Real effects of anticipated and unanticipated money: Some problems of estimation and hypothesis testing

The paper addresses two issues that arise in estimation and testing of the real effects of anticipated and unanticipated money. First it is shown that identification of the effects of unanticipated (or unperceived) monetary growth on real output is possible only if the a priori restriction is imposed that monetary growth does not depend on unanticipated (or unperceived) output. Second, the existing empirical work of Barro and others does not allow for three known channels through which money can affect real variables. These are (1) past and present anticipations of future monetary growth (the inflation tax channel), (2) expectations of monetary growth in a given period conditioned at various preceding dates (the Fischer-Phelps-Taylor effect) and (3) past and present revisions in forecasts of future monetary growth. The presence of the first of these would mean that alternative open-loop monetary growth rules have real effects. The presence of the other two implies that monetary feedback rules can have real effects. Omission of the first channel can lead to biased estimates of the effects of past anticipated monetary growth. Potentially serious observational equivalence problems are associated with the other two.