Jeffrey A. Busse, Jing Ding, Lei Jiang, Ke Wu

Read Full Paper

This paper uses the dynamic conditional correlation (DCC) model to estimate daily mutual fund betas, providing a more accurate reflection of changes in fund risk due to daily trading activity. The study finds significant evidence of market timing ability among actively managed U.S. equity funds, which is not captured by traditional approaches. Unlike traditional models, the daily beta estimates positively correlate with fund performance, particularly during down markets, where successful timers exhibit lower downside risk.

The study contrasts three daily beta measures: the DCC model, the constant conditional correlation (CCC) model, and rolling OLS regressions. The DCC model offers superior accuracy by incorporating time-varying volatility and correlation, while the CCC model only accounts for time-varying volatility. The DCC model’s ability to capture dynamic correlation between fund and market returns enhances its predictive accuracy over rolling-window betas.

Using daily fund returns instead of monthly estimates, the study employs a two-stage approach to measure market timing ability. This method, in line with Jiang, Yao, and Yu (2007), first estimates beta and then assesses its relation to market returns, which improves the detection of timing ability. The paper finds that successful timing is positively correlated with fund performance and is associated with lower downside risk. Furthermore, funds with high timing skill exhibit persistence in their timing ability over time and attract investor flows.

This research contributes to the literature on mutual fund market timing by demonstrating the effectiveness of daily betas in identifying timing ability, which traditional monthly models fail to capture. It also provides robust evidence linking market timing skill with performance and investor behavior.