There is a broad consent that the extent, quality as well as the success of innovation activity is critically dependent on the generation and the application of knowledge. It is also widely recognized that innovation processes have a pronounced regional dimension and that the conditions for innovation activity may differ considerably between geographic areas. We investigate the contribution of different inputs, particularly different knowledge sources, on regional patenting output in the framework of a knowledge production function. Earlier work (Fritsch & Franke, 2004) suggests that there are considerable differences with regard to the role of different types of knowledge for innovation activity. These different types of knowledge have different degrees of localization. We expect that the impact of the knowledge sources differs by field of as well as according to the type of institution (university vs. non-university public research institution). .One may, therefore, expects that the technology opportunity differs across sectors and geographical areas. However, little is known about such differences. The innovation indicator used in our analysis is the number of patents. The knowledge sources included are: •R&D employment, •R&D employment in small firms and large firms respectively, •Qualification of the regional workforce, •Qualification of employees in small firms and large firms respectively, •Size (number of employees, budget) of public research institutions by field of research. •Amount of research funds from private firms by field of study. •Amount of external research funds from public departments by field of study. •Amount of external research funds from the German Science Foundation (DFG) by field of study. The contribution of these knowledge sources is tested systematically on the level of German districts (Kreise). Information on the different knowledge sources is available for each district on a yearly basis. We test the influence of the knowledge sources in different specifications by including the respective information for the particular region and for adjacent regions. This will reveal the role of local knowledge and of interregional knowledge spillovers for innovation behavior. We expect that technology performance of regions is a path dependent process. The analysis is performed for German regions in the 1990s. Information on the yearly number of patents relates to the years 1995-2000. regional employment is taken from the German Social Insurance Statistics. Information on universities and their funds are taken from the German Hochschulstatistik which has been especially prepared for this analysis.
