Different kinds of literature have suggested inconsistent answers to motivations behind a firm’s publications strategy, which, unlike patents, does not afford property rights but rather increases the chances of expropriation. A key unanswered question in the literature is: do firms use publications as an instrument to recruit scientists? Drawing on the innovation literature, I argue that when a labor market shortage affords scientists higher bargaining power, firms tend to disclose more internal R&D via publications. To test this hypothesis, I use a novel dataset of 200 million job posts and 1.7 million firm-level publications. For identification strategy, I exploit the unanticipated rise of deep learning in AI, which resulted in a shortage of AI talents, and thus, increased scientists’ bargaining power. I demonstrate how human capital bargaining power increases firm-level R&D disclosure. This relationship is particularly salient when firms need highly trained scientists (e.g., PhD holders). Further, I document that AI job posts increase the number of AI publications, but only in the same fields of increased demands. Finally, using a novel methodology, I document that labor shortage did not increase the number of patent-paper pairs or simultaneous disclosure of the same research projects in both patents and papers.
Contact: Michael E. Rose