Stochastic Trends and Short-Run Relationships Between Financial Variables and Real Activity

This paper re-examines the relationship between financial variables and real activity in a unified statistical framework. Using the methods of cointegration and separation. we characterize the long-run and short-run relationships between three sets of variables and then use the framework to assess the predictive power of alternative financial variables for real activity. Three main results emerge from the analysis. First, we show that although two sets of variables may not share the long-run trend. the error correction terms from one set of variables may have important explanatory power for the variables in another set. Second, we show that some of the key variables discussed in the literature can be interpreted as error correction terms from another system. Third, comparing two key error correction terms, M2 velocity and the interest rate spread between commercial paper and Treasury bills, we find that M2 velocity appears to be a more consistent predictor of output than is the interest rate spread