Money and the business cycle: Another look

A study examined the role of money within a vector autoregressive (VAR) framework. To investigate the importance of omitted variable-types and omitted lags, 4 models were developed that differed with respect to some of the variables included. In each of the models, common lag lengths of 4 and 8 were considered. Recent VAR evidence has suggested that monetary shocks are not important for output fluctuations. The current study attributed these results to model misspecification due to both omitted variables and inappropriate lag lengths. It was found that money growth is Granger-causal prior to real output growth, with money innovations accounting for 22% of its forecast error variances, as opposed to 4% in Sims (1980), 0.3% in Fackler (1985), and 7% in Litterman and Weiss (1985). The fact that longer lags are important in capturing monetary influences is consistent with the established notion that monetary policy works with long and variable lags. The evidence suggests that research emphasis should again be placed on the question of why money leads output