• Selected
• Entire List

Abstract:
I investigate the impact of returns on broad styles, such as growth funds, on mutual fund flows. I test whether mutual fund investors pursue styles, as predicted by the style investing hypothesis of Barberis and Shleifer (2003). Although in the aggregate, style-level flows to style categories are positively (negatively) related to past returns on the given category (other categories), at the individual fund level this pattern disappears. In fact, after controlling for fund returns, flows are negatively related to style performance. Such patterns persist for three different style classifications I consider here. The findings go against the hypothesis of style investing, and are consistent with within-style return chasing and evaluating fund managers based on both fund-level and style-level returns.

Abstract:
This paper studies the optimal asset allocation to a security whose anomalous returns are possibly disappearing over time. We apply our framework to two anomalies, positive stock index autocorrelations and the January effect. We summarize the empirical evidence on these phenomena from the point of view of investors with differing beliefs on the behavior of the anomaly. Our results suggest that anomalies decay over time, and that accounting for this decay has a large effect on optimal portfolio weights. Somewhat surprisingly, we find that priors that downweight the historical magnitude of an anomaly can sometimes increase the future allocations to it.

anomalies, Bayesian analysis, the January effect, return autocorrelation, asset allocation

Abstract:
Mutual fund portfolio decisions depend on previous trades of other funds with outstanding past performance: funds mimic the mutual fund industry leaders. Controlling for other factors, for each dollar leaders invest in a given stock, the follower funds invest 15 to 30 cents in the subsequent quarter. Funds with poor past performance mimic more than funds with moderate performance; there is also some evidence that small and young funds mimic more than old and large ones. Leader trades that are likely to convey more information (more extreme trades, trades in more opaque stocks) elicit stronger responses. Mimicking makes economic sense: portfolios based on past trades of the leaders deliver significant 3 and 4 factor alphas of the order of 0.35% per month. Characteristics-matched returns indicate that a large portion of this abnormal performance can be attributed to stock momentum, which suggests a possible link between follow the leader behavior and momentum.

mutual funds, portfolio choice, information acquisition, momentum

Abstract:
In this paper, I identify "best idea" trades of fund managers. These trades account for about 30% of fund volume and outperform benchmarks and other fund trades by as much as 47 basis points per month. I identify such trades using an ex ante proxy that does not rely on any performance-related variables and is instead based on common trades of managers with similar information sets. The performance of best ideas is not explained by herding, short-term liquidity pressures, changes in the index composition, or funds' reaction to analyst revisions, and does not revert in the long term. The remaining fund trades (not best ideas) fail to beat passive benchmarks even before expenses, e.g., their average characteristic-adjusted return is statistically insignificant 0.03% per month. Finally, best ideas also improve funds' after-fee returns: Funds which participate in best idea trades beat other funds by up to 0.3% per quarter.

mutual funds, management companies, portfolio choice, skill

Abstract:
We study bargaining at the end of high-stakes poker tournaments, in which participants often negotiate a division of the prize money rather than bear the risk of playing the game until the end. This setting is ideal for studying bargaining: the stakes are substantial, there are no restrictions on the negotiations or the terms of a deal, outside options are clearly defined, there are no agency conflicts, and there is little private information. Even in this setting, we find that risk-reducing deals often are not completed or even proposed. As expected, we find that players are more likely to negotiate when the gains to trade are large and when the coordination costs are lower. Surprisingly, although the likelihood of a successful deal is increasing in the stakes, this relation is driven only by the tournaments with the very largest prizes. It is also puzzling that the success of a proposal depends on who makes it, but initiating a proposal does not affect the proposing player's payoff in a completed deal. Divisions of prizes are closely related to players' outside options, while at the same time one of two focal points are often chosen. We also find intriguing differences between two-player deals and deals with three or more players.

Bargaining, Negotiations, Poker

Abstract:
We study whether the success of bargaining and the agreed upon terms depend on the characteristics of the person who initiates negotiations ("the initiator"). We approach this question in the context of high-stakes online poker tournaments, in which participants often negotiate a division of the prize money rather than risk playing until the end. Although initiators typically are in a worse than average position and are less well known, negotiations initiated by better known and better performing agents are more likely to lead to an agreement. This would suggest that gains to trade depend on who the initiator is, but, surprisingly, initiating bargaining does not affect the initiator's payoff in a completed deal.

Additionally, we find strong evidence in support of Cramton, Gibbons, and Klemperer (1987), who argue that for bargaining to succeed the parties' stakes in an enterprise must be close to equal.

bargaining, negotiations, poker