Sources of Asymmetry in Production Factor Dynamics

The aim of this paper goes some steps beyond the specification and estimation of production technologies and nonlinear dynamic optimization rules by instrumental variables methods as initially proposed by Kennan (1979). The question, that remains unanswered by the mere estimation of the structural parameters, is which sources of asymmetry contribute most to the shape of the factor input cycles. To investigate that question, we solve a nonlinear rational expectations model using an extended version of the parameterized expectations algorithm (PEA) put forward by Den Haan and Marcet (1990) and Marcet and Marshall (1994). When we completely specify the generating processes of the model's forcing variables, the PEA solution method enables us to identify the importance of the two possible sources of asymmetry in factor input dynamics that we consider in this paper. One source is behavioral or internal asymmetry (asymmetry in the propagation mechanism) that results from facts that are typically incorporated in what is stated as the costs of adjustment (cf. Hamermesh and Pfann, 1996a). The second source considered is external non-linearity that is induced by the variables that drive the model. External non-linearity could, for example, be due to nonlinear pricing rules, including for example downward rigidities, or to asymmetrically distributed taste or productivity shocks. The differences between those two possible sources are important because the first, behavioral, source is partly endogenous and can be controlled for by the firm, while the second, external, source cannot. The estimates of the model's structural parameters are based on undetrended seasonally adjusted quarterly time series data of the Netherlands manufacturing sector for the period 1971.I - 1990.IV. Economic growth is considered to be a common trend that results from accumulated productivity shocks. We find that behavioral asymmetry, resulting from asymmetric adjustment costs, contributes substantially to the non-linearity of factor input dynamics: 53 percent of the total second order effect for capital and 48 percent for employment is accounted for by behavioral asymmetry. Our analysis shows that the significance of this finding is not simply an artifact due to the presence of external non-linearity. It is an important structural phenomenon that has to be accounted for when analyzing factor input dynamics at an aggregate level