Abstract
This article applies cointegration techniques to estimate a monthly demand system for meat in Italy. In contrast to existing studies where Engle and Granger's two step procedure and Triangular Vector Error Correction Model (TVECM) representations are usually exploited, it applies a cointegrated Vector Error Correction Model (VECM) where also prices and expenditure enter endogenously the system and the cointegration rank is not assumed to be known a priori but subject to inference. It highlights some of the advantages of using the VECM compared to the TVECM, including the possibility of testing the cointegration rank of the system and the (weak and strong) exogeneity of prices and expenditure within a well-specified statistical model. This may lead to remarkable improvements in the efficiency of parameters system estimates.