Abstract
Applied cointegration analysis has much to gain from strong links with economic theory. For example, the current generation of equilibrium macroeconomic models have simple predi tions for cointegrating vectors. These models also suggest that important information about the economic structure can be found in the short run dynamics, which most cointegration studies disregard. Simulations of a stochastic business cycle model show that tests of cointegrating vectors, forecasts, and variance decompositions based on long run assumptions can be sharpened by imposing even very simple economic restrictions.