ریسک اشتغال، انگیزه های جبران خسارت، و در نظر گرفتن ریسک مدیریتی : مدارک و شواهد از صنعت صندوق های سرمایه گذاری

We examine the influence on managerial risk taking of incentives due to employment risk and due to compensation. Our empirical investigation of the risk taking behavior of mutual fund managers indicates that managerial risk taking crucially depends on the relative importance of these incentives. When employment risk is more important than compensation incentives, fund managers with a poor midyear performance tend to decrease risk relative to leading managers to prevent potential job loss. When employment risk is low, compensation incentives become more relevant and fund managers with a poor midyear performance increase risk to catch up with the midyear winners.

We analyze mutual fund managers’ risk taking decisions in response to the incentives they face.1 In making their investment decisions, fund managers face two main incentives. First, they want to earn high compensation. Second, they want to keep their jobs, i.e., do not want to be laid off. We examine how these incentives, which we term ‘compensation incentives’ and ‘employment incentives’, respectively, determine the fund managers’ risk taking behavior. We show that it depends on the interim performance of the funds they manage: compensation incentives lead managers of funds with a poor interim performance to increase their fund's risk relative to managers of funds with a good interim performance. In contrast, employment incentives lead managers of funds with a poor interim performance to decrease their fund's risk relative to managers of funds with a good interim performance. We start our discussion with an analysis of the incentives for poorly performing fund managers (“losers”), before turning to a description of the incentives for well performing fund managers (“winners”). Brown, Harlow, and Starks (1996) are the first to examine risk taking incentives of mutual fund managers in a yearly tournament setting. They exclusively focus on implicit compensation incentives that arise due to the positive convex relationship between the inflow of new money into the fund and its past performance (e.g., Sirri and Tufano, 1998).2 As the fund family typically charges a fixed percentage management fee on its assets under management, the fund family profits from funds that reach a top position by the end of the year and eventually attract new inflows. However, the fund manager also benefits from reaching a top position by the end of the year since a fund manager's compensation typically depends on past performance as well as on the size of the fund managed by her (see, e.g., Khorana, 1996; Farnsworth and Taylor, 2006). As fund size is mechanically linked to inflows, there is also a convex relationship between the fund's past performance and the compensation of the fund's manager. Such a convex relationship could lead to yearly tournaments among fund managers: midyear losers increase their funds’ risk in the second part of the year since they have not much to lose from a further deterioration of their position in terms of inflows and eventually compensation, while increasing risk increases their chance of catching up with the midyear winners (Brown, Harlow, and Starks, 1996). However, there is an additional incentive that fund managers face and which is neglected in the previous literature on fund manager tournaments: fund managers are concerned about keeping their jobs. Fund managers care about employment incentives because losing their jobs would entail significant costs in terms of foregone income, loss in reputation, and loss of future job opportunities. We expect that these employment incentives are taken into consideration by fund managers in making portfolio risk decisions and thus, become vital in explaining risk taking strategies. If a manager takes on too high a risk (perhaps because of tournament incentives), then there is also a higher risk of poor performance and the probability of forced turnover is much higher for fund managers with poor past performance, i.e., particularly midyear losers face a serious threat of being laid off (Khorana, 1996; Chevalier and Ellison, 1999; Hu, Hall, and Harvey, 2000).3 Thus, if midyear losers follow risky strategies in the second part of the year, they increase the probability of achieving a performance that is so bad that it would eventually trigger job loss (Bloom and Milkwovich, 1998). Consequently, employment incentives should cause midyear losers to decrease their risk, ceteris paribus.4 Consider now the incentives for midyear winners, which are markedly different from those of midyear losers described above. On the one hand, compensation incentives lead midyear winners to try to lock in their leading position and play it safe rather than to increase their risk. On the other hand, employment incentives are of much less, if any, relevance for them. They face no serious threat of dismissal due to poor performance. Thus, unlike midyear losers, they have no reason to change their risk due to employment incentives. Our analysis so far shows that employment incentives and compensation incentives lead to diametrically opposite hypotheses regarding managerial risk taking. Compensation incentives should lead midyear losers to increase their risk relative to midyear winners. Employment incentives should lead midyear losers to decrease their risk relative to midyear winners. The relative strength of employment incentives and compensation incentives depends on the expected costs of job loss as well as on the expected increase in compensation due to reaching a top position. Market returns are a simple but ideal proxy for the relative strength of these two incentives because they capture the market environment the fund managers face. The reasoning is as follows: after bear markets aggregate inflows into funds are generally low (e.g., Karceski, 2002; Breuer and Stotz, 2007). Therefore, the fund manager attracts little new money for her fund by reaching a top position. As a consequence, even the size of the best performing funds grows only slightly, the fund family earns little additional fee income, and eventually is not very profitable. Both effects lead to weak compensation incentives for fund managers in bear markets: as the fund manager's personal compensation is positively related to fund size, the manager makes little additional income by reaching a top position. Furthermore, the bonus payments the fund manager receives depend heavily on the profitability of the fund family (Farnsworth and Taylor, 2006), which is low in bear markets. Based on these observations, Karceski (2002) argues that fund managers do not care much about outperforming other fund managers during bear markets, i.e., compensation incentives are weak in bear markets. In contrast, employment incentives are strong in bear markets. The low aggregate inflows into mutual funds after bear markets eventually lead to many fund closures, primarily of badly performing funds (Zhao, 2005b). Thus, the threat that the fund of a poorly performing manager will be closed down is more severe in bear markets, resulting in a higher probability of job loss. Chevalier and Ellison (1999) show that job loss is indeed more likely after bear markets than after bull markets.5 They conjecture that the reason for this finding is that fund companies are less profitable and have to cut back costs after low inflows and thus lay off fund managers. At the same time, fewer new funds are started (Zhao, 2005a) and a fund manager might face difficulties in finding a new job in the fund industry if she actually loses her job.6 Following this line of reasoning, we expect employment incentives to be strong and compensation incentives to be weak in bear markets. In contrast, we expect compensation incentives to be strong and employment incentives to be weak in bull markets for the following reason: aggregate flows into the mutual fund market are high after bullish markets. Thus, the additional inflows and eventually the compensation that a fund manager can capture by achieving a top position are high in this case. Furthermore, since fund families are typically more profitable in bull markets, the bonuses for well performing fund managers are particularly high (Farnsworth and Taylor, 2006). Consequently, compensation incentives are strong in this case. In contrast, employment incentives are weak: there are few fund closures after bull markets (Zhao, 2005b), making it unlikely that a manager loses her job due to a closure. Moreover, the probability of job loss for fund managers is generally lower in bull markets (Chevalier and Ellison, 1999). At the same time, a lot of new funds and even entirely new fund families are founded (Zhao, 2005a; Faff, Parwada, and Yang, 2006). Thus, the threat of dismissal is not severe because there are many alternative job options available even if a fund manager loses her job. Therefore, employment incentives are relatively weak in this case, while compensation incentives are strong. From this analysis, we conclude that which incentive dominates is contingent on the market performance. Compensation incentives are more likely to dominate in bull markets, while employment incentives are more likely to dominate in bear markets. Furthermore, we expect that the more bullish (bearish) the markets are, the more pronounced is the impact of compensation (employment) incentives. Thus, our main hypotheses are: Hypothesis 1. In bull markets compensation incentives dominate and managers of funds with a poor midyear performance increase fund risk more than managers of funds with a good midyear performance. Hypothesis 2. In bear markets employment incentives dominate and managers of funds with a poor midyear performance increase fund risk less than managers of funds with a good midyear performance. Hypothesis 3. The more bullish (bearish) the markets are, the more (less) pronounced is the increase in fund risk of managers of funds with a poor midyear performance relative to managers of funds with a good midyear performance. One important consideration in testing these hypotheses is estimating the managers’ ex ante risk choices. Thus, we use portfolio holdings data of US equity mutual funds (over the period 1980 to 2003) because holdings data allow us to capture the intended risk taking strategies of fund managers rather than the realized ones. The latter, captured from return data, might be partly driven by unexpected changes in stock risk. Our results support all three hypotheses. In bull markets midyear losers increase risk more than midyear winners. In bear markets midyear losers increase risk less than midyear winners. The extent of managers’ risk adjustment is larger the more bullish or bearish the market is. These findings show that ignoring the interplay between employment incentives and compensation incentives can easily yield misleading results. Our results have implications for fund investors and fund families as well as for regulatory authorities. Our findings are stable over time and hold after controlling for fund, fund family, and market segment characteristics that might influence the risk taking behavior of fund managers. They also hold after taking into account the impact of risk limits fund managers might face. Fund managers might be forced to adjust their risk, if they unintentionally exceed those limits. Such violations of risk limits can happen because risk cannot be predicted completely reliably and realized risk eventually deviates from intended risk. We do find that fund managers counterbalance such risk surprises by adjusting their risk. We also find some evidence consistent with the notion that managers with high tenure care less about compensation incentives than younger managers. However, all these effects do not change our main results. Our study is related to three strands of research. First, we contribute to the general literature on managerial risk taking in response to compensation incentives (e.g., Cohen, Hall, and Viceira, 2000; Rajgopal and Shevlin, 2002; Coles, Daniel, and Naveen, 2006). For the mutual fund industry, Brown, Harlow, and Starks (1996) provide empirical evidence that managers respond to implicit compensation incentives that arise due to the convex performance flow relationship. This finding is confirmed in several follow-up studies, like Koski and Pontiff (1999), Elton, Gruber, and Blake (2003), Qiu (2003), and Hu, Kale, Pagani, and Subramanian (2008). However, Brown, Harlow, and Starks (1996) and Kempf and Ruenzi (2008) report that the risk taking behavior of fund managers is not stable over time. Our study reconciles the partially contradictory evidence provided in earlier studies by showing that the relative strength of compensation incentives and employment incentives drives risk taking behavior of fund managers. We also show that not taking both incentives into consideration can lead to incorrect conclusions. Second, we contribute to the literature on employment incentives and risk taking. There are few empirical studies on this issue. Chakraborty, Sheikh, and Subramanian (2007) analyze the behavior of managers of industrial firms and show that managers who face high employment risk make less risky decisions than managers who face low employment risk. Looking at analysts’ behavior, Hong and Kubik (2003) and Clarke and Subramanian (2006) find that analysts who face greater employment risk issue more conservative forecasts. Chevalier and Ellison (1999, p. 429) examine the career concerns of fund managers and argue that “the desire to avoid termination is the most important career concern.” Our study contributes to this literature by showing that fund managers adjust risk in response to employment incentives in systematic ways. Finally, we contribute to the recent literature on the role of restrictions in the mutual fund industry (e.g., Almazan, Brown, Carlson, and Chapman, 2004) by showing that the response to risk surprises is an important factor driving managerial risk taking. To the best of our knowledge our study is the first to explicitly examine reactions of managers to unintended risk realizations. The rest of this article is organized as follows: the next section introduces the data and details how the intended risk taking of fund managers is calculated. In Section 3, we empirically examine the influence of compensation and employment incentives on managerial risk taking. In Section 4, we analyze other potential drivers of risk taking. In particular, we examine the robustness of our results by taking into account the impact of risk surprises the fund manager might face. In addition, we look at the impact of characteristics of the fund, the fund family, the market segment, and the fund manager on managerial risk taking. Section 5 concludes.

Mutual fund managers face various incentives that have an impact on their risk taking. While compensation incentives arising from the convex performance flow relationship are studied in great detail (e.g., Brown, Harlow, and Starks, 1996; Koski and Pontiff, 1999; Elton, Gruber, and Blake, 2003), there is little evidence on the impact of employment risk on the risk taking decisions of mutual fund managers. In this paper, we jointly examine compensation incentives and employment incentives. Using data on portfolio holdings of US equity mutual funds and stock returns from 1980 to 2003, we find that the way fund managers alter their risk in response to their midyear performance depends on the relative strength of their employment and compensation incentives. Our findings indicate that midyear losers increase their risk more than midyear winners in bull markets where compensation incentives would be more important. The opposite holds in bear markets, where employment incentives would dominate. These results reconcile contradictory results presented in earlier studies. Our results are neither driven by characteristics of the fund, the fund family, the market segment, or the fund manager nor by the reaction of managers to unexpected risk realization. We find that managers counterbalance risk surprises only if realized risk is higher than initially planned. This is consistent with the idea that fund managers face risk limits they must not or do not want to exceed by the end of the year. However, our main results are not affected by this response to risk surprises. Gaining a better understanding of the incentives driving fund managers’ behavior is important for fund investors and fund companies, as these incentives can lead to adverse managerial behavior. As Brown, Harlow, and Starks (1996) point out, risk adjustment of fund managers as a response to compensation incentives may not be optimal for fund investors. The same is true for risk adjustments due to employment incentives. They are not aimed at building a portfolio with optimal risk-return characteristics from the fund investor's point of view and can create additional trading costs, which eventually hurt performance (Bagnoli and Watts, 2000; Li and Tiwari, 2006; Huang, Sialm, and Zhang, 2008). James and Isaac (2000) show that risk changes due to such incentives can even lead to inefficient price formation in asset markets and might thus be of some interest from a regulatory point of view. Perhaps the most important implication of our study for future research on managerial risk taking is that temporal variations of compensation and employment incentives should not be neglected. Our findings suggest that ignoring such variations can easily deliver misleading results and eventually lead to erroneous conclusions. We believe that our results not only hold for managers of mutual funds, but also might have important implications for the behavior of managers of corporations in general. In the corporate world the business cycle might play a role similar to the role played by bull and bear markets in the mutual fund industry. For example, it is likely that employment risk is only a minor concern for managers in a boom period, while it might seriously impact their decisions in a recession. Analyzing the impact of business cycles on the incentives corporate managers face offers an interesting avenue for future research.