Based on a money market analysis using the cointegrated VAR model the paper demonstrates some possible pitfalls in macroeconomic inference as a direct consequence of inadequate stochastic model formulation. A number of questions related to concepts such as empirical and theoretical steady-states, speed of adjustment, feedback and interaction effects, and driving forces are addressed within the framework of the cointegrated VAR model. The interpretation and analysis of common driving trends are related to the notion of shocks or disturbances to a system, distinguishing between permanent and transitory, and anticipated and unanticipated effects