The Roles of Vehicle Currency and Real Effective Exchange Rates in the Trade of Every ASEAN Member With the EU-28

Abstract

A common limitation of the studies on the exchange rate-trade balance linkage is ignoring the USD’s role as a vehicle currency. Some recent papers have recognized this gap and proposed a hypothesis that the neglection of the USD’s role as a vehicle currency can restrict the significance of findings. This study aims to test that hypothesis in the trade between each of the 10 countries in ASEAN (Association of Southeast Asian Nations) with the whole EU-28. Specifically, this paper is the first to examine how the vehicle currency USD’s exchange rates, along with the real effect exchange rates, asymmetrically impact the trade balance of each ASEAN country with the EU-28 by employing the NARDL method to investigate whether the ignorance of the USD leads to fewer significant coefficients of exchange rates. The results indicate that the vehicle currency model (VCER) outperforms the real effective exchange rate (REER) model, which supports the hypothesis. In addition, the results are robust when controlling for the role of exchange rate volatility.

Keywords

asymmetric effects exchange rate vehicle currency trade balance ASEAN EU-28 exchange rate volatility

Introduction

ASEAN consists of 10 member countries: Brunei, Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand, and Vietnam. Most of them are emerging markets and developing countries with dynamic economies and high growth rates. Ranked at the eighth largest economy in the world, the entire ASEAN has approximately 640 million customers, which makes them an important partner in global trade (European Commission, 2021). In addition, ASEAN is respectively the largest trading partner of China (Harada, 2020), and the fourth largest trading partner of the US (Office of the United States Trade Representative, n.d.). Moreover, ASEAN is the EU’s third largest trading partner outside Europe, only behind China and the US (European Commission, 2021). Accordingly, realizing the prominence of ASEAN-EU trade, one priority of the EU is encouraging exportation to ASEAN countries (European Commission, 2021). Up to now, the EU-Singapore and EU-Vietnam free trade agreements already came into force, while the negotiation with Indonesia is ongoing and the negotiations with Malaysia, Philippines, and Thailand are delayed (European Commission, 2021). Besides, the EU has always been among the top trading partners of every ASEAN country.¹ Therefore, the ASEAN-EU trade has recently been received considerable attention.

In the trade between two partners, the currencies of the exporter or the importer, or the vehicle currency belonging to the third country can be used (Goldberg & Tille, 2008; Magee & Rao, 1980). The USD is the most dominant vehicle currency in global trade (Boz et al., 2020). It has been the world-leading vehicle currency for many decades and employed in around 88% of the global transactions, far exceeding the runner-up EUR (Bank for International Settlements, 2016, 2019; Krugman, 1980). Moreover, emerging market economies highly utilize the USD as a vehicle currency (Boz et al., 2020). Specifically, the share of USD in the exportation of emerging market economies in Asia is always more than 75%, and that of the Latin American counterparts is always nearly 87% (Boz et al., 2020). Furthermore, many high-income countries also substantially employ the USD as a vehicle currency. For instance, the EU-28 strongly utilizes the USD in trading with non-EU partners (Eurostat, 2021). Thus, the popularity of USD and its crucial role as a vehicle currency in global trade is undeniable as it is heavily used in the trade between any two countries, especially when they are not the US (Boz et al., 2020).

The exchange rate-trade balance relationship is a traditional topic in international finance and international economics. Based on the framework of the Marshall-Lerner condition and the J-curve effect, the literature is very abundant and available for a large number of countries (Bahmani-Oskooee & Hegerty, 2010; Bahmani-Oskooee & Ratha, 2004). Numerous studies were unable to find the significant impacts of exchange rates on trade balances due to the two main culprits: aggregation bias and the symmetric assumption of exchange rate-trade balance linkage (Usman et al., 2021). Recently, a new factor contributing to the lack of significant results was proposed: the neglection of the vehicle currency USD in the trade between any two non-US partners (Bao & Le, 2021b). Namely, in Vietnam’s bilateral trade with the countries in the EU and the UK, more significant results were detected when the USD’s role as a vehicle currency was incorporated (Bao & Le, 2021b). This finding raises a new question that whether the usefulness of including the USD’s role as a vehicle currency applies not only to the case of Vietnam but also the cases of other ASEAN countries. So far, this question has not been addressed by any research. In fact, very few studies have analyzed the exchange rate-trade balance linkage of an ASEAN member with respect to the EU-28 treated as a whole. To our knowledge, there are only two papers relating to this line of research: Bineau (2016) for Cambodia-EU trade, and Bahmani-Oskooee and Aftab (2017b) for Malaysia-EU trade, but they did not inspect the USD’s role as a vehicle currency. Hence, the trade of all ASEAN members with the EU-28 treated as a whole, along with the inclusion of the vehicle currency USD, has not been covered by any research.

Recognizing the importance of the USD as a vehicle currency, this paper tries to fill the gap unexamined by the existing studies about the ASEAN countries. Particularly, this paper explores the impacts of the vehicle currency USD’s exchange rates as well as real effective exchange rates on the trade balance of every ASEAN member with the whole EU-28. The main hypothesis to be tested in this paper is as follows:

Hm: The neglection of the USD’s role as a vehicle currency limits the significance of the findings. (In other words, the inclusion of the USD’s role as a vehicle currency can detect more significant impacts of exchange rates on ASEAN countries’ trade balances.)

To test this hypothesis, we compare the performance of the vehicle currency (VCER) model and the real effective exchange rate (REER) model. Moreover, for robustness check, we control for the role of exchange rate volatility and compare the performance of the extended VCER and REER models. The results support the hypothesis Hm.

This paper can contribute to the existing literature about the exchange rate-trade balance relationship in the cases of ASEAN countries. Namely, this paper is the first study to investigate the exchange rate-trade balance linkage for each of the 10 ASEAN countries in trading with the EU-28 treated as a whole. In addition, it is also the first to use the real effective exchange rate to fully capture the movement between an ASEAN country’s currency with all the currencies of the EU-28 rather than relying on the bilateral exchange rate between an ASEAN country’s currency with the EUR. Moreover, it is also the first paper to incorporate the role of USD as a vehicle currency in the trade of each ASEAN member with the entire EU-28, which allows more detailed findings and helpful policy recommendations when the impacts of real effective exchange rates and the vehicle currency USD’s exchange rates can be compared. Furthermore, this paper extends the results of Bao and Le (2021a) and reduce the aggregation bias which allows more significant findings. Last but not least, this paper enriches the literature of all ASEAN countries, especially for Brunei and Myanmar when hardly any research about these two countries has been published.

Literature Review

The exchange rate-trade balance nexus is a traditional topic that has been being inspected by numerous papers for a long time. Early studies tried to inspect if the devaluation of a country’s currency could stimulate its trade balance by validating the Marshall-Lerner condition. Accordingly, if they can prove that the absolute value of export demand elasticity plus the import analog is greater than one, the Marshall-Lerner condition is satisfied, and the devaluation policy is effective in boosting the trade balance (Wilson & Tat, 2001). The aforementioned approach is an indirect method to examine the influence of currency devaluation on the trade balance, and since the introduction of the term J-curve effect by Magee (1973), a large number of studies have evaluated the direct exchange rate-trade balance linkage (Soleymani & Chua, 2014). Bahmani-Oskooee (1985) first proposed the technique to detect the J-curve effect by observing the short-run pattern of the estimated exchange rate coefficients and proved that it existed in Greece, India, and Korea but not Thailand during the quarters from 1973 to 1980. Later, Rose and Yellen (1989) documented that the J-curve phenomenon is identified when the short-run exchange rate coefficients are negative and the long-run counterparts are positive. Grounded in the above-mentioned bases, manifold studies have scrutinized the relationship between exchange rates and trade balances of various countries and territories in the world, and their results are distinguishable (Bahmani-Oskooee & Hegerty, 2010; Bahmani-Oskooee & Ratha, 2004). Besides, the development of research has gone in line with the introduction of new econometric methods. Since the NARDL method was proposed by Shin et al. (2014), recent studies found strong evidences that exchange rates asymmetrically impact trade balances, and the use of NARDL method is superior to the ARDL one in analyzing the exchange rate-trade balance relationship (Iyke & Ho, 2018; Nusair, 2017).

Nearly all ASEAN countries have their own literature at different data levels. For example, at aggregate level, Arize (1994) contained five ASEAN members (Indonesia, Malaysia, the Philippines, Singapore, and Thailand) and reported that, among them, only Malaysia did not associate with the positive impact of currency depreciation on its trade balance during 1973Q1 to 1991Q1. However, when Yusoff (2010) used the VECM model and the 1977Q1 to 1998Q2 data, Malaysia’s trade balance with the rest of the world was encouraged when the ringgit depreciated. Nevertheless, when Bahmani-Oskooee and Kanitpong (2017) employed the ARDL and NARDL methods, they reported no reaction of Malaysia’s trade balance to the real effective exchange rate from 1975Q1 to 2016Q1. Thus, the findings of the same country vary with distinctive data, methods, and time frames. Accordingly, the results of Indonesia, the Philippines, Singapore, and Thailand reported by Bahmani-Oskooee and Kanitpong (2017) deviated from those of Arize (1994). For the case of Laos between 1993Q1 and 2010Q4, with the application of ARDL approach, Kyophilavong et al. (2013) identified no long-run impact of exchange rate on its trade balance. Lee (2018) examined the trade of Vietnam with the rest of the world proxied by her 15 largest trading partners in the period 1996Q1 to 2016Q4 by the ARDL technique and claimed that the depreciation of Vietnamese dong (VND) hurt Vietnam’s trade balance in the long run. In contrast, Phong et al. (2018) used 22 largest trading partners to represent the rest of the world, employed the ARDL method and the quarterly data spanning from 2000 to 2015, and demonstrated that the depreciation of VND facilitated Vietnam’s trade balance in a J-curve pattern.

Studies about the exchange rate-trade balance linkage at bilateral level are also available for most of the ASEAN countries, for instance, Cambodia (Bineau, 2016), Indonesia (Bahmani-Oskooee & Harvey, 2019), Laos (Kyophilavong et al., 2018), Malaysia (Bahmani-Oskooee & Harvey, 2010), the Philippines (Harvey, 2018), Singapore (Wilson & Tat, 2001), Thailand (Brahmasrene & Jiranyakul, 2002), and Vietnam (Lee, 2018). Remarkably, Bineau (2016) included the EU-28, along with other partners (Hong Kong, Indonesia, Japan, Korea, Malaysia, Singapore, Thailand, the UK, and the US) in his analysis of Cambodia’s bilateral trade in the quarters from 1998 to 2014. With the utilization of the Fully Modified OLS estimator, he concluded that the depreciation of riel improved Cambodia’s trade balance with all the partners. However, as the EU-28 consists of many countries, the bilateral exchange rate used by Bineau (2016) might not reflect the full movement of riel against all currencies of the EU-28. Thus, the real effective exchange rate between Cambodian riel and all currencies of the EU-28 can be a more suitable proxy. Some other notable papers relating to ASEAN members should also be listed. For example, Baharumshah (2001) assessed the relationship between the exchange rates and trade balances of Malaysia and Thailand with respect to the US and Japan during 1980Q1 to 1996Q4 by employing the VECM model. The results indicated that the depreciation of domestic currencies fostered Malaysia’s and Thailand’s trade balances with both the US and Japan, but the J-curve phenomenon was not supported. Wilson (2001) examined the impacts of exchange rates on the trade balance of Singapore, Malaysia, and Korea with respect to the US and Japan during 1970Q1 to 1996Q4. They reported that the depreciation of Malaysia’s currency encouraged its trade balance with the US, and no J-curve effect was detected.

Commodity-level studies are also available for some ASEAN countries, especially Malaysia. For example, Soleymani and Chua (2014) investigated the impacts of bilateral exchange rate on Malaysia’s trade balances with China in 52 industries during 1993Q1 to 2012Q4. The ARDL estimation results demonstrated that 18 industries were benefited from the depreciation of ringgit against yuan in the long run, but 13 industries were negatively affected. Again, Bahmani-Oskooee and Aftab (2018) focused on Malaysia-China trade in 59 commodities. Employing both ARDL and NARDL methods on the 2001M3 to 2015M12 data, they observed that the NARDL method was better than the ARDL counterpart in terms of providing more significant results. They also detected asymmetric short-run and long-run effects. Some other noticeable papers relating to Malaysia can be instanced as Bahmani-Oskooee and Aftab (2017a) for Malaysia-Thailand trade, and especially Bahmani-Oskooee and Aftab (2017b) for Malaysia-EU trade. Namely, applying both the ARDL and NARDL methods, Bahmani-Oskooee and Aftab (2017b) inspected the influences of EUR/MYR exchange rate on Malaysia’s trade balances with the whole EU-28 in 63 industries over the period 2000M5 to 2013M12. By choosing the EUR/MYR exchange rate, it could be suggested that the EUR was an important invoicing currency in Malaysia’s trade with all EU-28 countries even if nine of them (i.e., Bulgaria, Croatia, Czechia, Denmark, Hungary, Poland, Romania, Sweden, and the UK) did not employ the EUR as their official currencies. Nevertheless, the EUR cannot be considered vehicle currency² in the trade between Malaysia and the whole EU-28 because it belongs to the Eurozone which is a part of the EU-28. Hence, although the EUR/MYR is a good proxy, it may not fully capture the movement of MYR against all the currencies of the EU-28. Thus, the real effective exchange rate between MYR and all currencies of the EU-28 can be more appropriate. Apart from Malaysia, other countries such as the Philippines (Arize et al., 2017), Singapore (Arize et al., 2017), Thailand (Bahmani-Oskooee & Kanitpong, 2019), and Vietnam (Ho et al., 2021) have also been analyzed by the existing papers conducted at industry level.

While most of the existing literature about the exchange rate-trade balance nexus overlook the role of vehicle currency (Bao & Le, 2021b; Yang & Gu, 2016), a few papers have investigated this problem. Šimáková (2013) investigated the effects of EUR as a vehicle currency in Hungary-Czechia, Hungary-Poland, and Hungary-UK trade between 1997Q1 and 2012Q2, which revealed different responses of Hungary’s trade balance under the influences of the vehicle currency EUR’s exchange rates and the bilateral exchange rates. Yang and Gu (2016) acknowledged the crucial role of USD as a vehicle currency and discovered that China’s exportation and importation with Singapore were significantly affected by the depreciation of SGD or CNY against the USD during 1993M1 to 2013M12. Nhung et al. (2018) found that both the nominal bilateral exchange rate (JPY/VND) and the vehicle currency USD’s exchange rate (USD/VND) had significant impacts on Vietnam’s trade balance with Japan from 2001Q1 to 2017Q3. This result goes in line with the fact that roughly 90% of Vietnam’s trade value is invoiced in USD (Bao & Le, 2021b). Bao and Le (2021a) analyzed the trade between the entire ASEAN and the whole EU-28 in the period 2000Q1 to 2018Q1. They disclosed that the exchange rate between ASEAN’s currencies and the vehicle currency USD facilitated ASEAN’s overall trade balance with the EU-28, but the real effective exchange rate between them had no effect. Their findings reflect the importance of USD as a vehicle currency in the trade between ASEAN and the EU-28.

Methodology

A myriad of papers has relied on the standard two-country model to inspect the exchange rate-trade balance relationship (e.g., Iyke & Ho, 2017; Kyophilavong et al., 2013; Rose & Yellen, 1989). This study also employs the aforesaid model as follows:

l n T B_{i, t} = a_{i} + b_{i} . l n R E E R_{i, t} + c_{i} . l n Y_{i, t} + d_{i} . l n Y F_{i, t} + e_{i, t}

(1)

In equation (1), the subscript i represents each of the 10 ASEAN members, and t denotes time. Although equation (1) looks like panel data specification, it represents the 10 time-series regressions of the 10 ASEAN countries. Regarding the variables, $T B_{i}$ is the trade balance of the ith country with the whole EU-28, measured by the export/import ratio. In addition, $R E E R_{i}$ is the real effective exchange rate between the currency of the ith ASEAN member with all the currencies of the EU-28, and it is constructed in a manner that an increase signifies the depreciation of the ith country’s currency. Thus, if $b_{i} > 0$ , the depreciation of the ith country’s currency boosts its trade balance. Meanwhile, $Y_{i}$ symbolizes the real income of the ith country, proxied by its real GDP. Besides, $Y F_{i}$ represents the real GDP of the whole EU-28 which is defined in equation (2):

Y F_{i, t} = \prod_{j = 1}^{28} G D {P_{j, t}}^{ω_{j, i, t}}

(2)

In equation (2), $G D P_{j}$ indicates the real GDP of country j belonging to the EU-28, and $ω_{j, i}$ is the share it occupies in the total trade value of the EU-28 with the ith ASEAN member. It should be noted that $\sum_{j = 1}^{28} ω_{j, i} = 1$ for any value of i.

So as to investigate the role of USD as a vehicle currency, we use the model employed by Šimáková (2013) as well as Bao and Le (2021b):

l n T B_{i, t} = a'_{i} + b'_{i} . l n U S D_{i, t} + c'_{i} . l n Y_{i, t} + d'_{i} . l n Y F_{i, t} + e'_{i, t}

(3)

where $U S D_{i}$ denotes the real exchange rate between the USD and the currency of the ith ASEAN country. The increase of $U S D_{i}$ also indicates the depreciation of the ith country’s currency against the USD. Hence, when $b'_{i} > 0$ , the depreciation of the ith country’s currency against USD fosters its trade balance. Besides, the remaining variables in equation (3) are the same as those in equation (1).

In order to examine the asymmetric impacts of exchange rates on the trade balances, the variables $l n R E E R_{i, t}$ and $l n U S D_{i, t}$ are decomposed into the partial sums of positive and negative changes following the work of Shin et al. (2014):

R E E R_{i, t}^{+} = \sum_{g = 1}^{t} \max (Δ l n R E E R_{i, g}, 0)

(4)

R E E R_{i, t}^{-} = \sum_{g = 1}^{t} \min (Δ l n R E E R_{i, g}, 0)

(5)

U S D_{i, t}^{+} = \sum_{g = 1}^{t} \max (Δ l n U S D_{i, g}, 0)

(6)

U S D_{i, t}^{-} = \sum_{g = 1}^{t} \min (Δ l n U S D_{i, g}, 0)

(7)

In equations (4) and (5), $R E E R_{i, t}^{+}$ and $R E E R_{i, t}^{-}$ respectively denote the depreciation and appreciation of the ith country’s currency against all the currencies of the EU-28. Likewise, in equations (6) and (7), $U S D_{i, t}^{+}$ and $U S D_{i, t}^{-}$ respectively indicate the depreciation and appreciation of the ith country’s currency against the USD.

Equation (1), combined with the partial sums of positive and negative changes of $l n R E E R_{i, t}$ (i.e., $R E E R_{i, t}^{+}$ and $R E E R_{i, t}^{-}$ ), is turned into the error correction form based on the works of Pesaran et al. (2001) and Shin et al. (2014) as follows:

\begin{array}{l} Δ l n T B_{i, t} = α_{i} + \sum_{h = 1}^{p 1_{i}} (ρ_{i, h} . Δ l n T B_{i, t - h}) + \sum_{k = 0}^{p 2_{i}} (β_{i, k}^{+} . Δ R E E R_{i, t - k}^{+}) \\ + \sum_{l = 0}^{p 3_{i}} (β_{i, l}^{-} . Δ R E E R_{i, t - l}^{-}) + \sum_{s = 0}^{p 4_{i}} (γ_{i, m} . Δ l n Y_{i, t - s}) \\ + \sum_{n = 0}^{p 5_{i}} (δ_{i, n} . Δ l n Y F_{i, t - n}) + η_{i} . l n T B_{i, t - 1} + θ_{i}^{+} . R E E R_{i, t - 1}^{+} \\ + θ_{i}^{-} . R E E R_{i, t - 1}^{-} + π_{i} . l n Y_{i, t - 1} + ϕ_{i} . l n Y F_{i, t - 1} + ε_{i, t} \end{array}

(8)

In this paper, equation (8) is called the real effective exchange rate (REER) model because it allows evaluating both the short-run and long-run asymmetric impacts of real effective exchange rates on the trade balance of each ASEAN country with the whole EU-28. The NARDL method of Shin et al. (2014) is superior to the normal ARDL one in scrutinizing the impacts of exchange rates on trade balances and can provide more significant results (Nusair, 2017). In addition, the NARDL method possesses all the strengths of the ARDL method. Specifically, it permits the occurrence of I(1) and I(0) variables, and thus unit-root testing is not compulsory (Iyke & Ho, 2018). Shin et al. (2014) showed that equation (8) can be estimated by the conventional ARDL bound testing method of Pesaran et al. (2001). Particularly, the bound test will be conducted to identify the existence of cointegration among the variables. The null hypothesis of no cointegration H₀: $η_{i} = θ_{i}^{+} = θ_{i}^{-} = π_{i} = ϕ_{i} = 0$ is rejected if the bound-test F-statistic is higher than the upper bound connected with I(1) variables (Pesaran et al., 2001). In such case, the alternative hypothesis H₁: $η_{i} \neq θ_{i}^{+} \neq θ_{i}^{-} \neq π_{i} \neq ϕ_{i} \neq 0$ is supported, which signifies the presence of cointegration among the variables. Nevertheless, in case the bound-test F-statistic is lower than the lower bound connected with I(0) variables, H₀ cannot be rejected. Besides, if the F-statistic lies between the two bounds, no conclusion can be made. After the cointegration is identified, the REER model in equation (8) can be estimated for both short-run and long-run impacts. The estimated results are trustworthy if there is no autocorrelation, heteroskedasticity, and misspecification, which can be checked by Breusch-Godfrey, Breusch-Pagan, and Ramsey RESET tests. Moreover, the CUSUM (Cumulative Sum of Recursive Residuals) and CUSUMSQ (Cumulative Sum of Square of Recursive Residuals) tests will help verify if the estimated coefficients are stable, which further strengthens the reliability of the results.

Similarly, equation (2), together with the partial sums of positive and negative changes of $l n U S D_{i, t}$ (i.e., $U S D_{i, t}^{+}$ and $U S D_{i, t}^{-}$ ), is transformed into error correction model as follows:

\begin{array}{l} Δ l n T B_{i, t} = α'_{i} + \sum_{h = 1}^{q 1_{i}} (ρ'_{i, h} . Δ l n T B_{i, t - h}) + \sum_{k = 0}^{q 2_{i}} (β'_{i, k}^{+} . Δ U S D_{i, t - k}^{+}) \\ + \sum_{l = 0}^{q 3_{i}} (β'_{i, l}^{-} . Δ U S D_{i, t - l}^{-}) + \sum_{s = 0}^{q 4_{i}} (γ'_{i, m} . Δ l n Y_{i, t - s}) \\ + \sum_{n = 0}^{q 5_{i}} (δ'_{i, n} . Δ l n Y F_{i, t - n}) + η'_{i} . l n T B_{i, t - 1} + θ'_{i}^{+} . U S D_{i, t - 1}^{+} \\ + θ'_{i}^{-} . U S D_{i, t - 1}^{-} + π'_{i} . l n Y_{i, t - 1} + ϕ'_{i} . l n Y F_{i, t - 1} + ε'_{i, t} \end{array}

(9)

In this paper, equation (9) is regarded as the vehicle currency model (VCER) because it captures the short-run and long-run asymmetric effects of the vehicle currency USD’s exchange rates on the trade balances of the 10 ASEAN members with the whole EU-28. The estimation procedure of the VCER model is exactly the same as the REER model. Namely, in equation (9), the null hypothesis of the bound test for cointegration among the variables (i.e., H₀: $η'_{i} = θ'_{i}^{+} = θ'_{i}^{-} = π'_{i} = ϕ'_{i} = 0$ ) is tested against the alternative one (i.e., H₁: $η'_{i} \neq θ'_{i}^{+} \neq θ'_{i}^{-} \neq π'_{i} \neq ϕ'_{i} \neq 0$ ). After the cointegration is identified, long-run and short-run coefficients can be estimated. To ensure the reliability of the outcomes, Breusch-Godfrey, Breusch-Pagan, Ramsey RESET, CUSUM, and CUSUMSQ tests are also applied.

The 2000Q1 to 2018Q1 data³ used in this paper is collected from various sources including IMF’s Direction of Trade Statistics and International Financial Statistics, Eurostat, FRED (Federal Reserve Economic Data), OECD, ADB (Asian Development Bank), and General Statistics Office of Vietnam. All variables are in real terms and converted into indices in which the values of the base quarter 2000Q1 are set to 100. Due to the lack of quarterly GDP for Brunei, Cambodia, Laos, and Myanmar, annual data is retrieved and interpolated into quarterly data based on Kyophilavong et al. (2013, 2018).

Empirical Results and Discussion

Descriptive Statistics and Unit-Root Test

Before reporting the estimation results, it is useful to demonstrate some analyses of the descriptive statistics and unit-root tests of the variables. First, the descriptive statistics are presented in Table A1 in the Appendix section. For each ASEAN country, we have 73 observations of the variables lnTB_i,t, lnY_i,t, and lnYF_i,t. Regarding the exchange-rate variables (i.e., REER⁺_i,t, REER^–_i,t, USD⁺_i,t, and USD^–_i,t), there are 72 observations as one observation is lost due to the first differences of exchange rates (see equations (4)–(7)). Thus, as the sample size is below 80, we can apply the critical values of bound tests for cointegration given by Narayan (2005) which are the small-sample adjustments of the large-sample ones developed by Pesaran et al. (2001). In addition, although the variables REER⁺_i,t, REER^–_i,t, USD⁺_i,t, and USD^–_i,t do not have the symbol “ln,” they are still in natural logarithm since they are calculated as the partial sums of positive or negative changes in lnREER_i,t and lnUSD_i,t (see equations (4)–(7)). Accordingly, their coefficients in the REER and VCER models where lnTB_i,t is the dependent variable denote exchange rates’ elasticities.

The outcomes of unit-root tests are demonstrated in Table A2 in the Appendix section. The ADF and PP tests indicate no I(2) process. Moreover, the mixture of I(1) and I(0) variables is found in the majority of cases such as Brunei, Cambodia, Indonesia, Laos, Malaysia, Myanmar, and Vietnam. Therefore, the employment of NARDL method in this paper is appropriate. In addition, the NARDL method can also be applied for the cases with only I(1) variables (i.e., Philippines, Singapore, and Thailand), which is a notable advantage of ARDL-family methods (Odhiambo, 2009).

Estimation Results

The asymmetric impacts of exchange rates on the trade balances of the 10 ASEAN countries with the EU-28 between 2000Q1 and 2018Q1 are demonstrated in Table 1. For each of the 10 ASEAN countries, the real effective exchange rate (REER) and vehicle currency exchange rate (VCER) models are presented side by side. In addition, the symbol “POS_i,t” represents the variables REER⁺_i,t in the REER model and USD⁺_i,t in the VCER model, respectively. Similarly, the symbol “NEG_i,t” represents the variables REER^–_i,t in the REER model and USD^–_i,t in the VCER model, respectively. For each country’s models, the maximum lags are set to eight, and the optimal ones are selected by the Akaike information criterion (AIC). In case autocorrelation, heteroskedasticity, misspecification, etc. arise, we reduce the maximum lags and re-estimate the models. At first glance, the long-run relationship among the variables is found in nearly all cases, as evidenced by the significant bound-test F-statistics. Moreover, the outcomes are free from autocorrelation, heteroskedasticity and wrong functional form as none of the Breusch-Godfrey, Breusch-Pagan, and Ramsey RESET tests’ statistics are high enough to indicate such problems. Besides, the stability of the coefficients is affirmed when CUSUM tests are stable in all the cases and CUSUMSQ tests indicate the same results in the majority of them. Therefore, the NARDL estimation is trustworthy and can be used for further analyses.

Table 1.

NARDL Estimation Results of the Elementary REER and VCER Models.

	i = Brunei		i = Cambodia
	REER	VCER	REER	VCER
Long run
POS_i,t	3.07	−0.72	2.43***	3.30
NEG_i,t	−3.29	4.07**	−0.66***	7.36***
lnY_i,t	−4.11*	−3.63***	−5.14***	−11.45***
lnYF_i,t	10.98**	11.53***	6.48**	10.23***
Short run
ΔPOS_i,t	−3.46	−0.93	2.38***	2.36
ΔPOS_i,t−1	−6.89			9.46**
ΔPOS_i,t−2	−15.47**			11.10**
ΔPOS_i,t−3	−14.40**
ΔPOS_i,t−4	−4.31
ΔPOS_i,t−5
ΔPOS_i,t−6
ΔPOS_i,t−7
ΔNEG_i,t	1.68	5.25**	−0.64**	−2.10
ΔNEG_i,t−1	7.40			−4.37
ΔNEG_i,t−2	14.71**
ΔNEG_i,t−3	1.34
ΔNEG_i,t−4	1.89
ΔNEG_i,t−5
ΔNEG_i,t−6
ΔNEG_i,t−7
ΔlnY_i,t	2.35	3.09	−7.35***	−8.64***
ΔlnY_i,t−1	6.01	6.45	5.25**	5.48**
ΔlnY_i,t−2	2.91	0.47
ΔlnY_i,t−3	−3.21	0.64
ΔlnY_i,t−4	18.94**	11.43***
ΔlnY_i,t−5		−9.35**
ΔlnY_i,t−6		6.72
ΔlnY_i,t−7		8.13**
ΔlnYF_i,t	7.07	4.41	6.36**	11.17***
ΔlnYF_i,t−1	−14.15
ΔlnYF_i,t−2	1.10
ΔlnYF_i,t−3	−4.37
ΔlnYF_i,t−4	−22.47*
ΔlnYF_i,t−5
ΔlnYF_i,t−6
ΔlnYF_i,t−7
Constant	−44.93*	−42.31***	−0.51	10.80
Bound test	3.69**	10.90***	8.67***	11.64***
Adj—R²	.59	.67	.62	.70
Breusch-Godfrey	0.04	1.24	0.27	0.58
Breusch-Pagan	0.91	1.28	1.17	1.55
Ramsey RESET	2.74	0.18	2.25	0.70
CUSUM	S	S	S	S
CUSUMSQ	U	U	S	S
	i = Indonesia		i = Laos
	REER	VCER	REER	VCER
Long run
POS_i,t	0.61	0.64*	3.79*	−1.82
NEG_i,t	−0.53	−2.19***	−2.74*	3.34***
lnY_i,t	−2.41**	−1.22	−10.55**	2.54**
lnYF_i,t	−0.22	−0.58	14.09*	−2.32
Short run
ΔPOS_i,t	0.40	0.71	1.70	11.03
ΔPOS_i,t−1		0.07	−2.97
ΔPOS_i,t−2		−0.93	−3.47
ΔPOS_i,t−3		−0.83	−2.37
ΔPOS_i,t−4		1.68**	−1.08
ΔPOS_i,t−5		1.47**	3.33
ΔPOS_i,t−6
ΔPOS_i,t−7
ΔNEG_i,t	−0.32	−1.74***	−2.27**	−3.38
ΔNEG_i,t−1	−0.02	1.88**		−8.66*
ΔNEG_i,t−2	−0.35	2.75***		−2.46
ΔNEG_i,t−3	−0.26	2.54***		−12.26**
ΔNEG_i,t−4	−0.07	0.73
ΔNEG_i,t−5	−0.50*	0.36
ΔNEG_i,t−6	−0.90***
ΔNEG_i,t−7	−0.55*
ΔlnY_i,t	−1.60**	−1.99*	4.16	16.13***
ΔlnY_i,t−1		1.41	4.10	−2.64
ΔlnY_i,t−2		−0.35	10.94*	6.86
ΔlnY_i,t−3		−1.50	5.52	4.70
ΔlnY_i,t−4		−0.73	12.75**	4.32
ΔlnY_i,t−5			18.19***	10.36**
ΔlnY_i,t−6			1.01
ΔlnY_i,t−7			11.81*
ΔlnYF_i,t	−0.14	−1.59	−0.64	−0.25
ΔlnYF_i,t−1		2.60	−15.51***	−3.72
ΔlnYF_i,t−2		3.27	−12.19**	−3.70
ΔlnYF_i,t−3		2.51	−9.39*	−2.24
ΔlnYF_i,t−4		2.56	−0.39	7.67*
ΔlnYF_i,t−5		−0.35	−8.60*	0.91
ΔlnYF_i,t−6		2.29	−9.92**	−7.29*
ΔlnYF_i,t−7
Constant	10.98**	24.35	−10.94	4.48
Bound test	6.66***	6.15***	3.30*	5.68***
Adj—R²	.40	.53	.72	.70
Breusch-Godfrey	0.02	2.69	0.42	1.13
Breusch-Pagan	0.54	1.21	0.95	0.93
Ramsey RESET	0.19	0.32	2.34	1.85
CUSUM	S	S	S	S
CUSUMSQ	S	S	S	S
	i = Malaysia		i = Myanmar
	REER	VCER	REER	VCER
Long run
POS_i,t	0.43*	1.18*	0.25*	0.13**
NEG_i,t	0.29	1.74*	0.98	1.97***
lnY_i,t	−0.97***	0.57	−0.78	−0.10
lnYF_i,t	2.56***	−1.12	0.51	10.09***
Short run
ΔPOS_i,t	0.24*	0.32***	0.38	0.51**
ΔPOS_i,t−1			−0.04	0.15
ΔPOS_i,t−2			0.28	0.41**
ΔPOS_i,t−3			−0.48***	−0.45**
ΔPOS_i,t−4			−0.20
ΔPOS_i,t−5			−0.14
ΔPOS_i,t−6			−0.47***
ΔPOS_i,t−7
ΔNEG_i,t	−0.13	0.48***	3.86*	−5.14
ΔNEG_i,t−1	−0.50
ΔNEG_i,t−2	−0.29*
ΔNEG_i,t−3	−0.71*
ΔNEG_i,t−4	−0.73*
ΔNEG_i,t−5
ΔNEG_i,t−6
ΔNEG_i,t−7
ΔlnY_i,t	−0.58**	−0.04	5.62*	15.94***
ΔlnY_i,t−1	0.75***	0.58***
ΔlnY_i,t−2
ΔlnY_i,t−3
ΔlnY_i,t−4
ΔlnY_i,t−5
ΔlnY_i,t−6
ΔlnY_i,t−7
ΔlnYF_i,t	1.42**	−0.31	3.11	4.63
ΔlnYF_i,t−1			2.31	−7.29
ΔlnYF_i,t−2			−4.04	−10.32*
ΔlnYF_i,t−3			11.30*	1.98
ΔlnYF_i,t−4				−10.35**
ΔlnYF_i,t−5				−8.39**
ΔlnYF_i,t−6				−6.91
ΔlnYF_i,t−7
Constant	−1.69	1.88	7.27	−44.29***
Bound test	3.58*	3.32*	6.81***	12.07***
Adj—R²	.53	.41	.51	.63
Breusch-Godfrey	1.16	2.05	1.95	0.38
Breusch-Pagan	0.84	1.14	0.46	0.90
Ramsey RESET	2.43	1.17	0.05	0.52
CUSUM	S	S	S	S
CUSUMSQ	S	S	U	U
	i = Philippines		i = Singapore
	REER	VCER	REER	VCER
Long run
POS_i,t	3.83**	7.78**	−0.31	−5.12***
NEG_i,t	1.14	0.33	−0.60	−6.58***
lnY_i,t	−17.24***	−5.71***	1.85	5.57***
lnYF_i,t	10.30***	1.70	−3.88	3.02**
Short run
ΔPOS_i,t	−0.60	1.96	0.11	−5.16*
ΔPOS_i,t−1	−3.95***	−0.94	−0.65	3.32
ΔPOS_i,t−2	−4.46***	−2.65**	−0.46	9.94***
ΔPOS_i,t−3	−2.96***	−1.78	−0.31	−0.95
ΔPOS_i,t−4	−2.06**		0.35	7.81**
ΔPOS_i,t−5	−1.74**		1.36***	1.42
ΔPOS_i,t−6
ΔPOS_i,t−7
ΔNEG_i,t	0.07	0.18	−0.19	−2.66
ΔNEG_i,t−1	0.13			5.01*
ΔNEG_i,t−2	2.07**			9.11***
ΔNEG_i,t−3				3.17
ΔNEG_i,t−4				4.20
ΔNEG_i,t−5
ΔNEG_i,t−6
ΔNEG_i,t−7
ΔlnY_i,t	−2.49***	−3.27***	0.58	5.09**
ΔlnY_i,t−1	4.33***	−0.50		−2.53
ΔlnY_i,t−2	2.42***	−1.20		−8.54**
ΔlnY_i,t−3		−2.57**		−1.42
ΔlnY_i,t−4				−5.65*
ΔlnY_i,t−5
ΔlnY_i,t−6
ΔlnY_i,t−7
ΔlnYF_i,t	2.16	0.93	0.11	5.02***
ΔlnYF_i,t−1	−5.40***		4.35***	−0.49
ΔlnYF_i,t−2	−5.52***		2.55*	0.41
ΔlnYF_i,t−3	−4.67**			0.79
ΔlnYF_i,t−4	−3.35			−1.56
ΔlnYF_i,t−5				1.11
ΔlnYF_i,t−6				−5.43**
ΔlnYF_i,t−7
Constant	20.41**	12.07*	4.46	−36.98***
Bound test	7.62***	4.84***	2.89	3.12
Adj—R²	.54	.30	.57	.69
Breusch-Godfrey	1.43	0.60	0.00	0.05
Breusch-Pagan	1.03	1.16	0.77	1.01
Ramsey RESET	1.09	0.00	0.52	1.21
CUSUM	S	S	S	S
CUSUMSQ	S	S	S	S
	i = Thailand		i = Vietnam
	REER	VCER	REER	VCER
Long run
POS_i,t	0.33	3.20***	1.96***	3.25
NEG_i,t	0.25	−1.19***	−0.69	3.08**
lnY_i,t	−1.81***	−3.13***	−1.67***	−4.43***
lnYF_i,t	2.72*	4.55***	4.66*	3.12
Short run
ΔPOS_i,t	−0.84*	−0.59	−0.001	0.40
ΔPOS_i,t−1	−0.97*	−2.75**	−1.30**	−6.34***
ΔPOS_i,t−2	−0.61	−2.32**	−1.10**	−2.49
ΔPOS_i,t−3	−0.77	−1.88**
ΔPOS_i,t−4		−1.23
ΔPOS_i,t−5		−2.17***
ΔPOS_i,t−6
ΔPOS_i,t−7
ΔNEG_i,t	1.30*	−1.34*	−0.35	0.99***
ΔNEG_i,t−1	0.34	1.53*
ΔNEG_i,t−2	−1.04	1.35
ΔNEG_i,t−3	−0.38	1.65*
ΔNEG_i,t−4	−0.44	0.93
ΔNEG_i,t−5	−1.68***	2.46***
ΔNEG_i,t−6	−1.35**
ΔNEG_i,t−7
ΔlnY_i,t	0.16	−0.38	−0.86***	−1.01***
ΔlnY_i,t−1	−1.59*	2.49
ΔlnY_i,t−2	−2.46***	2.46
ΔlnY_i,t−3	−1.70	3.11*
ΔlnY_i,t−4	−2.30*	3.09*
ΔlnY_i,t−5	−0.20	4.21***
ΔlnY_i,t−6	1.18	2.37*
ΔlnY_i,t−7		1.39
ΔlnYF_i,t	−0.39	2.52	1.09	−1.34
ΔlnYF_i,t−1	−3.09	−5.69**	4.55*	8.67***
ΔlnYF_i,t−2	−0.69	−0.16	−4.12*
ΔlnYF_i,t−3	0.29	−0.78
ΔlnYF_i,t−4	−5.56**	−3.49**
ΔlnYF_i,t−5	0.08	2.30
ΔlnYF_i,t−6	−2.97	−4.35**
ΔlnYF_i,t−7	−6.62**
Constant	0.46	−3.70	−4.47	3.57
Bound test	4.20**	5.44***	10.57***	6.69***
Adj—R²	.60	.74	.79	.81
Breusch-Godfrey	0.01	0.01	0.05	1.02
Breusch-Pagan	1.48	0.90	1.11	0.85
Ramsey RESET	0.20	0.27	0.71	2.67
CUSUM	S	S	S	S
CUSUMSQ	U	S	U	U

Note. POS_i,t represents REER⁺_i,t in the REER model and USD⁺_i,t in the VCER model, respectively. NEG_i,t represents REER^–_i,t in the REER model and USD^–_i,t in the VCER model, respectively. The outcomes of Breusch-Godfrey, Breusch-Pagan, and Ramsey RESET tests are reported in terms of F-statistics. The results of CUSUM and CUSUMSQ tests are “S” (stable) or “U” (unstable). The significance levels 10%, 5%, and 1% are denoted by *, **, and ***.

Following the works of Bahmani-Oskooee & Fariditavana, 2016; Rose & Yellen, 1989), the J-curve effect is identified when the long-run coefficients of exchange rates are positive and their short-run counterparts are negative or insignificant. Thus, in the REER model, the J-curve effect is witnessed in the trade of Laos, Myanmar, the Philippines, and Vietnam with the EU-28. When the role of USD as a vehicle currency is captured in the VCER model, the Philippines and Thailand are associated with the J-curve phenomenon. Noticeably, the Philippines experience the J-curve effect in both the REER and VCER models. Additionally, 1% depreciation of Philippine peso (PHP) against the USD fosters the Philippines’ trade balance with the EU-28 by around 7.78% in the long run. However, the same amount of PHP depreciation against the currencies of the EU-28 only boosts the Philippines’ trade balance by around 3.83%. This implies that the Philippines’ trade balance is more sensitive to the fluctuation of the vehicle currency USD’s exchange rate (USD/PHP) than the real effective exchange rate.

The Marshall-Lerner condition can be supported if the depreciation of a country’s currency stimulates its trade balance in the long run (Rose, 1991). Thus, in this paper, if the long-run coefficients of REER⁺_i,t in the REER modelor USD⁺_i,t in the VCER model are positive, the Marshall-Lerner is supported for the ith country. The first row of Table 2 demonstrates that in six ASEAN countries including Cambodia, Laos, Malaysia, Myanmar, the Philippines, and Vietnam the depreciation of their currencies against the EU-28’s currencies improves their trade balances. Moreover, when the role of USD as a vehicle currency is considered, five ASEAN countries (i.e., Indonesia, Malaysia, Myanmar, the Philippines, and Thailand) satisfy the Marshall-Lerner condition. Especially, Cambodia, Myanmar, the Philippines, and Vietnam have the Marshall-Lerner condition supported in both REER and VCER models. Moreover, while devaluation policy seems ineffective for promoting the trade balances of Indonesia, Malaysia, and Thailand in case the real effective exchange rates are employed, it becomes effective when the vehicle currency USD is utilized. Thus, the choice of invoicing currencies could influence the effectiveness of the devaluation policy.

Table 2.

Summary of Long-Run Impacts of Exchange Rates on the Trade Balance of ASEAN Countries With the EU-28.

	REER model	VCER model
The depreciation of domestic currency boosts the trade balance	Cambodia, Laos, Malaysia, Myanmar, Philippines, Vietnam	Indonesia, Malaysia, Myanmar, Philippines, Thailand
The depreciation of domestic currency lowers the trade balance		Singapore
The appreciation of domestic currency boosts the trade balance		Cambodia, Brunei, Laos, Malaysia, Myanmar, Vietnam
The appreciation of domestic currency lowers the trade balance	Cambodia, Laos	Indonesia, Singapore, Thailand

To test the main hypothesis Hm (which is mentioned in the Introduction section), we compare the performance of the REER and VCER models by counting the number of significant exchange rates’ coefficients in each model. From Table 3, it can be observed that the VCER model outperforms the REER model since all ASEAN countries have significant short-run and long-run exchange rates’ coefficients. Meanwhile, in the REER model, the number is only six. Thus, the main hypothesis of this paper is supported. Specifically, the neglection of the USD’s role as a vehicle currency is a factor contributing to the lack of significant results. This finding is analogous to Bao and Le (2021a) for ASEAN-EU overall trade. Thus, the USD’s role as a vehicle currency is too important to be overlooked in ASEAN-EU trade at various levels of analysis. Thus, the vital role of USD as a vehicle currency should not be neglected when inspecting the exchange rate-trade balance relationship between any two partners different from the US.

Table 3.

The Number of Significant Exchange Rates’ Coefficients in the Elementary REER and VCER Models.

	REER model	VCER model
Long-run	6	10
Short-run	10	10
Both long-run and short-run	6	10

So as to detect the asymmetric influences of exchange rates on the trade balances, there are two main methods: observation or Wald test (Bahmani-Oskooee & Baek, 2019; Iyke & Ho, 2018). When the long-run coefficients denoting the depreciation and appreciation of a country’s currency are clearly distinguishable, the long-run asymmetry can be detected by observation. For example, in the VCER model of Indonesia, the long-run coefficient denoting the depreciation of rupiah against the USD (i.e., .64) is obviously different from the appreciation counterpart (i.e., −2.19), which indicates the presence of long-run asymmetry. Nevertheless, in the case of Malaysia, as the two long-run coefficients in the VCER model (i.e., 1.18 and 1.74) are positive and not so different in size, it is difficult to judge whether the long-run asymmetry exists. Hence, the Wald test is needed, and Table 4 shows the Wald test results for the cases that are unclear and thus cannot be concluded by the observation method. In the case of Malaysia, for instance, the Wald test’s F-statistic is low, and there is no long-run asymmetric effect accordingly. Additionally, the observation method can also be used to detect the short-run asymmetry when the lag lengths of ΔPOS and ΔNEG are different, or their coefficients are distinct at the same lag. In all the cases except Malaysia’s VCER model, short-run asymmetries are identified by observation. Regarding Malaysia’s VCER model, there is no evidence of short-run asymmetry because the Wald test’s F-statistic (i.e., 1.92) is not significant.

Table 4.

Long-Run and Short-Run Wald Test for the Asymmetric Effects of Exchange Rates in Some Cases That the Observation Method Cannot Be Used.

	REER model		VCER model
	Country	F-statistics	Country	F-statistics
Long-run			Malaysia	1.92
			Myanmar	25.96***
			Singapore	8.74***
Short-run			Malaysia	1.92

Note. The asterisk *** indicates p-value below 1%.

Besides, Table 5 lists all the long-run and short-run asymmetries, which reinforces the suitability of the NARDL method as the majority of ASEAN countries have short-run and long-run asymmetric effects of exchange rates on their trade balances with the EU-28. Specifically, all the 10 ASEAN members witness short-run asymmetries in the REER model while nine of them experience short-run and long-run ones in the VCER model.

Table 5.

Cases With Long-Run and Short-Run Asymmetries.

	REER model	VCER model
Long-run asymmetries	Cambodia, Laos, Malaysia, Myanmar, Philippines, Vietnam	Brunei, Cambodia, Indonesia, Laos, Myanmar, Philippines, Singapore, Thailand, Vietnam
Short-run asymmetries	Brunei, Cambodia, Indonesia, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand, Vietnam	Brunei, Cambodia, Indonesia, Laos, Myanmar, Philippines, Singapore, Thailand, Vietnam

Robustness Check

In this section, we examine the robustness of the estimation results of the elementary models presented in Table 1 and the outcomes displayed in Table 3 by incorporating the volatility of exchange rates. It is well documented that the exchange rate volatility can impact a country’s export and import (Bailey et al., 1987; Hayakawa & Kimura, 2009; McKenzie, 1999). Thus, it can be used as a control variable for the trade balance (Singh, 2004). Moreover, the exchange rate volatility has been deemed a relevant factor affecting a country’s choices of invoicing currencies in its global trade (Bowe & Saltvedt, 2004; Liu & Lu, 2019; Witte, 2010). Therefore, we focus on the role of exchange rate volatility to conduct the robustness check of the elementary models to further scrutinize the impacts of real effective exchange rates and the vehicle currency USD’s exchange rates on the ASEAN countries’ trade balances with the EU-28.

Exchange rate volatility can be measured in different ways (McKenzie, 1999). Following Medhora (1990) and Bahmani-Oskooee et al. (2014), we use the within-period standard deviation to compute exchange rate volatility. Namely, in this paper, the standard deviation of the monthly real exchange rates in each quarter is employed to construct the quarterly exchange rate volatility, which is specified in equations (10) and (11):

\begin{matrix} V O L_R E E R_{i, t} = \sqrt{\frac{\sum_{m = 1}^{3} {(R E E R_{i, t, m} - {\bar{R E E R}}_{i, t, m})}^{2}}{3 - 1}} \end{matrix}

(10)

\begin{matrix} V O L_U S D_{i, t} = \sqrt{\frac{\sum_{m = 1}^{3} {(U S D_{i, t, m} - {\bar{U S D}}_{i, t, m})}^{2}}{3 - 1}} \end{matrix}

(11)

In equation (10), $V O L_R E E R_{i, t}$ denotes the volatility of the real effective exchange rate between the currency of country i in ASEAN and the EU-28’s currencies in the quarter t. Besides, the subscript m denotes each of the 3 months in the quarter t. Similarly, in equation (11), $V O L_U S D_{i, t}$ indicates the volatility of the real exchange rate between the currency of the country i in ASEAN and the vehicle currency USD.

We incorporate the natural logarithm of $V O L_R E E R_{i, t}$ into the elementary REER model (which is specified by equation (8)) to form the extended REER model. Meanwhile, we add the natural logarithm of $V O L_U S D_{i, t}$ into the elementary VCER model (which is specified by equation (9)) to form the extended VCER model. The NARDL estimation results of the extended REER and VCER models are presented in Table 6. Similar to the estimation procedure of the elementary models, the maximum lags are set to eight, and the optimal lags are chosen by the AIC. For the cases suffering from problems such as autocorrelation, heteroskedasticity, and misspecification, we lower the maximum lags and the re-estimate them.

Table 6.

NARDL Estimation Results of the Extended REER and VCER Models.

	i = Brunei		i = Cambodia
	Extended REER	Extended VCER	Extended REER	Extended VCER
Long run
POS_i,t	−1.34	−4.05**	1.58**	4.57
NEG_i,t	−0.44	−0.99	−0.49**	7.80***
lnY_i,t	−1.93*	−0.39	−4.15**	−12.42***
lnYF_i,t	10.87*	9.50***	8.62*	11.29***
lnVOL_i,t	0.41	0.76***	0.17*	0.14*
Short run
ΔPOS_i,t	−7.97	−5.50**	2.45	1.81
ΔPOS_i,t−1	−4.00		−1.77	5.73
ΔPOS_i,t−2	−7.45			9.99**
ΔPOS_i,t−3	−4.23
ΔPOS_i,t−4	−5.25
ΔPOS_i,t−5	−4.38
ΔPOS_i,t−6
ΔPOS_i,t−7
ΔNEG_i,t	4.19	10.28**	−0.28	−2.54
ΔNEG_i,t−1	1.07		5.07***
ΔNEG_i,t−2	4.80		0.98
ΔNEG_i,t−3	−0.77		−0.44
ΔNEG_i,t−4	4.16		2.72*
ΔNEG_i,t−5	3.44		3.65***
ΔNEG_i,t−6
ΔNEG_i,t−7
ΔlnY_i,t	−0.19	4.20**	−4.25**	−8.33***
ΔlnY_i,t−1	9.49		4.60	5.37**
ΔlnY_i,t−2	3.55		−1.62
ΔlnY_i,t−3	1.70		4.82
ΔlnY_i,t−4	16.15**		−5.57*
ΔlnY_i,t−5	−1.37
ΔlnY_i,t−6
ΔlnY_i,t−7
ΔlnYF_i,t	6.54	11.26*	9.41**	11.93***
ΔlnYF_i,t−1	−16.68	4.20	−0.83
ΔlnYF_i,t−2	−15.20	−1.76	−3.61
ΔlnYF_i,t−3	−9.02	8.76*	−5.85*
ΔlnYF_i,t−4	−11.17		−7.86**
ΔlnYF_i,t−5	0.93		−9.93***
ΔlnYF_i,t−6
ΔlnYF_i,t−7
ΔlnVOL_i,t	0.01	0.10	−0.005	−0.003
ΔlnVOL_i,t−1	−0.87	−0.72***	−0.22*	−0.08
ΔlnVOL_i,t−2	−0.91	−0.33**	−0.14
ΔlnVOL_i,t−3	−0.43	−0.24**	−0.20**
ΔlnVOL_i,t−4	−0.002		−0.26***
ΔlnVOL_i,t−5	0.04
ΔlnVOL_i,t−6
ΔlnVOL_i,t−7
Constant	−59.09	−56.63***	−19.71	9.34
Bound test	4.29**	5.27***	8.70***	11.39***
Adj—R²	.51	.67	.77	.72
Breusch-Godfrey	0.47	0.14	0.27	0.55
Breusch-Pagan	1.48	1.26	0.51	1.48
Ramsey RESET	1.23	0.94	0.54	1.73
CUSUM	S	S	S	S
CUSUMSQ	U	U	S	U
	i = Indonesia		i = Laos
	Extended REER	Extended VCER	Extended REER	Extended VCER
Long run
POS_i,t	0.56	0.93**	−1.45	−4.07
NEG_i,t	−0.41	−0.97***	−1.37*	3.74***
lnY_i,t	−2.65***	−1.78**	−2.12	3.06***
lnYF_i,t	−0.40	−1.64*	2.01	−2.53
lnVOL_i,t	0.04*	−0.07**	−0.20*	−0.07
Short run
ΔPOS_i,t	0.40	0.75**	−1.17	6.27
ΔPOS_i,t−1
ΔPOS_i,t−2
ΔPOS_i,t−3
ΔPOS_i,t−4
ΔPOS_i,t−5
ΔPOS_i,t−6
ΔPOS_i,t−7
ΔNEG_i,t	−0.27	−0.78***	−1.11*	−9.33
ΔNEG_i,t−1	−0.07			−10.79**
ΔNEG_i,t−2	−0.37			−3.26
ΔNEG_i,t−3	−0.36			−14.39***
ΔNEG_i,t−4	−0.14
ΔNEG_i,t−5	−0.51*
ΔNEG_i,t−6	−1.05***
ΔNEG_i,t−7	−0.64**
ΔlnY_i,t	−1.89***	−1.43**	−1.72	16.80***
ΔlnY_i,t−1				1.25
ΔlnY_i,t−2				8.40*
ΔlnY_i,t−3
ΔlnY_i,t−4
ΔlnY_i,t−5
ΔlnY_i,t−6
ΔlnY_i,t−7
ΔlnYF_i,t	−0.28	0.28	6.71**	2.13
ΔlnYF_i,t−1		1.47
ΔlnYF_i,t−2		2.23**
ΔlnYF_i,t−3
ΔlnYF_i,t−4
ΔlnYF_i,t−5
ΔlnYF_i,t−6
ΔlnYF_i,t−7
ΔlnVOL_i,t	0.03*	−0.02**	−0.16*	−0.10
ΔlnVOL_i,t−1		0.02*
ΔlnVOL_i,t−2		0.02**
ΔlnVOL_i,t−3
ΔlnVOL_i,t−4
ΔlnVOL_i,t−5
ΔlnVOL_i,t−6
ΔlnVOL_i,t−7
Constant	12.88***	16.35***	5.38	3.85
Bound test	6.48***	9.40***	4.79***	6.22***
Adj—R²	.45	.51	.64	.67
Breusch-Godfrey	0.00	1.07	0.02	0.10
Breusch-Pagan	0.59	0.67	0.41	0.41
Ramsey RESET	1.61	0.25	0.26	2.48
CUSUM	S	S	S	S
CUSUMSQ	S	S	S	S
	i = Malaysia		i = Myanmar
	Extended REER	Extended VCER	Extended REER	Extended VCER
Long run
POS_i,t	0.49*	0.51**	0.09	1.69***
NEG_i,t	0.33*	0.80***	−0.09	3.61***
lnY_i,t	−1.03***	0.35	−1.24	−3.82**
lnYF_i,t	2.74***	−0.59	−3.46	17.24***
lnVOL_i,t	0.03	−0.01	0.27	−1.34**
Short run
ΔPOS_i,t	0.28*	0.33***	0.04	0.56**
ΔPOS_i,t−1				−0.99**
ΔPOS_i,t−2				−0.63
ΔPOS_i,t−3				−1.33***
ΔPOS_i,t−4				−0.46*
ΔPOS_i,t−5				−0.17
ΔPOS_i,t−6				−0.36*
ΔPOS_i,t−7
ΔNEG_i,t	0.06	0.53***	−0.05	−8.06**
ΔNEG_i,t−1	−0.62
ΔNEG_i,t−2	−0.23
ΔNEG_i,t−3	−0.76*
ΔNEG_i,t−4	−0.63
ΔNEG_i,t−5
ΔNEG_i,t−6
ΔNEG_i,t−7
ΔlnY_i,t	−0.72**	0.23	1.07	6.94
ΔlnY_i,t−1	0.81***
ΔlnY_i,t−2
ΔlnY_i,t−3
ΔlnY_i,t−4
ΔlnY_i,t−5
ΔlnY_i,t−6
ΔlnY_i,t−7
ΔlnYF_i,t	1.56**	−0.39	−1.78	−0.66
ΔlnYF_i,t−1				−14.17**
ΔlnYF_i,t−2				−15.69**
ΔlnYF_i,t−3				−5.19
ΔlnYF_i,t−4				−17.39***
ΔlnYF_i,t−5				−11.96**
ΔlnYF_i,t−6				−12.91**
ΔlnYF_i,t−7
ΔlnVOL_i,t	0.02	−0.007	−0.15	0.07
ΔlnVOL_i,t−1			−0.15	1.24***
ΔlnVOL_i,t−2				1.15***
ΔlnVOL_i,t−3				0.89***
ΔlnVOL_i,t−4				0.50**
ΔlnVOL_i,t−5				0.17
ΔlnVOL_i,t−6				0.25
ΔlnVOL_i,t−7
Constant	−2.15	1.86	13.80	−50.00**
Bound test	3.36**	5.25***	1.43	4.98***
Adj—R²	.47	.36	.40	.72
Breusch-Godfrey	0.69	0.63	0.06	0.18
Breusch-Pagan	0.86	1.56	0.35	0.72
Ramsey RESET	2.32	0.69	0.12	1.39
CUSUM	S	S	S	S
CUSUMSQ	S	U	U	S
	i = Philippines		i = Singapore
	Extended REER	Extended VCER	Extended REER	Extended VCER
Long run
POS_i,t	5.11***	8.83***	−0.29	−2.99***
NEG_i,t	0.27	−0.77	−0.58***	−3.94***
lnY_i,t	−18.12***	−8.20***	2.62***	3.54***
lnYF_i,t	6.73***	3.83	−3.29***	0.78
lnVOL_i,t	0.10	−0.21	−0.21***	0.03*
Short run
ΔPOS_i,t	−0.39	1.98	0.40	−4.92
ΔPOS_i,t−1	−5.18***	−1.11	−0.38	9.80***
ΔPOS_i,t−2	−5.53***	−5.33***	−0.45	10.86**
ΔPOS_i,t−3	−2.88***	−4.13**	−0.32	3.54
ΔPOS_i,t−4	−1.68**	2.27	0.69	8.66***
ΔPOS_i,t−5	−1.68**	3.35**	1.15**	9.12***
ΔPOS_i,t−6
ΔPOS_i,t−7
ΔNEG_i,t	0.65	0.04	−0.68***	−5.16**
ΔNEG_i,t−1	1.36*	−0.45		7.48**
ΔNEG_i,t−2	2.94***	−0.31		11.22***
ΔNEG_i,t−3		1.83		8.23**
ΔNEG_i,t−4		−2.13		7.44**
ΔNEG_i,t−5				6.41**
ΔNEG_i,t−6
ΔNEG_i,t−7
ΔlnY_i,t	−2.29***	−2.35*	0.33	6.97
ΔlnY_i,t−1	5.61***	3.30*	−0.87	−6.90**
ΔlnY_i,t−2	3.02***	1.85		−9.97***
ΔlnY_i,t−3		−1.22		−4.55
ΔlnY_i,t−4				−7.36**
ΔlnY_i,t−5				−6.68**
ΔlnY_i,t−6
ΔlnY_i,t−7
ΔlnYF_i,t	−0.79	2.45	−1.70	3.82*
ΔlnYF_i,t−1	−6.01***	−3.31*	4.04***	2.20
ΔlnYF_i,t−2	−3.95**	−3.62**	5.30***	1.40
ΔlnYF_i,t−3			2.30	4.69**
ΔlnYF_i,t−4			2.70
ΔlnYF_i,t−5			5.87***
ΔlnYF_i,t−6
ΔlnYF_i,t−7
ΔlnVOL_i,t	0.09***	−0.002	0.04	0.004
ΔlnVOL_i,t−1		0.12	0.24***	−0.06**
ΔlnVOL_i,t−2		0.07	0.20***	−0.03*
ΔlnVOL_i,t−3		0.14**	0.15***
ΔlnVOL_i,t−4		0.12***	0.09***
ΔlnVOL_i,t−5			0.06***
ΔlnVOL_i,t−6
ΔlnVOL_i,t−7
Constant	32.20***	17.35	10.53**	−29.11**
Bound test	6.62***	3.51*	5.48***	5.32***
Adj—R²	.61	.57	.72	.71
Breusch-Godfrey	0.37	1.68	0.82	0.15
Breusch-Pagan	1.18	0.78	0.66	0.83
Ramsey RESET	2.07	0.65	0.00	0.32
CUSUM	S	S	S	S
CUSUMSQ	S	S	S	S
	i = Thailand		i = Vietnam
	Extended REER	Extended VCER	Extended REER	Extended VCER
Long run
POS_i,t	0.64*	4.93***	1.82***	1.61
NEG_i,t	0.44**	−1.32***	−1.33***	3.19**
lnY_i,t	−1.75***	−3.38***	−0.82**	−4.82***
lnYF_i,t	2.21	2.97***	−1.84	5.91*
lnVOL_i,t	0.04	−0.14*	−0.35***	0.13
Short run
ΔPOS_i,t	−0.85	0.45	−0.08	−1.11
ΔPOS_i,t−1	−1.52**		−4.34***	−7.91***
ΔPOS_i,t−2	−0.89		−3.39***
ΔPOS_i,t−3	−0.95		−3.16***
ΔPOS_i,t−4			−1.86**
ΔPOS_i,t−5			−1.52*
ΔPOS_i,t−6			−1.85**
ΔPOS_i,t−7
ΔNEG_i,t	1.94**	−0.63***	−0.64	3.49**
ΔNEG_i,t−1	0.85		0.41	2.20
ΔNEG_i,t−2	−1.21*		−0.32
ΔNEG_i,t−3	−0.78		−0.98
ΔNEG_i,t−4	−0.80		−1.00*
ΔNEG_i,t−5	−1.79***		−0.72
ΔNEG_i,t−6	−1.71**		−0.89
ΔNEG_i,t−7	−0.82
ΔlnY_i,t	0.24	−0.08	0.18	−1.06***
ΔlnY_i,t−1	−1.56		1.18
ΔlnY_i,t−2	−3.04**		2.75**
ΔlnY_i,t−3	−1.88		2.94**
ΔlnY_i,t−4	−2.33*		2.14**
ΔlnY_i,t−5	−0.88		2.08**
ΔlnY_i,t−6	1.01
ΔlnY_i,t−7
ΔlnYF_i,t	−2.70	1.28	−6.58*	−1.00
ΔlnYF_i,t−1	−2.57	3.75***	10.92**	9.51***
ΔlnYF_i,t−2	−0.55	1.57	−0.13
ΔlnYF_i,t−3	−0.19		4.93
ΔlnYF_i,t−4	−6.08***		7.52**
ΔlnYF_i,t−5	−1.45		5.75*
ΔlnYF_i,t−6	−3.67		20.01***
ΔlnYF_i,t−7	−6.79**
ΔlnVOL_i,t	0.04	−0.01	−0.03	0.002
ΔlnVOL_i,t−1		0.03**	0.56***
ΔlnVOL_i,t−2			0.48***
ΔlnVOL_i,t−3			0.38***
ΔlnVOL_i,t−4			0.24***
ΔlnVOL_i,t−5			0.17***
ΔlnVOL_i,t−6
ΔlnVOL_i,t−7
Constant	2.17	2.87	35.38***	−0.07
Bound test	3.47*	6.87***	13.34***	6.97***
Adj—R²	.63	.52	.94	.82
Breusch-Godfrey	0.40	0.26	0.64	0.16
Breusch-Pagan	1.45	0.70	0.87	0.86
Ramsey RESET	0.00	0.06	0.06	1.47
CUSUM	S	S	S	S
CUSUMSQ	U	S	S	U

Note. POS_i,t represents REER⁺_i,t in the extended REER model and USD⁺_i,t in the extended VCER model. NEG_i,t represents REER^–_i,t in the extended REER model and USD^–_i,t in the extended VCER model. lnVOL_i,t denotes the natural logarithm of VOL_REER_i,t in the extended REER model and the natural logarithm of VOL_USD_i,t in the extended VCER model. The outcomes of Breusch-Godfrey, Breusch-Pagan, and Ramsey RESET tests are reported in terms of F-statistics. The results of CUSUM and CUSUMSQ tests are “S” (stable) or “U” (unstable). The significance levels 10%, 5%, and 1% are denoted by *, **, and ***.

From Table 6, we observe that the impacts of exchange rate volatility on ASEAN countries’ trade balances can be either positive, negative, or insignificant, which is analogous to the findings of the existing literature about its ambiguous effects on the trade flows of different countries (McKenzie, 1999; Sharma & Pal, 2018; Sugiharti et al., 2020). In addition, most of the ASEAN members witness the long-run significant impacts of exchange rate volatility on their trade balances, either in the extended REER model or the extended VCER model, except for Malaysia and the Philippines.

We notice high level of robustness when comparing the long-run exchange rates’ coefficients in Table 1 (which shows the estimation results of the elementary models) and Table 6 (which shows the estimation results of the extended models). Specifically, for the convenience of the readers, we report the summary of the long-run exchange rates’ coefficients in Table 7. Comparing the elementary and extended REER models, the coefficients of POS_i,t and NEG_i,t are similar in terms of sign when they are statistically significant in most of the cases except Laos. Meanwhile, regarding the elementary and extended VCER models, virtually all ASEAN countries have strongly robust coefficients in terms of sign and significance except Brunei. For example, Singapore’s trade balance is negatively affected by both the depreciation and appreciation of its currency against the vehicle currency USD in the elementary as well as extended VCER models. Additionally, the Philippines’ trade balance is fostered by the depreciation of its currency against the vehicle currency USD, as evidenced by the positive coefficients of POS_i,t in both the elementary and extended VCER models (i.e., 7.78 and 8.83, respectively).

Table 7.

Summary of Long-Run Exchange Rates’ Coefficients of the Elementary and Extended REER and VCER Models.

Variable	REER	VCER	Extended REER	Extended VCER
i = Brunei
POS_i,t	3.07	−0.72	−1.34	−4.05**
NEG_i,t	−3.29	4.07**	−0.44	−0.99
i = Cambodia
POS_i,t	2.43***	3.30	1.58**	4.57
NEG_i,t	−0.66***	7.36***	−0.49**	7.80***
i = Indonesia
POS_i,t	0.61	0.64*	0.56	0.93**
NEG_i,t	−0.53	−2.19***	−0.41	−0.97***
i = Laos
POS_i,t	3.79*	−1.82	−1.45	−4.07
NEG_i,t	−2.74*	3.34***	−1.37*	3.74***
i = Malaysia
POS_i,t	0.43*	1.18*	0.49*	0.51**
NEG_i,t	0.29	1.74*	0.33*	0.80***
i = Myanmar
POS_i,t	0.25*	0.13**	0.09	1.69***
NEG_i,t	0.98	1.97***	−0.09	3.61***
i = Philippines
POS_i,t	3.83**	7.78**	5.11***	8.83***
NEG_i,t	1.14	0.33	0.27	−0.77
i = Singapore
POS_i,t	−0.31	−5.12***	−0.29	−2.99***
NEG_i,t	−0.60	−6.58***	−0.58***	−3.94***
i = Thailand
POS_i,t	0.33	3.20***	0.64*	4.93***
NEG_i,t	0.25	−1.19***	0.44**	−1.32***
i = Vietnam
POS_i,t	1.96***	3.25	1.82***	1.61
NEG_i,t	−0.69	3.08**	−1.33***	3.19**

Note. The significance levels 10%, 5%, and 1% are denoted by *, **, and ***.

To check the robustness of our main hypothesis Hm (i.e., the neglection of the USD’s role as a vehicle currency can lead to fewer significant results), we look into Table 6 and count the numbers of significant exchange rates’ coefficients in the extended REER and VCER models. The outcome is displayed in Table 8. When controlling for the role of exchange rate volatility, the extended VCER model still outperforms its counterpart in both the short run and long run. Thus, the main hypothesis Hm is supported by the elementary as well as extended models (see Tables 3 and 8). This finding reflects the fact that the ASEAN countries heavily rely on the vehicle currency USD in their international trade with many partners including the EU-28. Therefore, the ignorance of the USD’s role as a vehicle currency when analyzing the exchange rate-trade balance relationship is a factor contributing to the lack of significant results (Bao & Le, 2021a, 2021b).

Table 8.

The Number of Significant Exchange Rates’ Coefficients in the Extended REER and VCER Models.

	Extended REER model	Extended VCER model
Long-run	7	10
Short-run	8	10
Both long-run and short-run	7	10

Conclusion

This paper is the first to investigate the impacts of the vehicle currency USD’s exchange rates, together with the real effective exchange rates, on the trade balances of all ASEAN countries with the whole EU-28. The main hypothesis of this paper is supported with high level of robustness. Namely, the empirical results demonstrate the usefulness and prominence of the USD as a vehicle currency where all ASEAN members witness the significant effects of the USD’s exchange rates on their trade balances in both the short run and long run. Thus, studies on the exchange rate-trade balance nexus between any two partners different from the US should not ignore the importance of USD as a vehicle currency because it dominates the global trade.

The findings of this paper show that the trade balances of several ASEAN members react distinguishably to the real effective exchange rates and the vehicle currency USD’s exchange rates. Hence, the effectiveness of the devaluation policy to boost trade balances can depend on the usage of different invoicing currencies. For example, Indonesia’s and Thailand’s trade balances are unresponsive to the real effective exchange rates in the long run, but they are positively affected by the vehicle currency USD’s exchange rates. Thus, the weak-currency policy can help Indonesia and Thailand to foster their trade balances with the EU-28 if they utilize more proportion of the vehicle currency USD. Moreover, as ASEAN countries highly utilize USD as a vehicle currency, it can influence not only their trade balances with the US but also with other partners such as the EU-28. With strong reliance on the USD, ASEAN countries’ policy-makers should carefully monitor the fluctuation of their currencies against the USD as it can simultaneously impact their trade with various partners that may altogether occupy higher percentage than the share of the US in their total trade values. For example, the vehicle currency USD can affect ASEAN countries’ trade with the partners in the Regional Comprehensive Economic Partnership (RCEP) (i.e., Australia, China, Japan, New Zealand, and South Korea) which accounted for the higher share than that of the US in ASEAN’s total trade values. Therefore, the policy-makers should pay attention to the usage of the invoicing currencies, especially the USD, in the trade with the aforesaid RCEP partners. In addition, they should collect more data about the usage of the invoicing currencies used in the trade of each ASEAN country with various partners and investigate the factors affecting the choice of each currency.

This paper controls for the role of exchange rate volatility in the extended REER and VCER models, which indicates the robustness of the results. However, the limitation of this paper is that it has not included other variables that can affect the trade balances. Future studies can test the main hypothesis proposed by this paper for other countries by expanding the models with more control variables.

Footnotes

Appendix

Table A3.

Summary of Some Notable Empirical Papers.

Papers	Focused countries	Partners	Methods and time	Models	Findings
Panel A: Studies focusing on ASEAN countries
Arize (1994)	Indonesia, Malaysia, Philippines, Singapore, Thailand	The rest of the world	Method: Engle-Granger, Johansen cointegration	TB = f(E)	The depreciation of currencies fosters the trade balances of all countries except Malaysia
Arize (1994)	Indonesia, Malaysia, Philippines, Singapore, Thailand	The rest of the world	Time: 1973Q1 to 1991Q1	TB = f(E)
Baharumshah (2001)	Malaysia, Thailand	Japan, the US	Method: Johansen-Juselius cointegration	TB = f(E, Y, YF)	The depreciation of domestic currencies fosters Malaysia’s and Thailand’s trade balances with Japan and the US
Baharumshah (2001)	Malaysia, Thailand	Japan, the US	Time: 1980Q1 to 1996Q4	TB = f(E, Y, YF)	No evidence of the J-curve effect is found
Wilson and Tat (2001)	Singapore	The US	Method: Engle-Granger, Johansen-Juselius cointegration	TB = f(E, Y, YF)	The exchange rate does not affect the trade balance of Singapore with the US
Wilson and Tat (2001)	Singapore	The US	Time: 1970 to 1996	TB = f(E, Y, YF)	The findings do not support the presence of J-curve effect
Brahmasrene and Jiranyakul (2002)	Thailand	Germany, Japan, Singapore, the Netherlands, the UK, the US	Method: Engle-Granger cointegration	TB = f(E, Y, YF)	The depreciation of Thai baht fosters Thailand’s trade balances with all the partners except Singapore
Brahmasrene and Jiranyakul (2002)	Thailand	Germany, Japan, Singapore, the Netherlands, the UK, the US	Time: 1990Q1 to 2000Q4	TB = f(E, Y, YF)
Yusoff (2010)	Malaysia	The rest of the world	Method: VECM	TB = f(E, Y, YF)	The depreciation of ringgit fosters Malaysia’s trade balance
Yusoff (2010)	Malaysia	The rest of the world	Time: 1977Q1 to 1998Q2	TB = f(E, Y, YF)	The evidence of J-curve effect depends on the exchange rate regimes
Kyophilavong et al. (2013)	Laos	The rest of the world	Method: ARDL	TB = f(E, Y, YF)	The depreciation of domestic currency does not improve the trade balance of Laos
Kyophilavong et al. (2013)	Laos	The rest of the world	Time: 1993Q1 to 2010Q4	TB = f(E, Y, YF)	J-curve effect is identified
Bineau (2016)	Cambodia	The EU-28, Hong Kong, Indonesia, Japan, Korea, Malaysia, Singapore, Thailand, the UK, the US	Method: FMOLS	TB = f(E, YF/Y)	The depreciation of domestic currency fosters Cambodia’s trade balance
			Time: 1998Q1 to 2014Q3		The overall panel data estimation results show no J-curve effect
			Time: 1998Q1 to 2014Q3		The J-curve effect is identified in Cambodia’s trade with Indonesia and Singapore
Bahmani-Oskooee and Aftab (2017b)	Malaysia	EU-28	Method: ARDL, NARDL	TB = f(E, Y, YF)	The NARDL method is superior to the ARDL method in detecting more significant results and more evidence of the J-curve effect
Bahmani-Oskooee and Aftab (2017b)	Malaysia	EU-28	Time: 2000M5 to 2013M12	TB = f(E, Y, YF)	The J-curve effect is identified in 12 out of the 63 industries
Bahmani-Oskooee and Harvey (2019)	Indonesia	12 Trading partners	Method: ARDL, NARDL	TB = f(E, Y, YF)	The NARDL method is superior to the ARDL method in detecting more significant results
Bahmani-Oskooee and Harvey (2019)	Indonesia	12 Trading partners	Time: 1990Q1 to 2015Q4	TB = f(E, Y, YF)	The J-curve effect is identified in Indonesia’s trade with the Philippines, Singapore, Thailand, the UK, and the US
Phong et al. (2018)	Vietnam	The rest of the world	Method: ARDL	TB = f(E, Y, YF)	The depreciation of Vietnamese dong fosters Vietnam’s trade balance
Phong et al. (2018)	Vietnam	The rest of the world	Time: 2000Q1 to 2015Q4	TB = f(E, Y, YF)	The J-curve effect is identified
Bao and Le (2021a)	ASEAN	The EU-28	Method: NARDL	TB = f(E, Y, YF)	The neglection of the USD’s role as a vehicle currency limits the significance of the results
			Time: 2000Q1 to 2018Q1	TB = f(V, Y, YF)	The real effective exchange rate demonstrates no impact on ASEAN’s trade balance with the EU-28
					The vehicle currency USD’s exchange rate positively affects ASEAN’s trade balance with the EU-28
					The J-curve effect is detected in the vehicle currency model
Bao and Le (2021b)	Vietnam	27 Countries in the EU and the UK	Method: NARDL	TB = f(E, Y, YF)	The inclusion of the USD’s role as a vehicle currency helps detect more significant results and additional cases of J-curve effect
Bao and Le (2021b)	Vietnam	27 Countries in the EU and the UK	Time: 2000Q1 to 2018Q1	TB = f(V, Y, YF)
Panel B: Studies focusing on other countries
Bahmani-Oskooee (1985)	Greece, India, South Korea, Thailand	The rest of the world	Method: Almon distributed lag	TB = f(E, Y)	This is the pioneering study to propose a method to detect the J-curve effect
Bahmani-Oskooee (1985)	Greece, India, South Korea, Thailand	The rest of the world	Time: 1973Q1 to 1980Q4	TB = f(E, Y)	The results indicate that J-curve effect is found in Greece, India, and South Korea
Rose and Yellen (1989)	The US	Canada, France, Germany, Italy, Japan, the UK	Method: Engle-Granger cointegration	TB = f(E, Y, YF)	This study provides the theoretical and empirical bases for the use of the standard two-country model in analyzing the impacts of exchange rates on trade balances. Namely, a country’s trade balance is modeled by the three independent variables: the exchange rate, the domestic income, and the partner’s income
			Time: 1960Q1 to 1985Q1 (some partners have different time frames)		This study is also the first to propose a new method to detect the J-curve effect: the negative exchange rates’ coefficients in the short run combined with the positive ones in the long run
					The findings indicate no J-curve effect
Singh (2004)	India	The rest of the world	Method: Johansen-Juselius cointegration, GARCH	TB = f(E, Y, YF, VOL)	Exchange rate volatility does not affect India’s trade balance
Singh (2004)	India	The rest of the world	Time: 1975Q2 to 1996Q3	TB = f(E, Y, YF, VOL)	The J-curve effect is not detected
Bahmani-Oskooee et al. (2005)	Australia	23 Partners	Method: ARDL	TB = f(E, Y, YF)	The positive effects of Australian dollar’s depreciation, alongside the J-curve effect, are found in the trade of Australia with Denmark, New Zealand, and South Korea
Bahmani-Oskooee et al. (2005)	Australia	23 Partners	Time: 1973–2001	TB = f(E, Y, YF)
Gomes and Paz (2005)	Brazil	The rest of the world	Method: VECM	TB = f(E, Y, YF)	The depreciation of Brazil’s currency fosters its trade balance, thus supporting the Marshall-Lerner condition
Gomes and Paz (2005)	Brazil	The rest of the world	Time: 1990M1 to 1998M12	TB = f(E, Y, YF)	The evidence of J-curve effect is detected
Narayan (2006)	China	The US	Method: ARDL	TB = f(E)	The depreciation of CNY against the USD fosters China’s trade balance with the US
Narayan (2006)	China	The US	Time: 1979M11 to 2002M9	TB = f(E)	No J-curve effect is identified
Šimáková (2013)	Hungary	Austria, Czechia, France, Germany, Italy, the Netherland, Poland, the UK	Method: Johansen-Juselius cointegration	TB = f(E, Y, YF)	This research captures the role of EUR as a vehicle currency in the trade of Hungary with Czechia, Poland, and the UK
Šimáková (2013)	Hungary		Time: 1997Q1 to 2012Q2	TB = f(V, Y, YF)	The Hungary-UK trade witnesses the existence of the J-curve effect
Šimáková and Stavárek (2014)	Czechia	Austria, France, Germany, Italy, Poland, and Slovakia	Method: Johansen-Juselius cointegration	TB = f(E, Y, YF)	This study is conducted at industry level
			Time: 1993 to 2013		The impacts of Czech koruna’s depreciation on Czechia’s trade balances vary with the partners and the industries
			Time: 1993 to 2013		No evidence of J-curve effect is found
Bahmani-Oskooee and Fariditavana (2015)	Canada, China, Japan, the US	The rest of the world	Method: ARDL, NARDL	TB = f(E, Y, YF)	This is the pioneering study to propose the asymmetric assumption about the exchange rate-trade balance relationship
			Time: 1971Q1 to 2013Q3		The NARDL method is superior to the ARDL method in finding more cases of the J-curve effect. When the ARDL method is used, only Canada and the US experience the J-curve effect. When the NARDL method is used, the J-curve effect is identified in Canada, China, and the US.
			Time: 1971Q1 to 2013Q3		Future studies should pay attention to the asymmetric linkage between change rates and trade balances
Bahmani-Oskooee et al. (2016)	Mexico	13 Partners	Method: ARDL, NARDL	TB = f(E, Y, YF)	The NARDL method is better than the ARDL method in providing more support for the J-curve effects
			Time: 1980Q1 to 2014Q4 (some partners have different time frames)		When the ARDL method is used, the J-curve effect is detected in Mexico’s trade with six partners: Brazil, Canada, France, Japan, South Korea, and the US
					When the NARDL method is used, the J-curve effect is detected in Mexico’s trade with ten partners: Brazil, Canada, France, India, Japan, Peru, South Korea, Spain, the UK, and the US
Iyke and Ho (2017)	Ghana	The rest of the world	Method: ARDL, NARDL	TB = f(E, Y, YF)	When the ARDL method is used, the impact of Ghana’s exchange rate on its trade balance is insignificant. Thus, no J-curve effect is detected.
			Time: 1986Q1 to 2016Q3		When the NARDL method is used, the impact of Ghana’s exchange rate on its trade balance becomes significant. The presence of the J-curve effect is also supported. Therefore, the NARDL method is better than the ARDL method in studying the exchange rate-trade balance relationship.
			Time: 1986Q1 to 2016Q3		When the NARDL method is used, the J-curve effect is detected in Mexico’s trade with ten partners: Brazil, Canada, France, India, Japan, Peru, South Korea, Spain, the UK, and the US
Nusair (2017)	16 European transition economies	The rest of the world	Method: ARDL, NARDL	TB = f(E, Y, YF)	The NARDL method is better than the ARDL method in studying the exchange rate-trade balance relationship
Nusair (2017)	16 European transition economies	The rest of the world	Time: 1994Q1 to 2015Q2 (some partners have different time frames)	TB = f(E, Y, YF)	The J-curve effect is detected in 12 out of the 16 European transition economies
Yazgan and Ozturk (2019)	33 Countries	For each of the 33 countries, its partners are the remaining 32 countries in the sample	Method: panel data estimation by the CCE, CCEP, CCEMG estimators	TB = f(E, Y, YF)	The depreciation of domestic currencies can foster the trade balances
Yazgan and Ozturk (2019)	33 Countries		Time: 1981Q1 to 2010Q2	TB = f(E, Y, YF)	The findings show the absence of J-curve effect
Usman et al. (2021)	Pakistan	The UK	Method: ARDL, NARDL	TB = f(E, Y, YF)	The NARDL method is better than the ARDL method in studying the exchange rate-trade balance relationship, which helps identify more cases of J-curve effect
Usman et al. (2021)	Pakistan	The UK	Time: 1980 to 2016	TB = f(E, Y, YF)	The findings indicate the presence of J-curve effect in 14 out of 28 industries

Note. E = exchange rates; Y = the incomes of the focused countries; YF = the incomes of the partners; V = vehicle currencies’ exchange rates; VOL = exchange rate volatility; ARDL = autoregressive distributed lag; NARDL = nonlinear autoregressive distributed lag; CCE = common correlated effects; CCEP = common correlated effects pooled; CCEMG = common correlated effects mean group; GARCH = generalized autoregressive conditional heteroskedasticity; FMOLS = fully modified ordinary least squares; VECM = vector error correction model.

Source: Authors’ review of the literature.

Data Availability Statement

Data is available in public repository. The authors confirm that all data underlying the findings are fully available without restriction.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the European Union’s Horizon 2020 research and innovation programmed under Marie Skłodowska-Curie grant agreement No. 734712; Ho Chi Minh City University of Law. The authors owe extensive gratitude to the editor and the anonymous reviewers for their constructive comments.

ORCID iD

Hoang Phong Le

Notes

References

Arize

A. C.

(1994). Cointegration test of a long-run relation between the real effective exchange rate and the trade balance. International Economic Journal, 8(3), 1–9. https://doi.org/10.1080/10168739400000001

Arize

A. C.

Malindretos

Igwe

E. U.

(2017). Do exchange rate changes improve the trade balance: An asymmetric nonlinear cointegration approach. International Review of Economics & Finance, 49, 313–326. https://doi.org/10.1016/j.iref.2017.02.007

Baharumshah

A. Z.

(2001). The effect of exchange rate on bilateral trade balance: New evidence from Malaysia and Thailand. Asian Economic Journal, 15(3), 291–312. https://doi.org/10.1111/1467-8381.00135

Bahmani-Oskooee

(1985). Devaluation and the J-curve: Some evidence from LDCs. The Review of Economics and Statistics, 67(3), 500–504. https://doi.org/10.2307/1925980

Bahmani-Oskooee

Aftab

(2017a). Asymmetric effects of exchange rate changes and the J-curve: New evidence from 61 Malaysia-Thailand industries. Review of Development Economics, 21(4), e30–e46. https://doi.org/10.1111/rode.12298

Bahmani-Oskooee

Aftab

(2017b). Asymmetric effects of exchange rate changes on the Malaysia-EU trade: Evidence from industry data. Empirica, 44(2), 339–365. https://doi.org/10.1007/s10663-016-9324-8

Bahmani-Oskooee

Aftab

(2018). Asymmetric effects of exchange rate changes on the Malaysia-China commodity trade. Economic Systems, 42(3), 470–486. https://doi.org/10.1016/j.ecosys.2017.11.004

Bahmani-Oskooee

Baek

(2019). Asymmetry cointegration and the J-curve: New evidence from Korean bilateral trade balance models with her 14 partners. Journal of the Asia Pacific Economy, 24(1), 66–81. https://doi.org/10.1080/13547860.2018.1469589

Bahmani-Oskooee

Fariditavana

(2015). Nonlinear ARDL approach, asymmetric effects and the J-curve. Journal of Economic Studies, 42(3), 519–530. https://doi.org/10.1108/jes-03-2015-0042

10.

Bahmani-Oskooee

Fariditavana

(2016). Nonlinear ARDL approach and the J-curve phenomenon. Open Economies Review, 27, 51–70. https://doi.org/10.1007/s11079-015-9369-5

11.

Bahmani-Oskooee

Goswami

G. G.

Kumar Talukdar

(2005). The bilateral J-curve: Australia versus her 23 trading partners. Australian Economic Papers, 44(2), 110–120. https://doi.org/10.1111/j.1467-8454.2005.00252.x

12.

Bahmani-Oskooee

Halicioglu

Hegerty

S. W.

(2016). Mexican bilateral trade and the J-curve: An application of the nonlinear ARDL model. Economic Analysis and Policy, 50, 23–40. https://doi.org/10.1016/j.eap.2016.02.003

13.

Bahmani-Oskooee

Harvey

(2010). The J-curve: Malaysia versus her major trading partners. Applied Economics, 42(9), 1067–1076. https://doi.org/10.1080/00036840701721158

14.

Bahmani-Oskooee

Harvey

(2019). The J-curve and bilateral trade balances of Indonesia with its major partners: Are there asymmetric effects? New Zealand Economic Papers, 53, 63–76. https://doi.org/10.1080/00779954.2017.1371207

15.

Bahmani-Oskooee

Hegerty

Zhang

(2014). Exchange-rate risk and UK-China trade: Evidence from 47 industries. Journal of Chinese Economic and Foreign Trade Studies, 7(1), 2–17. https://doi.org/10.1108/jcefts-04-2013-0011

16.

Bahmani-Oskooee

Hegerty

S. W.

(2010). The J- and S-curves: A survey of the recent literature. Journal of Economic Studies, 37(6), 580–596. https://doi.org/10.1108/01443581011086639

17.

Bahmani-Oskooee

Kanitpong

(2017). Do exchange rate changes have symmetric or asymmetric effects on the trade balances of Asian countries? Applied Economics, 49(46), 4668–4678. https://doi.org/10.1080/00036846.2017.1287867

18.

Bahmani-Oskooee

Kanitpong

(2019). Asymmetric effects of exchange rate changes on Thailand-China commodity trade: Evidence from 45 industries. Chinese Economy, 52(3), 203–231. https://doi.org/10.1080/10971475.2018.1548142

19.

Bahmani-Oskooee

Ratha

(2004). The J-curve: A literature review. Applied Economics, 36(13), 1377–1398. https://doi.org/10.1080/0003684042000201794

20.

Bailey

M. J.

Tavlas

G. S.

Ulan

(1987). The impact of exchange-rate volatility on export growth: Some theoretical considerations and empirical results. Journal of Policy Modeling, 9(1), 225–243. https://doi.org/10.1016/0161-8938(87)90010-x

21.

Bank for International Settlements. (2016). Triennial Central Bank Survey of foreign exchange and OTC derivatives markets in 2016. https://www.bis.org/publ/rpfx16.htm

22.

Bank for International Settlements. (2019). Triennial Central Bank Survey of foreign exchange and over-the-counter (OTC) derivatives markets in 2019. https://www.bis.org/statistics/rpfx19.htm

23.

Bao

H. H. G.

H. P.

(2021a). The role of vehicle currency in ASEAN-EU trade: A double-aggregation method. Journal of Asian Finance Economics and Business, 8(5), 43–52. https://doi.org/10.13106/jafeb.2021.vol8.no5.0043

24.

Bao

H. H. G.

H. P.

(2021b). Asymmetric impact of exchange rate on trade between Vietnam and each of EU-27 countries and the UK: Evidence from nonlinear ARDL and the role of vehicle currency. Heliyon, 7(6), e07344. https://doi.org/10.1016/j.heliyon.2021.e07344

25.

Bineau

(2016). Real exchange rate and bilateral trade balance of Cambodia: A panel investigation. Economic Bulletin, 36(2), 895–900.

26.

Bowe

Saltvedt

T. M.

(2004). Currency invoicing practices, exchange rate volatility and pricing-to-market: Evidence from product level data. International Business Review, 13(3), 281–308. https://doi.org/10.1016/j.ibusrev.2004.01.003

27.

Boz

Casas

Georgiadis

Gopinath

Mezo

H. L.

Mehl

Nguyen

(2020). Patterns in invoicing currency in global trade (Working Paper No. 20/126). International Monetary Fund. https://www.imf.org/en/Publications/WP/Issues/2020/07/17/Patterns-in-Invoicing-Currency-in-Global-Trade-49574

28.

Brahmasrene

Jiranyakul

(2002). Exploring real exchange rate effects on trade balances in Thailand. Managerial Finance, 28(11), 16–27. https://doi.org/10.1108/03074350210768130

29.

European Commission. (2021). Countries and regions – Association of South East Asian Nations (ASEAN). https://ec.europa.eu/trade/policy/countries-and-regions/regions/asean/

30.

Eurostat. (2021). Extra-EU trade by invoicing currency. https://ec.europa.eu/eurostat/statistics-explained/index.php/Extra-EU_trade_by_invoicing_currency

31.

Goldberg

L. S.

Tille

(2008). Vehicle currency use in international trade. Journal of International Economics, 76, 177–192. https://doi.org/10.1016/j.jinteco.2008.07.001

32.

Gomes

F. A. R.

Paz

L. S.

(2005). Can real exchange rate devaluation improve the trade balance? The 1990–1998 Brazilian case. Applied Economics Letters, 12(9), 525–528. https://doi.org/10.1080/13504850500076908

33.

Harada

(2020). ASEAN becomes China’s top trade partner as supply chain evolves. Nikkei Asia. https://asia.nikkei.com/Politics/International-relations/ASEAN-becomes-China-s-top-trade-partner-as-supply-chain-evolves

34.

Harvey

(2018). Bilateral J-curve between Philippines and trading partners: Linear and non-linear approach. Asian Economic and Financial Review, 8(2), 131–144. https://doi.org/10.18488/journal.aefr.2018.82.131.144

35.

Hayakawa

Kimura

(2009). The effect of exchange rate volatility on international trade in East Asia. Journal of the Japanese and International Economies, 23, 395–406. https://doi.org/10.1016/j.jjie.2009.07.001

36.

S. H.

Nguyen

T. T.

To-The

(2021). On the (a)symmetric effects of real exchange rate on trade flows: New evidence from US–Vietnam trade balance at the industry level. Journal of the Asia Pacific Economy. Advance online publication. https://doi.org/10.1080/13547860.2021.1884346

37.

Iyke

B. N.

S. Y.

(2017). The real exchange rate, the Ghanaian trade balance, and the J-curve. Journal of African Business, 18(3), 380–392. https://doi.org/10.1080/15228916.2017.1315706

38.

Iyke

B. N.

S. Y.

(2018). Nonlinear effects of exchange rate changes on the South African bilateral trade balance. Journal of International Trade & Economic Development, 27(3), 350–363. https://doi.org/10.1080/09638199.2017.1378916

39.

Krugman

(1980). Vehicle currencies and the structure of international exchange. Journal of Money Credit and Banking, 12(3), 513–526. https://doi.org/10.2307/1991725

40.

Kyophilavong

Shahbaz

Rehman

I. U.

Souksavath

Chanthasene

(2018). Investigating the relationship between trade balance and the exchange rate: The case of Laos’ trade with Thailand. Global Business Review, 19(3), 590–603. https://doi.org/10.1177/0972150917713869

41.

Kyophilavong

Shahbaz

Uddin

G. S.

(2013). Does J-curve phenomenon exist in case of Laos? An ARDL approach. Economic Modelling, 35, 833–839. https://doi.org/10.1016/j.econmod.2013.08.014

42.

Lee

(2018). The dynamics of the trade balance of Vietnam and its weak currency policy: A bilateral autoregressive distributed lag (ARDL) analysis. Journal of Applied Economics and Business Research, 8(2), 68–92. http://www.aebrjournal.org/uploads/6/6/2/2/6622240/joaebrjune2018_68_92.pdf

43.

Liu

(2019). Trade, finance and endogenous invoicing currency: Theory and firm-level evidence. Pacific-Basin Finance Journal, 56, 21–44. https://doi.org/10.1016/j.pacfin.2019.05.007

44.

Magee

S. P.

(1973). Currency contracts, pass-through, and devaluation. Brookings Papers on Economic Activity, 1, 303–325. https://doi.org/10.2307/2534091

45.

Magee

S. P.

Rao

R. K. S.

(1980). Vehicle and nonvehicle currencies in international trade. American Economic Review, 70(2), 368–373. https://www.jstor.org/stable/1815500

46.

McKenzie

M. D.

(1999). The impact of exchange rate volatility on international trade flows. Journal of Economic Surveys, 13(1), 71–106. https://doi.org/10.1111/1467-6419.00075

47.

Medhora

(1990). The effect of exchange rate variability on trade: The case of the West African Monetary Union’s imports. World Development, 18(2), 313–324. https://doi.org/10.1016/0305-750x(90)90055-3

48.

Narayan

P. K.

(2005). The saving and investment nexus for China: Evidence from cointegration tests. Applied Economics, 37(17), 1979–1990. https://doi.org/10.1080/00036840500278103

49.

Narayan

P. K.

(2006). Examining the relationship between trade balance and exchange rate: The case of China’s trade with the USA. Applied Economics Letters, 13(8), 507–510. https://doi.org/10.1080/13504850500400488

50.

Nhung

N. C.

Anh

B. T.

Hue

L. T.

Huyen

N. T. C.

(2018). The impact of exchange rate movements on trade balance between Vietnam and Japan: J curve effect test. VNU Journal of Science: Economics and Business, 34(2), 17–27. https://doi.org/10.25073/2588-1108/vnueab.4154

51.

Nusair

S. A.

(2017). The J-curve phenomenon in European transition economies: A nonlinear ARDL approach. International Review of Applied Economics, 31(1), 1–27. https://doi.org/10.1080/02692171.2016.1214109

52.

Odhiambo

N. M.

(2009). Energy consumption and economic growth nexus in Tanzania: An ARDL bounds testing approach. Energy Policy, 37(2), 617–622. https://doi.org/10.1016/j.enpol.2008.09.077

53.

Office of the United States Trade Representative. (n.d). Association of Southeast Asian Nations (ASEAN). https://ustr.gov/countries-regions/southeast-asia-pacific/association-southeast-asian-nations-asean

54.

Pesaran

M. H.

Shin

Smith

R. J.

(2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Economics, 16, 289–326. https://doi.org/10.1002/jae.616

55.

Phong

L. H.

Bao

H. H. G.

Van

D. T. B.

(2018). Testing J-curve phenomenon in Vietnam: An autoregressive distributed lag (ARDL) approach. In Anh

Dong

Kreinovich

Thach

(Eds.), Econometrics for financial applications: Studies in computational intelligence (Vol. 760, pp. 491–503). Springer International Publishing.

56.

Rose

A. K.

(1991). The role of exchange rates in a popular model of international trade. Does the ‘Marshall-Lerner’ condition hold? Journal of International Economics, 30, 301–316. https://doi.org/10.1016/0022-1996(91)90024-z

57.

Rose

A. K.

Yellen

J. L.

(1989). Is there a J-curve? Journal of Monetary Economics, 24(1), 53–68. https://doi.org/10.1016/0304-3932(89)90016-0

58.

Sharma

Pal

(2018). Exchange rate volatility and India’s cross-border trade: A pooled mean group and nonlinear cointegration approach. Economic Modelling, 74, 230–246. https://doi.org/10.1016/j.econmod.2018.05.016

59.

Shin

Greenwood-Nimmo

(2014). Modelling asymmetric cointegration and dynamic multipliers in a nonlinear ARDL framework. Festschrift in honor of Peter Schmidt: Econometric methods and applications. In Sickels

R. C.

Horrace

W. C.

(Eds.), Festschrift in honor of Peter Schmidt (pp. 4281–4314). Springer.

60.

Šimáková

(2013). Estimation of the J-curve effect in the bilateral trade of Hungary. Ekonomická revue, 16(3), 183–191. https://doi.org/10.7327/cerei.2013.09.07

61.

Šimáková

Stavárek

(2014). Exchange-rate impact on the industry-level trade flows in the Czech Republic. Procedia Economics and Finance, 12, 679–686. https://doi.org/10.1016/s2212-5671(14)00393-1

62.

Singh

(2004). Testing J-curve hypothesis and analysing the effect of exchange rate volatility on the balance of trade in India. Empirical Economics, 29(2), 227–245. https://doi.org/10.1007/s00181-003-0162-8

63.

Soleymani

Chua

S. Y.

(2014). How responsive are trade flows between Malaysia and China to the exchange rate? Evidence from industry data. International Review of Applied Economics, 28(2), 191–209. https://doi.org/10.1080/02692171.2013.858666

64.

Sugiharti

Esquivias

M. A.

Setyorani

(2020). The impact of exchange rate volatility on Indonesia’s top exports to the five main export markets. Heliyon, 6(1), e03141. https://doi.org/10.1016/j.heliyon.2019.e03141

65.

Usman

Bahmani-Oskoee

Anwar

Ullah

(2021). Is there J-curve effect in the trade between Pakistan and United Kingdom? Asymmetric evidence from industry level data. The Singapore Economic Review. Advance online publication. https://doi.org/10.1142/s0217590821500089

66.

Wilson

(2001). Exchange rates and the trade balance for dynamic Asian economies—Does the J-curve exist for Singapore, Malaysia, and Korea? Open Economies Review, 12, 389–413. https://doi.org/10.1023/a:1017982901034

67.

Wilson

Tat

K. C.

(2001). Exchange rates and the trade balance: The case of Singapore 1970 to 1996. Journal of Asian Economics, 12(1), 47–63. https://doi.org/10.1016/s1049-0078(01)00072-0

68.

Witte

M. D.

(2010). Currency invoicing: The role of ‘Herding’ and exchange rate volatility. International Economic Journal, 24(3), 357–374. https://doi.org/10.1080/10168730903510666

69.

Yang

(2016). Effects of exchange rate variations on bilateral trade with a vehicle currency: Evidence from China and Singapore. Journal of International Money and Finance, 68, 50–73. https://doi.org/10.1016/j.jimonfin.2016.06.010

70.

Yazgan

M. E.

Ozturk

S. S.

(2019). Real exchange rates and the balance of trade: Does the J-curve effect really hold? Open Economies Review, 30, 343–373. https://doi.org/10.1007/s11079-018-9510-3

71.

Yusoff

M. B.

(2010). The effects of real exchange rate on trade balance and domestic output: A case of Malaysia. The International Trade Journal, 24(2), 209–226. https://doi.org/10.1080/08853901003652377