Why-Do-Closed-end-Country-Funds-Trade-at-Enormous-Premiums-during-

Document Sample
Why-Do-Closed-end-Country-Funds-Trade-at-Enormous-Premiums-during- Powered By Docstoc
					Why Do Closed-end Country Funds Trade at Enormous Premiums during Currency Crises?
Nandini Chandar and Dilip Kumar Patro* Faculty of Management, Rutgers University First draft: May 1998 Current draft: September 19, 1999 Comments Welcome We investigate the response of U.S. traded country fund premiums to currency crises in related foreign (local) markets. Our analysis includes 25 currency crises over the past decade involving 18 funds investing in 12 emerging markets, and 7 funds investing in 6 developed markets. We find that fund premiums and the volatility of the premiums increase dramatically in response to a currency crisis, both for emerging and developed markets funds, and that these effects dissipate slowly over time. Our results indicate that differential risk exposures of country fund shares and net asset values (NAVs) are important in explaining these fund premiums. These effects are exacerbated during a crisis. While the NAV returns show sensitivity to changes in the local market index, share returns are sensitive to changes in both local and world market indices. Therefore, in response to a currency crisis, when local stock markets decrease in value, fund NAVs react more strongly than their share prices which have a strong global component. We also show that the high premiums observed during currency crises are not due to the reluctance of investors to trade and realize losses.

JEL Classification: G12, G15 Keywords: Country Funds, Currency Crisis, Risk Exposures

*

Both authors are from the Faculty of Management at Rutgers University. Please address correspondence to: Dilip K. Patro, Rutgers University, Management Education Center, 111 Washington Street, Newark, NJ 07102, Tel: (973)353-5709, Fax: (973)353-1233, Email: dilipk@andromeda.rutgers.edu. Dilip Patro would like to thank Don Cassidy for kindly providing parts of the data.

1. Introduction
Over the past decade, interest in foreign equities has increased dramatically in the U.S., as investors seek higher returns and greater diversification benefits that international investments can provide. Emerging stock markets, in spite of their volatility and risk, have been particularly attractive due to their potential for high growth and low correlation with developed markets. The recent Asian currency crisis has renewed attention on the risks inherent in emerging markets investing.1 The response of U.S. traded closed-end country fund (hence, country fund) prices to these local economic events has been puzzling. Contrary to expectations in the financial press that these funds will sell at deep discounts subsequent to these crises, country funds have actually traded at enormous premiums in response to these events.2 Further, this phenomenon is pervasive and occurs with

consistent regularity across both developed and emerging markets funds.3 The purpose of this paper is to investigate the behavior of country funds’ premiums during periods of currency crises in the local (foreign country) market by documenting empirical regularities about country fund premiums and how they respond to fundamental volatility shocks. We then discuss possible explanations for these

observations in the light of prior research related to country funds.

1

For an excellent overview of various aspects of the Asian currency crisis see articles listed on Nouriel Roubin’s web page at http://www.stern.nyu.edu/~nroubini/asia/AsiaHomepage.html. 2 For instance, see Mclean (1997), who states: “If only the world always made sense. Then the best way to bet on a comeback in Southeast Asia would be obvious: buy closed-end country funds. After all, they must be really cheap now given the big meltdown in these markets.” She also quotes an emerging markets fund manager who considers the observed premiums “a massive anomaly.” 3 Country fund share prices and their net asset values differ. The difference between the share price and the net asset value is referred to as the fund premium. Therefore a discount is treated as a negative premium.

1

Specifically, our objectives in this paper are: (1) to document the behavior of country fund premiums during local currency crises over the ten years (1988-1997); i.e., to examine the responses of NAVs and prices to these events, and their contributory effects on fund premiums; and to investigate the process and speed of adjustment; (2) to examine the relative volatility of NAVs and prices during the event period and to investigate whether this is different from the general findings in the literature (e.g., Pontiff (1997)) relating to non-event periods; (3) to determine whether different relationships are observed for NAVs and prices of emerging markets and developed markets funds during periods of currency crisis. We identify 25 currency crises affecting twelve emerging market (emerging market) countries, and six developed market (developed market) countries over the period January 1988 to December 1997. Figure 1 plots the fund premiums for each country fund over a 50-week period surrounding a currency crisis in the related local country, and illustrate the very interesting and dramatic pattern that we explore. We see that, in

overwhelming majority of these cases, country fund premiums show a significant increase in the aftermath of a currency crisis. This increase in premiums is, however, more

pronounced and dramatic for emerging market funds when compared with developed market funds. In general, both NAVs and prices of these funds decline, but NAVs decline more rapidly and dramatically than prices, thus resulting in an increase in fund premiums. Our paper makes two important contributions to the literature on country funds. Prior research on country funds has focused on time periods when economies were relatively stable.4 In this paper, we investigate the response of country fund premiums to
4

Lee, Shleifer and Thaler (1990) and Bodurtha, Kim and Lee (1995), provide a review of this literature.

2

economic shocks in foreign countries. We show how the determinants of fund price and NAV returns change in response to destabilizing events. This is an important

consideration and caveat for researchers who examine historical fund data over a relatively long period of time. Second, we relate our findings to prior research on closed-end funds in general and country funds in particular by providing assessments and complementary explanations for the existence of fund premia. In the process, we document one more empirical regularity relating to these funds that is puzzling. We relate our findings to prior research on country funds that provide evidence in non-extreme settings. Our study complements the literature on the pricing of country funds (Brosner-Neal, Brauer, Neal, and Wheatley (1990); Bodurtha, Kim and Lee (1995); Errunza, Senbet and Hogan (1998)); and studies of relative volatilities of fund NAVs and prices (Pontiff (1997)). The paper is organized as follows. In Section 2, we review the existing literature on closed-end fund puzzles. Section 3 discusses the data. The methodology and the empirical results are reported in Section 4. The last section concludes.

2. The Closed-End Fund Puzzle: Existing Explanations
Closed-end country funds are funds that issue a fixed number of shares and invest the proceeds largely in equity securities from a particular (local) country. The shares of these funds are traded on the secondary markets and, unlike those of open-end funds, the shares cannot be redeemed by the shareholders at their net asset values (NAVs). In the U.S., these funds trade at their U.S dollar price.

3

Several factors have accelerated international interest in emerging stock markets. These include developments such as the expansion of the European Common Market to include Greece and Portugal; the replacement of communist economies with market economies in Eastern Europe; the development of active stock markets in China, Brazil, and Chile; and the lifting of some barriers to international ownership of equity in countries such as China, India, Indonesia, Korea, Malaysia, Mexico, Philippines, Singapore, Taiwan and Thailand. In the U.S., there has been increased interest in emerging market fund investments, as evidenced by a significant increase in the number of country funds over the past decade.5 Country funds have provided a simple vehicle for investor participation in foreign markets by providing a managed diversified portfolio without the barriers or transactions costs that direct investment can entail. These funds have been the most important avenues for emerging market investment and global diversification (see, Chang, Eun and Kolodny (1995)). Bailey and Lim (1992) and Bekaert and Urias (1996), however, cast doubt on the diversification potential of country funds. If markets are efficient, frictionless, and internationally integrated, the price at which these funds trade must equal their NAV in U.S. dollar terms. In reality, however, many country funds and domestic closed-end funds trade, on average, at a statistically significant discount to their NAV ((Bodurtha, Kim and Lee (1995), Hardouvelis, La Porta, and Wizman (1994); and Lee, Shleifer, and Thaler, (1991)).6

5

Whereas only 4 closed-end country equity funds traded on the NYSE at the end of 1984, there are currently over 97 such funds investing in the equity securities of over 31 countries. 6 Lee, Shleifer and Thaler (1991) summarize the findings related to the existence and cross-sectional and longitudinal variation of the significant discount that is generally characteristic of U.S. closed-end funds as a four-part anomaly.

4

Hypotheses that have been suggested to explain these phenomena may be categorized as (1) rational explanations; i.e., those relating to market frictions caused by factors such as agency costs, transactions costs, deferred taxes on unrealized capital gains, illiquidity of assets; and (2) non-rational explanations like the investor sentiment hypothesis, which argues that individual investors, the predominant owners of closed-end fund shares are noise traders, who tend to trade on sentiment (see, DeLong, Shleifer, Summers and Waldman, (1990)). Discounts and premiums are therefore hypothesized to vary with investor sentiment. Many emerging markets funds historically traded at large premiums (see, Bodurtha, Kim and Lee (1995)) that existed in the face of foreign investment restrictions, as funds become unique vehicles for investing in emerging countries (Bosner-Neal, Brauer, Neal, and Wheatley (1990); Errunza, Senbet and Hogan (1998)). However, the rapid deregulation and liberalization of emerging stock markets, have led to large drops in the premiums of closed-end funds invested in those countries. Many of them subsequently started trading at discounts. Hardouvelis, La Porta and Wizman (1994) and Bodurtha, Kim and Lee (1995) find empirical support for the noise trader model in explaining country fund discounts during non-crisis periods. Kramer and Smith (1995) argue that the “investor sentiment”

hypothesis is inconsistent with Mexican funds and other Latin American funds trading at huge premiums subsequent to the Mexican currency crisis in December 1994. They

indicate that the hypothesis can be justified only by viewing these premiums as investor optimism, which does not seem plausible in this context. They instead hypothesize that individual investors who hold country funds are loss-averse, and therefore would be

5

willing to pay a premium for these funds even in the face of great pessimism, as their disutility for realizing losses is so great. Frankel and Schmuckler (1996) propose that the Mexican fund premiums following the Mexican currency crisis of 1994 can be explained by the “divergent expectations” hypothesis, which could result if Mexican investors had information that was superior to that of international investors. We examine the same phenomena that Kramer and Smith (1993), and Frankel and Schmucker (1996) do. Our study is however more comprehensive than these prior studies in that we consider all major currency crises over the past ten years, whether they relate to developed or emerging economies and provide a more complete documentation of country fund responses to these events. We then assess our findings with respect to existing explanations for country fund premiums, which include the “investor sentiment,” “lossaversion,” and “divergent expectations” models suggested by earlier researchers.

3. Data
Our sample includes all NYSE traded country funds investing in countries that have experienced a currency crisis between 1988-1997, the period in which the popularity of country funds increased substantially. We first identify the events that are described as currency crises. The events and dates are obtained by searching the International

Monetary Fund’s Report on Exchange Arrangements and Restrictions, the Financial Times, Bloomberg and the Wall Street Journal.7 Our resulting sample consists of 18 funds

7

The event dates are reported as particular dates in some instances, while in others a window of time is indicated. We use the earliest date of the reported crisis to define our event period.

6

investing in 12 emerging market countries, and seven funds investing in six developed market countries. Table 1 provides details of the currency crises during this period, including the nature of the event reported, the reported dates, the related country fund(s), the fund’s ticker symbol and date of inception. Panel A of Table 1 contains event data relating to emerging market funds, while Panel B describes events relating to developed market funds. The weekly (Friday closing) U.S. dollar share prices, NAVs and dividends for each fund are from obtained from Lipper Analytical Services and Bloomberg for the period of January 8, 1988 to December 26, 1997, a total of 522 weeks.8 The share returns are adjusted for stock splits, dividends and rights offerings. The local market indices for the different countries and the world market portfolio are the weekly Morgan Stanley Capital International (MSCI) indices, which do not include dividends. For the country funds, the exchange rates for the currency of the country of origin with the U.S. dollar are obtained from the Federal Reserve Board. Trading volume data for a subset of the sample is obtained from the online Microsoft Investor database.

4. Empirical Results
A. Country fund premiums before and during a currency crisis We begin by presenting some descriptive statistics for country fund premiums. Table 2 presents the statistics for the emerging market funds (Panel A) and the developed

8

Lipper Analytical is a premier source of U.S. mutual fund data. Print publications such as The Wall Street Journal and New York Times, obtain closed-end fund data from Lipper Analytical.

7

market funds (Panel B) for the entire sample period.9 The average premiums range from 30.23 percent to negative 18.56 percent. As is well known, most emerging market and developed market funds have sold at an average discount during this period. Some

emerging market funds have also sold at average premiums, although the standard deviation of premiums for this group is the greatest. The distribution of premiums is tighter for the developed market funds, excluding funds investing in Italy and Spain. The following section provides a detailed empirical analysis of the response of country fund premiums to local currency crises. In general, the event windows corresponding to the currency crises are not well defined. They are more tightly contained in some cases, and tend to dissipate over a longer period in others. Moreover, there are also instances of multiple crises reported within a few days or weeks of the crisis initially reported. In order to have a uniform basis for comparison across funds, we use a 25-week window starting with the earliest reported event date.10 The non-event period used for comparison is the 25-week period preceding the onset of the event. Further, when multiple events over a relatively short time period are reported, the earliest event date in Table 1 is used to define the event period. The last five columns of panels A and B of Table 2, report comparison of mean country fund premiums, for the 25 weeks prior to the crisis with premiums in the 25-week period following the initial announcement of the currency crisis. Statistical significance of changes in premiums are evaluated using t-statistics from the regression of the premiums for each fund on a constant and dummy variable which is zero for the 25 weeks before the

Classifications of countries as emerging market or developed market are as contained in Solnik (1996). To the extent that the real event period is shorter than the 25-week period we have adopted, the power of our statistical tests will be reduced. We will explore this issue again later.
10

9

8

crisis and one for the 25 weeks of the crisis. Since the residuals from these regressions are auto-correlated due to the persistence of the premiums, the t-statistics are corrected for serial correlation using 13 Newey-West lags. A significantly positive coefficient on the dummy variable implies that the premium during the crisis is significantly higher than premium before the crisis for the fund. In addition, comparison is made across the entire sample of 25 events using both a paired t-test and a Wilcoxon signed rank test. In an overwhelming majority of the cases, for both emerging market and developed market funds, the country fund has experienced a statistically significant increase in premiums following the onset of the crisis. Only two emerging market funds—the Brazil Fund, and the Czech Republic Fund have decreases in premiums (t-statistics of minus 3.87 and minus 1.64 respectively), while the New South Africa showed no significant change in premiums in response to the crisis. In response to the currency crisis that threatened the ERM, one developed market fund, the Italy fund, shows a significant decrease in premiums (t-statistic of –2.63) in contrast to the increase in premiums of other developed market country funds in response to the event. The Growth Fund of Spain does experience an increase in fund premiums following the first currency crisis. However, statistical significance is not obtained due to the high standard deviation of the premiums following the crisis. The Italy Fund and The Growth Fund of Spain, experience no significant changes in premiums in response to a second crisis. As discussed below, this is a result of the increased relative volatility of NAVs following the crisis. In terms of the overall sample, the paired t-tests and the z-statistics indicate that both the mean and the standard deviation of fund premiums are significantly higher during the crisis. This is true

9

for both emerging market and developed market funds.11

The results show that the

increase in premiums subsequent to the onset of a currency crisis is a pervasive phenomenon, and is experienced by country funds investing in both emerging and developed markets. It is well know that premiums on closed-end funds move together. Therefore, to verify that the high premiums during crisis are not simply because premiums are going up for all closed-end funds in that period, the regressions above were re-run by including the premium on an index of 42 domestic closed-end funds as an additional independent variable. The results are similar and are therefore not reported.12 Interestingly, we also find that the correlation of the changes in premiums of sample funds and U.S. domestic funds during the crisis period is low. Figure 1 provides graphs depicting the premiums on each fund for the 50-week period surrounding the event date. These reinforce earlier findings that premiums on most of these funds increase in the event period.13

B. Excess volatility of share prices and NAVs Prior research has found that the volatility of fund price returns is significantly higher than the volatility of NAV returns both for domestic funds (see, Pontiff (1997)) and for country funds (see, Hardouvelis et al. (1994)). Under the noise trader hypothesis, this “excess volatility” is attributed to the behavior of small investors, as an explanation of country fund discounts. We investigate whether changes in fund premiums during currency crises are caused by changes in relative volatility of prices and NAV.
Paired t-tests and z-tests for sub-samples of emerging market and developed market funds indicated results similar to that obtained from the overall sample, and are therefore not reported. 12 The list of 42 domestic funds used along with results of the empirical test is available from the authors.
11

10

Table 3 panel A (B) presents measures of volatility of country fund share and NAV returns prior to and during currency crises for emerging market (developed market) country funds. In response to the crisis, most funds have experienced an increase in volatility in terms of both share prices and NAV. Table 3 presents the ratio of standard deviations of fund price returns and NAV returns both during crisis-periods and pre-crisis periods. It is apparent that most funds have also experienced an increase in these ratios. For the aggregate sample, this is confirmed by paired t-tests (price volatility 4.78; NAV volatility 4.90; p-values of 0.00) and z-statistics (price volatility 3.86; NAV volatility 3.99; p-value of 0.00). Table 3 also compares the relative volatility of fund share returns and NAV returns prior to and during the crisis.14 In the case of a majority of emerging market funds, while share price returns are more volatile that NAV returns prior to the crisis, the relative volatility decreases to less than one during the crisis. The average ratio has declined significantly (t-ratio 2.69; p-value < .01) from 1.21 in the pre-crisis period to 0.98 during the crisis period. Thus the local market volatility increases significantly relative to fund market volatility in the period following a currency crisis. On the other hand, the developed market funds on average do not experience a significant change in relative volatility of share returns and NAV returns. It appears that for emerging market funds, premiums during crises may be partially explained by excess volatility of the underlying assets in relation to fund prices. While this phenomenon observed for emerging market funds may be consistent with the noise trader hypothesis

All results not reported are available from the authors upon request. Relative volatility is measured as the ratio of the standard deviations of fund share price returns and fund NAV returns.
14

13

11

(Lee et al. (1991)), it cannot be extended to developed market funds, which do not demonstrate increased relative volatility of underlying assets. The excess volatility of underlying assets relative to fund price observed in the case of emerging markets funds is counter to Pontiff’s (1997, p. 155) finding that “the average fund has monthly return variances that are 64 percent greater than the variance of its NAV return.” Thus, the relative volatility of fund NAV and price during currency crises appear to be different from those that are found in normal periods. Further, these patterns persist over the comparatively long event windows in our study.

C. Responsiveness of fund prices to NAV changes In this section, we document the relative sensitivity of country fund shares to changes in NAVs, during the currency crisis and compare it with normal patterns of responsiveness, using the following model:
rp, t = α p + β n rn , t + β n −1rn −1, t + β p rp −1, t + e p ,t

(1) where: rp,t = share return for each fund for the period t. rn,t = NAV return for each fund for period t For each fund, therefore, we regress weekly share returns on a constant, NAV returns, lagged NAV returns and a lagged share return. Table 4 presents the coefficients on NAV returns before and during the crisis. It does not appear that there are significant differences in how fund share prices respond to NAV changes during a crisis. In most cases, the coefficients are far lower than one, suggesting that fund prices are relatively “sticky” with respect to NAV changes. This is consistent with the results in Hardouvelis,

12

LaPorta and Wizman (1994), and Bodurtha, Kim, and Lee (1995). It is also consistent with the hypothesis that investors under-react to news that affects fundamentals, as demonstrated for domestic closed-end funds (see Pontiff (1997). Klibanoff, Lamont and Wizman (1998), find that there is an increase in the sensitivity of prices to contemporaneous NAV changes in the presence of some dramatic news affecting a country fund. They infer that although investors underreact to mundane NAV changes, the extent of underreaction decreases significantly when investors observe more dramatic news. Table 4 shows that prices underreact to NAV changes for most funds, both before and during the crisis. However, the onset of the crisis does not show any significant difference in the extent of this underreaction, both for developed and emerging markets funds. During the crisis, the coefficient on NAV returns is higher for 17 funds and lower for 17 funds than the pre-crisis levels. Average coefficients of prices to NAV changes remain virtually unchanged in response to the crisis (average coefficient is 0.56 before and during the crisis) and the difference is statistically insignificant.15 Therefore, unlike Klibanoff et al. (1998), we do not find that the extent of underreaction of prices to fundamentals decreases in the aftermath of the crisis.

D. Sensitivity of fund returns to local and world market returns and exchange rates Alternate explanations for country fund premiums have focused on market frictions and segmentation, which result from barriers to capital flows, costly arbitrage, and transactions costs. International asset pricing models (e.g., Errunza and Losq (1985)) and models of country fund pricing (e.g., Errunza, Senbet and Hogan (1998)) extend the use

13

of the domestic CAPM in an international context by suggesting that the expected return on country fund shares is determined by their covariances with the world market portfolio. The international CAPM, however, applies only in an integrated world capital market (Errunza and Losq (1985)). In reality, factors such as transactions costs, psychological, legal, and cultural barriers, differential taxation, political risk and exchange risks are obstacles to arbitrage strategies and serve to segment the markets. The effect of financial market segmentation is that securities with the same risk characteristics, but listed in two different markets have different prices. Previous research (e.g., Eun and Janakiraman (1986), Hietala (1989), and Bailey and Jagtiani (1994)) also shows that investment restrictions cause market segmentation, and can have substantial effects on share prices. Due to these investment barriers, the determinants of country fund prices and their fundamentals may vary. The funds’ net asset values (NAVs), are determined by their covariances with both the world market portfolio and their domestic market portfolio. Errunza, Senbet, and Hogan (1996) suggest that when there is imperfect substitution between a country fund and its underlying assets, the fund sells at a discount or premium depending on the extent of international investor access to the domestic market, the degree to which the securities in core markets like the U.S. span the securities in the emerging markets, and the degree of comovement in the country fund universe. While it is possible that market segmentation plays an important role in explaining the increase in premiums on country funds that we observe, we do not directly test this hypothesis.16 Testing the market segmentation hypothesis would require sophisticated

15

We also ran the above regression using another lagged NAV and a lagged premium. The results obtained are similar and hence not reported. 16 In our discussions in this paper, “increase in premiums” subsume reductions in discount.

14

tests that examine if there are differences in prices of risk. Instead, we show that the risk exposures for fund prices and NAVs are different, and these differences are magnified during a currency crisis. During periods of crisis, the local stock markets are affected substantially which in turn affects the NAVs. The shares are however held by investors who are diversified globally and are therefore more insulated from country-specific idiosyncratic risk. These effects therefore manifest as large premiums for the country funds. In order to assess whether fund shares and NAV have significantly different risk exposures to the U.S., global and local markets, we perform separate regressions of fund share price returns and fund NAV returns on local and global market indices. Further, if purchasing power parity (PPP) does not hold between countries, foreign exchange risk will affect expected returns on securities (Adler and Dumas 1983).17 Generally, the

empirical evidence indicates that PPP is often violated in the short-run, therefore, we include an exchange rate factor in our regressions. We utilize a multi-factor model such as those represented by equations (2) and (3) below. The factors we use to test this hypotheses are based on the literature discussed above, which suggests that three factors—the return on the home market index, the return on the world market index and the changes in exchange rates could potentially determine the expected returns on securities in international capital markets. We use the world

If PPP held exactly for an asset, the real returns measures in two different currencies would exactly be the same, since the exchange rate would simply mirror differential inflation rates.

17

15

market index, excluding the U.S., since we want to test if the risk exposure to the world market is a global phenomenon or a U.S. phenomenon.18 Table 5 reports the results from fund-specific regressions of share returns (Panel A), and NAV returns (Panel B) on MSCI world index returns (excluding the U.S), local market index returns, and changes in local country exchange rate vis-à-vis the U.S. dollar. Specifically, we estimate the following models for each fund separately: Panel (2) Panel (3)
where: rp,t = total return on the shares (including dividends) of country fund “i” between t-1 and t; rn,t= total return on the NAV (including dividends) of country fund “i” between t-1 and t; rw,t = return on the Morgan Stanley Capital International (MSCI) world market index (excluding the U.S); rh,t = return on the MSCI index, corresponding to the country of origin of the fund; rx,t = logarithm changes in the exchange rate of the local currency with the U.S. dollar.

A:

rp,t

= =

αn+βn,wrw,t αn+βn,wrw,t

+βn,hrh,t+βn,xrx,t

+en,t

B:

rn,t

+βn,hrh,t+βn,xrx,t

+en,t

The results reported in Panel A of Table 5 clearly indicate that fund share prices are determined by both global and local market returns. In contrast, in Panel B, we see that NAV returns are determined almost exclusively by returns on local indices. Only in one case, the Jakarta Growth Fund, do we find a significant risk exposure to the world markets. While shares and NAVs have significant risk exposures to local indices, NAV returns are more highly sensitive to local market returns as evidenced by the substantially higher local market betas in panel B. The adjusted R-squares for the NAV model are also
18

We also estimate these models using the world market index, which includes the U.S. The results are

16

much higher than for the share price returns model, suggesting that fund prices add a layer of noise to the underlying asset values.19 The U.S. market index is highly positively correlated with the world market index (correlation of 0.67). Therefore, it is not surprising that when the U.S. market index is used in place of the world market index, as Bodurtha, Kim and Lee (1995) do, the factor is highly significant. In order to ascertain whether country fund prices have a global component that is independent of the U.S. markets, we used the MSCI world index excluding the U.S.–this index has a correlation of only 0.36 with the U.S. index. While share price returns (Panel A) are sensitive to both the world index and the local index, the NAV returns (Panel B) are dependent only on the performance of the local index. In fact, the average beta on the world market index for the NAV returns model in Panel B is zero. The changes in exchange rates is often insignificant for both shares and NAVs. The Table provides strong evidence that there is a global component to fund share prices that is independent of the U.S. market. Therefore, unlike Bodurtha, Kim and Lee (1995) who interpret the sensitivity of share returns to the U.S. market index as investor sentiment, we attribute our findings to the structure of international capital markets. These results are reinforced in panel C of Table 5, where changes in premiums are regressed on MSCI world index (excluding the U.S.), local returns and exchange rates. It is evident that there is a strong global component that explains changes in premiums. In addition, premium changes have mostly a negative relationship with changes in local

very similar and therefore not reported. 19 In additional tests, we replaced the world market index with the U.S. index and obtained essentially the same results (not reported here).

17

market indices and a positive correlation with world index returns. The local component is significant in explaining changes in these premiums for 14 of these 25 funds. In order to further highlight the impact of the crisis, Table 6 breaks down the results obtained in Table 5 into pre-crisis (Panel A) and crisis (Panel B) period betas. In the pre-crisis period, share returns have average betas of 0.38 and 0.58 on the world and local markets respectively. NAV returns are determined mostly by local market conditions (beta of 0.73) and are virtually independent of the world market (beta of –0.04). During the crisis, the exposures to global markets become increasingly important in explaining fund share returns (global index beta of 0.67 vs. local index beta of 0.42), while NAV returns are still largely determined by local market conditions (global index beta of 0.05 vs. local index beta of 0.75). Thus, while NAV returns have the same sensitivity to the local index before and during periods of currency crises, share returns are more sensitive to the world market index during the crisis. We therefore find that the differences risk exposures of fund prices and NAVs are exacerbated during periods of currency crises. Table 6 also presents results of the above model using fund premiums as the dependent variable. The effects discussed above are again apparent. The global sensitivity of premiums has, on average, increased dramatically during the crisis (beta on world index of 0.78 vs. 0.39 in the pre-crisis period), while the local market index remains less important in explaining fund premium changes. It is also interesting to note that the sensitivity of share prices and premiums to exchange rate changes increases dramatically during the crisis, while exchange rates remain less important in explaining NAV returns.

18

E. Discussion of results and other potential explanations We discuss our results in the light of other hypotheses that have commonly been used to explain country fund premiums. One is the “investor sentiment” hypothesis, where “sentiment” is used to refer to general investor optimism or pessimism that is essentially unrelated to fundamentals. When U.S. investors hold country fund shares, a change in their sentiment will be reflected in fund prices and not in their local NAVs. Lee, Shleifer and Thaler (1991) suggest that small investors, who predominantly hold closed-end fund shares, tend to trade on sentiment and act as noise traders. Fund prices and premiums therefore vary with investor sentiment. Hardouvelis et al. (1994), Bodurtha, Kin and Lee (1995) and Frankel and Schmukler (1996) find support for the noise trader model in explaining country fund discounts, suggesting that U.S. investors, on average, tend to underestimate the fundamental values of fund assets. The results in this paper are not generally consistent with the investor sentiment hypothesis. We find that fund premiums increase dramatically in response to a currency crisis in the local country, and that this pattern is persistent and pervasive. For emerging markets funds in particular, the decrease in relative volatility of prices and NAVs may explain the existence of crisis premia. As Kramer and Smith (1995) argue, it is difficult to conceive of the rise in fund premiums following a currency crisis in the related country as evidence of investor optimism about the country. Further, we find that the fund shares have a global component that is independent of the U.S. market, making it difficult to justify sentiment based explanations. Pontiff (1996) argues that arbitrage costs have implications for closed-end fund premiums, and funds more difficult to replicate are likely to deviate more from their

19

NAVs. Clearly, that is what we find when comparing emerging and developed markets. The high arbitrage costs itself may segment markets. Thus arguments based on arbitrage costs are consistent with those related to market segmentation. However, we do not find any evidence that the emerging and developed market funds are different in how they respond to changes in fundamentals. Therefore, although arbitrage costs may be

important, we show that the structure of international capital markets is also very important. Kramer and Smith (1995) suggest that the increase in premiums of the Mexican country funds following the currency crisis in Mexico in late 1994 is due to the reluctance of investors to realize their losses. Practitioners (e.g., McLean (1997)) express similar views. If this were the case, we should observe a decrease in the trading volume of these funds during the crisis. To test this hypothesis, Table 7 presents an analysis of trading volumes for nine Asian funds in response to the 1997 Asian currency crisis. Mean daily trading volume 25 weeks prior to the crisis and 25 weeks during the crisis are compared by means of a regression of the trading volume on a constant and a dummy variable, which is 0 before the crisis and 1 during the event period. The t-statistics from these regressions are positive and significant, with the exception of Thai Capital Fund. These indicate that, on average, trading volumes go up significantly during a crisis. Therefore, it does not appear then that investors are reluctant to trade and realize losses. Another rationale suggested (Frankel and Schmuckler (1996)) is that the premiums are due to asymmetric information and reflect the superior information that the local investors have over foreign investors. This would imply that any such premium observed should be temporary and should reverse within a relatively short time. In contrast, the

20

premiums we observe in our study are persistent over a comparatively long window. We thus conclude that the premiums are due to the differences in risk exposures of the shares and the NAVs. The recent Asian crisis has focused attention on the world financial markets. Emerging markets, especially, are of interest since our knowledge of international financial markets is primarily drawn from studies of developed markets.20 While there is a natural inclination to think of currency crises as rare events, banking and currency crises have become increasingly common, particularly in the developing world. These considerations are increasingly important as international financial markets become more integrated. By addressing issues relating to the behavior of international securities in U.S. markets, we provide a better understanding of the risks and benefits of international investments.

5. Conclusions
In this paper, we investigate the response of U.S. traded country fund premiums to currency crisis in the related local country. We first document some stylized facts relating to such phenomena that have occurred in both developed and emerging markets over the past decade. Then, we relate our findings with prior research on how fund prices and NAVs react to changes in fundamentals. Our study indicates that country funds trade at enormous premiums in response to currency crises. Further, we find that these premiums

20

Some recent studies that focus on emerging markets equity include Bailey and Chung (1994), which examines determinants of equity returns in Mexico; Bekaert and Harvey (1997) examine volatility in emerging equity markets; and Bailey and Jagtiani (1994) analyze the effect of ownership restrictions in Thailand. Prior research generally has not considered the effect of macro-economic shocks on equity prices in international markets. The study that is closest to ours is the one by Frankel and Schmukler (1996) which examines the effect of the Mexican currency crisis on three Mexican country funds.

21

exist for several weeks following the onset of a currency crisis and dissipate over a relatively long time period. We compare our observations with prior research both in a rational and nonrational framework. Non-rational models propose that asset prices reflect investor

underreaction or overreaction to news relating to fundamentals. We find evidence that fund prices underreact to NAV changes. However, contrary to Klibanoff, Lamont, and Wizman (1998), we do not find evidence that this underreaction decreases during the crisis. We also find that the premiums are not caused by investors’ reluctance to trade and realize losses. As a rational explanation for the observed crisis premia, we find that fund shares are sensitive to global markets, while NAVs are driven by local market conditions. This divergence in risk exposure is exacerbated during a currency crisis, when fund NAVs continue to be sensitive largely to local market returns, while the relative sensitivity of fund prices and hence premiums to global markets increases dramatically. In response to a currency crisis, therefore, when the local stock markets decrease in value, the NAV reacts more strongly than fund prices do. While fund prices also fall in response to the crisis, they react to this event in a global context. Their reactions therefore, tend to be more moderate. The more dramatic decline of NAVs relative to fund prices results in an increase in premiums or a decrease in discounts following a currency crisis. We interpret our results to complement existing explanations by indicating that differential risk exposure has an important role in explaining fund prices and premiums and these effects may be magnified during periods of currency crises.

22

Our research has implications for policy discussions relating to the role of foreign investors in local economies. Following the recent Asian financial crisis, politicians and journalists have commonly argued that foreign investments have a destabilizing effect on local stock prices. This has led to increasing calls for controls on capital flows, intended to shield a country from high volatility caused by “hot money flows.” If foreign investors had such a destabilizing effect, we should expect to see increases in volatility and declines in fund prices relative to their NAVs soon after the onset of these crises. Instead we find that NAVs decline more rapidly and dramatically than prices and also tend to be more volatile during this period. Therefore, consistent with the conclusions of recent research (Choe, Kho, and Stulz (1998)), we do not find evidence that foreign investors have a destabilizing effect on local markets. Further, since the fund shares have a strong global component, U.S investors are protected to some extent from crisis-induced volatilities in local market prices. Therefore, country funds may be a better vehicle than open-ended mutual funds for investor participation in foreign equities. Several issues are unresolved and offer potential for further research. These

include, tests of market segmentation effects in order to determine whether the premiums observed are caused by differences in priced risk. It is also puzzling that even for funds investing in very open, developed markets, the NAVs are affected almost entirely by local market conditions. This could be due to factors such as investor choice (French and Poterba (1991)), which needs further exploration. Finally, the issue of the relationship between currency crisis and contagion in financial markets is still relatively unexplored.

23

References
Adler, Michael and Bernard Dumas, 1983, International Portfolio Choices and Corporation Finance: A synthesis, Journal of Finance 38, 925-984. Bailey, Warren, and Y.P. Chung, 1994, Exchange Rate Fluctuations, Political Risk, and Stock Returns: Some Evidence from an Emerging Market, Journal of Financial and Quantitative Analysis 30, 541-562. Bailey, Warren, and J. Jagtiani, 1994, Foreign Ownership Restrictions and Stock Prices in the Thai Capital Market, Journal of Financial Economics 36, 57-87. Bailey, Warren, and Joseph Lim, 1992, Evaluating the Diversification Benefits of the New Country Funds, Journal of Portfolio Management, Spring, 74-80. Bekaert, Geert, Campbell R. Harvey, 1997, Emerging Equity Market Volatility, Journal of Financial Economics 43, 29-77. Bekaert, Geert, and Michael Urias, 1996, Diversification, Integration and Emerging Market Closed-end Funds, Journal of Finance 51, 835-869. Bodurtha, James N. Jr., Dong-Soon Kim and Charles M. C. Lee, 1995, Closed-end Country Funds and U.S. Market Sentiment, Review of Financial Studies 8, 881-919. Bonser-Neal, Catherine, Greggory Brauer, Robert Neal, and Simon Wheatley, 1990. International Investment Restrictions and Closed-End Country Fund Prices, Journal of Finance 45, 523-547. Chang, Eric., Cheol Eun and Richard Kolodny, 1995, International Diversification Through Closed-end Country Funds, Journal of Banking and Finance 19, 1237-1263. Choe, Hyuk, Bong-Chan Kho, and Rene M. Stulz, 1998. Do Foreign Investors Destabilize Stock Markets? Working Paper, The Ohio State University. DeLong, J. Bradford, Andrei Shleifer, Lawrence Summers and Robert Waldman, 1990, Noise Trader Risk in Financial Markets, Journal of Political Economy 98, 703-739. Errunza, Vihang, and Etienne Losq, 1985, International Asset Pricing under Mild Segmentation: Theory and Tests, Journal of Finance 40, 105-124. Errunza, Vihang, Lemma Senbet and Ked Hogan, 1998, The Pricing of Country Funds From Emerging Markets: Theory and Evidence, International Journal of Theoretical and Applied Finance 1, 111-143.

24

Eun, Cheol., and S. Janakiramanan, 1986, A Model of International Asset Pricing with a Constraint on the Foreign Equity Ownership, Journal of Finance 41, September, 897-913. Frankel, Jeffrey, and Sergio Schmukler, 1996, Country Fund Discounts, Asymmetric Information, and the Mexican Crisis of 1994: Did Locals turn Pessimistic before International Investors? University of California, Berkeley NBER Working Paper 5714. French, Kenneth R., Poterba, James M., 1991, Investor Diversification and International Equity Markets, American Economic Review 81, 222-226. Hardouvelis, Gikas, La Porta, Rafael, Wizman, Thierry, 1994, What Moves the Discount on Country Equity Funds?. The internationalization of equity markets. Frankel, Jeffrey A., ed., National Bureau of Economic Research Project Report series. Chicago and London: University of Chicago Press. p 345-97. Klibanoff, Peter, Owen Lamont, and Thierry Wizman, 1998, Investor Reaction to Salient News in Closed-End Country Funds, Journal of Finance 53, 673-699. Kramer, Charles, and Todd Smith, 1995, Recent Turmoil in Emerging Markets and the Behavior of Country Fund Discounts: Removing the Puzzle of the Pricing of Closed-End Mutual Funds, International Monetary Fund Working Paper 95/68, July. Lee, Charles M. C., Andrei Shleifer and Richard Thaler, 1991, Investor Sentiment and the Closed-End Fund Puzzle, Journal of Finance 46, 75-109. McLean, Bethany, 1997, Case Closed. Fortune 136(11), December 8, 262. Pontiff, Jeffrey, 1996, Costly Arbitrage: Evidence from Closed-End Funds, The Quarterly Journal of Economics 111, 1135-51. Pontiff, Jeffrey, 1997, Excess Volatility and Closed-End Funds, American Economic Review 87, 155-169. Solnik, Bruno, 1996, International Investments, third edition, Addison-Wesley: New York.

25

Table 1 Currency Crises in Emerging and Developed Countries: January 1988-December 1997
The following table describes events identified as “currency crises” in major financial newspapers such as the Wall Street Journal and the Financial Times, or in IMF’s Annual Report on Exchange Arrangements and Restrictions. The sample includes 25 funds: 18 from emerging markets and 7 from developed markets. The classification of stock markets as “emerging” or “developed” is consistent with the IFC’s definition of an emerging market (Solnik, 1996). Panel A: Currency Crises in Emerging Markets Country Dates of Nature of Event Reported Country Fund(s) Fund IPO Date Crises investing in the country Symbol of Fund and listed before the crisis Argentina 11/27/92 Currency crisis in Argentina Argentina Fund AF 10/22/91 Brazil 1/16/89 Brazilian currency devalued by 14.1% Brazil Fund BZF 3/31/88 1 6/30/89 Exchange rate is devalued by 10.7% Czeck 5/26/97 Cabinet purges in the wake of currency crises Czeck Republic Fund CRF 9/30/94 Republic and stock market scandals India 7/1/91 Exchange rate was devalued from Rs. 21 per India Growth Fund IGF 8/12/88 U.S. $ to Rs. 23.12 per U.S. $ 7/3/911 Exchange rate was further devalued Indonesia 10/3/97 Rupiah falls 9.3% against U.S. $ Indonesia Fund IF 3/1/90 Since Asian currency crisis Jakarta Growth Fund JGF 4/10/90 Mid-July 1997 7/21/971 Rupiah tumbles 7% Korea Since Asian currency crisis Korea Fund KF 8/22/84 Mid-July 1997 Korea Investment Fund KIF 2/24/92 Korea Equity Fund KEF 11/24/93 Fidelity Advisor Korea 10/25/94 Malaysia Since Asian currency crisis Malaysia Fund MF 5/8/87 Mid-July 1997 Mexico 12/19/94 Currency devalued by 15% Mexico Fund MXF 6/3/81 12/21/941 12/27/94 1/10/95 3/7/95 Since Mid-July 1997 4/27/96 7/2/97 July 19971 1/26/94 Peso allowed to float Peso plunges 8.26% Mexican stock market plummets, and Peso loses ground Mexican Peso free falls Asian currency crisis Mexico Equity and Income Fund Emerging Mexico Fund MXE MEF 8/14/90 10/10/90

Philippines

First Philippine Fund

FPF

11/8/89

South Africa Thailand Turkey

Rand has lost 17% of its value since 2/15/96 Baht allowed to float Currency plunges by 40% Currency devalued by 13%

New South Africa Fund Thai Fund Thai Capital Fund Turkish Investment Fund

NSA TTF TC TKF

3/11/94 2/17/88 5/22/90 12/5/89

Panel B: Currency Crises in Developed Markets Country Dates of Nature of Event Reported Crises France 9/17/921 European currency markets reeled as France’s central Bank was forced to intervene to support the franc. July 19932 Currency crisis nearly destroys ERM Ireland 11/30/921 Ireland faces a worsening currency crisis 2 1/30/93 Irish Punt devalued by 10% 8/10/93 Punt plunges nearly 3.5% in 3 days July Currency crisis nearly destroys ERM 1993 Italy 1/10/901 Lira devalued by 3.8% 9/13/922 Lira devalued by 7% 9/17/92 Lira withdrawn from ERM July 19933 Currency crisis nearly destroys ERM 1/10/954 3/15/95 Since midJuly 1997 9/12/92 11/22/921 5/13/932 July 1993 9/9/92 9/16/921 July 19932
1

Country Fund(s) investing in the country France Growth Fund

Fund Symbol FRF

IPO Date of Fund 5/10/90

Irish Investment Fund

IRL

3/30/90

Italy Fund

ITA

2/27/86

Singapore Spain

Italian Lira falls Lira lost 15% against developed market. Asian currency crisis Peseta devalued by 4.8% Peseta devalued by 6% Currency devalued by 8% Currency crisis nearly destroys ERM Sterling approaches floor rate within ERM. British membership in ERM suspended. Currency crisis ensues Currency crisis nearly destroys ERM

Singapore Fund Spain Fund Growth Fund of Spain

SGF SNF GSP

7/24/90 6/21/88 2/14/90

United Kingdom

United Kingdom Fund

UKM

8/14/87

The dates indicated are the earliest dates that a currency crisis was reported during the period. In many instances, this occurrence was followed by subsequent crises. When these crises occur within six months of the original crisis, they are not considered to be separate events for purposes of empirical examinations. 1, 2, 3, etc denote different crisis in the same country.

Table 2 Closed-end Country Fund Premiums: Descriptive Statistics This Table presents descriptive statistics of the weekly premiums on the sample of closed-end country funds for the period January 1988-December 1997, and compares the distributions of country fund premiums before and during the crisis.. The premium on a fund is calculated as the percentage difference of the share price and dividends (if any) and the net asset values (NAVs). The t-statistic in the last column is from regressions of fund premium on a constant and a dummy variable which is zero for the 25 weeks before the crisis and 1 for 25 weeks subsequent to the initial onset of the crisis. The t-statistics are corrected for serial correlation using 13 Newey-West lags. A significantly positive coefficient on the dummy variable implies that the premiums during the crisis are significantly higher on average compared to the premiums before the crisis for the fund. Note that FRF2 refers to the second event for France Growth Fund and similarly IRL2, ITA2, ITA3, ITA4, SNF2 and UKM2 refer to subsequent events for these funds. Panel A: Emerging Markets Funds Country Fund Premiums Descriptive Statistics Response of Country Fund Premiums to Currency Crises Premiums before the crisis Fund Symbol1 AF BZF CRF IGF IF JGF KF KIF KEF MF MXF MXE MEF FPF NSA TTF TC TKF Average No of obs. 324 506 172 481 408 404 521 308 215 522 514 385 378 426 201 513 398 422 Mean (%) 3.00 -6.52 -8.84 2.02 12.63 2.22 30.23 3.32 -0.16 -1.71 -8.60 -4.20 -4.97 -16.27 -18.56 10.40 -2.44 6.03 Min(%) -19.25 -54.92 -23.68 -31.24 -21.04 -30.34 -18.38 -16.49 -19.09 -24.85 -33.19 -25.03 -25.11 -33.61 -23.99 -20.89 -24.02 -36.10 Max(%) 38.39 35.11 10.42 54.89 49.13 34.94 127.26 48.47 54.76 77.81 31.01 46.39 40.70 63.16 0.00 94.55 66.67 100.26 Std. (%) 11.10 17.38 7.32 16.03 13.75 10.63 30.09 11.89 11.07 13.55 10.27 12.95 13.25 10.33 3.59 21.18 15.90 20.65 Mean (%) 5.76 -26.50 -12.84 -14.40 -8.98 -12.32 1.95 -3.33 -1.17 -9.80 -4.99 2.88 -2.39 -16.75 -20.30 19.30 15.28 3.26 -4.74 Premiums during the crisis Std. (%) 6.42 6.37 4.21 8.29 19.55 13.24 14.25 11.11 13.85 15.97 5.92 8.64 7.93 5.72 1.85 10.95 10.15 22.02 10.36 2.01* -3.87* -1.64 1.16 4.67* 4.67* 2.22* 5.98* 3.92* 4.77* 7.96* 7.73* 16.98* 8.07* 0.76 7.56* 8.93* 7.60* t-statistic for difference of before and after means

Std. (%) Mean (%) 4.51 5.16 2.95 4.26 2.01 1.55 9.23 6.64 7.50 2.92 2.68 4.61 2.50 2.01 2.30 9.26 7.49 8.44 4.78 11.64 -41.81 -16.15 -9.22 24.20 11.60 20.98 24.24 26.65 24.18 12.90 27.59 28.40 -2.57 -19.83 47.32 46.98 52.67 14.99

Table 2: Continued
Panel B: Developed Market Funds

Country Fund Premiums Descriptive Statistics Fund Symbol1 No of obs. Mean (%) Min(%) Max(%) Std. (%) Symbol

Response of Country Fund Premiums to Currency Crises Premiums before the crisis Premiums during the t-Statistic crisis for difference of before and after means Mean (%) Std. (%) -10.14 3.23 8.38* 2.15 4.93 4.06* -19.57 3.19 1.08 -10.61 2.99 3.26* 5.98 15.82 1.40 -0.30 6.88 4.06* 4.03 5.52 -2.63* -11.03 2.78 -0.24 2.52 4.13 8.07* 0.28 3.10 6.77* 4.50 3.58 4.24* -13.74 2.92 0.63 -5.99 2.82 5.65* -8.18 3.19 4.01* -2.26 2.99 5.41* -4.16 4.54 6.28 7.71

FRF IRL ITA

400 406 520

-14.41 -14.65 -9.73

-33.23 -33.42 -31.82

23.08 7.53 41.68

SGF SNF GSP UKM

389 493 412 512

-1.57 1.23 -15.34 -13.70

-27.58 -23.51 -28.06 -26.07

36.28 144.55 13.53 2.56

7.71 FRF FRF2 5.53 IRL IRL2 10.28 ITA ITA2 ITA3 ITA4 10.14 SGF 27.83 SNF SNF2 5.25 GSP GSP2 4.99 UKM UKM2

Average Average (full sample) Paired t-test for full sample1 [p-value] z-statistic1 for full sample [p-value] 1 The name of the fund corresponding to these fund ticker symbols are reported in Table 1 * Denotes significance at p < .05

Mean (%) -17.67 -5.48 -22.31 -14.64 -3.87 -10.63 9.47 -10.56 -8.33 -7.87 -0.06 -14.42 -13.79 -11.72 -8.24 -9.34 -6.83 5.348* [0.00] 4.333* [0.000]

Std. (%) 2.25 2.91 5.91 2.92 11.28 3.29 5.04 4.41 2.45 3.80 3.55 2.37 2.82 2.24 3.31 3.90 4.38 4.16* [0.00] 4.316* [0.000]

Table 3 The Impact of Currency Crises on the Volatility of Country Fund Share and NAV Returns This Table compares the relative volatility of fund share price returns and NAV returns before and during the currency crises. In addition, the relative increase or decrease of each of these, i.e., share price returns and NAV returns are compared by examining independently the ratio of share return variances (NAV return variances) prior to and during the crisis. Panel A: Emerging Markets Funds Before the crisis During the crisis Relative return variances for share prices and NAV Fund Std. Dev. Std. Dev. of Ratio of Std. Std. Dev.of Std. Dev. of Ratio of Std. ratio of share return ratio of NAV return Symbol of Share variances* NAV Dev. of Share NAV Dev. of variances* 2 Returns ((4)/(2))2 Returns shares and Returns Returns shares and ((3)/(1)) NAVs NAVs (1) (2) (1)/(2) (3) (4) (3)/(4) AF 5.24 3.65 1.44 4.31 2.41 1.78 0.67 0.44 BZF 4.22 2.35 1.79 6.78 8.01 0.85 2.58 11.59 CRF 4.10 2.43 1.69 3.77 2.26 1.67 0.84 0.87 IGF 4.35 3.29 1.33 4.50 5.15 0.87 1.07 2.46 IF 2.87 2.85 1.01 9.25 11.49 0.81 10.40 16.23 JGF 2.54 2.53 1.00 8.48 10.80 0.78 11.15 18.19 KF 6.66 6.36 1.05 12.72 12.73 1.00 3.65 4.00 KIF 6.07 7.17 0.85 11.81 14.55 0.81 3.79 4.11 KEF 6.46 7.19 0.90 10.02 12.39 0.81 2.41 2.97 MF 3.20 2.93 1.09 7.64 9.43 0.81 5.71 10.38 MXF 3.99 3.58 1.11 8.19 10.83 0.76 4.21 9.14 MXE 3.78 2.03 1.86 11.08 10.12 1.09 8.59 24.91 MEF 4.23 3.31 1.28 10.05 10.60 0.95 5.65 10.25 FPF 3.41 3.01 1.13 4.58 6.83 0.67 1.81 5.16 NSA 3.00 2.20 1.37 2.05 2.12 0.97 0.46 0.93 TTF 5.41 4.61 1.17 7.27 8.22 0.88 1.80 3.17 TC 3.33 4.26 0.78 8.99 6.06 1.48 7.30 2.02 TKF 5.88 5.88 1.00 8.75 14.13 0.62 2.21 5.77 Average 1.21 0.98 (emerging markets)

Table 3: Continued Panel B: Developed Markets Funds Before the crisis Fund Symbol During the crisis Relative return variances for share prices and NAV Ratio of Std. ratio of share return ratio of NAV return Dev. of variances* variances* shares and (3)/(1) (4)/(2) NAVs (3)/(4) 1.78 2.06 1.38 2.51 2.19 0.80 1.32 0.68 1.59 1.34 1.98 1.45 4.52 1.24 0.85 1.25 2.37 1.75 1.46 0.66 0.91 0.78 0.63 1.11 0.96 3.58 6.82 1.48 1.51 0.91 1.66 0.96 1.45 1.25 1.30 0.96 1.44 0.94 1.28 1.43 1.40 1.69 1.53 1.69 0.82 1.65 3.61

Std. Dev. of Std. Dev. of Ratio of Std. Std. Dev.of Std. Dev. of Share NAV Dev. of Share NAV Returns Returns shares and Returns Returns (1) (2) NAVs (3) (4) (1)/(2) FRF 2.99 2.05 1.45 4.29 2.41 FRF2 3.27 2.16 1.52 4.85 1.93 IRL 4.74 2.34 2.02 3.89 2.96 IR2 2.60 2.28 1.14 3.66 2.74 ITA 6.57 1.75 3.75 7.31 1.62 ITA2 2.96 2.76 1.07 4.56 3.65 ITA3 5.10 2.98 1.71 4.15 2.84 ITA4 3.25 3.16 1.03 2.58 3.34 SGF 2.28 1.73 1.32 4.32 4.52 SNF 3.27 2.85 1.15 4.02 2.72 SNF2 3.93 1.93 2.03 3.85 2.32 GSP 3.01 2.80 1.08 3.43 2.75 GSP2 2.84 1.69 1.68 2.76 1.91 UKM 2.91 1.85 1.57 3.45 2.41 UKM2 2.60 2.43 1.07 3.38 2.20 Average 1.57 (developed markets) Average 3.97 3.16 1.38 6.08 6.01 1.28 2.35 (full sample) t-statistic** 4.78 4.90 1.25 [p-value] [0.00] [0.000] [0.21] z-statistic*** 3.86 3.99 1.581 [p-value] [0.00] [0.00] [(0.11] *All ratios greater than 2.05 are significant at p<.05 **Paired t-statistic tests the hypothesis that the volatility of share prices and NAV prior to and during the crises are equal. ***Wilcoxon z-statistic comparing the price and NAV volatility prior to and during the crisis.

Table 4 Speed of Adjustment of Shares Prices to NAV Changes This Table provides the coefficients on NAV returns (βn) derived from an OLS regression of price returns on a constant, NAV returns (rn,t), lagged nav returns (rn-1,t), and lagged price returns (rp-1,t). Specifically, the empirical model is: rp , t = α p + β n rn , t + β n −1rn −1, t + β p rp −1, t + e p, t Panel A: Emerging Markets Funds Before the crisis During the crisis Symbol t-stat t-stat βn βn AF 0.07 0.29 0.60 0.92 BZF 1.58 3.18* 0.38 2.78* CRF 0.15 0.49 0.98 3.70* IGF 0.41 1.35 0.54 4.21* * IF 0.59 5.04 0.42 2.33* * JGF 0.74 6.46 0.54 4.82* KF 0.48 1.76 0.76 3.20* KIF 0.48 1.65 0.78 8.05* KEF 0.29 1.20 0.61 6.63* * MF 0.60 2.60 0.49 4.71* MXF 0.94 6.63* 0.61 4.25* * MXE 1.16 4.05 0.51 1.33 MEF 1.02 9.33* 0.42 1.25 * 0.41 3.99* FPF 0.99 5.57 * NSA 0.56 5.37 0.24 1.23 0.33 1.95* TTF 0.73 3.81* TC 0.42 3.73* 0.57 1.81 TKF 0.05 0.29 0.28 2.70* Panel B: Developed Markets Funds Symbol t-stat t-stat βn βn * FRF 0.70 2.76 0.53 1.17 FRF2 0.90 2.73* 1.39 2.45* IRL 1.05 2.58* 0.64 3.74* IR2 -0.02 -0.10 -0.06 -0.34 ITA 0.93 1.01 1.86 2.49* * ITA2 0.60 2.12 0.04 0.18 0.59 2.30* ITA3 0.68 2.02* ITA4 0.39 1.55 0.34 1.70 0.54 2.53* SGF 0.78 3.14* SNF 0.19 0.75 0.40 1.16 SNF2 0.15 0.30 0.62 1.70 0.59 2.08* GSP 0.54 2.53* GSP2 0.23 0.72 0.28 0.54 UKM 0.29 0.50 0.46 1.23 UKM2 -0.08 -0.33 0.81 2.13* Average 0.56 0.56 t-statistic* -0.03 [p-value] [0.97] z-statistic** -0.089 [p-value] [0.92] 1 Significant at p < .05, *t-statistic compares the average coefficient on NAV returns before and during the crisis, **z-statistic is obtained from the Wilcoxon signed-rank test comparing the coefficient on NAV returns prior to and during the crisis.

Table 5 Risk Exposures of Country Fund Share/NAV Returns to the World Market Index Return (Excluding the U.S), Local Market Index Return and Exchange Rate Changes The models estimated are of the form:
r p ,t = α
p

+ β p ,w r w ,t + β p ,h r h , t + β p ,x r x, t + e p ,t

where, rp,t is the total return on the shares (including dividends) of a country fund between time t-1 and t, rw,t is the return on the Morgan Stanley Capital International (MSCI) world market index excluding the US, rh,t is the return on the MSCI index corresponding to the country of origin of the fund and rx,t is the logarithm changes in the exchange rate of the country with the U.S. dollar. Panel A (B) presents the results of estimating the model using share returns (NAV returns) as the dependent variable. Panel A: Dependent Variable = Share Returns t-stat t-stat t-stat Adj R2 Symbol Const. t-stat βp,h βp,x βp,w AF 0.00 -0.19 -0.02 -0.13 0.60* 8.78 0.00 -0.94 0.36 * * BZF 0.00 -0.77 0.41 2.97 0.53 10.48 0.00 -0.01 0.40 * * CRF 0.00 -1.07 0.25 1.14 0.58 4.88 -0.66 -2.99 0.24 IGF 0.00 -0.98 0.28 1.31 0.41* 5.60 0.71 1.21 0.13 IF 0.00 -1.12 0.57* 3.38 0.42* 4.89 -0.08 -0.46 0.21 * JGF 0.00 -1.25 0.54 3.68 0.47* 6.29 -0.10 -0.64 0.30 * * 4.68 0.66 10.18 -0.04 -0.09 0.35 KF 0.00 -0.77 0.58 * * KIF 0.00 -0.82 0.33 2.55 0.57 4.73 -0.10 -0.32 0.39 KEF -0.01 -1.91 0.64* 2.66 0.56* 10.28 -0.23 -0.70 0.47 * * MF 0.00 -0.08 0.42 2.87 0.57 7.37 0.29 0.80 0.19 MXF 0.00 -0.73 0.56* 3.67 0.75* 10.92 0.00 0.13 0.43 * * 2.36 0.47 3.93 0.01 0.19 0.19 MXE 0.00 -0.12 0.46 * * MEF 0.00 -1.05 0.43 1.98 0.66 4.63 0.01 0.20 0.29 FPF 0.00 -0.98 0.37 1.87 0.36* 2.57 0.36 1.26 0.28 * NSA 0.00 -1.11 -0.02 -0.17 0.63 7.55 -0.28 -1.28 0.29 TTF 0.00 -0.67 0.34* 2.23 0.48* 8.42 -0.10 -0.64 0.21 * * TC 0.00 -1.45 0.53 4.32 0.43 7.74 -0.30 -1.92 0.33 * * TKF 0.00 -1.12 0.35 2.35 0.34 7.28 0.05 0.36 0.25 FRF 0.00 -1.22 0.40* 2.65 0.72* 7.60 -0.52* -3.75 0.29 * * * 4.79 0.23 2.65 -0.28 -2.68 0.16 IRL 0.00 0.44 0.45 ITA 0.00 -0.10 0.40* 3.18 0.50* 7.73 -0.20 -1.43 0.20 SGF 0.00 -0.84 0.18 1.40 0.48* 3.64 -0.13 -0.38 0.09 * * 3.47 0.41 4.57 -0.21 -1.40 0.13 SNF 0.00 -0.46 0.55 GSP 0.00 -0.70 0.29* 3.49 0.62* 9.20 -0.20* -2.08 0.38 * * * UKM 0.00 0.47 0.29 2.90 0.73 7.34 -0.49 -4.27 0.25 Average 0.00 -0.74 0.38 2.62 0.53 6.77 -0.10 -0.87 0.27

Table 5: Continued Panel B: Dependent Variable = NAV Return Symbol const. t-stat t-stat βn,w βn,h AF 0.00 1.52 0.03 0.57 0.62* BZF 0.00 -0.26 0.09 0.76 0.71* * CRF 0.00 -0.02 0.17 2.19 0.58* IGF 0.00 -1.18 0.13 1.07 0.58* IF 0.00 -0.64 0.03 0.42 0.74* JGF 0.00 -0.83 0.02 0.30 0.71* KF 0.00 0.00 0.07 1.31 0.76* KIF 0.00 -0.93 0.06 0.67 0.83* KEF 0.00* -2.44 -0.09 -0.96 0.80* MF 0.00 -0.80 -0.07 -1.71 0.90* MXF 0.00 -0.15 -0.02 -0.18 0.80* MXE 0.00 -0.25 0.07 0.60 0.57* MEF 0.00 -0.69 0.02 0.14 0.74* FPF 0.00 -1.42 -0.04 -0.36 0.72* NSA 0.00 -0.18 -0.05 -0.98 0.68* TTF 0.00 -1.36 -0.05 -1.64 0.86* TC 0.00 -1.21 -0.05 -1.14 0.70* TKF 0.00 -0.93 -0.05 -0.81 0.87* FRF 0.00 0.18 -0.05 -1.50 0.81* IRL 0.00 0.78 -0.02 -0.53 0.67* ITA 0.00 0.28 0.00 -0.07 0.75* -2.15 0.62* SGF 0.00 -1.27 -0.14* SNF 0.00 0.56 0.02 0.42 0.78* GSP 0.00 0.19 0.04 1.13 0.75* UKM 0.00 -0.71 0.00 -0.06 0.68* Average 0.00 -0.47 0.00 -0.10 0.73

t-stat 22.40 16.54 12.15 10.39 17.76 13.65 21.26 13.53 22.27 17.43 9.15 7.32 9.55 8.68 10.94 63.35 20.35 35.28 24.10 21.06 35.28 7.03 22.38 27.92 11.49 19.25

βn,x 0.00 0.00 -0.32* -0.24 -0.02 0.04 0.15* -0.04 0.08 0.21 0.00 0.02 0.01 0.30 0.05 0.10* -0.02 0.03 0.22* 0.16* 0.12* 0.58* 0.13* 0.01 0.19* 0.07

t-stat -0.50 0.59 -3.60 -0.49 -0.31 0.39 2.71 -0.57 0.46 1.38 0.13 1.07 0.46 1.37 0.38 5.82 -0.40 0.55 6.21 3.64 2.95 2.33 3.02 0.24 3.45 1.25

Adj R2 0.86 0.65 0.71 0.32 0.82 0.83 0.72 0.80 0.89 0.79 0.54 0.37 0.47 0.79 0.67 0.87 0.81 0.89 0.77 0.79 0.82 0.33 0.76 0.79 0.60 0.71

Panel C: Dependent Variable = Changes in Premiums Symbol Const. t-stat t-stat βpd,w AF 0.00 -0.64 -0.07 -0.40 BZF 0.00 -0.60 0.32* 2.33 CRF 0.00 -1.01 0.06 0.34 IGF 0.00 -0.20 0.18 0.84 IF 0.00 -0.84 0.67* 3.07 3.34 JGF 0.00 -0.96 0.57* * KF 0.00 -0.62 0.60 3.26 KIF 0.00 -0.21 0.25 1.64 KEF 0.00 -0.72 0.67* 2.42 * MF 0.00 0.17 0.52 3.22 3.01 MXF 0.00 -0.56 0.52* * MXE 0.00 0.00 0.45 1.98 MEF 0.00 -0.47 0.40 1.59 FPF 0.00 -0.33 0.36* 2.10 NSA 0.00 -1.16 0.02 0.29 TTF 0.00 -0.26 0.49* 2.56 TC 0.00 -0.97 0.59* 4.87 * TKF 0.00 -0.32 0.44 2.25 FRF 0.00 -1.33 0.36* 2.89 * IRL 0.00 0.18 0.39 5.20 ITA 0.00 -0.21 0.35* 2.91 * SGF 0.00 -0.22 0.30 2.54 * SNF 0.00 -0.34 0.71 3.05 GSP 0.00 -0.75 0.22* 2.82 * UKM 0.00 0.61 0.27 2.78 Average 0.00 -0.47 0.39 2.44 * Significant at p < .05

βpd,h -0.03 -0.18* -0.01 -0.20* -0.38* -0.26* -0.10 -0.29* -0.26* -0.37* -0.05 -0.15 -0.10 -0.35* -0.03 -0.46* -0.30* -0.64* -0.08 -0.38* -0.23* -0.13 -0.45* -0.11 0.02 -0.22

t-stat -0.43 -3.77 -0.11 -2.05 -2.67 -2.13 -1.43 -3.56 -3.69 -3.69 -0.41 -0.71 -0.44 -2.21 -0.32 -5.90 -4.60 -7.77 -0.96 -5.31 -3.40 -0.92 -3.57 -1.80 0.23 -2.46

βpd,x 0.00 -0.01 -0.30 1.03 -0.09 -0.18* -0.14 0.03 -0.24 0.11 0.00 -0.02 -0.01 0.06 -0.28 -0.30 -0.40* 0.06 -0.64* -0.38* -0.30* -0.68 -0.30 -0.18 -0.56* -0.15

t-stat -0.12 -0.34 -1.47 1.65 -0.55 -2.57 -0.27 0.09 -0.48 0.25 -0.01 -0.93 -0.16 0.17 -1.69 -1.46 -2.93 0.30 -5.15 -4.26 -2.07 -1.82 -1.39 -1.76 -5.39 -1.29

Adj R2 -0.01 0.07 -0.01 0.02 0.12 0.11 0.02 0.09 0.12 0.06 0.03 0.03 0.01 0.16 0.01 0.09 0.14 0.32 0.09 0.15 0.05 0.01 0.03 0.02 0.06 0.07

Table 6 Relation of Country Fund Share/NAV Returns to the World Market Index Return (Excluding the US), Local Market Index Return and Exchange Rate Changes: Pre-Crisis Vs. Post-Crisis This Table estimates a model similar to that contained in Table 5. The models are estimated separately for pre-crisis and crisis periods. The models estimated are of the form: r p , t = α p + β p , w rw , t + β p , h r h , t + β p , x rx , t + e p , t where, rp,t is the total return on the shares (including dividends) of a country fund between time t-1 and t, rw,t is the return on the Morgan Stanley Capital International (MSCI) world market index excluding the US, rh,t is the return on the MSCI index, corresponding to the country of origin of the fund and rx,t is the logarithm changes in the exchange rate of the country with the U.S. dollar. In Panel A.1 (B.1), the model is estimated using share return (NAV return) as the dependent variable for the pre-crisis period. In Panel A.2 (B.2), the model is estimated using share return (NAV return) for the 25 weeks that are defined as the crisis period. Results not reported for KF, KIF and KEF in crisis period due to missing data. Panel A.1: Before Crisis: Dependent Variable = Share Return t-stat t-stat t-stat Adj R2 Symbol const. t-stat βp,h βp,x βp,w AF -0.01 -0.68 -0.91 -1.83 0.27 1.32 2.80 0.89 0.07 * CRF -0.01 -1.30 0.20 0.43 0.41 2.52 -0.58 -1.26 0.04 IF 0.00 0.41 0.29 0.43 0.31 0.97 0.95 0.54 0.00 JGF 0.00 0.73 0.02 0.07 0.58* 3.81 -0.01 0.00 0.44 KF -0.01 -0.96 1.25 1.96 0.48 1.07 -0.47 -0.45 0.35 KIF -0.01 -1.20 1.02* 2.41 0.60 1.81 -1.05 -1.48 0.37 KEF -0.01 -1.20 1.60* 2.51 0.48 1.41 -0.69 -0.68 0.49 * MF 0.00 -0.63 0.01 0.02 0.85 2.69 3.01 1.61 0.45 MXF 0.00 -0.03 1.00* 2.84 0.89* 3.44 -2.23 -1.05 0.58 * MXE 0.00 0.57 0.81 1.65 0.62 2.74 -2.03 -0.85 0.25 MEF 0.00 -0.10 0.89 1.84 0.85* 3.65 -1.38 -0.52 0.48 * * FPF 0.00 0.04 0.27 0.55 0.81 4.12 0.92 2.18 0.46 * NSA 0.00 0.17 0.18 0.29 0.66 2.16 0.24 0.46 0.23 TTF 0.01 0.69 0.53 0.73 0.69* 4.17 -0.76 -0.18 0.36 * 3.83 -0.24 -0.07 0.20 TC 0.00 -0.37 -0.03 -0.06 0.32 TKF 0.01 0.38 0.24 0.27 0.10 0.57 0.33 0.24 -0.12 FRF 0.01 1.24 0.30 1.39 1.07* 5.00 -1.18* -3.02 0.52 IRL 0.00 0.26 0.56 1.49 0.30 0.79 -1.14 -1.18 0.38 ITA 0.01 1.27 0.36 0.30 0.59 0.71 1.32 1.02 0.06 SGF 0.00 -0.43 0.45 1.27 0.55 1.58 -0.86 -1.09 0.18 SNF -0.01 -0.62 -0.15 -0.43 0.63 1.77 -0.52 -0.63 0.12 GSP -0.01 -0.95 0.01 0.05 0.46 1.09 0.42 0.62 0.19 * UKM 0.00 -0.18 -0.10 -0.36 0.73 3.31 0.70 1.63 0.57 Average 0.00 -0.13 0.38 0.77 0.58 2.37 -0.10 -0.14 0.29

Table 6: Continued Panel A.2: Before Crisis: Dependent Variable = NAV Return Symbol const. t-stat t-stat βn,w βn,h AF 0.00 0.60 0.00 0.04 0.50* CRF 0.00 0.70 0.04 0.28 0.69* IF 0.00 1.22 0.03 0.15 0.82* JGF 0.00 1.41 0.06 0.24 0.75* KF 0.00 0.53 -0.04 -0.15 0.83* KIF 0.00 0.21 -0.01 -0.04 1.04* KEF 0.00 0.44 -0.22 -0.76 0.96* MF 0.00 -1.68 0.09 0.48 0.85* MXF 0.00 0.25 -0.02 -0.10 0.94* MXE 0.00 1.56 -0.03 -0.21 0.53* MEF 0.00 0.07 0.01 0.06 0.86* FPF 0.00 0.37 -0.02 -0.09 0.94* NSA 0.00 -0.23 -0.17 -1.45 0.81* * TTF 0.00 1.31 -0.23 -3.07 0.78* TC 0.00 -0.71 -0.04 -0.24 0.69* * TKF 0.00 0.12 -0.80 -2.29 0.70* -2.06 -0.04 -0.46 0.83* FRF 0.00* IRL 0.00 -1.59 -0.01 -0.07 0.61* * ITA 0.00 1.79 0.61 4.00 0.51* SGF 0.00 -0.34 0.05 0.24 0.60* SNF -0.01 -1.46 -0.38 -1.88 0.80* GSP 0.00 -1.30 0.11 0.76 0.78* UKM 0.00 0.11 0.06 0.18 0.05 Average 0.00 0.06 -0.04 -0.19 0.73 βn,x 1.31* -0.04 0.00 0.46 0.06 -0.32 0.06 -0.64 -0.05 -0.06 0.15 2.91* 0.11 -1.09 1.06 -0.18 0.46* 0.33 -0.55* 0.74 0.80* 0.17 0.50 0.07 Adj R2 0.97 0.71 0.84 0.83 0.91 0.95 0.93 0.90 0.77 0.75 0.79 0.90 0.81 0.99 0.89 0.76 0.93 0.91 0.67 0.52 0.61 0.90 -0.02 0.79

t-stat 17.53 10.07 8.18 6.02 5.63 8.80 6.51 13.37 6.42 5.46 7.26 10.03 8.35 28.30 8.66 7.02 9.54 7.86 5.87 3.17 5.46 8.09 0.19 8.60

t-stat 2.78 -0.19 0.00 0.71 0.12 -0.85 0.12 -1.28 -0.04 -0.07 0.12 14.88 0.49 -1.61 0.73 -0.15 4.88 1.70 -2.52 1.34 3.29 1.09 0.75 1.14

Table 6: Continued Panel A.3: Before Crisis: Dependent Variable = Change in Premiums Symbol const. t-stat t-stat t-stat βn,w βn,h AF -0.01 -0.79 -0.98 -1.83 -0.25 -1.19 CRF -0.01 -1.50 0.14 0.36 -0.24 -1.54 IF 0.00 0.04 0.23 0.51 -0.47* -2.13 JGF 0.00 0.11 -0.04 -0.16 -0.15 -1.35 KF -0.02 -1.00 1.18 1.68 -0.32 -0.76 * 2.61 -0.44 -1.82 KIF -0.01 -1.60 0.91 KEF -0.02 -1.21 1.66 2.10 -0.48 -1.21 MF 0.00 0.00 -0.06 -0.10 0.03 0.09 * 4.49 -0.04 -0.23 MXF 0.00 -0.25 0.95 MXE 0.00 0.04 0.84 2.00 0.09 0.46 * MEF 0.00 -0.19 0.84 2.51 0.00 -0.01 FPF 0.00 -0.10 0.23 0.44 -0.12 -0.52 NSA 0.00 0.22 0.30 0.56 -0.11 -0.39 TTF 0.01 0.60 0.91 1.06 -0.12 -0.60 TC 0.00 -0.17 -0.03 -0.05 -0.45* -2.99 * TKF 0.01 0.22 0.98 0.85 -0.63 -2.66 FRF 0.01 1.69 0.28 1.30 0.20 0.93 IRL 0.00 0.54 0.45 1.77 -0.26 -1.01 ITA 0.01 0.88 -0.21 -0.18 0.19 0.23 SGF 0.00 -0.18 0.36 1.44 -0.04 -0.22 SNF 0.00 -0.01 0.20 0.53 -0.16 -0.48 GSP -0.01 -0.66 -0.08 -0.39 -0.30 -0.75 UKM 0.00 -0.26 -0.14 -0.38 0.59* 3.90 Average 0.00 -0.16 0.39 0.92 -0.15 -0.62 βn,x 1.52 -0.46 0.88 -0.42 -0.53 -0.79 -0.72 3.29 -2.09 -2.07 -1.54 -1.70* 0.10 0.30 -2.26 0.64 -1.36* -1.11 1.68 -1.49* -1.19 0.23 0.17 -0.22

t-stat 0.45 -1.09 0.68 -0.49 -0.51 -1.55 -0.56 1.86 -1.37 -1.03 -0.76 -4.56 0.21 0.06 -0.51 0.31 -4.25 -1.55 1.38 -3.26 -1.45 0.39 0.33 -0.75

Adj R2 0.24 -0.03 0.18 -0.08 0.28 0.60 0.53 0.13 0.23 0.02 0.09 0.23 -0.14 -0.01 0.24 0.32 0.36 0.10 -0.04 0.13 0.07 -0.01 0.18 0.16

Table 6: Continued Panel B.1: During Crisis: Dependent Variable = Share Return Symbol const. t-stat t-stat βp,w βp,h AF 0.00 -0.05 0.03 0.05 0.59 CRF 0.00 -0.14 0.43 0.87 0.46 IF -0.01 -0.67 0.77 0.81 0.40* JGF -0.02 -1.65 0.98 1.81 0.39* KF -0.01 -0.64 2.09* 2.14 0.59* KIF -0.01 -0.63 0.67 1.07 0.62* KEF 0.00 -0.27 0.36 0.62 0.55* MF 0.03 0.84 1.52* 2.51 0.56* MXF 0.00 -0.07 0.29 0.39 0.47* MXE -0.01 -0.66 1.88 1.51 0.04 MEF -0.01 -0.27 1.19 1.15 0.24 * FPF -0.01 -1.23 0.99 2.57 -0.12 NSA 0.00 -0.38 -0.11 -0.38 0.34 TTF -0.02 -0.98 0.56 1.13 0.05 TC -0.02 -2.10* 0.79 1.70 0.18 TKF -0.04 -1.49 1.52 1.26 0.24 FRF 0.00 -0.24 -0.93 -1.32 0.76* IRL 0.00 0.17 1.15* 2.39 -0.15 ITA 0.00 -0.30 0.55 0.65 0.98 SGF -0.01 -1.22 0.41 0.61 0.37 SNF -0.01 -0.99 -0.64 -1.12 0.49* GSP 0.00 -0.77 0.30 0.56 0.77* UKM -0.01 -1.05 0.70 1.16 0.93* Average -0.01 -0.64 0.67 0.96 0.42 βp,x 16.68* -0.60 -0.14 -0.16* 0.00 -0.04 -0.11 2.17 0.44 0.70 0.85 0.92* -0.30 3.03 0.86 -0.44 -0.77* -0.32 0.23 0.24 -0.07 -0.21 -0.88 0.96

t-stat 1.67 1.56 3.00 3.38 7.05 6.45 9.52 3.59 3.87 0.31 1.46 -0.85 1.20 0.54 1.94 1.83 3.28 -0.47 0.78 1.57 2.53 3.72 2.35 2.62

t-stat 2.03 -1.51 -0.82 -3.84 0.02 -0.23 -0.90 0.98 1.13 1.25 1.58 2.71 -0.45 1.98 0.66 -0.56 -2.36 -0.57 0.23 0.28 -0.41 -0.72 -3.51 -0.13

Adj R2 0.28 0.14 0.29 0.51 0.77 0.73 0.74 0.51 0.26 -0.01 0.07 0.31 -0.08 0.15 0.22 0.17 0.23 0.27 0.01 0.08 0.08 0.47 0.27 0.28

Table 6: Continued Panel B.2: During Crisis: Dependent Variable = NAV Return Symbol Const. t-stat t-stat βn,w βn,h AF 0.00 0.59 0.00 0.03 0.51* * CRF 0.00 0.17 0.27 2.32 0.53* IF -0.01 -1.24 0.19 0.66 0.74* -2.54 -0.21 -0.65 0.75* JGF -0.02* KF 0.00 0.73 0.13 0.31 0.77* KIF -0.01 -1.84 0.45 1.55 0.89* KEF 0.00 -0.52 -0.10 -0.30 0.76* MF -0.01 -0.73 0.20 0.49 0.97* MXF -0.01 -0.35 0.66 0.58 0.68* MXE -0.01 -0.43 0.05 0.04 0.46 MEF -0.02 -0.85 0.28 0.28 0.56 FPF 0.00 -1.43 -0.09 -0.60 0.78* NSA 0.00 0.73 0.07 0.63 1.08* TTF 0.00 -0.57 -0.12 -0.55 0.84* TC -0.01 -1.56 -0.25 -0.89 0.57* TKF -0.01 -1.49 0.14 0.25 0.90* FRF 0.00 0.63 -0.24 -1.92 0.95* IRL 0.00 -0.53 0.04 0.19 0.65* ITA 0.00 1.86 -0.09 -1.22 0.68* SGF -0.01 -1.49 0.08 0.28 0.75* SNF 0.00 -0.66 0.10 0.35 0.66* GSP 0.00 -0.22 -0.25 -1.07 0.82* UKM 0.00 0.87 -0.16 -1.10 0.99* Average -0.01 -0.47 0.05 -0.01 0.75 βn,x 0.03 -0.39 -0.05 -0.05 0.09 -0.05 -0.01 -0.23 0.24 0.53 0.55 0.08 -0.11 0.22 0.02 -0.22 0.29* -0.17 0.28 0.75 0.20 -0.08 0.02 0.08

t-stat 10.22 9.43 10.29 8.93 13.49 34.31 33.56 7.64 2.05 1.68 1.79 13.86 8.15 21.57 9.75 15.96 13.76 4.35 4.98 4.94 6.35 8.13 6.18 10.93

t-stat 0.02 -2.86 -1.33 -1.93 1.24 -1.38 -0.29 -0.34 0.49 1.21 1.05 0.68 -0.40 0.42 0.03 -0.92 4.71 -0.71 1.00 0.92 1.71 -0.68 0.09 0.12

Adj R2 0.88 0.83 0.92 0.91 0.97 0.98 0.98 0.87 0.32 0.12 0.20 0.92 0.85 0.95 0.74 0.94 0.92 0.84 0.51 0.52 0.86 0.83 0.79 0.77

Table 6: Continued Panel B.3: During Crisis: Dependent Variable = Change in Premiums Symbol const. t-stat t-stat t-stat βn,w βn,h AF 0.00 -0.17 0.03 0.04 0.12 0.29 CRF 0.00 -0.14 0.15 0.38 -0.08 -0.37 IF -0.01 -0.29 0.92 0.64 -0.42 -1.75 JGF -0.01 -0.38 1.40 1.62 -0.40 -1.86 KF -0.02 -0.88 2.77 1.50 -0.25* -2.11 KIF 0.00 -0.22 0.31 0.39 -0.34* -3.18 KEF 0.00 -0.21 0.77 0.74 -0.30* -3.55 * * 2.13 -0.47 -2.31 MF 0.05 0.88 1.76 MXF 0.01 0.25 -0.44 -0.49 -0.21 -0.79 MXE -0.01 -0.38 2.35* 2.25 -0.55 -1.49 MEF 0.01 0.33 1.10 0.85 -0.35 -0.72 FPF -0.01 -0.86 1.07* 3.43 -0.89* -7.32 * NSA 0.00 -0.70 -0.14 -0.64 -0.60 -2.78 * TTF -0.02 -0.79 1.02 1.52 -1.20 -7.69 TC -0.01 -0.64 1.48 1.52 -0.58* -3.59 * TKF -0.03 -0.73 1.75 0.75 -0.94 -3.38 FRF 0.00 -0.38 -0.62 -1.02 -0.15 -0.75 IRL 0.00 0.43 0.90* 2.36 -0.65* -2.62 ITA -0.01 -0.75 0.57 0.56 0.57 0.40 SGF -0.01 -0.66 0.37 0.75 -0.41 -1.77 SNF 0.00 -0.69 -0.72 -1.32 -0.17 -0.72 GSP 0.00 -0.58 0.48 0.84 -0.05 -0.23 UKM -0.01 -1.26 0.76 1.39 -0.03 -0.07 Average 0.00 -0.38 0.78 0.88 -0.36 -2.10 * Significant at p < .05 βn,x 18.83* -0.16 -0.12 -0.16* -0.10 0.01 -0.12 2.70 0.20 0.25 0.33 0.81* -0.14 4.30 1.29 -0.23 -0.96* -0.11 0.13 -0.52 -0.26 -0.11 -0.83* 1.09 Adj R2 0.05 -0.10 0.13 0.31 0.12 0.24 0.19 0.34 -0.05 0.23 -0.02 0.64 0.30 0.56 0.18 0.39 0.43 0.09 -0.07 0.09 0.13 -0.10 0.25 0.19

t-stat 2.21 -0.44 -0.45 -2.51 -0.44 0.05 -0.63 0.94 0.31 0.41 0.41 2.95 -0.32 1.63 0.59 -0.15 -3.39 -0.22 0.13 -0.72 -1.33 -0.34 -2.57 -0.17

Table 7 Trading Volumes for Asian Funds Prior to and During Local Currency Crises This Table compares the average daily trading volume of nine Asian funds, for the 25 weeks prior to the Asian currency crisis with that for the 25 week crisis period. Average trading volume for the 25 weeks subsequent to the crisis period is also compared with the crisis period volumes. The tstatistics reported are derived from a regression of the trading volumes on a constant and a dummy variable which is 0 for the pre-crisis period and 1 for the crisis period. Symbol Pre-Crisis Volume Crisis Period Volume Change Post-Crisis Volume Change in Jan 97-June 97 July 97-dec 97 in Volume Jan 97-June 98 Volume Mean Std. Dev. Mean Std. Dev. Pre-Crisis t-Statistic Mean Std Dev. to Crisis for Period difference 24556.00 2.69* 54668.00 40951.06 26536.00 3.66* 20540.00 20918.75 294596.00 1.86* 436564.00 375676.97 40700.00 2.50* 45500.00 41005.47 49536.00 2.46* 70600.00 67022.73 63992.00 2.70* 42040.00 35738.42 65464.00 4.06* 36972.00 26984.57 30032.00 3.30* 50064.00 33796.00 52.00 0.007 21124.00 16565.28 Crisis to Post-Crisis Period 13620.00 -16892.00 -144020.00 -31752.00 -9044.00 -63748.00 -59096.00 -26116.00 -15696.00 t-statistic for difference 1.13 -2.07* -0.85 -1.83 -0.39 -2.65* -3.63* -2.56* -2.69*

IF 16492.00 13194.50 41048.00 43686.18 JGF 10896.00 9845.79 37432.00 34850.24 KF 285988.00 231734.88 580584.00 756449.12 KIF 36552.00 28548.91 77252.00 76174.94 KEF 30108.00 42541.41 79644.00 91250.22 MF 41796.00 28555.57 105788.00 114757.38 FPF 30604.00 24582.23 96068.00 76768.87 TTF 46148.00 24805.43 76180.00 38020.13 TC 36768.00 24003.26 36820.00 23996.84 *Significant at p<.05 or better

Figure 1: Premiums on Country Funds In all the graphs, country fund premiums are plotted for a period of 25 weeks before and 25 weeks after a currency crisis in a country.
AR G ENTIN A
30.00 25.00 20.00 15.00 10.00 5.00 0.00 -9 -1 -25 -17 8 16 24 -5.00 -30.00
EVEN T D ATE

IN DIA
10.00 0.00 -7 -25 -16 3 12 21 -10.00 -20.00 50.00 40.00 30.00 20.00 10.00 0.00 -10.00
-2

M EXIC O

9

-20.00

EVEN T D ATE

-1

EVEN T D ATE

B R AZIL
0.00
12 21 -7 -2 -1

IN DONES IA
100.00 10.00 0.00 -25

P H ILLIPIN ES

-10.00 -20.00 -30.00 -40.00 -50.00 -60.00

5

6

3

80.00 60.00 40.00 20.00 0.00
-5 -2 -1 16

-17

20

-3

5

4

16
21

-10.00 -20.00 -30.00
5 6

-20.00
EVEN T D ATE

5

EVEN T D ATE

EVEN T D ATE

C ZEC H R EP UB LIC
0.00 -9 -1 -25 -17 8 16 -5.00 -10.00 -15.00 -20.00 -25.00
EVEN T D ATE

K O REA
80.00 60.00 40.00 20.00 0.00 -25 -19 -13 -7 -1 -20.00 -40.00
EVEN T D ATE

SO U TH AFR ICA
0.00
-2 -1 12 -7

-5.00 -10.00 -15.00 -20.00 6 -25.00 -30.00

5

6

EVEN T D ATE

3

24

-9

-1

8

T H AIL AN D
100.00 80.00 60.00 40.00 20.00 0.00 20.00 10.00 0.00

F R A N C E -T W O E V E N T S
10.00 5.00 0.00 -25 -17 16 24 -25 -21

S IN G A P O R E

-17

-13

8

4 18

-9

-1

-9

-5

-10.00 -20.00

-5.00 -10.00 -15.00

20

5

4

-3

9

-30.00
E VE N T D AT E

-2

-1

E VE N T D AT E

E VE N T D AT E

M A LA Y S IA
80.00 60.00 40.00

IR EL A N D - T W O E VE N T S
0.00 -25 -17 16 24 -9 -1 8 -10.00 -20.00 20.00 10.00 0.00

S P A IN

20.00 0.00 -25 -17 16 24 -9 -1 8 -20.00 -30.00 -40.00
E VE N T D AT E

-2

-1

-20.00 -30.00
E VE N T D AT E

E VE N T D AT E

TUR K EY
120.00 100.00 80.00 60.00 40.00 20.00 0.00 -20.00 50.00

IT AL Y -F O U R E V E N T S
5.00 0.00

-2

-1

40.00 30.00 20.00 10.00 0.00 -10.00 -25 -18 -11 -4 -20.00 -30.00

-1

-10.00

11

18

-1

U .K -T W O E VE N T S

11

25
25

5

8

1

-4

4

-25

-18

-11

-5.00 -10.00 4 11 18 25 -15.00 -20.00

-9

-1

8

16

E VE N T D AT E

24

5

7

E VE N T D AT E

E VE N T D AT E

-4

4

8


				
DOCUMENT INFO
Shared By:
Tags: Why-D, o-Clo
Stats:
views:30
posted:12/21/2009
language:English
pages:46
Description: Why-Do-Closed-end-Country-Funds-Trade-at-Enormous-Premiums-during-