US–Malaysia Trade at Commodity Level and the Role of the Real Exchange Rate

Abstract

Previous studies that were concerned with the impact of depreciation of the ringgit on the Malaysian trade balance employed data either between Malaysia and rest of the world or between Malaysia and each of her major trading partners. Specifically, the bilateral trade balance between Malaysia and the US is shown to be insensitive to the real bilateral ringgit–dollar rate. In this article we wonder if disaggregating trade flows between Malaysia and the US by commodity could help us to discover any significant effects that the real exchange rate could have. We consider 101 industries that export from US to Malaysia and 17 industries that import from Malaysia. While majority of the industries showed short-run sensitivity to the real bilateral exchange rate, short-run effects lasted into the long run almost in half of the industries in both group.

Keywords:

• Malaysia
• US
• commodity trade
• bounds testing

JEL CLASSIFICATION:

• F31

Notes

1. For a detailed review of foreign exchange policy as well as other policies implemented by the Malaysian government see Bahmani-Oskooee and Harvey (2010).

2. For some studies estimating the Marshall–Lerner condition see Houthakker and Magee (1969), Goldstein and Khan (1978, 1985), Noland (1989), Deyak et al. (1989, 1990), Marquez (1990), Reinhart (1995) and Razafimahefa and Hamori (2005).

3. See their footnote 13 on page 558.

4. For other applications of this approach see Halicioglu, F., (2007), Narayan et al. (2007), Tang (2007), Mohammadi et al. (2008), Wong and Tang (2008), De Vita and Kyaw (2008) and Payne (2008).

5. This contradicts Burda and Gerlach (1992), who argued that durable goods are relatively more sensitive to exchange rate changes than non-durable ones.

6. Note that the critical values of the t-ratio for significance of ECM _t–1 are non-standard. New critical values are tabulated by Banerjee et al. (1998) for small as well as large sample sizes and are used here. Here we use the critical value of –3.24 at the 10% significance level for sample size of 25 observations since next size is 50.

7. When a dummy variable is included in the error-correction model, Pesaran et al. (2001), footnote 17, p. 307, argue that their asymptotic theory and associated critical values hold only if the fraction of the period in which the dummy takes a non-zero value tend to zero with the sample size of T. Like their example in which this fraction was almost 0.20, in our case it is 0.27. Thus, there is no urgent need to modify the critical values.