Abstract
The present research shows how one can deal with stationary plus trend trade balance variables in a trade model whenever the rest of the variables contain a unit root. Data are used in a monthly and a quarterly basis from January 1980 to June 2011 and applied to four countries (Germany, France, Italy and United Kingdom). It is proved that an error correction mechanism suits better when detrending trade balances once they have been verified to have a deterministic trend.
Acknowledgement
The author gratefully acknowledges Carol Alexander's helpful comments and ideas on an earlier work, which was the basis for this paper.