In a simple model of segmented markets and exogenous liquidity shock, the closed-end country fund premium is negatively affected by the illiquidity in the host market where shares of the country fund are traded, and positively affected by the illiquidity in the home market where the underlying assets are traded. To the extent that expected and unexpected liquidity affects asset prices and returns, the closed-end country fund premium should reflect the difference between the illiquidity of the

more »

... nd shares and its underlying assets. Using the Amihud measure of illiquidity, we examine the model prediction for U.S.-traded single country closed-end funds, and find a strong association between the fund premium and illiquidity in both the host and the home markets. Moreover, this relation is much stronger for funds investing in emerging markets where market segmentation is more likely to be binding. These funds are also more sensitive to the systematic liquidity factor, providing additional evidence that the country fund premium may be partially explained by the liquidity risk premium. Illiquidity and Closed-End Country Fund Discounts Financial assets with similar or even the same payoffs can often have different liquidity. Since liquidity is a key feature of the capital market and the macroeconomic environment, an important question to ask is how liquidity affects asset prices. This paper tries to answer that question by investigating whether the fluctuation in the closed-end country fund (CEF) premium is related to liquidity over time. Both the theoretical and empirical aspects of the interaction between liquidity and asset prices have been studied extensively. In a theoretical model that formally relates asset prices to liquidity, Kyle (1985) shows that the asset price is negatively related to a measure of market "depth" known as Kyle's lambda. Allen and Gale (1996) argue that the illiquid asset's price is given by the smaller amount of the asset's long-term fundamental value and the amount determined by the supply and demand of cash (liquidity). 1 There is also extensive empirical evidence that liquidity affects asset prices. The positive return-illiquidity relation across different stocks has been documented in studies such as Amihud and Mendelson (1986), Subrahmanyam (1996), and Brennan, Chordia, and Subrahmanyam (1998) . Amihud (2002) studies this relation over time and confirms that the expected excess stock return increases with the expected but decreases with the unexpected illiquidity in the stock market. Pastor and Stambaugh (2003) , on the other hand, find that expected stock returns are significantly related to liquidity betas and provide evidence that liquidity is a priced state variable. 2 Since liquidity matters for asset prices and liquidity risk is significantly priced by investors, it is not surprising that assets with similar payoffs but different liquidity levels or liquidity loadings command different prices. For example, on-the-run Treasury bonds are much more liquid and more expensive than their off-the-run counterparts even though they have very similar cash flows and characteristics, and Treasury bonds are often priced differently from similar government agency bonds even after controlling for coupon payment and default risk. 3 In the same vein, the closedend fund price is likely to differ from the fund's underlying net asset value (NAV) if the fund shares and its underlying assets have different liquidity levels or different liquidity loadings. A closed-end fund is a firm that issues a fixed number of shares and uses the proceeds to invest in the shares of other companies. Unlike an open-end fund, a closed-end fund maintains a fixed number of shares that are traded as a stock on an exchange. Closed-end funds announce their portfolio net asset value per share (NAV) at regular intervals (usually weekly or daily) and it is observed that their share prices typically trade at different levels (discounts) to their NAV. 4 Many explanations have been offered for the existence and the behavior of the discrepancy between the fund share price and its NAV. Within the rational framework, management fees, agency costs, tax effects, market segmentation, and mis-valuation of underlying (illiquid) assets, have been invoked to explain the puzzle. For example, Bonser-Neal, Brauer, Neal, and Wheatley (1990) find a significant relation between the country fund price-NAV ratios and the announcements of changes in foreign investment restrictions. While Malkiel (1977) finds no correlation between US discounts and management expense as a proportion of NAV, Ross (2002) argues that management fees can account for the magnitude of the closed-end fund discount. In an empirical examination of UK closed-end funds, Gemmill and Thomas (2002) find that the size of discounts is significantly related to proxies for arbitrage bounds and the freedom of managers to increase fees in the presence of costly arbitrage. Within the behavioral finance literature, investor sentiment is used as an alternative explanation. 5 This explanation is based on the observation that the U.S. closed-end fund shares are mainly held by individual investors, and that many small investors are irrational and driven by sentiment. Elton, Gruber, and Busse (1998) , however, find no empirical support for small investor sentiment as a priced factor for either common stocks or closed-end funds, and Dimson and Minio-Kozerski (1999) point out that the behavioral explanation is inconsistent with the empirical evidence on the UK closed-end fund market, which is largely dominated by institutional owners. Gemmill and Thomas (2002) reject the hypothesis that noise-trader risk is the cause of the long-run discount, but they find that the retail-investor flows into mutual funds are associated with fluctuations in discounts. In general, however, none of the above explanations provides a satisfactory account of the time-varying behavior of the discount. In this paper, we argue that the time-varying behavior of the closed-end fund discount provides a unique and interesting laboratory to study the effect of liquidity on asset prices. The set of closed-end funds has an important feature, i.e., the fund shares and its underlying assets are traded separately and are typically associated with different liquidity. To the extent that liquidity risk 4 The closed-end funds are generally issued at a premium. Weiss (1989) and Hanley, Lee, and Seguin (1994) provide empirical evidence of closed-end fund premium at the issuance, and initial price stabilization behavior provided by the lead underwriters. Cherkes (2003) argues that this special feature of buyers paying the IPO costs via IPO overpricing with the underwriters providing prolonged after-market price support as a supplement to the IPO over-pricing is neither anti-competitive nor predatory. empirical evidence that the noise trader model is consistent with the closed-end fund discount, in general, and the country fund discount, in particular.