated with high delinquency and prepayment. One important source of productivity growth is technology adoption. In a case study, I use Quicken Loans’ adoption of Rocket Mortgage online platform in late 2015 as an exogenous technology shock, and find that competitors respond by hiring more loan officers, especially males, to compete with Quicken Loans in local mortgage markets.