Do Institutional Investors Have Timing Ability? New Evidence from Daily Trades

Abstract

Mutual funds generate returns for their investors either through superior stock selection skill or through their ability to time the equity market. Despite the substantial extant literature on market timing ability of mutual funds that uses both returns-based and holdings-based measures, the evidence seems to be mixed. While the returns-based measures used in several studies (e.g., Treynor and Mazuy (1966), Henriksson and Merton (1981), Becker, Ferson, Myers, and Schill (1999), Jiang (2003)) generally find no timing ability or even negative timing ability among mutual funds, recent work by Jiang et al. (2007) using holdings-based measures has found evidence of positive timing ability in mutual funds. Holdings-based measures constructed using quarterly holdings have several advantages over the returns-based measures of timing (see Jiang et al. (2007)). However, since holdings-based measures rely on quarterly data, they suffer from several limitations. First, given that mutual fund managers are required to disclose their holdings on a quarterly basis, they can engage in window dressing (e.g., He, Ng, and Wang (2004)) that can distort the timing measures. Second, since quarterly data does not provide information about the intra-quarter trading, quarterly-holdings-based timing measures do not capture any short-term timing ability of mutual funds related to their intra-quarter trades. Goetzmann, Ingersoll, and Ivkovich (2000) show that if mutual funds engage in daily market timing, timing measures using data from longer horizon can result in underestimation of timing ability. Our study makes several contributions to the extant literature on timing ability of mutual funds by addressing the limitations of quarterly holdings data and providing new evidence through the use of high-frequency trade-by-trade data. First, we examine the timing ability at different frequencies by testing the correlation of daily trades with future market returns at different horizons ranging from the next day up to one year. This allows us to determine if mutual funds possess short-term or long-term timing ability. Second, since we have data on both buys and sells of mutual funds, our paper is the first to separately examine the timing ability of their buy and sell trades. It is possible that mutual funds have asymmetric timing ability on buys and sells since their trades can be affected by their being forced to buy and sell in response to investments and redemptions by their investors (Alexander, Cici, and Gibson, 2007). It is conceivable that positive and negative timing abilities across buys and sells can offset each other and therefore one may not observe any timing ability at the aggregate fund level. Interestingly, our preliminary results show that mutual funds have positive timing ability for sells and negative timing ability for buys. We further show that this pattern of timing ability can be partially explained by quarter-end month behavior, and capital flows. Finally, our study aims to examine the fund characteristics such as size, turnover, fees, portfolio concentration, and factor loadings that are associated with timing ability. Being the first study that examines the timing ability of mutual funds through their trades, our research provides important references for mutual fund researchers, managers, and investors. Despite the evidence on actively managed mutual funds underperforming their benchmarks, there continues to be significant investment in the active management industry. Our research sheds light on this puzzle by showing that mutual funds may be able to add value through their timing abilities provided they do not need to provide continuous liquidity to their investors.