| Summary: | My dissertation consists of three chapters covering mutual fund terminations, survivorship bias, and smart money effect, respectively. The first chapter examines domestic equity fund terminations through the determinants of the hazard function. It finds that the lagged 3-month return, the standard deviation of return, the Sharpe ratio, and the price to book ratio best explain equity fund closures. Due to the interaction between time and the lagged 3-month return, we reject the proportionality assumption of the semi-parametric Cox model, thus invalidating the use of this model to estimate the hazard function of mutual funds. In addition, we find that different fund categories exhibit distinct hazard functions. The second chapter revisits survivorship bias within domestic equity mutual funds from the middle of 1998 to the middle of 2004. Using Morningstar Principia and tracking both disappearing funds and name changes, we report a statistically significant difference in performance between the complete sample and the survivorship-biased sample, both in raw returns and in risk-adjusted returns. We correlate this potential for a disproportional weighting of better performing funds to market conditions and show that changes in stock market indexes and interest rates have a bearing on the level of fund survivorship bias. The third chapter reexamines the relationship between mutual fund performance and new cash into the fund. We first estimate the major determinants of fund flows and performance cross-sectionally. Then controlling for both size and style, we show that positive new cash flow portfolios have significant risk-adjusted returns in the subsequent period. Granger causality test demonstrates that a causal relationship exists between fund performance and new cash flows for several groups of funds. Specifically, we find that fund performance Granger causes new cash flows and therefore find a "naïve" money explanation rather than a "smart" money explanation.
|
|---|
