The Dual Effects of Information Technology Clusters: Learning and Selection

Clusters are hubs of innovation. Economic developers have initiated a great deal of policies to promote clusters and encourage technological advancement. Despite the efforts of previous studies to quantify the cluster innovation relationship, they failed to differentiate two major mechanisms, such as learning (firms improving in clusters) and selection (less innovative firms being forced out of clusters). This study differentiates and quantifies learning and selection using nonparametric estimation and quantile regression. Using employment and establishment data as well as patent data from Maryland from the years 2001 to 2012, this study detects both mechanisms in information technology clusters. Learning in clusters increases patent filing by 15.7% to 17.2% for firms that have ever filed for patents. At the same time, firms filing fewer than one patent per year are 40% less likely to survive in clusters than they are in nonclusters. Magnitudes of both effects should be considered in the design of cluster policies