On the dynamics of product and process innovations: A bivariate random effects probit model

Based on a stochastic dynamic model of a firm's optimal innovative behavior we derive a simultaneous equation system for product and process innovations with intertemporal spillover effects. We estimate various versions of the model with dichotomous Innovation data at the firm level by using a bivariate dynamic random effects probit model. The data set, provided by the Ifo-Institute, covers the period between 1981 and 1989 and includes 586 firms of the West German manufacturing sector. It turns out that a firm's probabilities of product and process innovations depend positively on dynamic spillover effects even if one controls for firm size, market concentration, demand expectations, labor cost, unobserved heterogeneity and potential endogeneity of the explanatory variables.