Are Active Managers A Drag On Investor Wealth?

Are Active Managers A Drag On Investor Wealth? Evidence From An Option-Based Estimation

University of Melbourne – Faculty of Business and Economics; Financial Research Network (FIRN)

Manchester Business School; New York University – Stern School of Business

October 29, 2015

Abstract:

We estimate an option-based value of a fund manager’s conditional market timing skill in bear market states. We combine this value with alpha based estimates of selection skill to give an overall valuation of active management. At the aggregate level, we estimate that the benefit arising from the option value of active fund management in bad times can be large enough to cover its unconditional overall cost. Our analysis suggests that by taking account of the option premium delivered by managers’ bear market timing skills, the longstanding mutual fund underperformance puzzle could be largely rationalized.

Are Active Managers A Drag On Investor Wealth? Evidence From An Option-Based Estimation – Introduction

Does the actively managed mutual fund industry destroy value? Would the typical investor’s overall welfare increase if his or her portfolio were moved to passively managed funds? A significant mutual fund literature documents negative alphas relative to passive benchmarks and show that the cost of active management appears to far exceed its benefit (e.g., French 2008, Lewellen 2011).1 These data seem to stand at odds with theory such as Berk and Green (2004), where equilibrium in efficient markets should imply aggregate expected alpha that is positive gross of fees and zero net of them.

On the other hand, Moskowitz (2000), Staal (2006) and Kosowski (2011) find that mutual fund managers tend to perform better than passive benchmarks in bad times, and argue that the value added by active management may have been largely underestimated. Kacperczyk, Nieuwerburgh and Veldkamp (2014) find that the managers who exhibit stock picking skills in expansions also show positive market timing abilities in recessions. Further, the model of Glode (2011) demonstrates how aggregate underperformance of mutual funds and the expected outperformance in bad states of the economy can arise simultaneously. The natural open question for research is one of magnitude: can estimates of the benefit of state-dependent performance in bad times be economically sufficient to rationalize active mutual fund investing?

We implement a novel approach to estimating the value of actively managed mutual funds’ conditional stock selection and market timing skills in bear states. Building on Merton (1981)’s insight on the call option character of a manager’s market timing ability and extending his formula, we estimate the value of a fund manager’s conditional performance as the sum of the option value of the manager’s timing skill in bear states plus the conditional alpha as a measure of selectivity skill.3 In contrast to recent evidence suggesting that the positive conditional performance is due to managers’ counter-cyclical stock selection skills (e.g., Kosowski 2011, Glode 2011), our estimates highlight the relative importance of the option value of market timing success. This result would be consistent with Kacperczyk, Nieuwerburgh and Veldkamp (2014).

The other significant research question whether this option-based valuation of enhanced performance in bad times can be large enough to compensate for the cost of lower unconditional alphas. To investigate, we compute option-adjusted alphas against typical passive k-factor benchmarks. The unadjusted aggregate 4-factor alpha of our sample is large and negative, ?0.08% per month (t-stat=?2.78). In stark contrast, we find the option-adjusted alpha is statistically indistinguishable from zero. This finding suggests that the benefit of the service provided by fund managers may actually fall in line with its cost, consistent with predictions of Moskowitz (2000) and Glode (2011).

These findings hold not only at the aggregate level but also at the portfolio level. Sorting funds by prior-year return into decile portfolios along the lines of Carhart (1997), we find that option-adjusted alphas for the ten deciles are not statistically different from zero. Moreover, the Gibbons-Ross-Shanken test (GRS-stat=0.55, p-value= 99%) suggests that the option-adjusted alphas for ten decile portfolios are jointly equal to zero.