Risk Shifting and Mutual Fund Performance

TL;DR: In this article, the authors investigated the performance consequences of risk shifting, as well as the economic motivations and the mechanisms for risk shifting using a holdings-based measure of risk shifts, and found that funds that increase risk perform worse than funds that keep stable risk levels over time.

Abstract: Mutual funds change their risk levels significantly over time This paper investigates the performance consequences of risk shifting, as well as the economic motivations and the mechanisms of risk shifting Using a holdings-based measure of risk shifting, we find that funds that increase risk perform worse than funds that keep stable risk levels over time In addition, funds that expect higher benefits from risk shifting are more likely to increase risk and perform particularly poorly after increasing risk Our results are consistent with the notion that agency problems, rather than the ability to take advantage of changing investment opportunities, are the likely motivation behind risk shifting behavior

Figures

Table 7: Performance Consequences Using Alternative Risk Shifting Measures This table reports the abnormal monthly returns of portfolios of mutual funds sorted according to the most recent risk shifting measure. We compute six different measures of risk shifting: the standard deviation of the fund returns using all holdings; the proportion of the portfolio invested in cash and other non-equity securities; the standard deviation of the returns using only equity positions; the systematic risk, defined as the CAPM beta; and the idiosyncratic risk, defined as the standard deviation of the residuals from the CAPM and Carhart regressions, respectively. All returns are expressed in % per month. Portfolios 1 to 5 are defined similarly to Table 2 with different cutoffs for each RS measure: {±1%,±2.5%} for all holdings; {±2.5%,±7.5%} for cash proportions; {±1%,±2%} for equity holdings; {±.05,±0.15} for CAPM beta; and {±1%,±2%} for both idiosyncratic risk measures. The significance levels are abbreviated with asterisks: One, two, and three asterisks denote significance at the 10, 5, and 1% level, respectively.

Table 3: Mechanisms of Risk Shifting. This table summarizes the initial levels (Panel A) and the changes (Panel B) in the fund characteristics that determine the total risk of a fund. The initial levels correspond to the average characteristics over the prior 36 months and the changes correspond to the difference between the most recent characteristics and the average characteristics over the prior 36 months according to the most recent risk shifting measure. Portfolios 1 to 5 are defined in Table 2 with cutoff RS measures of {±1%,±2.5%}.

Figure 1: Persistence of Risk Shifting. The figure depicts the average risk shifting level for the five risk shifting portfolios four quarters before and eight quarters after the portfolio formation.

Table 4: Performance Consequences of Risk Shifting This table reports the abnormal monthly returns of portfolios of mutual funds sorted according to the most recent risk shifting measure. The table summarizes the intercepts from factor regressions based on the CAPM, Fama-French, Carhart, and Ferson-Schadt models. All returns are expressed in % per month. Portfolios 1 to 5 are defined in Table 2 with cutoff RS measures of {±1%,±2.5%}. The significance levels are abbreviated with asterisks: One, two, and three asterisks denote significance at the 10, 5, and 1% level, respectively.

Table 5: Longer-Term Performance Consequences of Risk Shifting This table reports Carhart-adjusted monthly returns of portfolios of mutual funds sorted according to the prior risk shifting measure. The performance of the fund portfolios is computed based on the risk shifting measures of funds over the prior four quarters. All returns are expressed in % per month. Portfolios 1 to 5 are defined in Table 2 with cutoff RS measures of {±1%,±2.5%}. The significance levels are abbreviated with asterisks: One, two, and three asterisks denote significance at the 10, 5, and 1% level, respectively.

Table 1: Summary Statistics This table summarizes the characteristics of the mutual funds in our sample over the period between 1980 and 2006.

Figure 3: Cumulative Distribution Function. We report the cumulative distribution functions of the fund rankings for mutual funds in different risk shifting groups. We sort mutual funds into the five portfolios according to their most recent risk shifting measure. At the end of every quarter we compute the cumulative distribution function of the percentile ranking of funds based on the Carhart alphas over the subsequent 12-month period for each of the five portfolios. Finally, we depict the average levels of the quarterly cumulative distribution functions. The figure shows the cumulative distributions for three different groups of mutual funds. The curve RS1 corresponds to funds in portfolio (1) that reduce their risk by more than 2.5%, RS5 to funds that increase their risk by more than 2.5%, and RS3 for funds that change their risk by less than 1%. All returns are expressed in % per month.

Table 2 summarizes the characteristics of mutual funds sorted according to the RS measure. Funds in Portfolio 1 decrease risk on average by 5.60% per year, which is approximately 26% of their realized volatility over the prior 36 months. On the other hand, funds in Portfolio 5 increase risk by 4.75% per year, which is also approximately 26% of their prior realized volatility. Thus, funds exhibit significant changes in their overall risk levels over time.

Table 8: Performance of Funds by Risk Shifting and Other Fund Characteristics This table reports the frequency distribution and the Carhart alphas of portfolios of mutual funds sorted according to risk shifting and other fund characteristics that are related to trading costs and prior fund performance (turnover, prior year flow, prior year return, and realized volatility over prior 3 years). Mutual funds are first sorted into two equal-sized groups according to whether the characteristic is above or below its median value. In a second step, funds are further divided into five groups according to their risk shifting level. The table reports the average proportion of funds that fall into one of the ten groups and the Carhart alphas of the ten portfolios and the differences in the Carhart alphas between selected portfolios. All returns are expressed in % per month. Portfolios 1 to 5 are defined in Table 2 with cutoff RS measures of {±1%,±2.5%}. The significance levels are abbreviated with asterisks: One, two, and three asterisks denote significance at the 10, 5, and 1% level, respectively.

Table 6: Performance of Funds with Different Risk Shifting Incentives This table reports the frequency distribution and the future Carhart alphas of portfolios of mutual funds sorted according to risk shifting and fund characteristics that are related to risk shifting incentives (expense ratio, age, size, and fund family size). Mutual funds are first sorted into two equal-sized groups according to whether the characteristic is above or below its median value. In a second step, funds are further divided into five groups according to their risk shifting level. The table reports the average proportion of funds that fall into one of the ten groups and the Carhart alphas of the ten portfolios and the differences in the future Carhart alphas between selected portfolios. All returns are expressed in % per month. Portfolios 1 to 5 are defined in Table 2 with cutoff RS measures of {±1%,±2.5%}. The significance levels are abbreviated with asterisks: One, two, and three asterisks denote significance at the 10, 5, and 1% level, respectively.

Figure 2: Future Performance of Portfolios by Risk Shifting Measure. This figure depicts the abnormal monthly returns of portfolios of mutual funds sorted according to the most recent risk shifting measure. The figure summarizes the intercepts from factor regressions based on the CAPM, Fama-French, Carhart, and Ferson-Schadt models. All returns are expressed in % per month.