Abstract
A study by McNown, Sam, and Goh [2018, Applied Economics, 50(13), 1509–1521] has shown that the autoregressive distributed lag (ARDL) bounds test proposed by Pesaran, Shin, and Smith [2001, Journal of Applied Econometrics, 16(3), 289–326] may draw incorrect conclusions on the status of the cointegration test, if the ARDL bounds test is not implemented correctly. We assess the long-run relationship between US exports and imports as well as between its eight major trading partners (Brazil, Canada, China, France, Germany, Japan, Mexico, and the United Kingdom) by applying the newly developed bootstrap ARDL test by McNown et al. (2018). The results show cointegration if exports are used as the dependent variable, but not when imports are being considered as the dependent variable. This suggests that the cointegration result is sensitive to the choice of the dependent variable. We have similar findings when we examine the US and its major trading partners. No long-run relationship exists between exports and imports in the case of the US, which concurs with the finding of Fountas and Wu [1999, International Economic Journal, 13(3), 51–58]. The results also suggest that the US attempts to reduce its bilateral imbalances through targeted trade policies may not be appropriate.
Get full access to this article
View all access options for this article.