Exchange rate uncertainty, UK trade and the euro

Abstract

The impact of exchange rate uncertainty on the disaggregated imports of the UK is investigated by focusing on 15 major manufacturing categories. Contrary to several previous theoretical and empirical studies, this paper finds no evidence that exchange rate uncertainty has a positive impact on international trade. An important implication of this study for UK macroeconomic policy is that adoption of the euro would have, in terms of its effect on exchange rate risk, a positive impact on the country's trade and economic welfare.

Acknowledgements

We thank the Research Enhancement Fund of The Nottingham Trent University for financial support. We would also like to thank participants of the 3rd International Conference on Money, Investment and Risk and of the 6th International Conference in Macroeconomic Analysis and International Finance for their comments. We also thank seminar participants at the Department of Economics of the University of Cyprus for their useful and constructive suggestions. The usual disclaimer applies.

Notes

As of September 2004 Charalambos Pattichis will be joining the School of Business at the American University of Beirut.

 Previous empirical studies show that there are no qualitative differences in using nominal or real exchange rate volatility (see Qian and Varangis, 1994; McKenzie and Brooks, 1997).

 West et al. (1993) show that GARCH (1,1) relatively performs better than other alternative ARCH-type models.

 Hsieh (1989) shows that standardized residuals obtained from various (G)ARCH-type models using the standard normal density are highly leptokurtic.

 See Holly (1995) who also finds a similar result for this measure of exchange rate volatility.

 Doornik et al. (1998) recommend the use of trace statistics, since these tests select the cointegrating rank more consistently than the alternative maximum eigenvalue statistics.

 See Engle et al. (1983) for a detailed explanation of the property of weak exogeneity. In the framework of the cointegrated VAR system, Johansen (1992) discusses the testing procedures.

 The Newey–West adjusted variances of the OLS estimators are calculated as

where

,

with lag window

and T = sample size in a common notation

in which the OLS estimator b _T is asymptotically approximated by the true value β as

. For the lag window, a Parzen kernel is generally recommended in the literature with a window size which does not exceed one-third of available observations, in order to ensure a positive semidefinite variance matrix (see Hamilton, 1994).

 Note that by setting all changes in the short-run equal to zero, the cointegrating vector for each sector (not reported, but available upon request) can be obtained from Equation 4 as: im _t  = ϕ₁ rp _t  + ϕ₂ y _t  + ϕ₃ v _t , where ϕ₁ = −θ/ϕ, ϕ₂ = −λ/ϕ, and ϕ₃ = −ξ/ϕ. It should be noted that the coefficients of the volatility measure in the cointegrating vectors were statistically insignificant. Note also that the feedback coefficients lie within a range between −0.01 and −0.23 and have the expected signs with statistical significance in almost all sectors, but indicate slow adjustment of the past disequilibrium in import trade volumes.