Berk and Green (2004) make a theoretical argument that performance persistence should not exist since new money flows into well-performing mutual funds and there are diseconomies of scale, or because successful funds capture excess returns by raising fees. We find that performance prediction continues when we examine samples of larger and larger funds and that past performance predicts future performance for holding periods up to three years. Funds that outperform index funds of the same risk can be identified. We find that expense ratios are lower for large funds, and decrease as funds get larger or perform well.

As a result of poor asset allocation decisions by 401(k) participants, 72% of all plans now offer target date funds, and participants heavily invest in them. Here, we study the characteristics and performance of TDFs, providing a unique view by employing data on TDFs holdings. We show that additional expenses charged by TDFs are largely offset by the low-cost share classes they hold, not normally open to their investors. Additionally, TDFs are very active in their allocation decisions and increasingly bet on nonstandard asset classes. However, TDFs do not earn alpha from timing or their selection of individual assets.