Essays in Mutual Funds and Stock Market Open Access
Kim, Jung Jae (Spring 2026)
Abstract
This dissertation investigates how institutional investors and information frictions shape asset prices, mutual fund behavior, and market stability. Across three essays, it combines theoretical modeling, causal inference, and machine learning methods to provide new evidence on the role of passive investing and investor information in financial markets.
Chapter 1 studies the impact of passive institutional ownership on stock return volatility. A theoretical model is developed in which information acquisition costs vary with stock liquidity, generating heterogeneous effects of passive ownership. The model predicts that passive ownership amplifies volatility more strongly in illiquid, small-cap stocks where informed trading is limited. Using U.S. equity data from 1990 to 2019, the results provide strong empirical support for this prediction.
Chapter 2 examines the information set used by mutual fund investors. Using machine learning techniques, the study analyzes fund- and stock-level characteristics that drive investor capital allocation decisions. The findings show that investors rely heavily on fund-level signals such as past flows and returns, while largely ignoring stock-level characteristics that are important predictors of future performance.
Chapter 3 develops a machine learning framework to predict mutual fund fire-sale events and their asset pricing implications. The results show that stocks exposed to predicted fire sales earn significant abnormal returns, highlighting a channel through which fund flows affect asset prices.
Table of Contents
Chapter 1
Ownership, Liquidity, and Volatility: The Role of Active and Passive Institutions
1.1 Introduction
1.1.1 Contribution to the Literature
1.2 Theoretical Framework and Hypotheses Development
1.2.1 Environment
1.2.2 Main Predictions
1.3 Data
1.3.1 Institutional Investors Ownership
1.4 Passive versus Active Ownership and Volatility
1.4.1 Baseline Panel Regressions
1.4.2 Quantile Regression
1.5 Quasi-Experimental Evidence: Index Reconstitutions
1.5.1 Identification Strategy
1.5.2 Instrumental Variable Estimation
1.6 Conclusion
Appendix A: Equilibrium Derivations
Appendix B: Robustness
Chapter 2
Machine-Learning the Information Set of Mutual Fund Investors
2.1 Introduction
2.2 Data
2.2.1 Fund Flow and Performance
2.2.2 Stock, Fund, and Family Characteristics
2.3 Pre-Analysis: Univariate Sorts
2.4 Method
2.4.1 Boosted Regression Trees
2.4.2 Relative Importance Measure
2.4.3 Out-of-Sample R²
2.4.4 Implementation
2.5 Results
2.5.1 Which Information Matters to Investors
2.5.2 Model Evaluation
2.5.3 Predicting Fund Returns
2.6 Conclusions
Chapter 3
Predicting Mutual Fund Fire-Sales: A Machine Learning Approach
3.1 Introduction
3.2 Data
3.2.1 Mutual Fund and Stock Data
3.2.2 Stock Data
3.3 Empirical Methods
3.3.1 The Price Pressure Measure
3.3.2 Two Prediction Problems
3.3.3 Look-Ahead Bias and its Correction
3.3.4 Boosted Regression Trees
3.3.5 Implementation
3.4 Asset Pricing Results
3.4.1 Portfolio Construction
3.4.2 FIT-Sorted Portfolio Returns
3.4.3 Residual-Sorted Portfolio Returns
3.4.4 Long-Run Return Dynamics
3.4.5 Robustness
3.5 Economic Channel
3.5.1 SHAP-Based Feature Importance
3.5.2 Feature Importance for Flow Prediction
3.5.3 Characteristic–Return Relationship
3.6 Conclusion
About this Dissertation
| School | |
|---|---|
| Department | |
| Degree | |
| Submission | |
| Language |
|
| Research Field | |
| Keyword | |
| Committee Chair / Thesis Advisor | |
| Committee Members |
Primary PDF
| Thumbnail | Title | Date Uploaded | Actions |
|---|---|---|---|
|
|
Essays in Mutual Funds and Stock Market () | 2026-04-02 13:46:20 -0400 |
|
Supplemental Files
| Thumbnail | Title | Date Uploaded | Actions |
|---|