Clusters, Heritage and the Microfoundations of Spillovers - Lessons from Semi-Conductors

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

Knowledge industries tend to cluster geographically, which has long been thought to promote spillovers of knowledge through local employees of competing firms exchanging know-how and changing employers. Employees of incumbent firms also sometimes found their own spinoff firms, which can be a source of knowledge spillovers. Intellectual Merit: The main purpose of the proposed work is to assess the strength of these alternative spillover mechanisms and their effect on firm performance in the context of the semiconductor industry, which famously clustered in Silicon Valley and secondarily in Boston, New York, and Los Angeles. Sorting out the importance of these alternative mechanisms is critical; if clusters promote knowledge exchange and employee mobility then clustering is socially beneficial whereas if spinoffs are the primary conduit for clustering then the benefits of clusters do not extend beyond spinoffs and their 'parents'. Using an annual buyers' guide, data are collected on the location and products of all semiconductor producers from the inception of the industry through 1987. Firm sales data are collected to identify the market leaders. The pre-entry experience of each producer is traced using the annual buyers' guides to identify prior producers of other electronics producers and trade journals and web sources are used to identify spinoffs and their parents. The patents assigned to each semiconductor producer and the inventors of the patents are collected to trace the mobility of the inventors over time. An analysis is conducted of the influence of location and other factors on whether semiconductor producers entered producing semiconductor products at the technological frontier, and if not whether they later shifted to produce frontier products. The determinants of the mobility of inventors are analyzed, including whether mobility was greater for inventors in clusters. Various new semiconductor products were developed in the sample period, and analyses are conducted of the factors influencing whether firms adopted new products, including the role of location. All of these decisions are in turn related to the performance of firms, as reflected in their sales and longevity. Broader Impact: Understanding whether clusters improve firm performance and the main mechanisms behind regional knowledge diffusion can yield important broader impacts for both strategy and policy. If direct knowledge sharing and inter-firm employee mobility are promoted by clustering and in turn contribute to clustering, then clustering has clear welfare benefits and would also benefit firms located in clusters. Alternatively, if spinoffs are the primary conduit for clustering, then external economies do not go beyond spinoffs and their 'parents' and clustering provides no social benefits per se nor any private benefits to firms located in clusters.