A Kalman filter approach to Chinese mutual funds' market timing abilities

Abstract

This paper examines the Kalman filter model’s abilities to capture the market timing skills of Chinese mutual fund managers. The Kalman filter approach provides a useful measure of timing skills since it allows for estimation of time series of portfolio beta. Using this measure, I compare the performance of the Kalman filter to that of the OLS-based market timing models developed by Treynor and Mazuy (1966) and Henriksson and Merton (1981). The major finding is that the Kalman filter model produces the least false positives among them. The OLS-based models generate the false positives at too high rate. The conclusion coincides with that of Mamaysky et al. (2008) who develops the Kalman filter approach to examine the timing abilities of the U.S. mutual funds. The Kalman filter detects the significant market timing behavior in Chinese mutual funds during the periods 2006-2010 and 2014-2018.