does exchange rate risk affect exports asymmetrically? asian evidence

Journal of International Money and Finance 28 (2009) 215–239

Contents lists available at ScienceDirect

Journal of International Moneyand Finance

journal homepage: www.elsevier .com/locate/ j imf

Does exchange rate risk affect exports asymmetrically?Asian evidence

WenShwo Fang a, YiHao Lai b, Stephen M. Miller c,*

a Department of Economics, Feng Chia University, 100 WenHwa Road, Taichung, Taiwanb Department of Finance, Da-Yeh University, No. 112, Shanjiao Road, Dacun, Changhua, Taiwanc Department of Economics, 4505 S. Maryland Parkway, Box 456005, University of Nevada, Las Vegas, NV 89154-6005, USA

JEL classification:C32F14F31F41

Keywords:ExportsExchange rate riskAsymmetric hedgingDCC bivariate GARCH-M model

* Corresponding author. Tel.: þ1 702 895 3969;E-mail address: [email protected] (S.M.

0261-5606/$ – see front matter � 2008 Elsevier Ldoi:10.1016/j.jimonfin.2008.11.002

a b s t r a c t

This paper tests the hypothesis of asymmetric effects of exchangerate risk with a dynamic conditional correlation bivariateGARCH(1,1)-M model. The asymmetry means that exchange raterisk (volatility) affects exports differently during appreciations anddepreciations, which may reflect exporter’s asymmetric riskperception and hedging behavior. Using bilateral exports fromeight Asian countries to the US, the real exchange rate risksignificantly affects exports for all countries, negative or positive,in periods of depreciation or appreciation. Thus, policy makers canconsider the stability of the exchange rate in addition to itsdepreciation as a method of controlling export growth.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Students of international trade recognize that exports respond to a number of variables such asforeign income as well as the real exchange rate and its volatility. While a consensus exists on theeffects of foreign income and the real exchange rate on exports, uncertainty reigns, both theoreticallyand empirically, on the effect of real exchange rate volatility. Initial analyses of the issues argue that riskadverse traders respond negatively to high exchange rate volatility, lowering trading volumes (e.g.,Ethier, 1973). A number of subsequent papers, however, illustrate theoretically how the effect ofexchange rate volatility on trade volumes proves ambiguous (Bailey et al., 1986, 1987; De Grauwe,1988;Franke, 1991; Dellas and Zilberfarb, 1993).

fax: þ1 702 895 1354.Miller).

td. All rights reserved.

mailto:[email protected]

www.sciencedirect.com/science/journal/02615606

www.elsevier.com/locate/jimf

WenShwo Fang et al. / Journal of International Money and Finance 28 (2009) 215–239216

Empirical ambiguity holds as well. Cushman (1983, 1986), Kenen and Rodrik (1986), Thursby andThursby (1987), Pozo (1992), Chowdhury (1993), Caporale and Doroodian (1994), and Doroodian(1999) all report that higher exchange rate volatility lowers trade volumes.1 Other researchers find few,or no, significant effects such as Hooper and Kohlhagen (1978), International Monetary Fund (1984),Bailey et al. (1986, 1987), and Korary and Lastrapes (1989). Sometimes researchers discover a significantpositive relationship. But such findings do not receive much exposure, probably because they runagainst conventional wisdom.

The Asia growth miracle produced a small literature, attempting to understand the causes of thatgrowth. The typical view suggests that the rapid growth of exports provides one of the main engines ofthe rapid growth in real GDP (World Bank, 1993). More recently, Dooley et al. (2004) argue that‘‘exporting to the USA is Asia’s main concern. Exports mean growth.’’ (p. 309). Consequently, this paperexamines the effects, if any, of real exchange rate volatility on exports from eight Asian economies –Indonesia, Japan, Korea, Malaysia, Philippines, Singapore, Taiwan, and Thailand to the United States.

The relationship between exchange rate risk and exports receives considerable attention in theliterature, since the collapse of fixed exchange rates in the early 1970s. Ethier (1973) argues thatexchange rate risk could lower exports due to profit risk. De Grauwe (1988), however, suggeststhat exporters might increase exports to offset potential revenue losses. Broll and Eckwert (1999) notethat the price of an option to export increases with risk. Pozo (1992) uncovers a negative effect ofexchange rate risk on UK exports to the US. Chowdhury (1993) and Arize (1995, 1996, 1997) findnegative effects of exchange rate risk on US, European, and G7 exports. Weliwita et al. (1999) and Fangand Thompson (2004) provide evidence of negative effects for Sri Lanka and Taiwan. Arize et al. (2000,2003) conclude that exchange rate risk generates a negative effect on LDC exports, using a movingsample standard deviation model. In contrast, Asseery and Peel (1991) find positive effects formultilateral exports except for the UK. Kroner and Lastrapes (1993) discover positive effects for France,Germany, and Japan, but negative effects for the UK and the US. McKenzie and Brooks (1997) reportpositive effects for Germany and the US. Klaassen (2004) reports no effect of monthly bilateral USexports on other G7 countries.

In a survey of the literature on the effect of exchange rate volatility on international trade flows,McKenzie and Melbourne (1999) conclude that most empirical studies commonly fail to reveala significant relationship, and where a significant relationship exists, it is positive or negative, seem-ingly at random. They argue that exchange rate volatility produces different effects in different marketsand suggest using bilateral trade flows or disaggregated export market specific data to test the effect.Wang and Barrett (2007) employ sectoral level, monthly export data from Taiwan to the US and findthat exchange rate volatility affects agricultural trade flows, but not trade in other sectors.

In a companion paper to the current paper, Fang et al. (2006) examine the net effects of exchangerate changes and exchange rate risk (volatility) on exports in eight Asian countries. That is, exchangerate depreciations theoretically raise exports, but the associated increase in exchange rate volatilitymay reduce exports, either offsetting, or completely reversing, the original effect of the depreciation.They do not consider, however, the possibility of asymmetric effects in their specification. Theyconclude ‘‘Exchange rate risk generates a negative effect for six of the countries, resulting in a negativenet effect in Indonesia, Japan, Singapore, and Taiwan and a zero net effect in Korea and Thailand.’’(p. 611).

While a variety of theoretical and empirical models attempt to isolate quantitatively importanteffects of exchange rate risk on exports, all work proceeds under the assumption of symmetry, meaningthat no difference exists between the risk effects of exchange rate appreciation and depreciation. Tseand Tsui (1997) find that a depreciation shock produces a greater effect on future volatility in exchangerates than an appreciation shock of the same magnitude. This asymmetric volatility becomes mostvisible during foreign exchange market crises when a large depreciation in the exchange rate associateswith a significant increase in market volatility. Risk-averse exporters behave differently when facingdifferent degrees of foreign exchange market volatility. Thus, different risk effects may emerge underconditions of exchange rate depreciation and appreciation.

1 Some authors consider export volume; others, import volume; and a few, both export and import volumes.

WenShwo Fang et al. / Journal of International Money and Finance 28 (2009) 215–239 217

Economically, a number of hypotheses exist to explain asymmetric market volatility and itsasymmetric effects.2 First, asymmetric responses may occur due to central bank intervention, whichgenerates uncertainty in the market about the true (long-run) value of the exchange rate (McKenzie,2002).3 When the currency falls by a large enough amount, the central bank steps in to sell its holdingsof foreign reserves and stem the depreciation. This intervention may in itself create uncertaintyregarding the true (long-run) value of the exchange rate. McKenzie (2002) argues that ‘‘alternatively,the intervention may be motivated by market uncertainty in which case the central bank is trying togive an ambiguous market direction.’’ (p. 247). Irrespective as to the exact nature of the uncertainty,sales of foreign exchange reserves in the event of exchange rate depreciations create uncertainty in themarket such that the following period’s volatility may also prove high. In contrast, when the exchangerate appreciates the central bank may choose to rebuild its foreign reserves or else leave the market toits own devices. In either case, the element of uncertainty surrounding positive purchases by thecentral bank differs from that when the exchange rate falls and the central bank sells foreign exchangereserves. Thus, shocks to the market of equal magnitude will generate different responses dependingon whether they associate with an exchange rate appreciation or depreciation. More recently,Fratzscher (2006) documents that actual interventions raise exchange rate uncertainty and volatility inG3 countries. Using Australia as an example, McKenzie (2002) supports his asymmetric hypothesis andsuggests that intervention may do more harm than good in volatile markets.

Second, asymmetric responses may occur due to asymmetric pricing-to-market behavior, whichinvolves adjusting export prices based on the degree of competition in foreign markets (Krugman,1987; Froot and Klemperer, 1989; Marston, 1990; Sercu, 1992; Knetter, 1994; Goldberg, 1995; Kanas,1997; Mahdavi, 2000). In Froot and Klemperer’s (1989) two-period duopoly model, second-periodprofit depends on first-period market share. To maximize the present value of profits in domesticcurrency, the first-order conditions show that the firm may offer lower prices when market shareexhibits value. Knetter (1994) argues that exporting firms with market share objectives do notpermit foreign currency prices to increase when the domestic currency appreciates. Rather than passon the effects of the appreciation by raising foreign currency prices and risk losing sales volume and,thus, market share, exporters maintain foreign currency prices and effectively reduce domesticprofit margins. In contrast, pricing-to-market occurs to a lesser extent with domestic currencydepreciations. In this case, exporters with market share objectives maintain rather than attempt toincrease their profit margins, thus lower foreign currency prices, so that sales volume and marketshare grow. Thus, cash flows increase to a lesser degree with domestic currency depreciation, thandecrease with domestic currency appreciation. Examining Japanese manufacturers pricing behavior,Marston (1990) finds asymmetric pricing-to-market, which is greater with yen appreciations.Focusing on the automobile industry, Goldberg (1995) also documents asymmetric pricing-to-market and shows that US import prices of German and Japanese automobiles are influenced lessby domestic currency depreciation, than appreciation. Both observations prove consistent with themarket share objective and asymmetric characteristics that Froot and Klemperer (1989) and Knetter(1994) describe.

Third, asymmetric responses may occur due to hysteretic behavior, or effects that persist after theoriginal causes of the effects no longer exist (Baldwin, 1988; Baldwin and Krugman, 1989; Dixit,1989). New exporters enter the market when the domestic currency depreciates. Their behaviorbecome hysteretic, if they remain in the market once the currency appreciates. Ljungqvist (1994)argues that hysteresis drives firms to maintain high-cost investments such as entry costs when the

2 Two popular theories exist to explain asymmetric volatility of equities: the leverage effect (Black, 1976; Christie, 1982) andthe volatility feedback effect (French et al., 1987; Campbell and Hentschel, 1992; Wu, 2001). Bekaert and Wu (2000) providea detailed review of the literature on these asymmetries. Also, numerous models focus on corporate behavior and investorsentiment showing that stock returns may respond asymmetrically to exchange rate shocks. Andren (2001) and more recentlyMuller and Verschoor (2006) review this literature. This study focuses on asymmetric effects of exchange rate risk on exports.

3 The true (long-run) value of the exchange rate is best represented by purchasing power parity (PPP). Relative PPP impliesthat deviations of the real exchange rate from its steady state equal zero. Lothian and Taylor (1996) document that two majorexchange rates – dollar–sterling and franc–sterling – prove mean-reverting and conclude that in the long run, PPP providesa useful empirical first approximation. The same conclusion appears again more recently in Koedijk et al. (2006).


domestic currency appreciates, creating an asymmetric competitive effect and, thus, asymmetricexposure. Facing new export competitors, the cash flows of existing exporters may not increase tothe degree that would occur without the new entrants. If exporters do not exit the market when thedomestic currency strengthens, the cash flows of the new and existing exporters probably decrease.These asymmetric competitive effects may not allow cash flows to increase with domestic currencydepreciations, yet cause cash flows to decrease with domestic currency appreciations. Christophe(1997) argues that firms remain even to the point of suffering operating losses. He attributesevidence of a reduction in the value of US multinational corporations during strong dollar periods tothe practice of hysteretic pricing.

Fourth, asymmetric responses may occur due to asymmetric hedging behavior, where firms takeonly one-sided hedges. Generally, firm managers perceive riskiness in terms of outcomes entailinga loss rather than in terms of the dispersion of outcomes, suggesting an asymmetry with positive andnegative changes. Andren (2001) argues that if the asymmetry of risk perception among managersmirrors managerial reactions to risks, then macroeconomic changes with adverse effects on firm valuewill elicit a managerial response, whereas beneficial changes will not. Koutmos and Martin (2003)attribute asymmetric exchange rate exposure over appreciations and depreciations within financialsectors to asymmetric hedging. In this study, we argue that, under the condition of US-dollar invoicingin most Asian trade contracts, since domestic currency depreciation improves export revenue indomestic currency, exporters may hedge against domestic currency appreciations yet remainunhedged against domestic currency depreciations. Thus, asymmetric hedging behavior generates anasymmetric effect of exchange rate risk on export revenue.

No empirical studies directly test whether exchange rate risk acts symmetrically or asymmetrically.This paper tests the hypothesis of asymmetric effects of exchange rate risk on export revenue, wherethe asymmetry measures possible differences in the exchange rate risk (volatility) effect when theexchange rate appreciates and depreciates. Evidence of significant exchange rate risk effects providesan explanation for the mixed results in previous empirical literature due to neglect of the asymmetricrisk effects and suggests that testing the effect of exchange rate volatility should consider exportersasymmetric motivations. Based on the asymmetric hedging hypothesis, we expect that the risk effectwill prove larger in depreciations than in appreciations, since adverse effects with appreciations aremitigated. If, however, exporters think that only losses are risky, then they could respond more toappreciations than to depreciations, implying that appreciations affect export revenue more thandepreciations. That is, the risk effect of appreciations exceeds that of depreciations. In the final analysis,the effect reduces purely to an empirical issue.

Whether asymmetric risk effects exist proves important to policy makers. Conventional wisdomargues that depreciation increases exports, but exchange rate risk induced by the depreciation can hurtexports. Thus, market intervention to stimulate exports may fail, if the authorities ignore the effects ofexchange rate risk. Fang and Thompson (2004) show that exports respond positively to depreciationsand negatively to risk effects in Taiwan, but the net effect only adds noise to export fundamentals. Fanget al. (2006) also find a significant positive response of exports to depreciations for 7 of 8 countries, butfind significant negative or significant positive effects of exchange rate risk on exports in 7 of 8countries.4 The existence of asymmetric risk effects, thus, further complicates and increases theuncertainty of trade policy. Thus, successful trade policy requires a full understanding and control ofexchange risk during periods of depreciation and appreciation.

To study the effects of exchange rate risk requires a measure of the unobservable exchange rate risk.Hodrick and Srivastava (1984) identify exchange risk as conditional and time varying. Moving standarddeviations of the exchange rate maintain the hypothesis of homoskedasticity while serving as a proxyfor heteroskedastic risk in Chowdhury (1993), Arize et al. (2000, 2003). This approach raises a logicalinconsistency and probably proves inadequate to capture fully exchange rate risk dynamics. Gener-alized autoregressive conditional heteroskedasticity (GARCH) models can successfully model rela-tionships between means and variances as in Bollerslev (1986, 1990), Engle et al. (1987), and Bollerslevet al. (1992). Our analysis specifies exchange rate risk as time-varying exchange rate volatility

4 Fang et al. (2006) implicitly assume that the effect of exchange rate risk is symmetric, however.

Table 1US share of total exports.

Country Share ratio (%)

Indonesia 16.0Japan 30.5Korea 26.5Malaysia 17.6Philippines 34.1Singapore 18.7Taiwan 32.8Thailand 18.7

Note: the data are obtained from Direction of Trade of the IMF, exports to the US/total exports.


constructed with a GARCH(1,1) process following Bollerslev (1986), such that a larger estimatedconditional variance indicates more risk.

This paper employs the bivariate GARCH-M model with dynamic conditional correlation (DCC)(Engle, 2002) in measuring the exchange rate risk effect on exports and testing for asymmetry.Engle’s DCC approach allows time-varying correlations between exports and the exchange rate. Itdiffers from previous studies that implicitly assume a constant correlation. This paper uses monthlytime-series data on bilateral exports from eight Asian countries – Indonesia, Japan, Korea, Malaysia,Philippines, Singapore, Taiwan, and Thailand – to the US for 1979–2003. The majority of existingstudies consider developed countries, but the eight Asian countries, except Japan, industrializedduring this period. Table 1 shows that the US accounts for a substantial portion of exports fromthese Asian countries. The average US share of total exports over the sample ranges from 16 percentfor Indonesia to 34 percent for the Philippines. The bilateral approach can avoid asymmetricresponses across exchange rates in highly aggregated data, and then focus on the asymmetric effectsof the exchange rate risk.5

Asian developing countries provide a good laboratory to study the effect of exchange risk on exportsin several respects. First, Klaassen (2004) finds no exchange rate risk effect on monthly bilateral USexports to other G7 countries, arguing that the exchange rate risk does not exhibit sufficient variabilityto uncover its effect on exports, and suggests studying the effect, using data on developing countries,for which much more volatile exchange rate risk may exist. Hausmann et al. (2006) document that thereal exchange rate of developing countries proves approximately three times more volatile than thereal exchange rate in industrial countries, mainly due to the level of development, the degree ofdiversification of the economy, and a much larger persistence of shocks to the variance of the rate itselfas captured by ARCH specifications.

Second, many analysts have long regarded exports as a major driving force of the rapid Asianeconomic growth. Dooley et al. (2004) note that three principal economic and currency zones exist inthe world today. Asia is a trade account region, the US is a center country, and Europe, Canada and LatinAmerica constitute a capital account region. They argue that Asian countries (except Japan) managetheir exchange rates against the dollar, and ‘‘exporting to the USA is Asia’s main concern. Exports meangrowth.’’ (p. 309). In contrast, countries in the capital account region float their currencies and focus onthe risk and return position of their international investments. Dooley et al. (2004) term the system ofAsian exchange rates as the ‘‘Revived Bretton Woods’’ system and claim that as long as export-ledgrowth continues, the system can persist.

Third, Wickham (1985) argues that if most of a developing country’s contracts are denominated inone major currency, a peg to this currency will prove the most suitable option to eliminate exchangerisk. McKinnon (2000) reports that much of East Asian commodity trade involves invoicing in USdollars. More specifically, US-dollar invoicing occurs for 98% of US worldwide exports and almost 90%

5 McKenzie and Melbourne (1999) suggest using bilateral trade data, since aggregate data probably dilute the true nature ofthe relationship and lessen the probability of deriving a significant result. Barkoulas et al. (2002) show that the source ofexchange rate uncertainty does matter in determining its effect on the volume of trade and argue that disaggregated datashould provide a better view of the effect of exchange rate risk on trade flows.


of US imports.6 McKinnon and Schnabl (2004), therefore, argue that the US-dollar invoicing providesthe East Asian economies the rationale for pegging solely to the dollar to hedge foreign exchange riskand stabilize exchange rates.

Fourth, in addition to the fact that Asian currencies seldom are used as the unit of account ordenomination in trade contracts, the opportunity to acquire forward cover or otherwise to hedgeexchange rate risk in Asian currencies generally proves limited. Eichengreen and Hausmann (1999)propose the ‘‘doctrine of original sin’’ of incomplete financial markets in developing countries.McKinnon and Schnabl (2004) argue that without a well-organized market in forward exchange(original sin), risk-averse exporters cannot conveniently hedge. Instead, the government provides aninformal hedge by pegging to the dollar. Calvo and Reinhart (2002) show ‘‘fear of floating’’ indeveloping countries. That is, many countries announce greater exchange rate flexibility but, in fact,still manage their exchange rates heavily, such as Indonesia, Korea, the Philippines, Singapore,Taiwan, and Thailand in our sample. Does a peg eliminate exchange rate volatility? According toBayoumi and Eichengreen (1998), trade dependence reduces exchange rate variability by promptingintervention. McKenzie (2002), however, argues that intervention creates uncertainty about the truevalue of the exchange rate in the market, resulting in higher volatility in the following periods.More recently, Fratzscher (2006) notes that most exchange rate interventions occur in secretwithout clear announcement, leading to widely different views about the scale, frequency, andlikelihood of future interventions across market participants, and thus, inducing higher marketvolatility.

In sum, Asian economies provide a valuable laboratory within which to examine the effects ofexchange rate risk on exports. This study extends prior research and contributes to the literature byanalyzing whether exporters react differently to exchange rate risk during depreciations andappreciations. Answers to this question prove important for our eight sample countries, except Japan,because many analysts regard exports as the major development strategy of the rapid Asianeconomic growth. Since most trade contracts involve US-dollar invoicing, Asian exporters arenaturally exposed to exchange rate risk. Furthermore, the absence of efficient forward exchangemarkets makes the task of hedging exchange rate risk for Asian emerging market exporters difficult.Even the reincarnation of the Bretton Woods arrangement does not guarantee that Asian exportersremain free from exchange rate risk. As such, understanding how exporters respond to exchange raterisk in different market scenarios provides valuable information for the monetary authorities, whomanage their exchange rates to mitigate the effects of unfavorable risk on exports and to achieve thetarget of export-led growth. In addition, such understanding also aids exporters in forecasting exportrevenue.

After testing the time-series properties of the variables and identifying the GARCH or ARCH effectsof the exchange rates, the empirical results of our bivariate GARCH-M DCC model provide strongsupport for the asymmetry hypothesis. In each country, positive depreciation effects exist along withnegative or positive exchange risk effects during depreciations or appreciations. These findings favorthe asymmetric hedging hypothesis and help to resolve existing empirical ambiguity in the literature.Further analyses show that for five of the eight countries, the magnitude of the difference between theexchange rate risk effects on exports during depreciations and appreciations proves significant, eithernegative or positive. Exporters in different economies may possess different risk perceptions. Theevidence supports the uncertainty of exchange rate policies designed to influence exports.

The rest of this paper is organized as follows. Section 2 specifies the analytical framework, whichincludes the main elements of the time-varying correlation bivariate GARCH-M model designed to testfor the asymmetric hypothesis of the exchange risk. Section 3 describes that data, analyzes the time-varying variances of exports and the exchange rates, and presents empirical results. Section 4 inves-tigates the asymmetric effects of exchange rate risk on exports. Section 5 summarizes the empiricalfindings and provides concluding remarks.

6 In contrast, although Japan is the second largest economy in the world, the Japanese yen is used much less in invoicingforeign trade. On the worldwide basis, Yen invoicing occurs for only about 36% of Japan’s exports, whereas US-dollar invoicingoccurs for 53% of Japan’s exports.


2. The bivariate GARCH-M model and testing for asymmetric effects

The nonstructural reduced-form export equation of Rose (1990), Pozo (1992), and Klaassen(2004) provides the building block for our empirical analysis of the asymmetric effects of exchangerate risk on Asian exports to the US.7 Real export revenue (x) depends on real foreign income (y),the real exchange rate (q), and real exchange rate risk (hq). Real export revenue equals nominalexport revenue in domestic currency deflated by the consumer price index (CPI). Foreign income,the US industrial production index, should produce a positive effect on exports. The real exchangerate, the domestic currency price of the US dollar times the ratio of US to domestic CPIs, shouldexhibit a positive effect on exports. The real exchange rate eliminates potential ambiguity fromadjusting price levels. The effect of exchange rate risk proves uncertain theoretically andempirically.

To capture dynamic adjustments, the following eclectic bivariate GARCH-M-DCC model providesthe framework for investigating and testing the asymmetric hypothesis.

Dlxt ¼ a0 þX2

i¼1

aiDlxt�i þX2

i¼0

biDlyt�i þX2

i¼0

ciDlqt�i þX2

i¼0

dih1=2q;t�i þ 3x;t and (1)

Dlqt ¼ s0 þ s13t�1 þX2

giMDi þ 3q;t ; where (2)
i¼1
3t jJt�1wStudent� tðvÞ; (3)

hx;t ¼ a0 þ a132x;t�1 þ a2hx;t�1; (4)

X2

hq;t ¼ b0 þ b132q;t�1 þ b2hq;t�1 þ

i¼1

liVDi; (5)

D2t ¼

�hx;t 00 hq;t

�; (6)

ht ¼ D�1t 3t ; (7)

Qt ¼ rxqð1� q1 � q2Þ þ q1ht�1h0t�1 þ q2Qt�1; and (8)

Rt ¼ diagðQtÞ�1QtdiagðQtÞ�1: (9)

Also, Dlxth100ðln xt � ln xt�1Þ, Dlyth100ðln yt � ln yt�1Þ, and Dlqth100ðln qt � ln qt�1Þ. The lagstructure of the mean equation of Dlxt is selected by the Schwartz Information Criteria (SIC) and3x;t is a white noise. hq;t is time varying exchange rate volatility estimated by the GARCH(1,1)process. The presence of the square root of hq;t , h1=2

q;t , in the mean equation of Dlxt constitutes thebivariate GARCH-M model. The MA component picks up serial dependence of Dlqt to ensure that3q;t is white noise. The residual matrix, 3t , conditional on the information set Jt�1 available at timet� 1 follows a bivariate Student-t distribution with degrees of freedom v. Our sample includes theAsian financial crisis in 1997, which exhibited dramatic movements in exchange rates in mostAsian countries. MDi and VDi are dummy variables employed to capture extraordinary exchange

7 Griffin and Stulz (2001) and Koutmos and Martin (2003) note that despite the plausibility of asymmetric exposure, it isextremely difficult to devise a structural model explicitly linking asymmetric exposure to asymmetric pricing-to-market,hysteresis, or asymmetric hedging. Our reduced-form approach that examines the relation between export revenue andexchange rate changes provides a promising alternative.


rate changes in the mean and the variance equations for Dlqt. hx;t measures the conditional vari-ance of exports. Conditions, ai > 0, bi > 0, li > 0, a1 þ a2 < 1 and b1 þ b2 < 1, imply positive andstable conditional variances of 3x;t and 3q;t . If a2 or b2 equal zero, the process reduces to anARCH(1). The matrix D2

t contains hx;t and hq;t along the principle diagonal and, thus, ht is thestandardized residual matrix. Qt equals the covariance matrix of ht , following a GARCH(1,1)process. rxq is the unconditional correlation matrix of exports and the exchange rates over thesample period. q1 and q2 positive and q1 þ q2 < 1 ensure that Qt is positively defined and mean-reverting. Rt equals the conditional correlation matrix composed of time-varying correlations. Eqs.(1)–(9) constitute the DCC estimator proposed by Engle (2002). If q1 ¼ q2 ¼ 0, then it reduces tothe Bollerslev (1990) constant conditional coefficient estimator.

Let F denote the parameters in D2t that includes all parameters in Eqs. (1)–(5) and Q denote the

parameters in Rt that includes q1 and q2. Then, the log likelihood function of bivariate t-distribution inthe maximization procedure is

LðF;QÞ ¼XT

t¼1

LtðF;QÞ;

where LtðF;QÞ ¼ ln Gðvþ22 Þ � ln Gðv2Þ � ln½pðv� 2Þ �1

2jDtRtDt j � vþ22 lnð1þ h0t D

�1R�1D�1htv�2 Þ

iand Gð$Þ repre-

sents the Gamma function.The model focuses on the effects of exchange rate movements on exports and the reduced-

form export equation includes depreciation and exchange rate risk as well as the rate of changeof foreign income as explanatory variables. The signs, magnitudes, and significance of the esti-mated coefficients (ci) in Eq. (1) provide a straightforward test of the relationship betweenexports and depreciation, where

Pci > 0 implies that depreciation improves exports. Also the

signs, magnitudes, and significance of the estimated coefficients of exchange rate risk ðh1=2q;t Þ in

Eq. (1) prove of interest. If exporters reduce their exports to minimize profit uncertainty duringperiods of exchange rate fluctuations, then

Pdi < 0. If, however, exporters intend to offset

potential losses or use options markets as a hedge, thenP

di > 0. As the equation constrains thedis to remain constant for the exchange risk variable during both appreciations and deprecia-tions, Eq. (1) implicitly assumes a symmetric response of the export revenue to the exchange raterisk.

To look for asymmetric effects, we test the hypothesis that di differs between appreciations anddepreciations. Let di ¼ d1i þ d2iD, where the dummy D¼ 1 for Dlqt < 0 (i.e., an appreciation) and 0 forDlqt � 0 (i.e., a depreciation). Eq. (1) becomes

Dlxt¼a0þX2

i¼1

aiDlxt�iþX2

i¼0

biDlyt�iþX2

i¼0

ciDlqt�iþX2

i¼0

d1ih1=2q;t�iþ

X2

i¼0

d2i

�Dh1=2

q;t�i

�þ3x;t : (1a)

The estimated relations divide as follows:

Depreciation : Dlxt ¼ ba0 þX2

i¼1

baiDlxt�i þX2

i¼0

bbiDlyt�i þX2

i¼0

bciDlqt�i þX2

i¼0

bd1ih1=2q;t�i þ 3x;t and

Appreciation : Dlxt

¼ ba0 þX2

i¼1

baiDlxt�i þX2

i¼0

bbiDlyt�i þX2

i¼0

bciDlqt�i þX2

i¼0

�bd1i þ bd2i

�h1=2

q;t�i þ 3x;t ;

wherePbd2i measures the difference in the effects of the exchange rate risk between appreci-

ations and depreciations. Eq. (1a) replaces Eq. (1) in estimating our bivariate GARCH-M models.Each of the models uses maximum likelihood estimation under a t-distribution and the Berndtet al. (1974) (BHHH) algorithm in the programs coded by RATS 5.0 of Estima.

Response asymmetries may arise from different conceptual sources. If export adjustments andexchange rate volatility come from symmetric distributions, then similar co-movements between


export growth and exchange rate volatility during depreciations and appreciations should occur. Theevidence (see Table 2) suggests, however, that the two markets exhibit non-symmetric distributions forall sample countries.

In Asia, asymmetric effects may arise from differences in export revenue (profit) expectationsamong exporters about the potential effect of exchange rate movements. McKinnon (2000) reports thatmost Asian trade uses US-dollar invoicing. As a consequence, currency mismatches create additionalproblems for exporters in Asia. That is, because export contracts get signed and denominated in USdollars m periods before required payments occur, Asian exporters must convert US dollars into localcurrencies at time t when they receive payment. Thus, both the current exchange rate and its volatilityaffect export revenue in domestic currency, despite the fact that the trade price and quantity getdetermined at time (t�m) (note that no dollarization exists in Asia). Exporters of small, open Asianeconomies experience more transaction risk than exporters in large countries, such as the US. Forexample, the emergence of a sudden appreciation of the domestic currency against the US dollar at, ornear, time t erodes export revenue in domestic currency, leading to a decrease in profit or an outrightloss, because of the exchange loss. We can paraphrase Calvo and Reinhart’s (2002) ‘‘fear of floating’’ into‘‘fear of appreciation’’ and ‘‘love of depreciation.’’8

Therefore, the exchange rate risk during depreciations and appreciations plays different roles,influencing exports due to the dynamics of trade. Traditional wisdom argues that exports benefit fromexchange rate depreciation and suffer from appreciation. Under US-dollar invoicing and original sin,Asian exporters fear exchange rate appreciation, but love depreciation. A small exchange rate appre-ciation (depreciation) could trigger a larger decline (rise) in Asian exports due to the concern aboutprofit disappointments (surprises) rather than the specific magnitude of the appreciation. That is, theappreciation risk effect dominates due to asymmetric hedging behavior as discussed earlier in Section 1.

Accordingly, we expect that (1) exports may respond to exchange rate movement during depre-ciations or appreciations and (2) the risk effects may significantly differ between exchange ratedepreciations and appreciations. In reality, for Asian emerging market economies, we may observeboth types of asymmetric effects. For developed countries such as the US, since trade gets denominatedin domestic currency (the US dollar), no worries of currency mismatches and original sin emerge. Thatis, we expect only the first type of asymmetric effects, or none at all. This intuition possibly explainsKlaassen’s (2004) findings of no exchange rate risk effect between the US and its trading partners.

In Eq. (1a), statistical evidence consistent with the asymmetric effect exists. If eitherPbd1i, orP

ðbd1i þ bd2iÞ, (or both) significantly differs from zero and the two sums differ significantly from eachother (or

Pbd2i differs significantly from zero). Based on the hypothesis of asymmetric hedging, theestimates may prove positive or negative in periods of either depreciation or appreciation. If both sumsprove statistically insignificant, then the exchange rate risk causes no effect on exports.

3. Data and empirical results

For each of the eight countries, the bilateral export variable equals monthly seasonally adjusted realexport revenue for the US from January 1979 to April 2003 with a base year of 1995. All data come fromthe International Financial Statistics and Direction of Trade of the IMF, except for Taiwan, where the datacome from AREMOS.

Table 2 reports preliminary statistics for the natural logarithmic differences of exports and the realexchange rate. Every country experienced depreciation and export growth over the sample, on average.Thailand exhibits the highest average export growth at 1.031 percent with a depreciation of 0.196percent. Indonesia exhibits the highest monthly depreciation at 0.336 percent and an export growth of0.486 percent. Thus, depreciation positively associates with exports, on average. The unconditional riskmeasured by standard deviations shows that Indonesia exhibits the most volatile exchange rate and

8 Asian countries ‘‘love of depreciation’’ in that Dooley et al. (2004) argue for ‘‘.a periphery for which the developmentstrategy is export-led growth supported by undervalued exchange rates.’’ (p. 307), and McKinnon and Schnabl (2004) arguefor ‘‘.moral hazard in the sense that economic agents, whether domestic banks or firms, prefer to gamble rather than to hedgetheir bets in the foreign exchanges.’’ (p. 340).

Table 2Preliminary statistics for exports and the exchange rate.

Indonesia Japan Korea Malaysia

Dlxt Dlqt Dlxt Dlqt Dlxt Dlqt Dlxt Dlqt

Sample size 291 291 291 291 291 291 291 291Mean 0.486 0.336 0.218 0.020 0.542 0.123 0.617 0.254SD 23.561 6.257 5.263 2.792 10.886 2.785 9.815 2.085Maximum 112.428 56.678 15.506 6.801 41.158 34.325 36.894 14.890Minimum �120.641 �26.884 �18.577 �10.068 �42.280 �8.509 �32.974 �15.417Skewness �0.166 (0.144) 3.026 (0.144) �0.035 (0.144) �0.609 (0.144) �0.186 (0.144) 6.678 (0.144) 0.049 (0.144) 0.348 (0.144)Kurtosis 8.475 (0.287) 32.407 (0.287) 3.787 (0.287) 3.757 (0.287) 5.013 (0.287) 82.118 (0.287) 4.118 (0.287) 26.085 (0.287)J–B N 364.801* 10,929.82* 7.573* 24.945* 50.807* 78,061.06* 15.278* 6467.65*Q(3) 70.030* 11.934* 52.199* 27.323* 70.169* 59.985* 68.233* 13.182*Q(6) 77.207* 29.785* 66.728* 28.284* 90.065* 64.426* 70.957* 14.315*Q2(3) 62.163* 55.883* 14.311* 8.800* 44.415* 13.136* 19.944* 139.630*Q2(6) 62.257* 87.651* 16.013* 17.596* 47.158* 13.622* 26.883* 188.000*ADF(m) �21.005*(1) �14.494*(0) �9.673*(2) �12.641*(0) �19.635*(1) �12.047*(1) �18.864*(1) �13.875*(0)rxq 0.213 0.206 0.215 0.081

Philippines Singapore Taiwan Thailand

Sample size 291 291 291 291 291 291 291 291Mean 0.622 0.186 0.487 0.095 0.283 0.053 1.031 0.196SD 9.528 2.702 12.145 1.411 8.956 1.560 11.542 2.609Maximum 35.601 21.006 55.490 6.380 37.592 9.020 49.175 16.295Minimum �38.113 �8.687 �54.574 �4.995 �25.208 �6.546 �43.237 �15.911Skewness �0.050 (0.144) 2.577 (0.144) �0.218 (0.144) 0.069 (0.144) 0.407 (0.144) 0.109 (0.144) �0.144 (0.144) 1.872 (0.144)Kurtosis 5.418 (0.287) 20.495 (0.287) 6.618 (0.287) 4.950 (0.287) 4.645 (0.287) 7.954 (0.287) 6.404 (0.287) 24.106 (0.287)J–B N 71.019* 4033.18* 160.985* 46.330* 40.824* 298.168* 141.504* 5570.93*Q(3) 64.406* 8.400* 100.780* 17.620* 89.918* 14.133* 38.784* 23.865*Q(6) 66.996* 9.516 101.580* 20.500* 90.098* 22.365* 58.018* 28.645*Q2(3) 31.870* 6.203 59.289* 48.710* 36.352* 3.324 53.417* 129.850*Q2(6) 35.351* 8.823 59.721* 86.074* 39.742* 6.538 109.77* 187.150*ADF(m) �18.787*(1) �14.335*(0) �19.291*(1) �13.543*(0) �20.683*(1) �13.980*(0) �14.982*(1) �12.766*(0)rxq 0.259 0.046 0.018 0.110

Note: SD represents standard deviation; J–B N denotes Jacque–Bera normality test; Q(k) and Q2(k) are Ljung–Box statistics for the level and squared terms for autocorrelations up to k lags;and ADF(m) is the augmented Dickey–Fuller unit root test with lags m selected by the SIC criterion. The measure of skewness and kurtosis are normally distributed as N(0, 6/T) and N(0, 24/T),respectively, where T (=291) equals the sample size.* Denotes significance at the 5-percent level.** Denotes significance at the 10-percent level.

WenShw

oFang

etal./

Journalof

InternationalM

oneyand

Finance28

(2009)215–239

224


exports while Japan and Singapore exhibit the least volatile exports and exchange rates, respectively.Export volatility exceeds exchange rate volatility in every country.

Skewness statistics reject Dlxt symmetry at the 5-percent level for Taiwan and Dlqt symmetryfor every country, except Singapore and Taiwan. Kurtosis statistics for Dlxt and Dlqt imply that allseries are leptokurtic with fat tails. Jarque–Bera tests reject normality for all variables and coun-tries, suggesting the use of the Student-t distribution in model estimation. The Ljung–Box Q-statistic tests for autocorrelation, where the number of lags (k) affects its power. Tsay (2002)suggests choosing k¼ ln(T) where T equals the number of observations (291), implying k¼ 5.67.Thus, the autocorrelations tests run up to 6 lags. Ljung–Box Q-statistics indicate autocorrelation inDlxt and Dlqt for all countries. Ljung–Box Q-statistics for squared Dlxt and Dlqt suggest time-varying variances for both series in all countries, except for Dlqt in the Philippines and Taiwan. AnARMA process for mean and variance equations captures the dynamic structure to generate white-noise residuals. In the model, we employ an AR(2) process for the mean equation of Dlxt; an MA(1)process for the mean equation of Dlqt; and GARCH(1,1) processes for Eqs. (4) and (5), the twovariance equations.

Valid inference in GARCH models requires stationary variables. After selecting lag lengths by theSchwartz Information Criterion (SIC), the augmented Dickey–Fuller (ADF) test shows that Dlxt and Dlqt

exhibit stationarity [I(0) series] at the 5-percent level for all series.The correlation coefficient between the two monthly log differenced series ranges from 0.018 in

Taiwan to 0.259 in the Philippines. Fig. 1 shows the sample correlation coefficient using a movingwindow of 12 observations (i.e., 1 year). The horizontal line denotes the correlation coefficient. Thecorrelations change over time and, except for Japan, appear to increase in recent years for mostcountries, especially Indonesia and Korea. Engle (2002), Tsay (2002), and Tse and Tsui (2002)provide evidence that the estimation of a time-varying correlation GARCH model improves overthat of a constant correlation model. This paper applies Engle’s (2002) dynamic conditionalcorrelation bivariate GARCH modeling approach. The bivariate GARCH model consists of two sets ofequations. The first set of equations includes a bivariate GARCH(1,1) model for the conditionalvariances in Eqs. (1a)–(5) and the second set, a GARCH(1,1) model for the correlation coefficient inEqs. (6)–(9).

Preliminary examination shows that the standard univariate GARCH(1,1) model for Dlqt performsadequately for Japan, Singapore, and Taiwan.9 Unstable variance processes emerge, however, inIndonesia, Korea, Malaysia, the Philippines, and Thailand because the Asian financial crisis increasedexchange market volatility immediately. Neglecting structural breaks may bias upward GARCH esti-mates of the persistence in variance, vitiating the use of GARCH to estimate the mean equation. Perron(1989, 1997) suggests identifying break points by examining data and using dummy variables tocapture shifts in mean or variance processes. Fig. 2 shows time plots of the eight exchange rates,marking the break dates.

One-time shocks appear as single pulses in the depreciation series and as mean shifts in volatility.Dummy variables enter the mean equation for Indonesia and Thailand and the variance equations forIndonesia, Korea, Malaysia, the Philippines, and Thailand. In the mean equations, dummies forIndonesia are MD1¼1 for t¼ 1983:04, MD2¼1 for t¼ 1986:09, and 0 otherwise; for Thailand, MD1¼1for t¼ 1981:07, MD2¼1 for t¼ 1984:11, and 0 otherwise. In the variance equations, dummies forIndonesia are VD1¼1 for t� 1997:07, and 0 otherwise; for Korea VD1¼1 for t� 1997:07, and 0 other-wise; for Malaysia VD1¼1 for 1997:07� t� 1998:12, and 0 otherwise; for the Philippines VD1¼1 for1983:01� t� 1984:12, VD2¼1 for 1997:07� t� 1998:12, and 0 otherwise; for Thailand VD1¼1 fort� 1997:07, and 0 otherwise. The 1997 Asian crisis raised exchange rate volatility in Indonesia, Korea,Malaysia, the Philippines, and Thailand. The Philippines also experienced another volatile period from1983 through 1984.

The properties of the time varying variance and correlation in exports and exchange ratessuggest the bivariate GARCH(1,1)-M model with dynamic conditional correlation specified in Eqs.

9 This result appears reasonable, since Japan, Singapore, and Taiwan were not significantly affected by the Asian financialcrisis (see Fig. 2).

-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000

-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000

-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000

-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000-1.0

-0.5

0.0

0.5

1.0

1980 1985 1990 1995 2000

Indonesia Japan

Korea Malaysia

Philippines Singapore

Taiwan Thailand

Fig. 1. Twelve-period rolling correlations.


-40-20

020406080

1980 1985 1990 1995 2000

8304 8609 9707

-15-10

-505

10

1980 1985 1990 1995 2000

-10

0

10

20

30

40

1980 1985 1990 1995 2000

9707

-20

-10

0

10

20

1980 1985 1990 1995 2000

9707-9812

-10

0

10

20

30

1980 1985 1990 1995 2000

8301-8412 9707-9812

-8

-4

0

4

8

1980 1985 1990 1995 2000

Percentage Change of Real Exchange RateReal Exchange Rate

Indonesia Japan

Korea Malaysia


-8

-4

0

4

8

12

1980 1985 1990 1995 2000-20

-10

0

10

20

1980 1985 1990 1995 2000

8107 8411 9707

Taiwan Thailand

Fig. 2. Structural changes for exchange rates.


Table 3Engle and Ng test for asymmetry in volatility.

Variable Indonesia Japan Korea Malaysia Philippines Singapore Taiwan Thailand

Dlqt 3.263 0.118 7.690 0.972 2.454 3.346 2.871 1.599p-Value (0.353) (0.990) (0.053) (0.808) (0.484) (0.341) (0.412) (0.660)Dlxt 0.504 3.285 5.552 0.770 0.534 4.922 1.847 2.006p-Value (0.918) (0.350) (0.136) (0.857) (0.911) (0.178) (0.605) (0.571)

Note: the joint test statistic TR2 from regression bu2t ¼ f0 þ f1S�t�1 þ f2S�t�1

but�1 þ f3Sþt�1but�1 þ zt asymptotically follows a c2

distribution with 3 degrees of freedom under the null hypothesis of no asymmetric effects in volatility. Only in Korea do we findevidence of asymmetry at even the 10-percent level for Dlqt.


(1a)–(9) to investigate the asymmetric effect of exchange rate risk. This model assumes thatpositive and negative shocks generate the same effects on volatility, because they depend on thesquare of prior shocks as specified in Eqs. (4) and (5). The exchange rates and exports may responddifferently to positive and negative shocks. Brooks and Henry (2000) suggest using Engle and Ng’s(1993) diagnostic test to detect asymmetric volatility. If asymmetric volatility emerges, then weshould adopt asymmetric models such as GJR (Glosten et al., 1993) or EGARCH (Nelson, 1991).10

Table 3 reports Engle and Ng’s joint test for asymmetry in variance of the two variables Dlxt andDlqt in the eight sample countries. The test statistics formulated by calculating TR2 from thefollowing regression:

bu2t ¼ f0 þ f1S�t�1 þ f2S�t�1but�1 þ f3Sþt�1

but�1 þ zt : (10)

The TR2 statistic follows asymptotically a c2 distribution with 3 degrees of freedom under the nullhypothesis of no asymmetric effects in volatility. The standardized residuals buj;t ¼ b3j;t=h1=2

j;t , j¼ x,q,come from the univariate GARCH models for Dlqt and Dlxt , respectively, as follows:

Dlxt ¼ a00 þX2

i¼1

a0iDlxt�i þX2

i¼0

b0iDlyt�i þX2

i¼0

c0iDlqt�i þ 30x;t with (11)

hx;t ¼ a0 þ a132x;t�1 þ a2hx;t�1; and (12)

Dlqt ¼ s0 þ s13q;t�1 þX2

giMDi þ 3q;t ; with (13)
i¼1
hq;t ¼ b0 þ b132q;t�1 þ b2hq;t�1 þ

X2

liVDi: (14)
i¼1
The specification of the model in Eqs. (13) and (14) match those in our bivariate DCC-GARCHmodel, while the specification in Eqs. (11) and (12) omit the terms involving h1=2

q;t�i. In Eq. (10) S�tand Sþt are dummy variables that equal 1 and 0 for but < 0 and but � 0, respectively, and zt equals aniid error term.

Engle and Ng (1993) propose the test for sign and size bias in conditional heteroscedastic models.A significant parameter f1 indicates the presence of sign bias, meaning that positive and negativeshocks to but�1 affect the conditional variance differently. The significance of f2 or f3 suggests size bias,meaning that not only the sign, but also the magnitude, of the shock in Dlxt and Dlqt prove important.The joint test statistics for sign and size bias, based on the Lagrange Multiplier and performed as TR2 inTable 3 all indicate insignificance at the 5-percent level, though Dlqt in Korea proves significant at the

10 In fact, Cappiello et al. (2006) develop the Asymmetric Generalized DCC (AG-DCC) estimator for just this circumstance.


10-percent level. Generally, the DCC GARCH model (1a)–(9) proves adequate to examine the asym-metric effects of exchange rate risk on exports for all countries.

We first estimate the general model. Although neither autocorrelation nor hetero-skedasticity exist, insignificant coefficients make it difficult to gauge the risk effect. Table 4reports estimated coefficients and standard errors for a parsimonious version with insignificantvariables deleted. The advantages of the parsimonious specification include higher precision ofestimates from reduced multicollinearity, increased degrees of freedom, more reliable esti-mates, and greater power of tests. The insignificant likelihood ratio statistic, LR(k), at the 5-percent level suggests no difference between the general and the parsimonious models foreach country.

All estimates of the ARMA components and dummy variables in mean Eqs. (1a) and (2) aresignificant and the parameters in the two variance equations are positive. Every country exhibits time-varying variances for exports and exchange rates, suggesting the bivariate GARCH model. The signif-icance of l1 and l2 in Eq. (5) confirms the use of dummy variables to alleviate the effect of structuralbreaks. Volatility persistence for Dlxt varies from 0.177 in Taiwan to 0.981 in Indonesia, and for Dlqt

from 0.186 in Taiwan to 0.885 in Thailand. The two variance processes converge. Joint estimates of thedegrees of freedom of the t-distribution are significant, the hypothesis of the multivariate Student-tdistribution is not rejected.

Both q1 and q2 in the GARCH(1,1) process of Qt are significantly positive and q1 þ q2 < 1. The sum ofq1 þ q2 lie between 0.662 in the Philippines and 0.962 in Malaysia. Table 5 reports statistics ofconditional correlation coefficients between Dlxt and Dlqt , using the bivariate GARCH(1,1)-M-DCCmodel of Eqs. (1a)–(9). The average of the coefficients ranges from 0.011 in Malaysia to 0.201 in Japan.The mean or the median come close to the unconditional coefficient in Table 2. Values of the maximum,the minimum, and the standard deviation show that the coefficient is not constant. Fig. 3 plots thefitted conditional correlation coefficient between Dlxt and Dlqt . The plots illustrate that the correlationcoefficient fluctuates over time, similar to that of Fig. 1. This characteristic along with the non-zeroestimates for q1 and q2 suggests the use of the time-varying correlation coefficient model for eachcountry.

Bivariate Ljung–Box Q2ðkÞ statistics (Hosking, 1980) for standardized residuals and squared stan-dardized residuals of Dlxt and Dlqt do not detect remaining autocorrelation or conditional hetero-skedasticity at the 5-percent level. The bivariate GARCH-M DCC model in Eqs. (1a)–(9) adequatelyrepresent each country.

The marginal effect of US manufacturing income on exports proves significantly positive for allcountries. Seven of the eight Asian countries experience contemporaneous effects, three experienceone-month-lagged, and two experience two-month-lagged effects. The cumulative effect ranges from1.745 for Malaysia to 3.282 for Thailand. Different countries respond differently to the US economy.Generally, quick adjustments and large estimates reflect the small open-economy property of theseeconomies.

Depreciation exhibits the expected positive effect on exports for the eight countries studied, butthese effects prove insignificant in Malaysia and Singapore.11 The cumulative depreciation effect rangesfrom 0.226 for Singapore to 2.477 for Korea. Every country exhibits lower individual or cumulativedepreciation effect than the US income effect, except Korea.

Exchange rate risk possesses significant effects on exports for all countries, negative or positive inperiods of depreciation or appreciation.

4. Asymmetric effects of exchange rate risk

Table 6 reports results of the sum tests for the asymmetric effect of the exchange rate risk. Thelikelihood ratio (LR) statistic with a c2 distribution and one degree of freedom tests the significance of

11 Fang and Miller (2007) report similar findings for Singapore, using a bilateral GARCH-M model with constant variance.Abeysinghe and Yeok (1998), using OLS, find that appreciation does not diminish Singapore’s exports due to their high importcontent. Lower import prices reduce the cost of export production.

Table 4Estimates for dynamic conditional correlation bivariate GARCH-M model (1a)–(9).

Indonesia Japan Korea Malaysia Philippines Singapore Taiwan Thailand

Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E.

a0 1.832* 0.413 4.901* 0.123 1.007** 0.557 0.439 0.403 0.401 0.286 4.555* 0.277 4.622* 0.358 1.182* 0.316a1 �0.649* 0.048 �0.574* 0.044 �0.600* 0.050 �0.628* 0.048 �0.629* 0.046 �0.697* 0.049 �0.712* 0.047 �0.654* 0.051a2 �0.354* 0.046 �0.269* 0.038 �0.302* 0.046 �0.254* 0.050 �0.241* 0.044 �0.251* 0.048 �0.292* 0.044 �0.329* 0.045

b0 2.988* 0.674 1.387* 0.303 1.749* 0.742 1.621* 0.484 2.651* 0.596 1.295* 0.579 2.075* 0.739b1 1.745* 0.600 1.401* 0.429 1.554* 0.562b2 0.984* 0.282 1.207* 0.572

c0 0.362* 0.053 0.708* 0.128 1.229* 0.181 0.226 0.215 0.554* 0.086c1 0.304* 0.063 1.032* 0.115 0.738* 0.115 0.820* 0.228 0.540* 0.072c2 0.210* 0.074 0.286* 0.077 0.737* 0.090 0.288 0.205 0.215** 0.123 0.286* 0.087

d10 �1.461* 0.145 1.240* 0.184 �0.446** 0.240d11 �1.410* 0.036 �1.723* 0.262 0.706* 0.156 �3.479* 0.233 0.199* 0.030d12 �0.376* 0.072 2.392* 0.216 �0.626* 0.184 �0.996* 0.243 �2.339* 0.153

d20 0.579* 0.096 �0.627* 0.075 0.632* 0.250 0.372* 0.118d21 �0.400* 0.109 �0.334* 0.080 �0.783* 0.239 0.598* 0.202 �1.332* 0.075 0.758* 0.350 �0.813* 0.050d22 1.011* 0.187 �0.422* 0.092

s0 0.069 0.057 0.187 0.185 0.033 0.070 0.113** 0.063 0.005 0.095 0.043 0.071 0.087 0.091 �0.042 0.059s1 0.207* 0.065 0.308* 0.060 0.351* 0.052 0.177* 0.067 0.357* 0.059 0.235* 0.054 0.212* 0.061 0.211* 0.061g1 30.261* 1.508 6.072* 0.643g2 16.552* 0.495 15.066* 0.329

a0 1.642* 0.609 13.189* 1.496 36.998* 4.438 5.713* 1.277 8.213* 3.289 7.257* 1.082 42.448* 5.102 1.464* 0.504a1 0.079* 0.012 0.204* 0.094 0.404* 0.077 0.143* 0.028 0.249* 0.086 0.271* 0.013 0.177* 0.090 0.085* 0.008a2 0.901* 0.010 0.303* 0.023 0.793* 0.021 0.721* 0.073 0.675* 0.028 0.887* 0.007

b0 0.251* 0.043 6.112* 0.317 0.108* 0.020 0.796* 0.097 0.695* 0.178 0.268* 0.008 1.834* 0.182 0.078* 0.011b1 0.451* 0.074 0.191* 0.043 0.094* 0.017 0.333* 0.088 0.338* 0.104 0.086* 0.009 0.186* 0.060 0.087* 0.003b2 0.309* 0.047 0.781* 0.021 0.425* 0.070 0.766* 0.016 0.798* 0.013l1 12.008* 4.216 0.713* 0.255 42.343* 20.030 9.095** 4.961 12.431* 4.790

WenShw

oFang

etal./

Journalof

InternationalM

oneyand

Finance28

(2009)215–239

230

l2 16.388** 9.253

v 5.951* 1.074 6.835* 1.729 4.051* 0.409 5.170* 0.831 3.018* 0.202 7.315* 1.814 4.822* 0.704 6.433* 1.047

q1 0.135* 0.061 0.024* 0.004 0.111* 0.051 0.032* 0.001 0.200* 0.058 0.056** 0.029 0.073* 0.026 0.029* 0.014q2 0.717* 0.131 0.670* 0.024 0.712* 0.008 0.930* 0.001 0.462* 0.116 0.695* 0.171 0.755* 0.198 0.891* 0.022

rxq;t 0.110 0.201 0.112 0.011 0.169 0.044 0.014 0.064

Q2(6) 35.167 19.255 30.017 28.935 32.510 11.301 28.691 31.584Q2

2 ð6Þ 12.090 20.064 16.656 9.593 29.215 21.853 22.378 25.184

LR(k) 3.360(6) 4.510(7) 2.916(4) 10.166(8) 1.726(2) 2.002(8) 4.714(9) 5.926(3)

Note: Coef. and S.E.; mean coefficient and standard error. Q2(6) and Q22 ð6Þ equal the bivariate Ljung–Box statistics (Hosking, 1980) of the standardized and squared standardized residuals for

autocorrelations up to 6 lags. LR(k) equals the likelihood ratio statistic following a c2 distribution with the degree of freedom k (in the parentheses), which tests if the restricted simplemodel exhibits the same explanatory power as the unrestricted general model when we eliminate the k insignificant estimates.* Denotes significance at the 5-percent level.** Denotes significance at the 10-percent level.

WenShw

oFang

etal./

Journalof

International

Money

andFinance

28(2009)

215–239231

Table 5Statistics for dynamic conditional correlations.


Mean 0.110 0.201 0.112 0.011 0.169 0.044 0.014 0.064Median 0.106 0.202 0.134 0.021 0.183 0.053 �0.001 0.063Maximum 0.609 0.305 0.396 0.155 0.494 0.246 0.420 0.220Minimum �0.413 0.094 �0.453 �0.155 �0.392 �0.185 �0.270 �0.104Std. Dev. 0.197 0.029 0.127 0.065 0.118 0.066 0.116 0.052


the sumP

d1i,Pðd1i þ d2iÞ, or

Pd2i, whether the total influence of exchange rate risk on exports

equals zero for depreciations, for appreciations, or for the differences between the two sums.The coefficient sums of exchange rate risk in depreciation significantly differ from zero for all

countries except Singapore and Thailand. Five countries exhibit significant negative effects; oneexhibits a significant positive effect. The magnitude of the sum ranges from 0.614 in Malaysia to�3.479in Taiwan. The coefficient sums during appreciation significantly differ from zero for Japan, Korea,Philippines, Singapore, and Thailand. Three countries exhibit a significantly negative sum; two exhibita significantly positive sum. The magnitude ranges from 0.494 for the Philippines to �3.671 forSingapore. Generally, the exchange rate risk affects exports for all countries. The effect proves negativefor depreciations or appreciations in four countries – Indonesia, Japan, Singapore, and Taiwan. Itexhibits a mixed negative or positive effect for depreciations or appreciations for the other fourcountries – Korea, Malaysia, the Philippines and Thailand. In sum, all eight countries exhibit evidenceof our first test of asymmetry.

These findings support our argument that Asian exporters respond differently to exchange rate riskin periods of depreciation and appreciation due to currency mismatches in trade. Even Japan, the onlydeveloped economy in our sample, possesses negative effects of exchange rate risk. Indonesia andMalaysia provide noteworthy examples. Indonesia not only experiences the highest depreciation ratebut also the highest standard deviation (in Table 2), where in the period of depreciation it experiencesthe most significant negative effect of exchange rate risk (significance at the 5-percent, rather than the10-percent, level), providing strong evidence of unfavorable effect of excessive exchange rate volatilityon exports. In contrast, Malaysia adopted an official peg to the US dollar since September 1998, the onlyemerging market economy in our sample to do so. An official peg policy creates a predictable envi-ronment for trade that generates a positive effect of exchange rate risk in depreciations. For the otherfive emerging market economies, Korea, the Philippines, and Thailand suffered seriously from theAsian crisis, and Singapore and Taiwan did not. In Fig. 2, we see that the three Asian crisis foreignexchange markets display high standard deviations, which closely track each other and prove similar tothat of Japan. Facing volatile exchange rates, both the monetary authorities and exporters aggressivelymanage exchange risk, resulting in a positive or small negative risk effect. On the other hand, Singaporeand Taiwan exhibit the lowest standard deviations in Table 2. Low volatility may lull the authorities andexporters into neglecting risk and leads to pure negative effects. That is, for these two countries, therisk estimates achieve the highest levels among our eight sample countries.

We further test for our second form of significant asymmetry. Since the exchange rate risk provessignificant on exports either in depreciation or appreciation (or both), the difference between the twocoefficient sums,

Pd2i, determines the test. We see that

Pd2i significantly differs from zero for Japan,

Korea, Malaysia, the Philippines, and Singapore, denoting different magnitudes for the depreciationand appreciation risk effects. In sum, these five countries exhibit significant asymmetry. The differencebetween the two coefficient sums does not differ significantly from zero in Indonesia, Taiwan, andThailand. That is, no significant difference exists between the magnitudes of the depreciation andappreciation risk effects and, thus, no significant asymmetry. Nonetheless, these three countries stillexhibit our first form of asymmetry in that one of the depreciation or appreciation risk effects provessignificant while the other does not.

We conjecture that Asian exporters show more concern for foreign exchange market appreciationthan depreciation due to US-dollar invoicing and currency conversion in trade. That is, exporters reactasymmetrically to foreign exchange losses and gains. This tendency gets reflected in exports, leading to

-.6

-.4

-.2

.0

.2

.4

.6

.8

1980 1985 1990 1995 2000.08

.12

.16

.20

.24

.28

.32

1980 1985 1990 1995 2000

-.6

-.4

-.2

.0

.2

.4

.6

1980 1985 1990 1995 2000-.16

-.12

-.08

-.04

.00

.04

.08

.12

.16

1980 1985 1990 1995 2000

-.4

-.2

.0

.2

.4

.6

1980 1985 1990 1995 2000-.2

-.1

.0

.1

.2

.3

1980 1985 1990 1995 2000

-.3

-.2

-.1

.0

.1

.2

.3

.4

.5

1980 1985 1990 1995 2000-.12-.08-.04.00.04.08.12.16.20.24

1980 1985 1990 1995 2000

Indonesia Japan

Korea Malaysia


Taiwan Thailand

Fig. 3. Dynamic conditional correlations.


Table 6Tests of asymmetric effect of exchange rate risk.P

d1iPðd1i þ d2iÞ

Pd2i

Indonesia �0.376* �0.198 0.178LR statistic 5.434 (0.020) 1.687 (0.194) 0.600 (0.439)

Japan �1.411** �2.371* �0.960*LR statistic 3.444 (0.063) 7.385 (0.007) 9.327 (0.002)

Korea �0.793** 0.218* 1.011**LR statistic 3.693 (0.055) 15.107 (0.000) 3.509 (0.061)

Malaysia 0.614* �0.169 �0.783*LR statistic 3.951 (0.047) 0.243 (0.622) 4.274 (0.039)

Philippines �0.735** 0.494** 1.229*LR statistic 3.249 (0.071) 3.154 (0.076) 5.432 (0.020)

Singapore �2.339 �3.671* �1.332*LR statistic 2.087 (0.149) 4.635 (0.031) 4.277 (0.039)

Taiwan �3.479** �2.722 0.757LR statistic 3.323 (0.068) 1.968 (0.161) 1.070 (0.301)

Thailand 0.199 �0.663* �0.862LR statistic 0.256 (0.613) 4.657 (0.031) 1.840 (0.175)

Note: LR statistic equals the likelihood ratio statistic that follows a c2 distribution with one degree of freedom that testsPd1i ¼ 0,

Pðd1i þ d2iÞ ¼ 0 and

Pd2i ¼ 0. P-values appear in parentheses.

* Denotes significance at the 5-percent level.** Denotes significance at the 10-percent level.


a greater risk effect for appreciations than for depreciations. From the negative sum,P

d2i, in Table 6,we observe that exporters fear appreciation in Japan, Malaysia, and Singapore. In terms of exporter riskperception, exporters prove more sensitive to appreciation (an adverse change) than to depreciation(a beneficial change). The positive risk effect, however, leaves a positive difference

Pd2i in Korea and

the Philippines. In this case, exporters manage the adverse effects of appreciation aggressively andobtain a positive effect, but leave the beneficial effects of depreciation unmanaged.

Exchange rate depreciation (appreciation) exhibits the expected positive (negative) effect onexports (i.e., ci coefficients) in each country, except for Malaysia and Singapore. The effect of exchangerate risk can complement or offset such exchange rate effects, depending on the country, whether theexchange rate depreciates (appreciates), and whether the exchange rate risk increases (decreases).Assuming that larger exchange rate adjustments associate with higher exchange rate risk, we can drawthe following inferences from our estimates. If these Asian countries try to stimulate their exports bydepreciating their currencies, those attempts to stimulate exports receive significant reinforcementfrom the exchange rate risk in Malaysia, but offsetting effects in Indonesia, Japan, Korea, thePhilippines, and Taiwan.

Previous empirical results on the effects of exchange rate risk without distinguishing asymmetricresponses provide mixed results. As a comparison, we also estimate the symmetric-effect GARCHmodel in Eqs. (1)–(9). Table 7 reports estimates. Diagnostic tests support the statistical appropriatenessof the dynamic conditional correlation bivariate GARCH-M model. First, the positive effects of USmanufacturing income of the two models produce a reasonable match. Second, significant positivedepreciation effects exist for all eight countries in the symmetric effect model. Although they exhibitsimilar patterns in the two models, the effect proves insignificant in the asymmetric model forMalaysia and Singapore. That is, the symmetric model provides more evidence of positive depreciationeffects than the asymmetric model. Third, the cumulative exchange rate risk effect in the symmetricmodel proves significantly negative for three countries – Indonesia, Japan, and Taiwan – and notsignificant for the other five countries. These findings agree with the majority of prior studies, whichconclude with either a negative exchange rate risk effect or no effect. In contrast, the asymmetricmodel identifies significant negative exchange rate risk effects for all countries, except Malaysia, forappreciations, depreciations, or both. Malaysia along with Korea and the Philippines exhibit significant

Table 7Estimates for dynamic conditional correlation bivariate GARCH-M model (1)–(9).


Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E. Coef. S.E.

a0 1.691* 0.415 3.937* 0.240 0.761** 0.427 0.442 0.408 0.128 0.305 3.366* 0.099 10.165* 0.241 1.314* 0.370a1 �0.643* 0.048 �0.570** 0.044 �0.577* 0.049 �0.625* 0.047 �0.617* 0.042 �0.684* 0.049 �0.736* 0.070 �0.645* 0.047a2 �0.353* 0.047 �0.272* 0.041 �0.277* 0.042 �0.250* 0.050 �0.230* 0.043 �0.257* 0.035 �0.324* 0.047 �0.321* 0.045

b0 2.865* 0.667 1.212* 0.347 1.521* 0.651 1.176* 0.471 2.618* 0.566 1.539* 0.554 2.446* 0.570b1 1.828* 0.604 1.550* 0.415 1.579* 0.524b2 1.066* 0.336

c0 0.298* 0.082 0.562* 0.118 0.936* 0.106 0.474* 0.131c1 0.280* 0.069 0.453* 0.079 0.924* 0.195 0.380* 0.188 0.395* 0.133 0.419** 0.215 0.590* 0.256 0.780* 0.166c2 0.148** 0.076 0.325* 0.078 0.485* 0.130

d0 0.421* 0.083 �0.088 0.229 1.189* 0.188 0.664* 0.122 1.579* 0.071d1 �0.653* 0.086 �1.476* 0.080 0.741* 0.123 �1.959* 0.097 �0.253* 0.122d2 �0.962* 0.199 �1.308* 0.124 �3.804* 0.084 �4.973* 0.103

s0 0.072 0.056 0.174 0.189 0.033 0.069 0.117** 0.063 0.004 0.094 0.037 0.088 0.106 0.093 �0.042 0.059s1 0.202* 0.068 0.310* 0.058 0.351* 0.055 0.183* 0.068 0.356* 0.057 0.236* 0.053 0.218* 0.060 0.212* 0.062g1 30.258* 1.483 6.065* 1.206g2 16.037* 0.598 15.069* 1.198

a0 1.839* 0.701 13.528* 1.892 40.966* 6.764 5.722* 1.275 8.397* 1.820 4.106* 0.053 44.521* 5.117 1.559* 0.479a1 0.096* 0.015 0.182** 0.097 0.363* 0.118 0.139* 0.027 0.240* 0.053 0.173* 0.006 0.092 0.069 0.082* 0.012a2 0.887* 0.011 0.282* 0.075 0.797* 0.020 0.725* 0.028 0.793* 0.005 0.890* 0.010

b0 0.251* 0.044 6.164* 0.479 0.118* 0.023 0.796* 0.099 0.713* 0.146 0.309* 0.023 1.823* 0.083 0.083* 0.016b1 0.489* 0.087 0.172* 0.055 0.101* 0.027 0.357* 0.099 0.333* 0.067 0.099* 0.023 0.164* 0.031 0.100* 0.021b2 0.299* 0.048 0.761* 0.024 0.401* 0.059 0.732* 0.016 0.787* 0.017l1 10.869* 4.018 0.799* 0.307 34.865* 15.375 16.632* 5.160 10.868* 3.659l2 18.962 11.580

v 5.691* 0.934 7.131* 1.858 4.143* 0.461 5.069* 0.795 3.023 * 0.170 7.174* 0.579 5.164 * 0.946 6.105* 1.257

q1 0.160 ** 0.082 0.057 ** 0.032 0.099* 0.030 0.011 0.018 0.204** 0.110 0.061** 0.034 0.049* 0.023 0.040* 0.006q2 0.592* 0.198 0.730* 0.073 0.828* 0.005 0.984* 0.037 0.441* 0.120 0.649* 0.155 0.859* 0.115 0.952* 0.013

Q2(6) 32.658 20.294 30.200 28.275 36.108 8.848 28.183 36.001Q2

2 ð6Þ 13.299 20.950 14.066 10.643 15.949 20.672 16.552 23.231Pdi �0.232* �1.476** �0.088 0.227 0.097 �2.226 �6.932* �0.253

x2 (4.624) (3.273) (0.082) (0.642) (0.263) (1.968) (6.889) (2.478)LR(k) 4.788 (4) 3.414 (5) 5.621 (5) 7.844 (5) 1.962 (2) 3.606 (5) 2.874 (4) 5.858 (4)

Note: see Table 4. c2 statistics are in parentheses testing for the significance ofP

di .* Denotes significance at the 5-percent level.** Denotes significance at the 10-percent level.

WenShw

oFang

etal./

Journalof

International

Money

andFinance

28(2009)

215–239235


positive effects for appreciations or depreciations. The asymmetric model that allows differentresponses during depreciations and appreciations provides more evidence of the effect of exchangerate risk on exports.

Klaassen (2004) finds no exchange rate risk effect on monthly bilateral US exports to other G7countries. The present paper uses data on monthly bilateral exports from eight Asian countries to theUS – seven developing and one developed. Applying the newly developed dynamic conditionalcorrelation bivariate GARCH(1,1)-M model and allowing asymmetric responses, we find significantexchange rate risk effects for all countries studied. The main difference between Klaassen (2004) andour paper relates to the recognition of exporter risk perceptions. Klaassen (2004) and all previousresearchers assume implicitly symmetric risk behavior with respect to depreciations and apprecia-tions. In contrast, we argue explicitly for asymmetric responses to the risk of exchange rate changes.The evidence of significant risk effects in our study may suggest that exporter asymmetric hedgingbehavior plays its role in determining the risk effect on exports. And exchange rate risk does affectexports asymmetrically, at least, in Asian economies.

5. Summary and discussion

This paper applies dynamic conditional correlation bivariate GARCH-M model to examine theasymmetric effects of exchange rate risk on exports, using monthly bilateral exports from eight Asiancountries to the US over the period 1979–2003. The empirical results summarize as follows. For all theeight countries, foreign income affects exports positively and significantly with contemporaneous,one-month-lagged or two-month-lagged effects. Exchange rate depreciation exhibits the normalpositive effect, but proves insignificant in two countries. Real exchange rate risk (volatility) producessignificant effect on exports for all countries, negative or positive. Moreover, all countries also exhibitour first form of asymmetry with respect to exchange rate risk during appreciations and depreciationsof the exchange rates. For Japan, Korea, Malaysia, the Philippines, and Singapore, the effects ofexchange rate risk prove consistent with our second form of significant asymmetry. The other threecountries, Indonesia, Taiwan, and Thailand, exhibit significant effects of exchange rate risk on exportsduring depreciations or appreciations, but not both. Indonesia, Japan, and Taiwan respond negatively toexchange rate risk during depreciations. Korea and the Philippines respond negatively to exchange raterisk during appreciations and positively in appreciations. Malaysia exhibits only a positive exchangerate risk effect during depreciations.

Our findings strongly support the view that exchange rate risk affects exports asymmetrically. Suchresponse asymmetries may arise from different sources such as exporter asymmetric risk perception,US-dollar invoicing, original sin, fear of floating, fear of appreciation, love of depreciation, lack of activeor improper foreign exchange market intervention, and so on. Asymmetries could emanate from one ormore sources. This paper concerns itself with the existence, rather than the sources, of asymmetries.Future research can consider the specific sources of asymmetric effects.

In sum, the conventional assumption of a symmetric effect of exchange rate risk at the aggregatelevel appears invalid. Given our asymmetric effects, then unfavorable effects of exchange rate risk onexports prove significant in five countries – Indonesia, Japan, Korea, the Philippines, and Taiwan –during depreciations, but in only three countries – Japan, Singapore, and Thailand – during appreci-ations. Unfavorable effects of exchange rate risk exist during depreciations and favorable effects duringappreciations for Korea and the Philippines. The role of the exchange rate in determining exportrevenue may prove less predictable, given asymmetric effects. Consider the effect of depreciations inKorea and the Philippines. In Fig. 2, their currencies depreciated substantially against the dollar inrecent years, especially after the Asian crisis of 1997. Although both countries possess strong positivedepreciation effects (the highest estimates among the eight countries in Table 4), the asymmetric effectgenerates a negative exchange rate risk effect, leading to an uncertain net effect of the depreciation onexports. This last statement assumes that the recent depreciation associates with higher exchange raterisk, which appears to be the case from Fig. 2. The negative exchange rate risk effect could offset or evendominate the positive depreciation effect. For Malaysia, however, the asymmetric exchange rate riskeffect reinforces the positive effect of depreciation.


References

Abeysinghe, T., Yeok, T.L., 1998. Exchange rate appreciation and export competitiveness: the case of Singapore. AppliedEconomics 30 (1), 51–55.

Andren, N., 2001. Is macroeconomic exposure asymmetric? Arne Ryde Workshop in Empirical Finance, Lund University.Arize, A.C., 1995. The effects of exchange-rate volatility on U.S. exports: an empirical investigation. Southern Economic Journal

62 (1), 34–43.Arize, A.C., 1996. Real exchange-rate volatility and trade flows: the experience of eight European economies. International

Review of Economics and Finance 5 (2), 187–205.Arize, A.C., 1997. Foreign trade and exchange-rate risk in the G-7 countries: cointegration and error-correction models. Review

of Financial Economics 6 (1), 95–112.Arize, A.C., Osang, T., Slottje, D.J., 2000. Exchange-rate volatility and foreign trade: evidence from thirteen LDC’s. Journal of

Business and Economic Statistics 18 (1), 10–17.Arize, A.C., Malindretos, J., Kasibhatla, K.M., 2003. Does exchange-rate volatility depress export flows: the case of LDCs.

International Advances in Economics Research 9 (1), 7–19.Asseery, A., Peel, D.A., 1991. The effects of exchange rate volatility on exports: some new estimates. Economics Letters 37 (2),

173–177.Bailey, M.J., Trvias, G.S., Ulan, M., 1986. Exchange rate variability and trade performance: evidence for the big seven industrial

countries. Weltwirtschaftliches Archiv 122 (3), 466–477.Bailey, M.J., Trvias, G.S., Ulan, M., 1987. The impact of exchange rate volatility on export growth: some theoretical considerations

and the empirical results. Journal of Policy Modeling 9 (1), 225–244.Baldwin, R., 1988. Hysteresis in import prices: the beachhead effect. American Economic Review 78 (4), 773–785.Baldwin, R., Krugman, P., 1989. Persistent trade effects of large exchange rate shocks. Quarterly Journal of Economics 104 (4),

635–654.Barkoulas, J.T., Baum, C.F., Caglayan, M., 2002. Exchange rate effects on the volume and variability of trade flows. Journal of

International Money and Finance 21 (4), 481–496.Bayoumi, T., Eichengreen, B., 1998. Exchange rate volatility and intervention: implications of the theory of optimum currency

areas. Journal of International Economics 45 (2), 191–209.Bekaert, G., Wu, G., 2000. Asymmetric volatility and risk in equity markets. Review of Financial Studies 13 (1), 1–42.Berndt, E.K., Hall, B.H., Hall, R.E., Hausmann, J.A., 1974. Estimation and inference in nonlinear structural models. Annals of

Economic and Social Measurement 3 (4), 653–665.Black, F., 1976. Studies of stock prices volatility changes. In: Proceeding from the American Statistical Association, Business and

Economics Statistics Section, pp. 177–181.Bollerslev, T., 1986. Generalized autoregressive conditional heteroskedasticity. Journal of Econometrics 31 (3), 307–327.Bollerslev, T., 1990. Modeling the coherence in short-run nominal exchange rates: a multivariate generalized ARCH approach.

Review of Economics and Statistics 72 (3), 498–505.Bollerslev, T., Chou, R.J., Kroner, K.F., 1992. ARCH modeling in finance: a review of the theory and empirical evidence. Journal of

Econometrics 52 (1–2), 5–59.Broll, U., Eckwert, B., 1999. Exchange rate volatility and international trade. Southern Economic Journal 66 (1), 178–185.Brooks, C., Henry, O.T., 2000. Can portmanteau model nonlinearity tests serve as general model misspecification diagnostics?

Evidence from symmetric and asymmetric GARCH Models. Economics Letters 67 (3), 245–251.Calvo, G., Reinhart, C., 2002. Fear of floating. Quarterly Journal of Economics 117 (2), 379–408.Campbell, J.Y., Hentschel, L., 1992. No news is good news. Journal of Financial Economics 31 (3), 281–318.Caporale, T., Doroodian, K., 1994. Exchange rate variability and the flow of international trade. Economics Letters 46 (1), 49–54.Cappiello, L., Engle, R.F., Sheppard, K., 2006. Asymmetric dynamics in the correlations of global equity and bond returns. Journal

of Financial Econometrics 4 (4), 537–572.Chowdhury, A.R., 1993. Does exchange rate volatility depress trade flows? Evidence from error-correction models. Review of

Economics and Statistics 75 (4), 700–706.Christie, A.A., 1982. The stochastic behavior of common stock variances: value, leverage and interest rate effects. Journal of

Financial Economics 10 (4), 407–432.Christophe, S.E., 1997. Hysteresis and the value of the U.S. multinational corporations. Journal of Business 70 (3), 435–462.Cushman, D.O., 1983. The effects of real exchange rate risk on international trade. Journal of International Economics 15 (1-2),

45–63.Cushman, D.O., 1986. Has exchange rate risk depressed international trade? The impact of third country exchange rate risk.

Journal of International Money and Finance 5 (3), 361–379.De Grauwe, P., 1988. Exchange rate variability and the slowdown in growth of international trade. International Monetary Fund

Staff Papers 35 (1), 63–84.Dellas, H., Zilberfarb, B.Z., 1993. Real exchange rate volatility and international trade: a reexamination of the theory. Southern

Economic Journal 59 (4), 641–647.Dixit, A., 1989. Hysteresis, import penetration, and exchange rate pass-through. Quarterly Journal of Economics 104 (2),

205–227.Dooley, M.P., Folkerts-Landau, D., Garber, P., 2004. The revived Bretton woods system. International Journal of Finance and

Economics 9 (4), 307–313.Doroodian, K., 1999. Does exchange rate volatility deter international trade in developing countries? Journal of Asian Economics

10 (3), 465–474.Eichengreen, B., Hausmann, R., 1999. Exchange rates and financial fragility. In: New Challenges for Monetary Policy. Federal

Reserve Bank of Kansas City, pp. 329–368.Engle, R.F., 2002. Dynamic conditional correlation: a simple class of multivariate generalized autoregressive conditional het-

eroskedasticity models. Journal of Business and Economic Statistics 20 (3), 339–350.Engle, R.F., Ng, V., 1993. Measuring and testing the impact of news on volatility. Journal of Finance 48 (5), 1749–1778.


Engle, R.F., Lilien, D.M., Robins, R.P., 1987. Estimating time-varying risk premia in the term structure: the ARCH-M model.Econometrica 55 (2), 391–407.

Ethier, W., 1973. International trade and the forward exchange market. American Economic Review 63 (3), 494–503.Fang, W.S., Lai, Y.H., Miller, S.M., 2006. Export promotion through exchange rate changes: exchange rate depreciation of

stabilization? Southern Economic Journal 72 (3), 611–626.Fang, W.S., Miller, S.M., 2007. Exchange rate depreciation and exports: the case of Singapore revisited. Applied Economics 39 (3),

273–277.Fang, W.S., Thompson, H., 2004. Exchange rates risk and export revenue in Taiwan. Pacific Economic Review 9 (2), 117–129.Franke, G.,1991. Exchange rate volatilityand international trading strategy. Journal of International Money and Finance 10 (2), 292–307.Fratzscher, M., 2006. On the long-term effectiveness of exchange rate communication and interventions. Journal of Interna-

tional Money and Finance 25 (1), 146–167.French, K.R., Schwert, G.W., Stambaugh, R.F.,1987. Expected stock returns and volatility. Journal of Financial Economics 19 (1), 3–31.Froot, K.A., Klemperer, P.D., 1989. Exchange rate pass-through when market share matters. American Economic Review 79 (4),

637–654.Glosten, L., Jagannathan, R., Runkle, D., 1993. Relationship between the expected value and the volatility of the nominal excess

returns on stocks. Journal of Finance 48 (5), 1779–1801.Goldberg, P.K., 1995. Product differentiation and oligopoly in international markets: the case of the U.S. automobile industry.

Econometrica 63 (4), 891–951.Griffin, J.M., Stulz, R.M., 2001. International competition and exchange rate shocks: a cross-country industry analysis of stock

returns. Review of Financial Studies 14 (1), 215–241.Hausmann, R., Panizza, U., Rigobon, R., 2006. The long-run volatility puzzle of the real exchange rate. Journal of International

Money and Finance 25 (1), 93–124.Hodrick, R.J., Srivastava, S., 1984. An investigation of risk and return in forward foreign exchange. Journal of International Money

and Finance 3 (1), 5–29.Hooper, P., Kohlhagen, S.W., 1978. The effects of exchange rate uncertainty on the price and volume of international trade.

Journal of International Economics 8 (4), 483–511.Hosking, J.R.M., 1980. The multivariate portmanteau statistic. Journal of the American Statistical Association 75 (371), 602–608.International Monetary Fund, 1984. Exchange rate volatility and world trade. Occasional Paper 28, Washington, DC.Kanas, A., 1997. Is economic exposure asymmetric between long-run depreciations and appreciations? Testing using cointe-

gration analysis. Journal of Multinational Financial Management 7 (1), 27–42.Kenen, P.B., Rodrik, D., 1986. Measuring and analyzing the effects of short-term volatility in real exchange rates. Review of

Economics and Statistics 68 (2), 311–315.Klaassen, F., 2004. Why is it so difficult to find an effect of exchange rate risk on trade? Journal of International Money and

Finance 23 (5), 817–839.Knetter, M.M., 1994. Is export price adjustment asymmetric? Evaluating the market share and marketing bottlenecks

hypothesis. Journal of International Money and Finance 13 (1), 55–70.Koedijk, K.G., Lothian, J.R., van Dijk, M.A., 2006. Foreign exchange markets: overview of the special issue. Journal of International

Money and Finance 25 (1), 1–6.Korary, K., Lastrapes, W.D., 1989. Real exchange rate volatility and U.S. bilateral trade: a real approach. Review of Economics and

Statistics 71 (4), 708–712.Koutmos, G., Martin, A.D., 2003. Asymmetric exchange rate exposure: theory and evidence. Journal of International Money and

Finance 22 (3), 365–383.Kroner, K.F., Lastrapes, W.D., 1993. The impact of exchange rate volatility on international trade: reduced form estimates using

the GARCH in mean model. Journal of International Money and Finance 12 (3), 298–318.Krugman, P., 1987. Pricing to market when the exchange rate changes. In: Arndt, S.W., Richardson, J.D. (Eds.), Real Financial

Linkages among Open Economies. MIT Press, Cambridge, Mass, pp. 49–70.Ljungqvist, L., 1994. Hysteresis in international trade: a general equilibrium analysis. Journal of International Money and Finance

13 (4), 387–399.Lothian, J.R., Taylor, M.P., 1996. Real exchange rate behavior: the recent float from the perspective of the past two centuries.

Journal of Political Economy 104 (3), 488–509.Mahdavi, S., 2000. Do German, Japanese, and U.S. export prices asymmetrically respond to exchange rate changes? Evidence

from aggregate data. Contemporary Economic Policy 18 (1), 70–81.Marston, R.C., 1990. Pricing to market in Japanese manufacturing. Journal of International Economics 29 (3-4), 217–236.McKenzie, M.D., 2002. The economics of exchange rate volatility asymmetry. International Journal of Finance and Economics 7

(3), 247–260.McKenzie, M.D., Brooks, R.D., 1997. The impact of exchange rate volatility on German–U.S. trade flow. Journal of International

Financial Markets, Institutions and Money 7 (1), 73–87.McKenzie, M.D., Melbourne, R.M.I.T., 1999. The impact of exchange rate volatility on international trade flows. Journal of

Economic Surveys 13 (1), 71–106.McKinnon, R.I., 2000. The east Asian dollar standard, life after death? Economic Notes 29 (1), 31–82.McKinnon, R.I., Schnabl, F., 2004. The east Asian dollar standard, fear of floating, and original sin. Review of Development

Economics 8 (3), 331–360.Muller, A., Verschoor, W.F.C., 2006. Asymmetric foreign exchange risk exposure: evidence from U.S. multinational firms. Journal

of Empirical Finance 13 (4–5), 495–518.Nelson, D.B., 1991. Conditional heteroskedasticity in asset returns: a new approach. Econometrica 59 (2), 347–370.Perron, P., 1989. The great crash, the oil price shock, and the unit root hypothesis. Econometrica 57 (6), 1361–1401.Perron, P., 1997. Further evidence on breaking trend functions in macroeconomic variables. Journal of Econometrics 80 (2),

355–385.Pozo, S., 1992. Conditional exchange-rate volatility and the volume of international trade: evidence from the early 1990s.

Review of Economics and Statistics 74 (2), 325–329.


Rose, A.,1990. Exchange rates and the trade balance: some evidence from developing countries. Economics Letters 34 (3), 271–275.Sercu, P., 1992. Exchange rate risk, exposure, and the option to trade. Journal of International Money and Finance 11 (6), 579–593.Thursby, J.G., Thursby, M.C., 1987. Bilateral trade flows, the linder hypothesis, and exchange risk. Review of Economics and

Statistics 69 (3), 488–495.Tsay, R.S., 2002. Analysis of Financial Time Series. John Wiley & Sons, New York.Tse, Y.K., Tsui, K.C., 1997. Conditional volatility in foreign exchange rates: evidence from the Malaysia Ringgit and Singapore

Dollar. Pacific-Basin Finance Journal 5 (3), 345–356.Tse, Y.K., Tsui, K.C., 2002. A multivariate generalized autoregressive conditional heteroskedasticity model with time-varying

correlations. Journal of Business and Economic Statistics 20 (3), 351–362.Wang, K.L., Barrett, C.B., 2007. Estimating the effects of exchange rate volatility on export volumes. Journal of Agricultural and

Resource Economics 32 (2), 225–255.Weliwita, A., Ekanayake, E.M., Tsujii, H., 1999. Real exchange rate volatility and Sri Lanka’s exports to the developed countries,

1978–96. Journal of Economic Development 24 (1), 147–165.Wickham, P., 1985. The choice of exchange rate regimes in developing countries, 1978–96. International Monetary Fund Staff

Papers 32 (2), 248–288.World Bank, 1993. The East Asian Miracle: Economic Growth and Public Policy. Oxford University Press, Oxford.Wu, G., 2001. The determinants of asymmetric volatility. Review of Financial Studies 14 (3), 837–859.

does exchange rate risk affect exports asymmetrically? asian evidence

Documents