Test of the J-Curve Hypothesis Between the U.S. and Six Western European Countries and Policy Implications

Abstract

Applying the VECM and the GIRF, this article finds that when the relative CPI is employed, there is support for a J-curve for Germany and lack of evidence for a J-curve for Belgium, France, Italy, the Netherlands, and Spain. When the relative PPI is used, similar results are found. There are six different patterns of the response of the bilateral trade balance to a shock to the real exchange rate. Policy makers need to pay attention to different responses of the trade balance to real depreciation in the short run and long run.

KEYWORDS:

• GIRF
• J-curve
• VECM

Notes

Notes. *Means that the null hypothesis of no cointegration relation can be rejected.

**Indicates that the significance is at the 5% level.

Note: Figures in the parenthesis are t-statistics.

In Johansen's method, it estimates the Π matrix from an unrestricted VAR and determines whether the restrictions implied by the reduced rank of Π can be rejected.

According to the Jarque-Bera test, the errors for each of the countries exhibit multivariate normal distribution.

Conference bands for the GIRFs in Figure 1 based on the forecast error variance are not available in EVIEWS.