We investigate the performance of mutual funds that trade using private information. These funds are uniquely identified from a set of 2730 funds with 44,315 fund-periods between 1994 and 2005. We compare the alignment of fund trades with brokers’ recommendations, which we regard as “public information” in the universe of informed and uninformed mutual funds. Funds that systematically trade counter to the public information form a homogenous subset of the privately informed funds. By using private information that contradicts the public information, these funds exhibit a superior average performance. After we control for serial correlation in fund returns, we assess this advantage as being an economically significant 1.7% per annum. We also show empirically that smaller funds are better able to capture the benefit of private information.

An empirical issue is whether a mutual fund’s change in intertemporal risk is intentional or arises from risk mean reversion. Our methodology uses actual fund trades to identify funds that actively change risk. Funds that are statistically identified as trading to change return variance or tracking error variance do not exhibit risk mean reversion. Mostly, funds trade to reduce risk and, in particular, tracking error variance. This is most evident for funds that previously attained a low tracking error variance. We find no evidence of a relation between past performance and intended changes to return variance or tracking error variance.