Implicit incentives and reputational herding by hedge fund managers

Abstract

Several theories of reputation suggest that managers' incentives affect their propensity to engage in herding behavior. This paper investigates these theories by tracking hedge fund managers' herding behavior over their careers. I first examine managerial incentives for herding, and show that more senior managers that deviate from the herd have a significantly higher probability of failure and do not experience higher fund inflows than their less-senior counterparts. These implicit incentives should encourage managers to herd more as their careers progress. I find strong support for this hypothesis: using a number of proxies for herding, I show that more experienced managers herd more than less-experienced managers. Finally, I examine performance differences between more and less-experienced managers, and find that while more experienced managers underperform less-experienced managers, this underperformance does not appear to be caused by differences in herding. Overall, these results are in direct contrast with studies of mutual fund managers, reflecting important difference in implicit incentives between the two industries.

Introduction

A number of theoretical models hypothesize that an individual manager's risk-taking behavior is affected by the risk-taking behavior of other managers in his peer group, a situation that Graham (1999) denotes as “reputational herding.” Specifically, Graham's model predicts that managers with high reputation and salary will herd (i.e., behave similarly to other managers, a risk-averse stance) to protect their current level of status and pay. In a related vein, Prendergast and Stole (1996) argue that managers wish to acquire a reputation for learning, and when faced with new information each period, less-experienced managers overreact to show they have good information. By contrast, more experienced managers under-react so as not to signal that their actions in prior periods were wrong. Thus, assuming that experience and reputation are positively correlated, both models imply that managers will herd more by taking on less risk relative to their peers as they gain more experience.¹

A large body of empirical research in labor economics, financial economics, and management supports this theory.² For example, the empirical results of Graham (1999) (investment newsletters) and Li (2002) (security analysts) indicate an increase in herding over time. Also, in studies of CEOs and top managers, Katz, 1982, Hambrick and Mason, 1984, Finkelstein and Hambrick, 1990, Jaggia and Thosar, 2000, Bertrand and Schoar, 2003 obtain similar results. However, some empirical research finds contradictory evidence; see, for example, Chevalier and Ellison (1999) (mutual funds), Hong, Kubik, and Solomon (2000) (securities analysts), and Lamont (2002) (macroeconomic forecasters) who show that herding decreases in managerial experience.³

Of course, it is difficult to directly observe the reputational concerns of a manager. One way to work around this issue is to focus on the implicit herding incentives for all managers in the industry. If the implicit incentives appear to reward high reputation (and highly paid) managers for protecting their status and pay, one would expect managers to herd more (i.e., decrease risk relative to their peers) as they gain experience, as predicted by Graham, 1999, Prendergast and Stole, 1996. In their analysis of mutual funds, Chevalier and Ellison (1999) take this approach, and show that manager termination is more performance-sensitive for young managers than for old; in other words, that high reputation (and presumably highly paid) managers need not worry too much about protecting their status and pay by herding, since it is the young managers that are most frequently terminated for poor performance. Hence, Chevalier and Ellison (1999) predict that young managers are more likely to herd while old managers are more likely to deviate from the herd, a prediction that is borne out in their empirical analysis.

Using a similar approach to Chevalier and Ellison (1999), I study the implicit incentives and reputational concerns of hedge fund managers. I focus on two implicit incentives present in the hedge fund industry: the desire to avoid termination and the desire to increase capital inflows. I first show that a manager's termination is highly sensitive to his experience level and his propensity to engage in herding behavior.⁴ More experienced managers are much more likely to be terminated for deviating from the herd than less-experienced managers. The second implicit incentive, to increase market share, is measured using net fund cash flows. Using a fixed-effects regression that follows individual managers over time, I show that more senior managers that deviate from the herd do not attract higher cash inflows than less-experienced managers. This finding, taken together with the above result that deviating from the herd by more senior managers increases the probability of termination, provides a strong incentive for more experienced managers to herd more as their careers progress.

I next test how managers behave in light of these implicit incentives: namely, do more experienced managers herd more? Using a fixed-effects regression, I document a positive relationship between manager experience and herding. This result is statistically significant, consistent across several measures of herding, and robust to a number of control variables. This analysis provides strong empirical support for the theoretical work of Graham, 1999, Prendergast and Stole, 1996.

Finally, I examine the impact of herding behavior on risk-adjusted performance. Although risk-adjusted performance worsens as managers gain experience, this lower performance does not appear to be caused directly by the propensity for more experienced managers to herd more.⁵

The paper is organized as follows. Section 2 describes the data. Section 3 documents that more experienced managers are significantly more likely than less-experienced managers to be terminated for deviating from the herd. Section 4 shows that more experienced managers that deviate from the herd do not experience higher fund inflows than less-experienced managers (in fact, there is weak evidence that they experience net outflows). The results of 3 Determinants of manager termination, 4 Determinants of fund flows provide evidence of implicit incentives that encourage managers to herd more as their careers progress. Section 5 tests this hypothesis and shows that, indeed, more experienced managers do engage in more herding. Section 6 examines the relationship between fund performance, manager experience, and herding, and Section 7 concludes.