| Soren Johansen's (1988) cointegration method is only one of many methods developed since Clive Granger's (1983) development of cointegration analysis. It is, however, the only method to test for multiple cointegrating vectors in a data set and then test hypotheses on these vectors. Because of these aspects, Johansen's method has become the most popular cointegration method used in the analysis of long-run economic relationships.;Several Monte Carlo studies have been devised to study the performance of Johansen's method under a variety of circumstances and to compare his method to other cointegration techniques. Johansen's method has gained support from such studies as the ‘best’ estimator based on small biases and small size distortions. However, most of these studies have only analyzed Johansen's method at the single equation, bivariate level. As Johansen's method is used to test for multiple cointegrating vectors in a data set and then test hypotheses on these vectors, it would be more beneficial to study his method at this level, which is done here for the first time.;It is determined that Johansen's method is sensitive to sample size and the speed of adjustment to disequilibrium, but still outperforms Engle and Granger's (1987) Ordinary Least Squares method based on a comparison of size distortions using Johansen's H5 test and F-tests based on the estimates from Engle and Granger's OLS method. It is also determined that Johansen's method is highly sensitive to changes in the structure of the common trends. |
