The hazards of mutual fund underperformance: A Cox regression analysis

This paper investigates the process determining mutual funds' conditional probability of closure, i.e., their hazard function. Using a nonparametric approach to estimate the effects of a fund's age on its hazard rate, we find a distinctly non-linear, inverse U-shaped pattern in the relationship. Hence, young and very old funds are least likely to be closed down. A fund's relative performance and (less significantly) the level of return in the sector in which the fund operates are also identified as important factors in the closure decision. Results from semiparametric Cox regressions are compared with those from the discrete choice probit model used by Brown and Goetzmann Brown, S.J., Goetzmann, W., 1995. Performance persistence, Journal of Finance. Vol. 50, pp. 679–698]. Finally, we provide a complete summary of the fund attrition process by estimating the survivor function, indicating the proportion of funds that survive up to a given age, and we identify the effect of fund attrition on standard measures of persistence of fund performance.