Do exchange rate changes have symmetric or asymmetric effects on the trade balances of Asian countries?

ABSTRACT

Recently the linear ARDL approach was modified and a non-linear version of the same approach that is used mostly to assess asymmetric effects of some exogenous variables on the dependent variable was introduced. The non-linear model was recently used by one study to show that indeed exchange rate changes have asymmetric effects on the trade balance of a few advanced countries. The same was demonstrated for transition economies by another study. In this article, we provide additional asymmetric effects from seven Asian economies by showing that in most cases we find evidence of short-run and long-run asymmetric effects of exchange rate changes on the trade balance. Like other studies, our findings are country specific.

KEYWORDS:

• Non-linear ARDL approach
• asymmetry effects
• the trade balance
• the exchange rate
• Asian countries

JEL CLASSIFICATION:

• F31
• F41

Acknowledgements

This research was supported by NIDA Business School Research Grants in Collaboration with Distinguished Visiting Professor. We also would like to thank valuable comments of an anonymous reviewer. Remaining errors, however, are our own.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ For most recent review articles see Bahmani-Oskooee and Ratha (2004) and Bahmani-Oskooee and Hegerty (2010).

² This section closely follows Bahmani-Oskooee and Fariditavana (2015).

³ Note that Pesaran, Shin, and Smith (2001) provide an upper bound critical value by assuming all variables to be I(1) and a lower bound value by assuming all variables to be I(0). However, they argue that their upper bound critical values could also be used when we have combination of I(0) and I(1) variables.

⁴ For exact normalization procedure, see Bahmani-Oskooee and Fariditavana (2015).

⁵ See Shin et al. (2014, p. 291).

⁶ For some other applications of these two approaches, see Apergis and Miller (2006), De Vita and Kyaw (2008), Hajilee and Al-Nasser (2014), Halicioglu (2007, 2008), Verheyen (2013), and Dellattee and Lopez-Villavicencio (2012).

⁷ Using the normalized long-run coefficient estimates from Panel B and long-run model (1), we generate the error term, called ECM. We then replace the lagged level variables by ECM_t−1 and estimate this new specification after imposing the same optimum lags from Panel A. A significantly negative coefficient obtained for ECM_t−1 will support the convergence of variables towards their long-run equilibrium values or cointegration. Note that the t-ratio that is sued to judge significant of this estimate has non-standard distribution. Thus, Pesaran et al. (2001, p. 303) tabulate a lower bound and a higher bound critical values for this test as well. However, for sample sizes such as ours which are not large enough, we use critical values from Banerjee et al. (1989, Table 1) who originally introduced this test within Engle and Granger (1987) cointegration approach. Our approach is justified since for large samples Pesaran et al.’s upper bound critical values are the same as Banerjee et al.’s values.

⁸ Only in the results for Japan three negative coefficients are followed by three positive estimates, supporting the short-run J-curve hypothesis.

⁹ Note that in some cases like Korea, the NEG variable carries a significantly positive coefficient, implying that a depreciation will improve Korean trade balance in the long run. It also implies that the well-known Marshall–Lerner condition is satisfied. However, since the POS variable carries a negative coefficient, the Marshall–Lerner condition is not met. Clearly, these different elasticity conditions could also contribute to asymmetric effects.

¹⁰ As for the other diagnostic statistics, both the LM and RESET statistics are insignificant in most linear and non-linear models. Almost all estimates in both models are stable reflected by CUSUM and CUSUMSQ tests.

Additional information

Funding

This research was supported by NIDA Business School Research Grants in Collaboration with Distinguished Visiting Professor.