In this paper we propose the use of a sequential multivariate approach to test for convergence. These tests allow us to reconcile the time series literature with the cross-sectional dimension which is basic when testing for convergence in regional blocs. In addition, this methodology helps to avoid the problem due to the limited dataset length. We apply multivariate unit root tests in two stages. First, we test for non-convergence without identifying the countries within the group that effectively converge and, in a second stage, the countries that converge are identified. The SURE technique allows for the existence of correlations across the series without imposing a common speed of mean reversion. The empirical results for Mercosur countries show that there is evidence of catching up of the small countries towards Brazil and, to some extent, Argentina. In contrast, the evidence of catching-up among the larger countries is weaker.

JEL classification: F15, C22, C15