Modern Innovation Theory suggests the concept of localized technological progress which criticizes the traditional neoclassical concept of the sources and effects of technological change, the firm's choice of technique and as a consequence the concept of the production function. In a short-cut, localized technological progress refers to the fact that new technological opportunities are generated and explored by firms under different, firm specific conditions which leads to intra-sectoral heterogeneity in the production techniques applied and the technical efficiency performed. Neoclassical formulations just contrary further the view that firms adjust in a rather similar fashion to some externally given and growing technological opportunities represented by a known parametric isoquant which in equilibrium implies technical and performance homogeneity.
From the empirical point of view the neoclassical approach applies well elaborated statistical procedures to estimate sectoral parametric production functions. Those methods principally rely on the assumptions of equilibrium and firm homogeneity; industry structures nevertheless to be observed are then interpreted mainly as a transitory phenomenon or can be traced back to some market breakdowns. A switch of the theoretical basis towards modern innovation theory does necessarily imply that for the detection of industry structures the neoclassical tools are not appropriate any more. A smooth production function applied by all firms of a sector does not exist.
Based on this, our paper presents a non-parametric method, Data Envelopment Analysis (DEA), suitable to detect industry structures based explicitely on the technical heterogeneity of firms and therefore of the respective production functions. With this method we show how vertical and horizontal intra-sectoral structures can be determined and how one can track the development of those structures over time. In addition we provide (a) an improved procedure to detect "dominant techniques" and (b) a modified DEA-model which helps to resolve the problem that in the standard version of DEA all best-practice firms (techniques) are considered as equally performing. Those new aspects help to get an even better account of industrial dynamics.
Contact:
Uwe Cantner and Georg Westermann
University of Augsburg
Department of Economics
Universitätsstr. 16
86159 Augsburg