Fund Managers “Double Down” To Ouperform by Jonathan Rhinesmith, Harvard.edu
I demonstrate that when investment fund managers “double down” on positions that have run against them, they outperform. Specifically, I find that a portfolio formed of the U.S. equity positions that hedge fund managers add to after recent stock-level underperformance generates significant annualized risk-adjusted outperformance of between 5% and 15%. This finding is not the result of a simple reversal effect, of a fund’s best ideas (large positions), or of the general informativeness of fund trades. My results are consistent with a career risks mechanism for this phenomenon. By adding to a losing position — the opposite of window dressing — managers are making their losses particularly salient. I demonstrate in a panel regression that investment managers avoid adding to losing positions. Furthermore, managers outperform by more when they double down after greater past losses in a position. These findings suggest a position-level limits-to-arbitrage effect. Even when an asset decreases in price for non-fundamental reasons, some of the investment managers with the most relevant knowledge of that asset may be particularly hesitant to add to their positions because they have already suffered losses in that asset.
“If the security you are considering is truly a good investment, not a speculation, you would certainly want to own more at lower prices.” Seth Klarman, Margin of Safety
Suppose an investment fund manager buys a stock for $10 that she thinks is worth $15. The stock proceeds to decline in value to $7, while the market remains flat. As an econometrician, one cannot easily tell if the stock’s fundamental value has dropped, or if the stock price movement was just noise, making the stock a better buy at $7 than it was at $10. If one believes the investment manager has skill, perhaps the investment manager can tell the difference. Yet even if the stock is a more attractive buy now, the fund manager may be hesitant to add to, or to “double down” on, her existing position. Her investors already know she has suffered substantial losses in the stock, and adding to the position will make those losses even more salient. The manager would effectively be employing reverse window dressing; instead of substituting out losing positions for winning stocks, she is making her losing positions even bigger. If such an effect were indeed at work, one would expect that fund managers would only “double down” in the most promising of situations, and that the corresponding positions would outperform.
In line with this reasoning, I find that in a sample of the long U.S. equity positions of hedge fund managers from January 1, 1990 through December 31, 2013, a portfolio formed of the positions that hedge fund managers add to following recent stock-level underperformance generates significant annualized risk-adjusted outperformance of between 5% and 15%. In turn, positions that managers double down on after greater position-level losses outperform by more than those that managers double down on after smaller losses. I demonstrate in panel regressions that managers avoid doubling the portfolio weights of losing positions. I also find tenuous evidence that managers facing more fund-level career risk, proxied by poor trailing manager-level returns, are particularly hesitant to substantially add to a losing position, relative to managers facing less career risk. While I cannot definitively prove that career risk is driving managers’ hesitancy to double down, my results are generally consistent with this mechanism.
I construct a variety of control portfolios to demonstrate that “doubling down” is not explained by mechanical return effects or by previously identified asset pricing phenomena. In particular, my finding is not the result of a simple reversal effect, of a fund’s best ideas (large positions), or of the general informativeness of fund trades. Funds exit, rather than double down on, most of the positions in which they suffer losses.