Search

In this dissertation, I use empirical methods to examine the relative performance of hedge funds, as well as their capacity constraints. The research is motivated by the increasingly high allocation of assets to hedge funds combined with limited regulation of the industry, making it an important area for academic research. By introducing new methods to examine hedge fund performance and capacity constraints, I aim to add new knowledge to the increasing discussion by industry participants and financial commentators and to provide novel insight for further research. This research takes the form of three studies. The first study seeks to identify which investor types are most informed. Specifically, I examine the informativeness of quarterly disclosed portfolio holdings across four institutional investor types: hedge funds, mutual funds, pension funds, and private banking firms. I find that overweight positions outperform underweight positions only for hedge funds. By decomposing holdings and stock returns, I find that hedge funds are superior to other institutional investors; both at picking industries and stocks, and that they are better at forecasting long and short-term returns. The second study, proposes a novel method to investigate capacity constraints in the hedge fund industry. I introduce the concept of cohort size, which is measured by the total assets of all hedge funds applying similar strategies. Together, these funds impact on opportunity and execution costs, so that the total cohort size, rather than simply the individual fund size, is associated with fund performance. The study finds cohort size to be negatively related to future quarterly returns. Finally, I introduce the cohort model, used to assess relative hedge fund manager skill. The model tested uses the correlation of monthly returns to locate cohorts, and forms peer benchmarksby averaging returns across the cohort. The advantages of the cohort model are that it is able to address the omitted variable problem present in factor models, and it is better able to disaggregate skill from factor exposures common to particular investment strategies. Consistent with improved identification of manager skill, cohort alpha shows stronger persistence over longer horizons.