Innovation and institutional ownership revisited: an empirical investigation with count data models

Abstract

By discriminating between a lazy manager and a career concerns hypothesis, Aghion et al. (Am Econ Rev 103(1):277–304, 2013. doi:10.1257/aer.103.1.277) try to disentangle the link between innovation and institutional ownership. Citation-weighted patent counts are used as a proxy for innovation, which motivates the use of count data models. A replication in a narrow sense confirms their empirical results which are mainly based on Poisson models (i.e., with a single set of regression coefficients). However, when extending the model framework by count data hurdle models, it is shown that the two hurdle parts do not coincide—as they should under the Poisson model—but lead to different results. Nevertheless, a remarkably stable positive correlation of citation-weighted patents and institutional ownership across all model specifications can be shown.