An analysis of closed-end funds discounts viewed from a lack of redemption perspective

CEF funds are invested in a portfolio of assets and shareholders are entitled to future cash flows generated from the portfolio. Investors cannot directly sell (buy) the fund shares back to (from) investment companies but can trade shares in secondary markets. Details of the CEF’s portfolio are publicly disclosed quarterly and the NAV is computed daily and reported in the financial press.

The discount is the difference between the exchange-traded share price and the underlying NAV and is computed as ln(Price/NAV). The CEF trades at a discount when the NAV exceeds the share price and at a premium when the share price exceeds the NAV. We refer to the discount as 12% when the computed value of the discount is −12%. Premium is treated as a negative discount.

The effect of investor sentiment on asset prices is a widely debated issue. Lee et al. (1991) propose investor sentiment as they argue that tax or illiquidity explanations are inadequate to explain the size and time variation in CEF discounts. LST and others find evidence supporting the investor sentiment hypothesis (see Chopra et al. 1993; Bodurtha et al. 1995; Baker and Wurgler 2006, 2007; Hwang 2011). Other studies, however, provide no evidence supporting the investor sentiments hypothesis (see Chen et al. 1993; Lemmon and Portniaguina 2006; Qui and Welch 2006; Cherkes et al. 2009). Pontiff (1996) posits that high arbitrage costs prevent arbitrageurs from bringing the CEF share prices in line with their NAVs. Gemmill and Thomas (2002) provide evidence that the behavior of CEF discount in UK is consistent with costly arbitrage but not with investor sentiment hypothesis. In addition to investor sentiments and transactions costs a number of other hypotheses for CEF discounts are offered in the literature including: liquidity, agency, signaling, taxes, dividends, and managerial ability.

The CEF discount can be viewed as the sum of the present value of all future dead-weight loss due to agency problems. CEFs with entrenched manager tend to have high degree of agency problem and high discount (Barclay et al. 1993). As the CEF discount widens, potential arbitrage profit motivates arbitrageurs to acquire and liquidate the CEFs.

The agency costs hypothesis argues that the CEF discount makes up the difference between the managerial expense charges and the benefit from unobservable management ability (Berk and Stanton 2007). If the managerial expense exceeds the benefits due to managerial ability the CEF may trade at a discount. CEFs with higher agency problem may exhibit larger discount but the CEF managers may act to limit the discount by adopting a managed dividend policy so as to minimize the potential for arbitrage (Wang et al. 2011). Bradley et al. (2010) suggest that the CEF discount is endogenously related to shareholder activism. Funds with large discounts are more susceptible to activists who can profit from buying and liquidating the funds. Since the level of activism can be expected ex ante the market prices reflecting such expectation could shrink the discount. Wermers et al. (2008) show that poor CEF performance helps predict replacement of the fund manager and that the market’s expectation of management replacement would lead to a reduction in the fund discount.

Our use of the CRSP index is reasonable as CEF share price responds to broad market movements. Our results are robust with regard to the choice of value or equal weighting. There is the possibility of an omitted index in the context of international funds and bond funds by using the CRSP index. Bodurtha et al. (1995) show that closed-end fund prices are more correlated to the market where the fund shares are traded compared to the market where the portfolio assets are traded.

Brickley and Schallheim (1985) report CAR of 14% for 16 CEFs announcing open-ending during 1962–1982 period. Brauer (1984) reports CAR of 5.18% in the month of open-ending and 4.1% for the following month based on a sample of 14 CEFs open-ending during 1963–1980.

We limit our post-announcement period to 26 weeks because the sample size starts to deteriorate quickly. In our sample, the termination period ranges from three days to more than 1 year. Seventy-five percent of the sample funds were liquidated within 30 weeks. The discount during the first 20 weeks exhibits a small downward trend with high degree of fluctuations.

Over the 1989–2012 period, we have an average of 31 foreign bond funds, 257 domestic bond funds, 46 domestic equity funds, and 35 foreign equity funds which are used to compute the IDISC for each of the groups.

Proxies for investor sentiment that have been recently used in studies include; risk measurement based on consumer sentiment from surveys (Lemmon and Portniaguina 2006; Hwang 2011), spread between returns on small-capitalization stocks and returns on large-capitalization stocks (Baker and Wurgler 2007), and the aggregate flow between bond funds and equity funds (Ben-Rephael et al. 2012). Qui and Welch (2006) and Lemmon and Portniaguina (2006) report that sentiment measures obtained from surveys are uncorrelated with the average CEF discount.

The NAV of CEFs holding illiquid asset may be overstated by the transaction costs if the fund portfolio is to be liquidated. Cherkes et al. (2009) propose that investors can avoid large transaction costs of directly trading the illiquid asset, by trading in CEFs that invest in illiquid assets. Because illiquid assets are likely to be associated with low CEF portfolio turnover, we examine but not report, the portfolio turnover as a proxy for asset illiquidity.