This dissertation contains three essays on the structure of hedge funds and the behavior of hedge fund managers and investors. In the first, I study the dynamics of changes in the leverage choices made by hedge fund managers and the implications for investors. Using SEC and commercial hedge fund database information, I examine the time-series and cross section of hedge fund leverage and find that, overall, hedge fund leverage changes are not driven by certain return-driven motivations, such as levering up before profitable times or in order to take advantage of concentrated investment opportunities. Instead, hedge fund leverage changes are driven primarily by risk mitigation measures, which may be imposed by external actors, such as prime brokers. Hedge funds lever down during volatile times and when risk spreads are high. Interestingly, when hedge fund leverage is perturbed by fund flows, I find no evidence that managers trade to return to a target leverage within four quarters. Thus overall hedge fund leverage may be the result of the fund's flow history, rather than predetermined leverage targets. In the second study, I examine hedge fund inceptions. New hedge funds may come into being because managers observe high demand for certain types of funds and create them to absorb this demand, or alternatively as a result of an innovative investment idea. I create empirical proxies that allow me to distinguish the two types of inceptions and show that funds that came about because of the supply of managerial investment from those that came about because of investor demand. Inceptions of the former type outperform those of the latter by a significant 4 to 5% annually over the first five years. In my third study, I use a special environment, a platform of hedge fund separate accounts, as a laboratory to examine the role of share restrictions and third-party evaluation in hedge fund returns. Separate accounts are tied to existing funds but have much lower share restrictions and feature third-party return evaluation. By comparing separate account and main fund returns for the same funds, I find that a reduction in share restrictions leads to a performance penalty of 1.7% per year. Also, the high liquidity and third party evaluation leads to reported returns that feature 33% less serial correlation. This latter result suggests that managers in the main fund use discretion in reporting practices to induce serial correlation in the reported returns of their main fund, which would lead to artificially good performance in risk-adjusted evaluation metrics for those funds.