How Effective Is Arbitrage of Foreign Stocks? The Case of the Malaysia Exchange-Traded Fund

Multinational Business Review, Fall 2003 by Hughen, J Christopher

Due to the currency controls, the !Shares Malaysia Fund did not trade on the American Stock Exchange on September 2. The next day, the fund announced the suspension of the creation of new shares and expressed doubts about the feasibility of creation unit redemptions. Figure 1 (on the next page) shows the premium fluctuations and share creation/redemption around the suspension of arbitrage. The dots on the graph show the timing of share creations, which are represented by dots in the positive section of the graph, and share redemptions, which are shown as dots in negative territory.7

As the Asian financial crisis spread between july 1997 and August 1998, the iShares Malaysia Fund traded at prices far from NAVs, and institutional investors created hundreds of thousands of shares over this time to take advantage of the price differentials. After the suspension of share creation in the iShares Malaysia Fund, the fund price quickly moved to a premium below -30 percent and later traded at a premium of 37 percent.

In the absence of arbitrage, international ETF premiums may become as volatile as the premiums on closed-end country funds. Figure 2 (on the next page) supports this assertion with a comparison of the premiums on the iShares Malaysia Fund and the premiums on the Malaysia Fund, the only closed-end fund that invests exclusively in this country. This exhibit reflects weekly NAV data over the period from September 4, 1998 to july 16, 1999. While the premiums on the country fund were larger than the ETF premiums, the premiums often moved in the same direction and followed a similar trend. The data on the iShares Malaysia Fund is consistent with the fund-facilitated arbitrage feature of ETF preventing the extreme premium fluctuations associated with country funds.

Table 1 (two pages ahead) provides information on the size of daily premiums over three different periods. From the time this fund started trading to the last trading day while the ringgit was fixed, the premiums were small-80 percent of the daily premiums were between 0.37 percent and 2.24 percent. After the ringgit was allowed to float, the absolute value of the premiums was generally larger. However, most of the premiums were ephemeral since the frequent occurrence of arbitrage helped to align prices in the domestic and foreign markets. After share creation and redemption was suspended by the fund company, there were extended periods of large premiums. From 9/3/98 to 1/15/99, over 10 percent of the premiums were less than -20 percent and 10 percent of the premiums were above 18.91 percent. On November 27, 1998, the fund again allowed the redemption of shares but only in ringgits. Eventually, share creations and redemptions were fully restored, and the large premiums on the iShares Malaysia Fund disappeared.

TESTS AND RESULTS

To formally test how the arbitrage suspension affects the ETF premiums, a time series regression model is estimated and shown in Table 2. The dependent variable is the absolute value of the iShares Malaysia Fund premium. An autoregressive error model with two lags is used to correct for serial correlation.