ON THE CLOSED-END FUNDS DISCOUNTS/PREMIUMS IN THE CONTEXT OF THE INVESTOR SENTIMENT THEORY

Abstract

The existence of closed-end funds discounts/premiums, although an issue largely studied, it is still puzzling both academics as well as practitioners. As it is well known, they result from the difference between the value of the shares of the fund, determined by the market, and their net asset value (the market value of the securities held by the fund, less the liabilities). Taking into account that the closed-end fund shares are traded on the stock exchange, as well as the assets included on their portfolios, no discrepancies would be expected (at least theoretically) between the market value of the funds and their net asset values, since the market should be able to adjust and correct the prices, due to the fact that the information is widely diffused.In attempt to explain this “puzzle” several theories have been suggested. On one hand, those based on rational factors, such as: potential tax liabilities due to unrealized capital gains, the dividend policy, the fund portfolio composition, agency costs and management performance and, on the other hand, those based on behavioral factors, such as the investor sentiment theory. This latter framework, at least theoretically is, in our view, the one that seems to better explain almost all the features of the “puzzle”, trying not only to explain the existence of discounts but also the existence of premiums and their behavior among the funds themselves and over time.In this context, we developed our research trying to explain the existence and persistence of the discounts/premiums. We also investigate the correlation between the discounts/premiums of those funds among themselves and each other over time, the mean reversion of the discounts/premiums, as well as the predictability power of the fund shares and of the net asset value returns. It was also our objective to search for the relevance of the investor sentiment theory in order to explain the discounts/premiums, so that we used Brauer's (1993) methodology and the signal extraction technique of French and Roll (1986).We also carried out (as far as we know, for the first time) a panel data analysis in order to check how much of the discounts/premiums variability is due to the presence of “noise traders”.This research was based on a sample of North-American closed-end funds, which invest mainly on stocks and/or bonds traded on the NYSE or on the AMEX, during the period from January 1987 to June 1999 (inclusive). The data was collected from the Wiesenberger database.From the results that we got, we noticed that there seems to exist an indication of the presence of “noise traders” on the closed-end funds market which, in turn, seems to confirm the assumptions of the investor sentiment theory: the discounts/premiums were positively correlated, were mean reverted and had some predictability power in terms of fund share returns but not so much in relation to their net asset value returns. Nevertheless, we observed that the estimated proportion of the variance of standardized weekly discounts changes, explained by the investor sentiment on the total period studied, was only 8.6%. Also, the results from the panel data analysis seem to suggest the relatively low importance of the investor sentiment theory to explain those discounts/premiums.