Kybernetika

This paper extends the method proposed in \cite{Lee:92} for quarterly nonstationary data, considering the estimation and testing for seasonal cointegration relationships when dealing with strongly seasonal monthly data. The testing procedure is based on the maximum-likelihood estimation of the `error correction mechanism' for the vector of series considered. Finite sample critical values for the cointegration test statistics at every frequency of interest are obtained by Monte Carlo simulations. Finally, tests are applied to Spanish production indexes data.