Closed-End Funds and Turnover Restrictions

Past studies have found that investors can earn higher returns than a benchmark by purchasing shares of closed-end funds with discounts or selling shares with premiums. These studies either ignored the impact of transaction costs or used equally weighted portfolio strategies without controls on turnover or transaction costs. We examined whether constraining the holdings of individual funds and turnover has any bearing on the excess returns earned by closed-end equity funds over a benchmark return. We found that when transaction costs were low, portfolios with frequent rebalancing and loose turnover constraints outperformed the benchmark and other portfolios in the period we studied. We found, in contrast, that when transaction costs were moderate to high, portfolios with less-frequent rebalancing and tight turnover constraints outperformed the benchmark and other portfolios. The implication for the portfolio manager is that excess returns may be achieved in a variety of trading-cost environments with the proper mix of policy variables. A popular view is that closed-end funds should be purchased for their discounts. If a closed-end fund trades at a discount, an investor can purchase shares at a price below net asset value or, at least, purchase shares cheaply. Selling these shares less cheaply at a later date can lead to excess returns.This view is supported by empirical findings. A number of studies have shown that managed portfolios of closed-end funds can outperform a benchmark when discounts and premiums form the basis for fund selection. One recent study found that outperformance could be achieved even in an environment of high transaction costs by selling short the funds with the deepest discounts. These studies did not control turnover and holdings of funds, however, which is an important practical issue that we attempt to address in this article.We examined whether constraining the amount of turnover and holdings of funds, thereby controlling trading costs, has any bearing on the excess returns earned by a managed portfolio of closed-end funds. We used an expected-return-optimization model that selects portfolios of closed-end funds in an environment defined by transaction costs. The expected return of each fund depends on its current discount. Our methodology allowed us to control trading costs by tightening or loosening the turnover constraints together with limiting the holdings of each fund.We back tested the model for the period January 11, 1991, to September 1, 2000, for various holding periods ranging from 1 to 13 weeks and various transaction-cost environments ranging from 0 to 4 percent rates. The data consisted of U.S.- and U.K.-traded closed-end equity funds. The sample contained 128 funds at the start in 1991 and ended with 206 funds in 2000. The benchmark was a capitalization-weighted portfolio of all funds in the sample.We found that portfolios with frequent rebalancing and loose constraints on turnover outperformed the benchmark and other portfolios when transaction costs were low. Lengthening the time between rebalancing and constraining the amount of turnover evidently prevent the model from capitalizing on the opportunities available in discounts. We also found, surprisingly, that when transaction costs were moderate to high, portfolios with less-frequent rebalancing and more-stringent restrictions on turnover outperformed the benchmark and other portfolios. In this case, constraints serve to protect the model from “wrong” decisions (i.e., being too aggressive on funds with high expected returns that subsequently produce low actual returns). The penalties for these decisions are more severe when transaction costs are high.The practical implication of the second finding for portfolio managers is that in an environment of high transaction costs, excess returns on discounted funds can still be achieved but only with the proper mix of policy variables. Specifically, the best-performing strategies in an environment of moderate-to-high (3–4 percent rate) transaction costs consist of tight constraints on turnover (about once a year) and infrequent rebalancing (about a month and a half between re-optimizations). Thus, not only is restricting turnover important when transaction costs are high, so also is limiting the frequency at which the model can reach “incorrect” decisions. This finding is in sharp contrast to the best mix of policy variables used in a low-transaction-cost environment.