Exploring the Robustness of the Oil Price-Macroeconomy Relationship

This paper reexamines the oil price-macroeconomy relationship with rolling Granger causality and structural stability tests. It finds that the relationship broke down amidst the falling oil prices and market collapse of the 1980s, suggesting misspecification of the oil price rather than a weakened relationship. Some proposed respecifications of the oil price yield considerable improvements, although they are not sufficient to achieve Granger causality of output unless interest rates are excluded from the VAR. There is some support for the explanation that oil prices affect the economy indirectly by inducing monetary policy responses, but this is incomplete and some evidence of misspecification remains. This paper was originally prepared for the U.S. Department of Energy Conference on the Macroeconomic Impacts of Oil Price Shocks, Washington DC, October 3-4, 1996. Financial support from Oak Ridge National Laboratory is gratefully acknowledged. I thank Andrew Filardo and Jeff Fuhrer for helpful comments. The views expressed are those of the author, and do not necessarily represent those of the Federal Reserve System. 1. Tests are of the hypothesis that all oil price coefficients are zero in 8 lag quarterly VARs which include GDP deflator inflation, the 3-month Treasury bill rate, and import inflation in addition to unemployment or output growth and an oil price. The two oil prices used are nominal log-differences and real log-levels (see the data appendix for exact definitions). The samples (before observations are lost to lags) begin in 1948:1 and run through 1995:2. 2