ABSTRACT

Using innovation research as a platform, we show that studies proposing new covariates rarely include previously identified determinants. Yet, only a sparse set of these hypothesized economic variables provide material, independent information about patents and citations. We then show that the inferences of previous studies can differ when including or excluding key economic determinants of innovation. Additional tests reveal that common solutions, including fixed effects and plausible shocks, do not mitigate the need to include previously identified innovation covariates. Rather than randomly selecting control variables, our analysis offers a framework for researchers to incorporate prior economic determinants of innovation.