Likelihood-Based Cointegration Analysis in Panels of Vector Error Correction Models

We propose in this paper a likelihood-based framework forcointegration analysis in panels of a fixed number of vector errorcorrection models. Maximum likelihood estimators of thecointegrating vectors are constructed using iterated GeneralizedMethod of Moments estimators. Using these estimators we constructlikelihood ratio statistics to test for a common cointegrationrank across the individual vector error correction models, bothwith heterogeneous and homogeneous cointegrating vectors. Thecorresponding limiting distributions are a summation of thelimiting behavior of Johansen (1991) trace statistics. We alsoincorporate both unrestricted and restricted deterministiccomponents which are either homogeneous or heterogeneous. Theproposed framework is applied on a data set of exchange rates andappropriate monetary fundamentals. The test results show strongevidence for the validity of the monetary exchange rate modelwithin a panel of vector error correction models for three majorEuropean countries, whereas the results based on individual vectorerror correction models for each of these countries separately areless supportive.