Data Exchanges Among Firms (with Sharique Hasan), 2021, Digital Business, 1(2), 1-11

Working Papers:

Which Investors Drive Factor Returns? [SLIDES]

Different investors hold different portfolios. To explain this phenomenon, I build a model in which investors have different information processing capabilities. The model predicts that highly capable investors specialize in factor timing, hold more volatile and dispersed portfolios, and reduce average risk premia and volatility. Using novel empirical measures of investors' capabilities and information choices, I find that hedge funds are the most capable investors, while insurance companies and pension funds are the least. Variation in factor timing ability is the primary driver of these differences. Investors' portfolios exhibit properties consistent with the model's predictions. Using a demand system approach, I show that hedge funds have the greatest per-dollar impact on both risk premia and volatility. I estimate that a $1 trillion outflow from the hedge fund industry increases factor risk premia by an average of 0.45 pp per annum, an effect concentrated among risk factors with the highest expected returns.

Hedge Fund Incentives, Risk Taking, and Asset Prices

I present a model in which hedge fund managers maximize their expected compensation subject to leverage constraints. This allows me to explore the impact of hedge funds' prototypical contract structure on their dynamic risk allocations and on asset prices in general. One implication of the model is that risk taking varies around a fund's distance to its high-water mark. My empirical work is consistent with the implications of the model in that hedge funds at and furthest from their high-water marks take on significantly greater levels of risk. This increased risk is accomplished in part by investing in more volatile securities. Further, as more hedge funds approach their high-water marks, aggregate hedge fund risk taking increases, the security market line flattens, and betting against beta returns increase, consistent with evidence that these funds increase investment in high beta securities. This work highlights the importance of misaligned preferences between financial intermediaries and investors in explaining asset prices.

Works in Progress:

Fund Alphas from the Cross-Section (joint with Andrew Patton and Brian Weller)

We propose a new technique to estimate alphas in mutual fund returns. Exploiting commonality among mutual fund strategies, we propose a PCA-based estimator to extract common factor loadings and alphas from the panel of mutual fund returns. Borrowing from the machine learning literature, we use matrix factorization and completion techniques. We find that this estimator extracts the correct number of factors in a variety of simulations and more reliably estimates mutual fund alphas than traditional approaches.