Essays on empirical asset pricing and household finance

Abstract

This dissertation contains three essays about the economic interpretability of machine learning models, the learning process of migrant workers to send international remittances, and the efficiency of ETF prices using ETF options. In the first essay, we examine how the return predictability of deep learning models varies with stocks’ vulnerability to investors’ behavioral biases. Using an extensive list of anomaly variables, we find that the long-short strategy based on deep learning signals generates greater returns for stocks more vulnerable to behavioral biases, i.e., small, young, unprofitable, volatile, non-dividend-paying, close-to-default, and lottery-like stocks. This performance of deep learning models becomes pronounced for stocks held by less sophisticated investors, when investor sentiment is high, and when disagreement is serious. These results suggest that deep learning models with nonlinear structures are useful for capturing mispricing induced by behavioral biases. In the second essay, we use data from a leading fintech firm in Korea, including transaction-level data on international remittance payments to developing Asia and referral data among users, to study the role of social networks in helping low-income workers optimize their use of new technology. Our study shows how experience using a desirable fintech feature--an order cancellation option--helps workers optimize the timing of the exchange rate applied to their transactions. Overall, we find that workers' personal experience using the fintech, and the experience of the worker's social network, increase the frequency and efficiency of workers' use of the feature. In the third essay, we provide evidence of significant frictions in the pricing of ETFs that can be detected using ETF options. First, the difference between call and put option-implied volatility, or IVspread, contains information about future ETF returns. This predictive power stems from the fact that ETFs with extreme IVspread face greater price pressure and become more illiquid. Second, the option-implied price of ETF is a useful predictor of future changes in NAV, particularly for ETFs that hold international stocks or illiquid securities. This predictability of NAV returns is linked to the noisiness of the benchmark index of the ETF. Our findings demonstrate significant inefficiencies in the ETF prices and the role of options market information in uncovering them.