The authors explore performance persistence in mutual funds using absolute and relative benchmarks. Their sample, largely free of survivorship bias, indicates that relative risk-adjusted performance of mutual funds persists; however, persistence is mostly due to funds that lag the S&P 500. A profit analysis indicates that poor performance increases the probability of disappearance. A year-by-year decomposition of the persistence effect demonstrates that the relative performance pattern depends upon the time period observed and it is correlated across managers. Consequently, it is due to a common strategy that is not captured by standard stylistic categories or risk adjustment procedures.

We explore performance persistence in mutual funds using absolute and relative benchmarks. Our sample, largely free of survivorship bias, indicates that relative risk‐adjusted performance of mutual funds persists; however, persistence is mostly due to funds that lag the S&P 500. A probit analysis indicates that poor performance increases the probability of disappearance. A year‐by‐year decomposition of the persistence effect demonstrates that the relative performance pattern depends upon the time period observed, and it is correlated across managers. Consequently, it is due to a common strategy that is not captured by standard stylistic categories or risk adjustment procedures.

We explore performance persistence in mutual funds using absolute and relative benchmarks. Our sample, largely free of survivorship bias, indicates that relative risk-adjusted performance of mutual funds persists; however, persistence is mostly due to funds that lag the S&P 500. A probit analysis indicates that poor performance increases the probability of disappearance. A year-by-year decomposition of the persistence effect demonstrates that the relative performance pattern depends upon the time period observed, and it is correlated across managers. Consequently, it is due to a common strategy that is not captured by standard stylistic categories or risk adjustment procedures.

Empirical analysis of rates of return in finance implicitly condition on the security surviving into the sample. We investigate the implications of such conditioning on the time series of rates of return. In general this conditioning induces a spurious relationship between observed return and total risk for those securities that survive to be included in the sample. This result has immediate implications for the equity premium puzzle. We show how these results apply to other outstanding problems of empirical finance. Long-term autocorrelation studies focus on the statistical relation between successive holding period returns, where the holding period is of possibly extensive duration. If the equity market survives, then we find that average return in the beginning is higher than average return near the end of the time period. For this reason, statistical measures of long-term dependence are typically biased towards the rejection of a random walk. The result also has implications for event studies. There is a strong association between the magnitude of an earnings announcement and the postannouncement performance of the equity. This might be explained in part as an artefact of the stock price performance of firms in financial distress that survive an earnings announcement. The final example considers stock split studies. In this analysis we implicitly exclude securities whose price on announcement is less than the prior average stock price. We apply our results to this case, and find that the condition that the security forms part of our positive stock split sample suffices to explain the upward trend in event-related cumulated excess return in the preannouncement period.

Empirical analysis of rates of return in finance implicitly condition on the security surviving into the sample. We investigate the implications of such conditioning on the time series of rates of return. In general this conditioning induces a spurious relationship between observed return and total risk for those securities that survive to be included in the sample. This result has immediate implications for the equity premium puzzle. We show how these results apply to other outstanding problems of empirical finance. Long‐term autocorrelation studies focus on the statistical relation between successive holding period returns, where the holding period is of possibly extensive duration. If the equity market survives, then we find that average return in the beginning is higher than average return near the end of the time period. For this reason, statistical measures of long‐term dependence are typically biased towards the rejection of a random walk. The result also has implications for event studies. There is a strong association between the magnitude of an earnings announcement and the postannouncement performance of the equity. This might be explained in part as an artefact of the stock price performance of firms in financial distress that survive an earnings announcement. The final example considers stock split studies. In this analysis we implicitly exclude securities whose price on announcement is less than the prior average stock price. We apply our results to this case, and find that the condition that the security forms part of our positive stock split sample suffices to explain the upward trend in event‐related cumulated excess return in the preannouncement period.