Empirical research typically builds on the findings of prior studies. New tests rely on previously identified determinants of the outcome variable studied as controls. Authors of a forthcoming RCFS paper, “Disregarding the Shoulders of Giants: Inferences from Innovation Research,” David Reeb and Wanli Zhao observe this process is often ad hoc. Using past research on corporate innovation as a platform, authors show that only a small subset of proposed determinants of innovation provide material, independent information about patents and citations. To do so, they use a data-driven approach – i.e., utilize various machine learning techniques to evaluate these covariates that have been argued or shown to affect innovation. After identifying the subset of determinants with high explanatory power (HEP), the authors show how including these HEP variables would have changed inferences in recent empirical studies and discuss how excluding them might affect future research. Commonly used econometric methods, including fixed effects and plausible shocks, do not change their findings on these previously identified innovation determinants. This important paper offers researchers a framework to select control variables for future research, rather than arbitrarily choosing a subset of control variables from prior studies.
Spotlight by Isil Erel
Photos courtesy of David Reeb and Wanli Zhao