Research

Working Papers

Technology Reliance, Data Quality and Mortgage Credit: Evidence from Automated Underwriting
(Job Market Paper)

Algorithmic evaluation of loan applications is crucial in assessing creditworthiness, but its effectiveness depends on the quality of borrower data. I study how changes in algorithmic underwriting in due cognizance of improved borrower data quality affects the efficiency of mortgage credit allocation and its economic consequences. I exploit an exogenous policy change in a major automated underwriting system’s assessment that reduced stringency in evaluating a subset of loan applications which was hitherto imposed due to data inaccuracies. Utilizing novel data and a differences-in-differences strategy, I find that the policy leads to an increase in mortgage credit access without any noticeable impact on credit risk. This change accounts for 26% of the increase in approval rates, with stronger effects where there is limited human interaction in the loan application process and where lenders have greater incentives for mortgage securitization. Further evidence suggests greater benefits for borrowers in market segments with pre-existing financial frictions as well as for racial minority borrowers. While homeownership rates rise with spillover effects through lower rent growth in the more exposed areas, exposed banks end up crowding out commercial credit. My findings highlight the significance of automated mortgage underwriting for creditworthy and credit-constrained borrowers, which has important economic implications.

Do Laboratory Actions Inform Investors’ Real Decisions? Evidence from Myopic Loss Aversion
(with John List, Asad Islam, Vy Nguyen and Kazi Iqbal)

Whether, and to what extent, behavioral anomalies uncovered in the lab can extend to natural environments by explaining decision making in the field remains of first order importance in economics and finance. We explore this in the context of myopic loss aversion (MLA). We use artefactual field experiments to elicit the extent of MLA exhibited by retail investors in constructed laboratory markets and link it to a proprietary, individual-level dataset of their private investment accounts. We find that MLA is associated with lower equity market investment levels and lower market beta of portfolios. The MLA effect is stronger for investors who are less experienced, who update themselves frequently about the market and who make changes to their portfolio frequently. Additional evidence shows that MLA investors react negatively to short term losses in their portfolio and their investments also perform poorly in the stock market.  

Work in Progress

Correcting Biases in Financial Market Decisions? Evidence from a Randomized Field Experiment Educating Retail Stock Traders
(with Asad Islam); Link to AEA Registration

Randomized Controlled Trial completed. Data Analysis in Progress. 

Participation in Stock markets has important implications for personal wealth building. However, individuals, especially unsophisticated investors may be disposed to behavioral biases which may potentially be suboptimal. Partnering with brokerages, we randomly assign individual non-professional, retail traders to receive training on finance and stock market essentials and study the implications for their decision making in the stock market. Our initial evidence suggests increased portfolio diversification and a change in portfolio composition for the group of treated traders who participated in the training program.

Peer to Peer Lending not “Peers” to Traditional Banking?

Using data from a large peer to peer crowdfunding platform, I show that such online platforms respond to credit gaps from closures of physical establishments of traditional financial intermediaries. Utilizing mergers between large banks operating in the same county to generate exogenous variation in the incidence of closures suggests that the results are causal. However, users do not permanently transition to such platforms as I find no strong evidence that borrowers in areas with higher closures are comparatively more likely to repeatedly use peer to peer lending services. Instead, I find that participation in such online lending platforms improves subsequent credit outcomes from traditional intermediaries for borrowers, especially those residing in areas that have experienced high merger induced closures. The results suggest despite financial technology disruption, some technological innovations in credit markets may be a temporary alternative in the event of negative credit supply shocks and highlights the relevance of traditional financial intermediaries. 

Firm Refinancing in Production Networks