亞洲新興市場的匯率風險效果

亞洲新興市場的匯率風險效果

研究成果報告(精簡版)

計 畫 類 別 ： 個別型

計 畫 編 號 ： NSC 96-2416-H-004-045-

執 行 期 間 ： 96 年 11 月 01 日至 97 年 07 月 31 日

執 行 單 位 ： 國立政治大學金融系

計 畫 主 持 人 ： 林建秀

計畫參與人員： 碩士班研究生-兼任助理人員：蔡岳昆

博士班研究生-兼任助理人員：許嫚荏

處 理 方 式 ： 本計畫涉及專利或其他智慧財產權，2 年後可公開查詢

中 華 民 國 97 年 10 月 06 日

Chien-Hsiu Lin July 17, 2008

Abstract

This paper investigates the impact of foregin exchange rate change on stock returns in the Asian emerging markets. Real exchange rate is used in this paper to capture the high in‡ationary e¤ect in the emerging markets. I …nd stock returns expose to exchange risk to a lesser extent during currency appreciation than during currency depreciation. An asymmetric exchange exposure embeded asset pricing model is then built to test with small and large size portfolios. My empirical results show that large …rms display nothing on ‡uctuations of exchange rates while small …rms show statistically signi…cant exchange exposure. The results support the argument that large …rms have advantages in scale of economy to pay less hedging cost, hence more motivated to use hedges to reduce exchange rate risk.

K eyw ords: Exchange exp osure; A sym m etry; A ppreciation-D epreciation cycle; the A sian em erging m ar-kets

JE L classi…cation : F31; G 12; G 15

1. Introduction

Foreign exchange risk is one of the important factors in international asset pric-ing. We start by thinking how asset prices should be determined in a world where …nancial markets are assumed to be completely integrated with global markets and people all over the world face the same consumption and investment set. In such a world, there are no barriers to international investment and products can ex-port and imex-port freely across di¤erent countries. Therefore, an asset should have

National Chengchi University. Address correspondence to Chien-Hsiu Lin, Department of Money and Banking, National Chengchi University, No. 64, Sec.2 Zhinan Rd., Wenshan District, Taipei City 11605, Taiwan, or email to [email protected]

the same price regardless of where it is traded. Under such completely integrated markets, the asset pricing model would contain only global pricing factors. On the contrary, if markets are assumed to be fully segmented, local pricing factors would determine what price assets should have. In other words, under completely segmented markets, the price of an asset depends on where it is traded.

Both extreme cases mentioned above cannot be directly applied into the real world because most of markets all over the world are neither completely integrated nor completely segmented. They are in between, called partially segmented mar-kets (see Bekaert and Harvey (1995)). When marmar-kets are partially segmented, asset prices are not the same across di¤erent markets and purchasing power parity (PPP) is violated. Under this circumstance, foreign exchange rate risk should be priced (see Solnik (1974), Stulz (1981) and Adler and Dumas (1983)). Therefore, compared with purely global pricing or purely domestic pricing models based on either completely integrated or completely segmented assumptions, asset pricing models under partially segmented markets would include foreign exchange risk pricing factors in addition to global and domestic pricing factors.

The goal of this paper is to investigate the impact of foreign exchange rate change on stock returns in the Asian emerging markets. There have been dif-ferent approaches to examining the signi…cance of foreign exchange rate risk in pricing assets. For example, Williamson (2001) …nds statistically competitive ef-fects of exchange rate shocks between Japan and the U.S. in a speci…cation that regresses the di¤erence in automotive industry returns between the two countries on the dollar/yen exchange rate return. Gri¢ n and Stulz (2001) …nd that weekly exchange rate shocks explain almost nothing of the relative performance of in-dustries. While longer horizons are used, the importance of exchange rate shocks

increases slightly. However, the available evidence is not su¢ cient to allow gener-alization about whether foreign exchange rate risk can be ignored in asset pricing in di¤erent market environments, such as emerging markets.

In this paper, I concentrate on the Asian emerging markets for the following reason. Erb, Harvey and Viskanta (1998) discovered that the Asian crisis had wide-spread impacts on currency valuation. In fact, many Asian emerging countries’ currencies severely declined in value during the crises. From the above evidence, we can conclude that the Asian crisis was a "regional" crisis. Since Asian emerging markets have experienced such kind of currency crises with overwhelming negative impact on their economies and stock markets, the fact may a¤ect the perception of investors with respect to the importance of foreign exchange rate risk, then putting more weight on exchange rate risk factor in pricing models. The evidence mentioned above indicates it might be clear to examine the in‡uence of foreign exchange rate change on stock returns in the Asian emerging markets rather than in other developed markets counterparts.

In this study, I choose six Asian emerging countries, Indonesia, Korea, Malaysia, the Philippines, Taiwan and Thailand as my sample countries. Both market lib-eralization in the beginning of 1990s and the 1997 Asian …nancial crisis would be covered in sample period in this paper. In this paper, I would use real ex-change rate instead of nominal exex-change rate to measure foreign exex-change rate changes. The main reason is that in the emerging markets, in‡ation is relatively larger and volatile than in the developed economies. Therefore, if we neglect ad-justing in‡ation change in exchange rates, the pricing of PPP deviations would be misleading. For example, if nominal exchange rate (US dollar/Foregin currency) decreases by 10 percent, we would think that local currency depreciation should

bene…t exporters or foreign assets holders. However, after accounting for relative high in‡ationary e¤ect in home country, currency depreciation bene…ts would dis-appear. Since real exchange rates are in‡ation adjusted, the change in the real exchange rate is a more correct measure of PPP deviations.

In the next section, I will show that during the sample period, the real exchange rates for all countries are observed to increase when the market is liberalized while decrease sharply after the …nancial crisis. And stock returns expose to exchange risk to a lesser extent during the liberalization period corresponding to currency appreciation than during the post crisis period related to currency depreciation. The asymmetric exchange exposure feature over appreciation-depreciation cycles thus provides us a helpful basis in incorporating foreign exchange risk into the asset pricing framework in the Asian emerging markets.

Next, an asymmetric exchange exposure embeded asset pricing model is built. Market integration e¤ect is adjusted in this model as well. And I would use the small and large size portfolios of all sample countries to test the validility of the model. The reason to test with the small and large portfolios is that large …rms are always regarded as involved with much more international trade or invest-ment activities than small …rms, making their returns highly integrated with that of global capital markets and expose to ‡unctuations of exchange rates. On the other hand, Nance, Smith and Smithson (1993) and Main (1996) argue that large …rms have more investment opportunities and advantage in scale of economy to pay less hedging cost, they are hence more motivated to use hedges to reduce exchange rate risk. Whether small …rms or large …rms command relatively large exchange exposure remains an open question and need empirical analysis to verify that. My empirical results show that the returns on large portfolios display

noth-ing on ‡uctuations of exchange rates. Small …rms of most countries, in contrast, show statistically signifcant exchange exposure. The empirical evidence leads to conclusion that large …rms’hedging behaviors sucessfully prevent themselves ex-posing to the ‡uctuation of currency value. However, only part of countries’small portfolio returns show asymmetric exchange exposure e¤ect, suggesting the causes behind asymmetric exchange exposure might be related to some …rms’characteris-tics, for instance, exporters or importers, which I did not capture in this empirical analysis.

My paper is organized as follows. In section 2, I would examine the asymmet-ric exchange exposure feature in the Asian emerging markets. Then I construct an asset pricing model encompassing asymmetric exchange exposure and market integration e¤ect in section 3. Empirical results are presented in section 4. Some concluding remarks are o¤ered in section 5.

2. Asymmetric Exchange Exposure

To examine the foreign exchange exposure, I follow the work of Adler and Du-mas (1984). They de…ne foreign exchange exposure in terms of a regression of asset value on the exchange rate. Other works include the research of Jorion (1990), Bondnar and Gentry (1993), Chow et al. (1997), Chamberlain et al. (1997), Wong (2000), Allayannis and Ofek (2001), Dahlquist and Robertson (2001), Dominguez and Tesar (2001), Bondnar and Marston (2004) all take related approaches. Basi-cally the foreign exchange exposure measured with this approach is called “resid-ual”exposure because controlling factors place in the regression other than foreign exchange change capture the common e¤ect between stock returns and foreign ex-change rate ex-changes (see Bondnar and Wong (2003)). In this paper, I include both

world and local market returns as controlling variables. The model setting is also consistent with asset pricing concept: as most of Asian emerging countries belong to partially segmented markets, stock returns should be in‡uenced by both global and local pricing factors.

The dealt with the exchange rate in this paper di¤ers from the past studies. As I addressed in the previous section, it is important to adjust in‡ation change from nominal exchange rate to extract correct pricing of foreign exchange exposure. Therefore, I use real exchange rate against US dollar as the measure of foreign exchange rate. The setting is reasonable because trade and capital ‡ow through current accounts and capital accounts of all sample Asian emerging countries show the United States is their largest source or destination country, and it is believed that sizable shares of …rms in these countries are exposed to the ‡uctuations in the US dollar. To avoid collinearity problem in regression, I don’t include other individual exchange rates. The detail calculation of real exchange rate is shown in appendix A.

2.1. Data Description

In this paper, I would measure stock returns’foreign exchange exposure in six Asian emerging countries, including Indonesia, Korea, Malaysia, the Philippines, Taiwan and Thailand. The …rm level returns, nominal exchange rate and consumer price index for each country are from Datastream. Sample periods for each country are varying depending on the availability of database. MSCI world market return is adopted as world market return measure. For local market return, returns for stock market index are used: the Jakarta Stock Exchange Composite Price index for Indonesia; the Korea Stock Exchange Composite for Korea; the Kuala

Lumpur Stock Exchange Composite for Malaysia; the Philippines Stock Exchange Composite Price Index for Philippines; the Taiwan Stock Exchange Weighted-price index for Taiwan; and the Bangkok S.E.T. Price Index for Thailand. The stock market index is from Global Financial Database. All returns are measured in US dollar and are in excess of 30-day Eurodollar deposit rate.

Figure 1 displays the real exchange rates of six sample countries from 1980 to 2007. Since the sample period covers the market liberalization and the 1997 …nancial crisis, I check if real exchange rate has signi…cant change after market liberalization and …nancial crisis. I partition the whole sample period into four sub-sample periods, de…ned as pre-liberalization, the liberalization period between market liberalization and the 1997 …nancial crisis, post-crisis period and recent period. Table 1 summarizes the break dates used to partition sub-sample periods for each Asian emerging country. For liberalization break date, I adopt Bekaert and Harvey’s discussion of the choices for o¢ cial liberalization dates (2000) to de…ne the market liberalization break date in my research.

[Insert Figure 1] [Insert Table 1]

Table 2 shows the sub-sample mean of real exchange rate for each country. From Table 2, we can see all six countries exhibit the similar pattern in real exchange rate change. Real exchange rates rise after market liberalization, mean-ing local currency appreciates after adjustmean-ing in‡ation change. But it decreases sharply during post-crisis period, showing large currency depreciation in this time. However, real exchange rate recovers gradually in the recent period. From what mentioned above, we can thus attribute the liberalization period and recent period

as currency appreciation cycles while post-crisis period as a depreciation cycle. [Insert Table 2]

2.2. Estimating Exchange Exposure

To estimate foreign exchange exposure, I conduct the following regression:

ri;t = i;wrw;t+ i;hrh;t+ i;xrx;t+ ui;t (1)

where ri;t is the excess return on stock i; rw;t is the excess return in the world market; rh;t is the excess return on local market; rx;t is the real exchange rate change. I proceed the above regression in each sub-sample period for the stocks in every country, and the results are summarized in Table 3. Table 3 reports the fraction of …rms for which we can reject the hypothesis that the coe¢ cient on the exchange rate change is zero. For all countries except Taiwan, we can see that compared to pre-liberalization period, fewer …rms in the liberalization period have the coe¢ cient on the exchange rate change statistically signi…cant from zero. The fact suggests that foreign exchange exposure has fallen when Asian emerging countries’ governments start to liberalize their markets. The evidence also contradicts our intuition that market liberalization would cause exchange rate become more volatile and thus have …rms exposed more to foreign exchange risk. The post-crisis period, however, shows the most exposure. Until the recent period, the high foreign exchange exposure is mitigated. As the pattern of real exchange rate suggests that the liberalization period and recent period belong to appreciation cycles but the post-crisis period corresponds to a depreciation cycle, one possible direction is to think there might exist asymmetric exchange exposure over appreication-depreciation cycle in the Asian emerging marekts.

[Insert Table 3]

To examine if asymmetric exchange exposure feature does exist in the Asian emerging markets and the cause behind it, I divide the sample into appreciation state and depreciation state. The regression can be written as follows:

ri;t = i;wrw;t+ ihrh;t+ 2 X j=1

j;i;xDj;trx;t+ ui;t (2)

where each Dj is a dummy variable:

D1= 1, rx 0, implying currency appreciation = 0; otherwise

D2= 1, rx< 0, implying currency depreciation = 0; otherwise

The estimated results for each country are reported in Table 4. From Table 4, we can observe that no matter in which subperiod, higher fraction of …rms exhibit exposure in currency depreciation than in appreciation. The evidence con…rms the existence of asymmetric exchange exposure over appreciation-depreciation cycle in the Asian emerging market. In Table 4, the number in parentheses represents the fraction of …rms with signi…cantly positive coe¢ cient on exchange change out of signi…cant part. We can see these ratios are generally high. From economic the-ory, we know that net importers or …rms with net US dollar liabilities should have positive coe¢ cient on exchange change while net exporters or frims with net US dollar assets have negative coe¢ cient. Therefore, the high ratios in parentheses imply that importers or …rms with US dollar liabilities command higher exchange exposure than exporters or …rms with US dollar assets in either appreciaton or depreciation cycle. This …nding indicates that exporters or …rms with US dol-lar assets perform more successful hedging works than importers or …rms with

US dollar liabilities. One possible reason for the di¤erence of hedging ability is that our sample period covers many large and unpredictable currency depreciation shocks, causing importers or …rms with US dollar labilities cannot proceed hedging strategies successfully.

[Insert Table 4]

2.3. Explanation Behind Asymmetric Exchange Exposure

The existence of asymmetric exchange exposure in the Asian emerging markets might be attributed to …rms’asymmetric heding behaviors. Asymmetric hedging occurs when …rms take one-sided hedges. For example, …rms with net long posi-tions may be inclined to hedge against domestic currency appreciaposi-tions yet remain unhedged against domestic currency depreciations. Alternatively, …rms with net short positions may be tended to hedge against domestic currency depreciations yet remain unhedged against domestic currency appreciations. If the hedging theory mentioned above can fully describe …rms’hedging behaviors in the Asian emerging markets, we should see net importers or …rms with net US dollar labilities expose to ‡uctuations of exchange rate in appreciation cycle while net exporters or …rms with net US dollar assets have higher exchange exposures in depreciation cycle. In other words, high porportion of signi…cantly positive coe¢ cient on exchange rate change should be observed when currency appreciates but low fraction of signi…cantly positive coe¢ cient should be seen when currency depreciates.

I then use the number in parentheses, the fraction of …rms with signi…cantly positive coe¢ cient out of signi…cant portion, to verify the above argument. The numbers in appreciaton cycle are high enough to prove that …rms with net long positions perform hedging strategies successfully. Therefore, only …rms with net

short positions expose to exchange rate change and bene…t from currency appre-ciation. However, the numbers in parentheses in depreciation cycle, though lower than those in appreciation cycle, are still very high. As asymmetric hedging be-havior theory suggests, …rms with net short positions should take hedges when currency depreciates, avoiding adverse e¤ect by depreciaton. But obviously, they are not! One possible explanation is that currency depreciaton shock comes too dramatically and unexpectedly to fail …rms’ hedging works. As a result, during depreciation cycle, we can observe both …rms with net long positions and net short positions expose to exchange rate ‡uctuations, inducing higher exchange exposure in depreciation cycle than in appreciation one.

The argument made above may serve as one of possible reasons to cause asym-metric exchange exposure in the Asian emerging markets. However, without de-tailly capture the industry structure and …rm characteristics of every country, we cannot extract exact sources behind this fact. Nevertheless, the discovery of asym-metric exchange exposure provides us a helpful basis to incorporate into an asset pricing model.

3. Exchange Exposure Embeded Asset Pricing Model

In this section, I would like to construct an asset pricing model encompass-ing asymmetric exchange exposure for Asian emergencompass-ing markets. Since the sample period cover market liberalization in the beginning of 1990s, in addition to em-bed asymmetric exchange exposure setting, we have to take into account market integration e¤ect as well. My model setting is as follows:

ri;t= i;0+ (1 Dc;t) i;1rh;t+ Dc;t( i;wrw;t+ i;h"h;t) + ( i;x+ Dx;t i;r)rx;t+ ui;t

2

i;t= i;0+ i;1u2i;t 1+ i;2 2i;t 1 (3)

where ri;t is the excess return on portfolio i, including small and large portfo-lios1 for each sample country; rw;t is the excess return in the world market; rh;t is the excess return on local market; rx;tis the real exchange rate change.

Dc and Dxare dummy variables:

Dc = 1, if after the o¢ cial liberalization date documented by Bekaert and Harvey (2000)

= 0; if before the o¢ cial liberalization date Dx= 1, rx< 0, implying currency depreciation

= 0; otherwise

"h;t is the residual from the regression of local market excess return on world market excess return after market liberalization. The regression can be shown as follows:

rh;t= 0+ 1rw;t+ "h;t (4)

Here, I propose that even the Asian emerging countries’governments o¢ cially liberalize their market to foreign investors, the market is still not compeletly in-tegrated with the world capital market. Therefore, besides global in‡uences, the asset returns would be a¤ected by domestic forces as well. To avoid double count-ing of world capital market impact, I regress local market excess return on the world market excess return, and then take the residual as an explanatory variable.

i;r is incorporated to estimate the asymmetric exchange exposure e¤ect. If there is no asymmetric exchange exposure in market, we should see _i;r is not statistically signi…cant from zero. Otherwise, _i;r appear to be statistically sig-ni…cant from zero. Moreover, since the volatility of stock returns in the Asian emerging markets is usually large and clustering compared to the developed coun-tries, I incorporate GARCH (1,1) to capture the above feature. The degree of volatility persistence is measured by 1+ 2and the unconditional variance, 2, is given by 0= (1 1 2). Existence of the unconditional variance requires that persistence is less than one.

The reason that I test the constructed asset pricing model with small and large size portfolios is that large …rms are always regarded as involved with much more international trade or investment activities than small …rms, making their returns highly integrated with that of global capital markets and expose to ‡unctuations of exchange rates. But another stream of research insists that large …rms have more investment opportunities and advantage in scale of economy to pay less hedging cost, they are hence more motivated to use hedges to reduce exchange rate risk. I would do empirical studies to verify which hypothesis is supported in the Asian emerging markets. I report the estimated results in Table 5.

[Insert Table 5]

4. Empirical Results

From the estimates of large portfolios in Table 5, we can see all countries ex-cept Thailand have interex-cept term ₀ statistically insigni…cant from zero, and the adjusted R2 _{are ranging from 0.76 to 0.919. The above evidence suggests that}

the constructed asset pricing model is robust for explaining returns variation of large size portfolios in most of Asian emerging countries. ₁ characterize local market risk exposure during pre-liberalization period. From Table 5, we observe all countries except Thailand show statistically signi…cant 1in explaining returns variation of large size portfolios. wand hrepresent the global market risk expo-sure and domestic market risk expoexpo-sure after market liberalization, respectively. Both of the above exposure estimates are statistically signi…cant at the 5% level for all countries. The signi…cance of _h implies that the Asian emerging markets keep partially segmented rather than fully integrated with global capital market even their government o¢ cially liberalize the markets. For the exchange exposure estimates, _xand _r, none of the countries shows these two coe¢ cients signi…cant in capturing returns variation of large size portfolios. The results also mean that large size portfolio exposes less to exchange rate ‡uctuations, con…rming the ar-gument that large …rms have advantages in scale of economy to pay less hedging cost, and then more motivated to use hedges to reduce exchange rate risk. For conditional variance parameter estimates, all countries except Thailand show sta-tistical signi…cance at the 5% level. Moreover, the sum of 1 and 2 is close to one, implying that the volatility of returns in the Asian emerging markets exhibits strong clustering feature.

The results of small portfolios show the estimates of intercept term are statis-tically insigni…cant for all countries except Malaysia. The adjusted R2 _{is ranging} from 0.125 to 0.665, lower than that in large size portfolio estimation. The rel-atively low adjusted R2 might be attributed to some unspeci…ed risk exposures existed in small size portfolios. From the estimated results of 1, we can de-duce that, except for Philippines, local market risk exposure can explain returns

variation of small portfolios in the pre-liberalization period. After market liber-alization, both global and domestic market risk exposures are obeserved to a¤ect returns variation of small size portfolios for all countries other than the Philip-pines. The most distinction between estimates of small and large portfolios comes from the estimates of exchange exposures, xand r. The estimates of large port-folios show that the exchange exposure is not signi…cant at all to explain returns variation of large size portfolio. However, the estimated results of small portfolios exhibit that Korea, Malaysia, Taiwan and Thailand exhibit signi…cant exchange exposures (either _xor _r) in explaining returns variation of small size portfolios. Therefore, we can conclude that compared to large …rms, small …rms expose much to currency shocks although they are usually regarded as dealing with fewer inter-national trade or investment activities. The relatively large exchange exposure of small …rms might be due to their lack of hedging budgets and motivation.

5. Conclusion

Many Asian emerging countries’ currencies severely declined in value during the 1997 …nancial crisis. Since Asian emerging markets have experienced such kind of currency crises with overwhelming negative impact on their economies and stock markets, investors may emphasize more on exchange rate risk in the Asian emerging markets than in the developed markets. Therefore, it is important to examine the impact of foreign exchange rate change on stock returns in the Asian emerging markets. To proceed, I use real exchange rate instead of nominal exchange rate to account for the high in‡ationary e¤ect in the Asian emerging markets. Real exchange rates for all sample countries are observed to appreciate after market liberalization while depreciate sharply after the 1997 …nancial crisis.

From the evidence metioned above, the liberalization period and recent period can be attributed as currency appreciation cycles while post-crisis period can be attributed as a depreciation cycle.

The empirical results show that stock returns expose to exchange risk to a lesser extent during appreciation cycles than during depreciation cycles. The asymmetric exchange exposure feature over appreciation-depreciation cycles thus provides us a helpful basis in constructing an asset pricing model in the Asian emerging markets. One possible reason causing asymmetric exchange exposure is …rms’ asymmetric hedging behaviors. Therefore, we should see only …rms with net short position expose to ‡uctuations of exchange rate in appreciation cycle while …rms with net long position have higher exchange exposures in depreciation cycle. However, currency depreciation shock comes too dramatically and unexpectedly so that …rms with net short position fail to hedge successfully. As a result, we can see higher exchange exposure is exhibited in depreciation cycle than in appreciation cycle.

Next, I construct an asset pricing model encompassing asymmetric exchange exposure for the Asian emerging markets. And then test the model with small and large size portfolios. The empirical results show that small …rms expose more to currency shocks than large …rms although they are usually regarded as involving with fewer international trade and investment activities. My results support the hypothesis that compared to small …rms, large …rms have advantages in scale of economy to pay less hedging cost, and then more motivated to use hedges to reduce exchange rate risk.

Appendix A. Derivation of Real Exchange Rate Change De…ne S_jr as the real exchange rate of currency j versus U$ S_jtr = Sjt P_Pjt_t

Sr

jt Pt= Sjt Pjt

where Sjt is the nominal exchange rate (U$/FCj), Ptis the price level in the US, Pjt is the price level in country j:

After taking log and di¤erence, we get ln Sr

jt + ln (Pt) = ln (Sjt) + ln (Pjt)

Thus, if we assume in‡ation in the US dollar is stable and non-stochastic, ln (Pt) = 0:

And real exchange rate change can be derived by the sum of nominal exchange rate change and local in‡ation rate change:

ln Sr

References

Adler, M., and B. Dumas, 1983. International Portfolio Choice and Corporation Finance: a Synthesis. Journal of Finance 38, 925–984.

Alder, M., and B. Dumas, 1984. Exposure to Currency Risk: De…nition and Mea-surement. Financial Management 13, 41-50.

Allayannis, G., and E. Ofek, 2001. Exchange Rate Exposure, Hedging, and the Use of Foreign Currency Derivatives. Journal of International Money and Finance 20, 273–296.

Bekaert, G., and C. R. Harvey, 1995. Time-Varying World Market Integration. Journal of Finance 2, 403-444.

Bodnar, G., and W.M. Gentry, 1993. Exchange Rate Exposure and Industry Char-acteristics: Evidence from Canada, Japan, and the US. Journal of International Money and Finance 12, 29–45.

Bodnar, G., and R. Marston, 2004. A Simple Model of Exchange Rate Exposure. Unpublished Working Paper. University of Pennsylvania, PA.

Bodnar, G., and M.H.F. Wong, 2003. Estimating Exchange Rate Exposures: Issues in Model Structure. Financial Management 32 (1), 35–67.

Chamberlain, S., J. Howe, and H. Popper, 1997. The Exchange Rate Exposure of U.S. and Japanese Banking Institutions. Journal of Banking and Finance 21, 871-892.

Chow, E. H., W. Y. Lee and M. E. Solt, 1997. The Exchange-Rate Risk Exposure of Asset Returns. Journal of Business 70, 105-123.

Dahlquist, M., and G. Robertsson, 2001. Exchange Rate Exposure, Risk Pre-mia, and Firm Characteristics. Unpublished Working Paper. Duke University, Durham, NC.

Dominguez, K., and L. Tesar, 2001. A Re-examination of Exchange Rate Exposure. American Economic Review Papers and Proceedings 91, 396-399.

Erb, C. B., C. R. Harvey, and T. E. Viskanta, 1998. Contagion and Risk. Emerging Markets Quarterly 2, 46-64.

Fama, E. F., and K. R. French, 1995. Size and Book-to-Market Factors in Earnings and Returns. Journal of Finance 50, 131-155.

Gri¢ n, J. M., and R.M. Stulz, 2001. International Competition and Exchange Rate Shocks: a Cross-Country Industry Analysis of Stock Returns. Review of Financial Studies 14, 215–241.

Jorion, P., 1990. The Exchange Rate Exposure of US Multinationals, Journal of Business 63, 331–345.

Mian, S., 1996. Evidence on Corporate Hedging Policy, Journal of Financial and Quantitative Analysis 31, 419–439.

Nance, D. R., C.W. Smith and C.W. Smithson, 1993. On the Determinants of Corporate Hedging, Journal of Finance 48, 391–405.

Solnik, B. H., 1974. An Equilibrium Model of the International Capital Market. Journal of Economic Theory 8 (4), 500-524.

Stulz, R., 1981. On the E¤ects of Barriers to International Investment. Journal of Finance 36, 923-934.

Williamson, R., 2001. Exchange Rate Exposure, Competitiveness and Firm Val-uation: Evidence from the World Automotive Industry. Journal of Financial Economics 59, 441–475.

Wong, M. H. F., 2000. The Association between SFAS 119 Derivatives Disclosures and the Foreign Exchange Risk Exposure of Manufacturing Firms. Journal of Accounting Research 38, 387-417.

.005 .010 .015 .020 .025 80 82 84 86 88 90 92 94 96 98 00 02 04 06 INDONESIA .04 .05 .06 .07 .08 .09 .10 .11 .12 80 82 84 86 88 90 92 94 96 98 00 02 04 06 KOREA 18 20 22 24 26 28 30 32 34 80 82 84 86 88 90 92 94 96 98 00 02 04 06 MALAYSIA 1.0 1.5 2.0 2.5 3.0 3.5 4.0 80 82 84 86 88 90 92 94 96 98 00 02 04 06 PHILIPPINES 1.2 1.6 2.0 2.4 2.8 3.2 3.6 80 82 84 86 88 90 92 94 96 98 00 02 04 06 TAIWAN 1.6 2.0 2.4 2.8 3.2 3.6 4.0 80 82 84 86 88 90 92 94 96 98 00 02 04 06 THAILAND

Country Indonesia

Korea

Malaysia

The

Philippines

Taiwan Thailand

Pre-liberalization Period

Liberalization

Break Date

Sep.

1989

Jan.

1992

Dec. 1988

Jun. 1991

Jan.

1991

Sep. 1987

Liberalization Period

Crisis Break

Date

Jul. 1997

Post-crisis Period

Recovering

Break Date

Jul. 2002

Recent Period

These liberalization break dates are documented by Bekaert and Harvey (2000) in their paper, "Foreign speculators and emerging equity markets."

Pre-liberalization

Period

Liberalization

Period

Post-crisis

Period

Recent Period

Indonesia 0.018172

0.018178

0.011578

0.017146

Korea 0.055803

0.084241

0.069314

0.094708

Malaysia 23.52612

27.80153

24.18272

27.02246

The

Philippines

1.685235 2.553728

2.297798

2.461327

Taiwan 1.982172

3.218701

2.930489

2.975091

Thailand 2.071356

2.748697

2.395549

2.805702

The equation estimated is:

r

r

r

u

r

=

β

+

β

+

β

+

Percent of firms rejecting

H

:

β

=

0

at the 5% level

Pre-liberalization

Period

Liberalization

Period

Post-crisis

Period

Recent Period

Indonesia ---- 2.67 0 1.69

Korea 6.91

1.57

3.68

4.22

Malaysia 10.53 7.6 15.71

6.78

The

Philippines

7.69 6.74

10.15

20.88

Taiwan 3.45

4.55

7.76

13.71

Thailand ---- 4.29

14.06

4.49

The equation estimated is

∑

+

+

+

=

r

r

D

r

u

r

β

β

β

Percent of firms rejecting

H

:

β

=

0

at the 5% level

( ): the number in parentheses represents the fraction of firms with significantly positive coefficient on exchange

change out of significant part

Pre-liberalization

Period

Liberalization Period

Post-crisis Period

Recent Period

Appreciation

Depreciation Appreciation

Depreciation Appreciation Depreciation Appreciation

Depreciation

Indonesia ----

----

0

5.33

(24.95)

1.9

(100)

0.95

(50.53)

1.69

(79.88)

8.78

(92.37)

Korea 3.23

(100)

5.53

(83.26)

1.57

(62.42)

5.09

(49.9)

1.72

(78.49)

6.99

(94.85)

3.79

(92.61)

5.72

(76.22)

Malaysia 5.26

(100)

5.26

(100)

0 9.94

(5.84)

3.45

(88.99)

10.34

(98.16)

6.06

(54.79)

9.67

(92.55)

The

Philippines

0 7.69

(100)

6.74

(100)

8.99

(100)

2.03

(50.25)

14.72

(96.54)

18.07

(97.79)

22.49

(91.06)

Taiwan 0 0 2.84

(100)

4.55

(100)

1.08

(39.81)

10.13

(100)

2.16

(81.02)

14.54

(85.08)

Thailand ----

---- 1.43

(0)

5.71

(75.13)

15.48

(82.75)

9.79

(81.82)

3.53

(95.47)

3.04

(73.68)

Large Portfolio Estimates

Panel A: conditional mean parameter estimates

t i t x r i t x x i ht h i t w w i t c t h i t c i t i

D

r

D

r

D

r

u

r

=

β

+

(

1

−

)

β

+

(

β

+

β

ε

)

+

(

β

+

β

)

+

Market

β

β

β

β

β

β

Adj.

R

Indonesia

-0.295 ----

1.029* 1.006* 0.186 0.146 0.919

Korea

-0.153 0.911* 1.416* 0.948* 0.116 -0.081 0.77

Malaysia

0.012 0.949*

0.827*

0.94* 0.218 0.072 0.849

The

Philippines

0.46 0.841*

1.106*

0.772*

0.205

0.255

0.76

Taiwan

0.232 1.139*

0.925*

0.992*

-0.306

0.309 0.905

Thailand

0.273* 0.672 0.993* 0.855* 0.149 0.245* 0.813

Panel B: conditional variance parameter estimates

2 1 , 2 , 2 , 1 , 0 , 2 ,t

=

+

+

α

α

u

α

σ

σ

Market

α

α

α

Indonesia

0.851* 0.261*

0.679*

Korea

0.355* 0.241*

0.76*

Malaysia

0.131* 0.25*

0.748*

The

Philippines

2.664* 0.346*

0.604*

Taiwan

0.276* 0.245*

0.745*

Thailand

3.704* 1.696*

-0.014

Small Portfolio estimates

Panel A: conditional mean parameter estimates

t i t x r i t x x i ht h i t w w i t c t h i t c i t i

D

r

D

r

D

r

u

r

=

β

+

(

1

−

)

β

+

(

β

+

β

ε

)

+

(

β

+

β

)

+

Market

β

β

β

β

β

β

Adj.

R

Indonesia

0.045 ---- 0.693*

0.731*

0.344 -0.032

0.582

Korea

0.62 0.656*

1.185*

0.979*

1.17*

-1.194*

0.399

Malaysia

1.006* 0.788* 1.088* 1.248* -0.345 0.811* 0.542

The

Philippines

2.7 0.5 0.444

0.112

1.836

-0.451

0.125

Taiwan

-0.094 1.485* 0.383* 0.82* 1.289* -1.311 0.665

Thailand

0.69 1.169*

0.565*

0.548*

0.965*

-0.717*

0.538

Panel B: conditional variance parameter estimates

2 1 , 2 , 2 , 1 , 0 , 2 ,t

=

+

+

α

α

u

α

σ

σ

Market

α

α

α

Indonesia

6.427* 0.411*

0.514*

Korea

10.908* 0.365*

0.533*

Malaysia

1.511 0.081*

0.896*

The

Philippines

93.081 -0.011

0.578

Taiwan

1.09 0.029

0.935*

Thailand

16.307* 0.438*

0.098