Competing macro-hypotheses in the United States: A Kalman filtering approach

Abstract

The empirical validity of alternative views about the short-run determinants of real output growth in the United States is investigated. A nested framework is used to test the macro rational expectations (MRE) hypothesis directly against two competing hypotheses - the neo-Keynesian view that anticipated and unanticpated monetary policy both matter,and Friedman's (1977) proposition that increased inflation uncertainty reduces real output at least temporarily. The unobservable explanatory variables are obtained from time-varying-parameter models of inflation and money growth, which generate forecast errors and their conditional variances consistent with rational expectations under a continuously chaning policy regime. The empirical results strongly support Friedman's view. The MRE hypothesis must be rejected since both anticipated and unanticipated monetary changes matter. The results prove robust across different model specifications and estimation techniques.