Test for Market Timing Using Daily Fund Returns

Using daily mutual fund returns to estimate market timing, some econometric issues, including heteroscedasticity, correlated errors, and heavy tails, make the traditional least-squares estimate in Treynor-Mazuy and Henriksson-Merton models biased and severely distort the t-test size. Using ARMA-GARCH models, weighted least-squares estimate to ensure a normal limit, and random weighted bootstrap method to quantify uncertainty, we find more funds with positive timing ability than the Newey-West t-test. Empirical evidence indicates that funds with perverse timing ability have high fund turnovers and funds tradeoff between timing and stock picking skills.