I conduct two studies that investigate the behavioral characteristics of mutual fund managers. First study, The Performance of Mutual Funds on Private Information, looks at the dimension of investment skills of fund managers. The investment skills of mutual fund managers can be assessed by their ability to generate private information. In this study, by investigating the simultaneous actions of fund managers and corporate managers, we estimate how much the actions of fund managers can be attributed to private information. Using the information of insiders' transactions as a proxy for the managers' private information, our performance measure, PS (Private Shares), captures variations in skills among fund managers, suggesting that the funds with higher PS outperform the funds with lower PS. The finding that PS is positively related to future fund performance is consistent with our conjecture that fund managers who actively trade on private information have better managerial skills than the ones that do not trade on private information.

In the second study, Impact of Religious Belief on Asset Management Industry, we investigate the effects of religion on the investing behavior of fund managers. We propose a measure of corporate social responsibility propensity (CSRP) by fund managers that captures the level of a manager's tendency to invest in firms that engage in socially responsible activities. Grounded in the basis of ethics and morality, religious belief is shown to have a positive impact on a fund manager's investment in firms with good corporate social responsibility (CSR) performance. The positive association between religiosity and CSRP is particularly strong in the sample of non-institutional funds. On the performance aspect, we find that funds in the highly religious region with a higher propensity to invest in socially responsible firms tend to exhibit future performance deterioration. Our results suggest that local religiosity has a significant impact on the investing behavior of fund managers.

Machine learning is increasingly gaining applications in Finance industry. In this dissertation, I use machine learning methods to predict mutual fund and hedge fund performances and address the issue whether mutual fund and hedge fund managers add value. Overall, machine learning methods tend to outperform OLS in terms of return prediction. From a machine learning point of view, mutual fund managers don’t add value while hedge funds do deliver risk-adjusted performance. Also, such outperformance of top hedge funds is persistent with three-year horizon. Furthermore, such outperformance provided by machine learning methods are not driven by fund characteristics. A regression of machine learning outperformance on macroeconomic variables show that machine learning models tend to perform better when the economy is in recession, when the market is bearish, and when the market and economic policy uncertainty are increased.

Recently, reinforcement learning (RL) is being explored by practitioners in trading. In particular, algorithmic trading provides an ideal setting for RL. In this dissertation, I trained an agent to place aggressive stock market orders with deep Q-network with multi-

step temporal difference. The back-test results show that, the aggressive order agent outperforms an agent with linear execution schedule by an average of 0.12 to 0.69 basis points on simulated orders in Asia Pacific stock markets. This shows that RL-based methods are capable of recognizing and utilizing market information and are promising to outperform traditional execution algorithms.

In summary, machine learning and reinforcement learning are promising to provide better performances in portfolio investment and trading.

This paper studies mutual funds' use of social media (Twitter) as a new marketing instrument and its effects on the behavior of mutual fund investors. Using hand-collected data from Twitter regarding 175 mutual fund families, I find that mutual fund families are more likely to introduce Twitter when they have star funds in the family and when they are introducing new funds. The average inception size of new funds after a family joined Twitter is significantly larger than before joining Twitter. Moreover, mutual fund families using Twitter receive higher subsequent flows relative to other similar non-Twitter families. These incremental flows disappear six months after the introduction of Twitter. The higher flows are not entirely driven by the flows to the newly launched funds, as pre-existing funds also experience higher flows after introducing Twitter. The flow effect is especially pronounced for young funds within a family. Overall, the evidence suggests that mutual funds strategically use the introduction of Twitter to capture investors' attention and attract flows.

Using a novel security-level data from SEC on US tri-party repo, this paper investigates how trading relationship impacts liquidity provision within the dealer-fund repo network. This paper documents a unique repo rate dynamic: in normal times, funds charge a premium to dealers with whom they have the strongest trading relationship; in market-wide liquidity shocks, these dealers are rewarded with lower repo rate markup and better immediacy. I exploit the 2016 Money Market Fund Reform as an exogenous liquidity shock to establish a liquidity insurance mechanism. As liquidity insurers are not easily replaceable, shown in the unexpected liquidation case of Charles Schwab Sweep Funds, costly search incentivizes dealers to engage in such stable quid pro quo relationship with money market funds.

This dissertation presents a collection of three essays on how investors process information, one of the most fundamental questions in finance. It employs advanced empirical techniques such as textual analysis and machine learning and spans various investor groups, including institutional investors, retail investors, and investors of emerging asset classes. The first essay “Does Partisanship Affect Mutual Fund Information Processing? Evidence from Textual Analysis on Earnings Calls” shows that partisan funds react stronger to information more consistent with their pre-existing beliefs. The second essay “Partisan Return Gap: The Polarized Stock Market in the Time of a Pandemic”, co-authored with Jinfei Sheng and Zheng Sun, explores how political beliefs affect asset prices during the COVID pandemic. The third essay “How Do Investors Value Technology in Cryptocurrency? Evidence from Textual Analysis”, co-authored with Jinfei Sheng and Yukun Liu, examines how investors evaluate and process information about new technologies within the cryptocurrency market. Overall, this dissertation offers valuable insights into how various factors shape investor information processing and asset prices.