Stationarity and cointegration tests: Comparison of Engle - Granger and Johansen methodologies

Engle-Granger methodology follows two-step estimations. The first step generates the residuals and the second step employs generated residuals to estimate a regression of first-differenced residuals on lagged residuals. Hence, any possible error from the first step will be carried into second step. The Johansen maximum likelihood methodology circumvents Engle-Granger methodology by estimating and testing for the presence of multiple cointegrating vectors through largest canonical correlations. The number of non-zero eigenvalues of Ψ of eq. 26 in the text will specify the number of cointegrating vectors. Some Monte Carlo evidence explores that Johansen procedure performs better than both single equation methods and alternative multivariate methods. In fact, evidence of this paper reveals, as well, that, as Engle-Granger yields some inconclusive outcome, the Johansen tests reach at least one cointegration relationship among variables for Canada, India, Italy, Japan, Turkey and the USA. Then, one may claim that Johansen methodology dominates the Engle- Granger methodology in cointegration analyses