Evolving localization patterns of company foundations Evidence from the German MST-industry

We investigate company foundations in the German micro technology industry (MST) by means of a spatial-temporal micro-geographic analysis. Our dataset includes the exact location (latitude, longitude) of all German MST firms, relevant research institutions and MST-patent owners including their date of inception for a timeframe of 12 years. Furthermore, we obtain employment data at a municipality level for three related industries. In order to deal with our unusual detailed data, we develop a new distance-based framework for a logistic regression that enables the usage of micro geographic data that is free of any spatial aggregation. A temporal component is integrated by means of a Markov chain so that our model allows both, a micro spatial and a micro temporal analysis. The model detects spatial dependencies on a firm level instead on an aggregated regional level. This allows an econometric modeling without the Modifiable Areal Unit Problem (MAUP), a well-known problem in spatial econometrics. Recently, the MAUP has been tackled by new metrics such as the Duranton-Overman-Index that however only allow the observance of (co-) localization but cannot include multiple variables. Our framework for a logistic regression is able to fill this methodological gap as it is distance-based, allows the observation of multiple variables and gives results for global models. Locations of company foundations are investigated with respect to their spatial proximity to similar firms, patent owner, related industries and research institutions and are benchmarked with the overall distribution of company foundations in Germany. We perform one investigation for the whole timeframe and three separate investigations with a three years period in order to test for temporal caused dynamics. The empirical results reveal some interesting findings that stand in line with well-established theories: Spatial proximity to other MST firms and to relevant research institutions has a clear positive influence on where a new firm is founded. This might be due to positive local spillover or to spinoffs from firms or research institutions. While this confirms that MST-firms depend on a high local innovation capacity, the proximity to owners of patents and to qualified workers in related industries has little or even a negative influence. Furthermore, the influence of the different actors is not constant over time but evolves with the industry's life cycle. The increasing importance of spatial proximity to firms stands in line with the self-augmenting process theory, while the parallel decreasing importance of proximity to research institutions suggests a shift from tacit to explicit knowledge. The analysis shows that the MST industry becomes more self-contained and that proximity to related industries is of decreasing importance.