An agency explanation of the closed-end discount

The difference between the market price of shares of closed-end investment companies and the net asset value of those shares is called the closed-end discount. A number of previous researchers have modeled the closed-end discount as a function of various economic factors and have limited success in explaining the variation of discounts. This dissertation combines the best of the previous explanations with some additional agency variables to form a two-step econometric model that explains the closed-end discount. The theory developed in this dissertation is based on the fact that many closed-end funds have provisions in their corporate charters that allow the funds to be protected from takeovers and liquidations. It is hypothesized that these provisions, which are agency factors, can affect the discount through two separate channels. First, the agency factors can affect fund expense ratios and fund turnover ratios, which can affect the discount. Second, the agency factors can affect the probability of "open-ending" a closed-end fund by means of a takeover. A two-step model is formulated and tested. In the first step, expense ratios and turnover ratios are modeled as a function of the agency factors and a set of additional variables. It is shown that the agency factors affect turnover ratios, but that the direction of the effects are counter to what was hypothesized. The agency factors had no effect on expense ratios. A modification to the original model was tested with results that were similar to the original model. The second step of the model used forecast residuals from the first step as measures of "excess" expenses and "excess" turnover, along with the agency factors and a set of economic variables, to explain fund discount in 1989. Excess expenses affect the discount in the hypothesized direction, but very little success was obtained with the agency factors. Modifications to the model showed that some of the agency factors were significant in explaining the discount and that the model may need to be reformulated before strong support for the agency factors can be found.