Kaleckian vs. Marxian specifications of the investment function: Some empirical evidence for the US

Following Lavoie et al. (2004), this paper empirically assesses four investment functions closing the Kaleckian baseline model in the long-run: (a) the Naive-Kaleckian specification without any long-run adjustment; (b) the Intermediate-Kaleckian specification with an endogenous adjustment of the normal utilization rate; (c) the Hysteresis-Kaleckian specification with an additional endogenous adjustment of autonomous investment; and (d) the French-Marxian specification with an exogenous normal utilization rate and endogenous autonomous investment. Confronting these specifications with data of the US manufacturing sector, we compare them with respect to the plausibility of the parameter estimates, the goodness of fit, the parameter stability, the out-of-sample performances and relative encompassing. We find the Intermediate-Kaleckian specification to be superior. For the Hysteresis-Kaleckian specification, we get implausible results which contradict Lavoie et al. (2004). Yet, their estimates seem to be biased due to endogeneity issues.