The investment-cash flow linkage revisited: Evidence from aggregate data and multivariate Granger-causality tests

Abstract

This paper reexamines the relationship between investment and cash flow in the United States employing an empirical approach which addresses some of the methodological problems of many previous studies. A restricted vector autoregressive model is specified for aggregate measures of investment and cash flow in which the optimal lag lengths of these as well as other variables are individually determined with the aid of Akaike's final prediction error (FPE) criterion. The model is then estimated using U.S. aggregate data (1957:I–1990:II) and the full information maximum likelihood (FIML) technique. Finally, multivariate Granger-causality tests are performed based on the estimated equations. The results suggest that, at the macro level, investment and cash flow are causally independent variables. However, consistent with both the accelerator and neoclassical investment models, output is found to exert “causal” influence on investment. The implications of these results for the current recession are discussed.