Essays on the Dynamic Strategies and Skill of Institutional Investors

Abstract

This dissertation studies the behavior of institutional investors, who control a large share of the world's investment capital, with the goal of shedding light on when and how those investors reveal information. Guided by economic intuition, I highlight instances in which the trades of fund managers are particularly informative. I focus on hedge funds and present evidence that in these instances funds' decisions predict future asset price movements.

These results demonstrate that fund managers possess valuable information. At the same time, my findings support a view of the world in which fund managers have more capital than what they allocate to opportunities with high expected returns. Hedge funds may be “smart” -- they may be able to identify mispriced securities -- while still delivering poor returns to their investors.

Chapter 1 presents evidence that price impact is an important consideration even at the quarterly time horizon of the trades I observe. If fund trades generate price impact, and if price impact is a function of volume, then funds should only be willing to trade a large share of volume when their information is compelling. Indeed, I find that hedge funds predict future stock returns when they purchase a large share of volume. I also provide evidence that the price impact of fund trades incorporates information into stock prices. If informative prices impact real economic decision making then these findings support the welfare relevance of the active management industry.

Chapter 2 shows that funds avoid adding to losing positions. When they do, however, they predict future stock-level outperformance. These results are consistent with a career risks mechanism, as adding to a losing position corresponds to reverse window dressing. They also suggest a position-level limits-to-arbitrage effect.

Chapter 3 demonstrates that hedge funds frequently buy back into stocks they have held in the past. This phenomenon occurs much more often than it would by chance. I use these findings to argue that fund managers develop company-specific expertise that persists over time. When funds establish expert positions after poor past stock-level performance, they predict future stock-level excess returns.