Abnormal Returns on CEFs and in Pre-and-Post-Credit-Crunch Periods

Abstract

1. Introduction It is a well-documented fact that closed-end funds (CEFs) trade at significant discounts. According to a thorough literature review provided by Dimson and Minio-Kozerski (1999), previous research examined both economic and behavioural explanations regarding the existence of CEFs' discount. Examining data on CEFs, there are a number of studies employing cointegration analysis. Gemmill and Thomas (2002) provide evidence that cointegration analysis indicates a strong link between CEFs' discounts and retail flows. According to the findings by Cheng et al. (1994), Gemmill and Thomas (2002) and Copeland (2007), if there is cointegration between the price of shares and the Net Asset Value (NAV) of CEFs, it is not a relationship implying a zero long run discount. Gasbarro et al. (2003) found strong evidence of cointegration between the price of shares and the NAV of CEFs however; their results indicate that mean-reversion is caused by changes in both share prices and NAVs. Previous research has also shown that, there is an opportunity for abnormal returns realization through exploitation of movements in the level of the discount. According to some researchers, efficiency in CEFs can be examined through Initial Public Offerings. Khorana et al. (2002) and Higgins et al. (2002) report, correspondingly, that rights offerings and new issues of shares are announced when funds trade at a premium. However, evidence is in favor of such strategies only when the price decline is substantial. There are also a number of papers, such as Brauer (1984 and 1988) and Brickley and Schallheim (1985), examining the opportunity for abnormal returns when open-ending pushes share price of CEFs to their NAV. Considering discount-based strategies, prior empirical evidences suggest the potential of abnormal returns realization (Thompson, 1978; Anderson, 1986; Cheng et al., 1994; Cakici et al. 2000 and 2002). Pontiff (1995) suggested that discounts' mean reversion is responsible for the correlation between fund discounts and future returns. In addition, Sias et al. (2001) have shown that, there is potential for abnormal returns by exploiting the mean reverting properties of CEFs' discounts. Under the joint hypothesis of risk neutrality and market efficiency, time series of share prices characterized as random walk processes. It is a well documented fact that, examining stocks trading in liquid financial markets that operate on highly efficient trading platforms, most time series of log share prices have a unit root. Facing lack of mean-reversion property, we cannot apply statistical arbitrage strategies that rely upon unconditional variance in order to realize excess returns. However, examining CEFs, we have found that, under certain market conditions, the discount is often stationary. Still, just evidence of discount stationarity is not sufficient information in order to successfully apply statistical arbitrage strategies. That is, given that the discount is a spread between the share price and the NAV of CEFs, we should at first identify the long-run relation, if any, between these variables. Next, given that NAVs are not trading, we should verify that the error correction mechanism is such that, in the short run, the discount narrows/widens due to changes (increases/falls) in share prices. Simultaneously, we should examine if the cointegrating vector is (1, -1) implying stationarity of discount. Present topic concerns the examination of weak form market efficiency, employing data from CEFs trading in Athens Stock Exchange (ASE). Our research extends the existing literature by considering if market performance, in ASE, alternates the mean-reverting properties of CEFs' discount and as a result affects realization of abnormal returns on CEFs. Reported results indicate that, examining an equally weighted portfolio of funds, there is no cointegrating relation between share prices and NAVs during the recent turmoil period due to the credit crisis while; moderate market performance characterizing pre-credit-crunch period ensures the mean-reversion of CEFs' discount and points to cointegration between the examined variables. …