Representative investors versus best clienteles: Performance evaluation disagreement in mutual funds

This paper develops a diagnostic tool for candidate performance measures that accounts for investor disagreement in mutual funds. We compare the evaluation for best clienteles, specified by an upper admissible performance bound, to the one for representative investors implicit in eleven models. The results show that linear factor models misrepresent best clientele alphas, with a disagreement that relates to fund characteristics. Consumption-based alphas are generally inadmissible. The manipulation-proof performance measure generates alphas that are sensitive to its risk aversion parameter and lack statistical precision. However, a reasonable parameter gives admissible values that reflect the alphas for the most favorable clienteles. • Paper develops and implements a diagnostic tool for candidate performance models that accounts for investor disagreement. • Tool is based on a comparison between commonly used performance measures and best clientele alphas. • Empirical results shows that most models misrepresent the value of mutual funds for favorable clienteles. • Disagreement is higher for funds with higher expenses, higher turnover, more longevity, older managers and smaller size