Timing Is Money: The Factor Timing Ability Of Hedge Fund Managers
VU vrije universiteit Amsterdam
VU University Amsterdam; Tinbergen Institute – Tinbergen Institute Amsterdam (TIA)
July 18, 2016
This paper studies the level, determinants, and persistence of the factor timing ability of hedge fund managers. We find strong evidence in favor of factor timing ability at the aggregate level, although we find ample variation in timing skills across investment styles and factors at the fund level. Our cross-sectional analysis shows that better factor timing skills are related to funds that are younger, smaller, have higher incentive fees, have a smaller restriction period, and make use of leverage. An out-of-sample test shows that factor timing is persistent. Specifically, the top factor timing funds outperform the bottom factor timing funds with a significant 1% per annum. This constitutes 13% of the overall performance persistence in hedge funds. The findings are robust to the use of an alternative model, alternative factors, and controlling for the use of derivatives, public information, and fund size.
Timing Is Money: The Factor Timing Ability Of Hedge Fund Managers – Introduction
Performance of the top hedge funds is persistent and cannot be explained by luck or sample variability (Kosowski, Naik and Teo, 2007). These findings are consistent with Jagannathan, Malakhov and Novikov (2010), who find that a portfolio of the top 33% of funds ranked on historical alpha t-statistics maintained its alpha in an out-of-sample test. Furthermore, they find little evidence for performance persistence among inferior funds, which support the interpretation that top funds have superior managerial talent and skills. In contrast to the results in hedge funds, Pastor and Stambaugh (2002) find that for the majority alphas of equity mutual funds are negative. The difference in performance (persistence) between hedge funds and mutual funds might be explained by the incentive fees for hedge fund managers, which enables hedge funds to attract more talented and skilled managers. Another possible explanation for the difference in performance can be related to the fact that hedge funds are less strictly regulated and therefore are more flexible to engage in, for example, short selling and leverage, and are better positioned to adjust their beta exposures to forecasts. The latter can be referred to as the timing ability of fund managers. In this paper, we study the extent of factor timing by hedge funds, its determinants, as well as its persistence.
It is well known that hedge funds employ dynamic strategies and have time-varying beta exposures on factors; see Fung and Hsieh (1997). Several studies followed examining the drivers of these dynamics. Patton and Ramadorai (2010) find that managers condition exposures on leverage, carry trade, as well as equity markets conditions. Fung et al. (2008) use a continuous rolling regression approach and find significant time variation in betas. Bollen and Whaley (2009) on the other hand study whether there are discrete changes in factor loadings. Schauten, Willemstein, and Zwinkels (2015) study whether the dynamic factor loadings can be explained by positive feedback dynamics.
We extend the literature in three important ways by combining factor investing with dynamic strategies. First, Chen and Liang (2007), among others, have shown that hedge funds possess market timing skills. Our study extends their paper and investigates whether hedge fund managers are able to time the Fung-Hsieh (2004) factors. Second, we examine the determinants of the timing ability of hedge fund managers. Ackermann et al (1999) and Liang (1999) find that hedge funds’ fee structure is a determinant of their performance. We extend their line of thinking and look into the cross-sectional determinants of factor timing skills. Finally, this study also tests whether factor timing skills can be a useful tool for investors to select hedge funds. Although results vary, a majority of studies finds a certain level of performance persistence for hedge funds; see e.g. Agarwal and Naik (2000). We extend this line of research by examining whether the performance persistence in hedge funds can be explained by persistence in market timing skills.
Our final data sample from Lipper TASS contains 3,124 hedge funds spread over nine investment styles over the period January 1994 to April 2014. We use the factor model of Fung and Hsieh (2004) to examine the factor timing skills of hedge fund managers at both an aggregate level and an individual fund level. The timing measure we employ is an extension of the famous Treynor and Mazuy (1966) measure.
Our main findings are as follows. At the aggregate level we find that hedge fund managers do possess factor timing skills, consistent with existing studies looking at market timing skills (Agarwal and Naik, 2000). We find especially strong timing skills for the market, size, and bond factors. Interestingly, we find substantially negative timing for the Emerging Markets factor. This might be caused by herding behavior of institutional investors in emerging markets, possibly due to information asymmetry (see Chloe, Kho, and Stulz, 1999).
At the individual level, the results suggest that funds with different investment styles show ample variation in timing skills on the different factors. The timing skills, though, appear not to be directly related to the style. Global macro funds, funds of funds, and long short equity funds show relatively more significant timing skills on the market factor than the funds of the other investment styles. Regarding the size factor, we find relatively more timing skills in convertible arbitrage funds, event driven funds, emerging market funds, and multi-strategy funds. Furthermore, only convertible arbitrage funds show evidence of emerging market factor timing skills. Finally, emerging market funds show the highest percentage of funds with timing skills on the credit-spread factor.
The results on the determinants of the factor timing ability of hedge fund managers suggest that better factor timing skills are related to funds that are younger, smaller, have higher incentive fees, have a shorter restriction period, and make use of leverage. There is, however, again quite some variation over the factors. For example, age and incentive fees are positively related to timing skills for the market and size factors.
Finally, we find that the factor timing skills of hedge fund managers have a certain degree of persistence. Whereas a long-short strategy based on the Fung-Hsieh alpha yields 7.2% annually, we find a significant out-of-sample alpha of almost 1% annually in the spread between the best and worst factor timing funds. In other words, approximately 13% of hedge fund outperformance is driven by factor timing, suggesting that the remaining 87% is driven by asset selection skills.
See full PDF below.