Implied Measures of Relative Fund Performance∗

We provide a methodology to examine the selection and market timing ability of fund managers. Conditional on portfolio weights in distinct investment classes, implied metrics for these two attributes are inferred from a cross-section of fund returns. Consequently, our procedure measures relative performance and circumvents the need to specify benchmark returns or peer funds. This technique is also robust to herding as well as window dressing, and mitigates survivorship bias over longer horizons. Empirically, the conditional infor-mation contained in classes defined by industry sectors, assets and geographical regions is of critical importance to fund evaluation. For each set of portfolio characteristics, we identify fund managers with moderate success at either selecting securities or timing-the-market. However, individual funds cannot overperform across all three criteria and evidence of investment skill dissipates at a higher performance threshold. ∗We thank Morningstar for providing us with the required data. In particular, the efforts of Stephen Murphy at Morningstar are greatly appreciated. The comments of Sheridan Titman significantly improved an earlier version of this paper. All remaining errors are our own. We also acknowledge the generous assistance provide