On nonparametric estimation of the Schumpeterian link between innovation and firm size: Evidence from Belgium,France, and Germany

This paper analyses the Schumpeterian link between innovative activity and firm size by means of the nonparametric Nadaraya-Watson estimator and of the partially linear approach by Speckman. Four data sets referring to the manufacturing industries of three European countries are available for the analysis. We demonstrate how nonparametric methods can produce more reliable conclusions than conventional methods. For this purpose, the roles of bandwidth choice, wild bootstrap, density estimation and trimming are studied. For the German data set of 1984 and for the French data set we find that small firms and large firms are more innovative than firms of intermediate size while the relation is rather hump-shaped for Germany 1989 and decreasing for Belgium. Including an additional parametric component into the estimations based on the French data contributes considerably to the explanation of innovative activity without affecting the U-shaped link between innovation and firm size.