Evaluation and Learning in R&D Investment
A. Frankel,D. Papanikolaou,Joshua L. Krieger,Danielle Li
DOI: https://doi.org/10.2139/ssrn.4315940
2023-05-01
Abstract:We examine the role of spillover learning in shaping the value of exploratory versus incremental R&D. Using data from drug development, we show that novel drug candidates generate more knowledge spillovers than incremental ones. Despite being less likely to reach regulatory approval, they are more likely to inspire subsequent successful drugs. We introduce a model where firms are better able to evaluate the viability of incremental drugs, but where investing in novel drugs helps firms learn about future projects. Firms appear to put more value on evaluation versus learning, and those patterns are in-part driven by the appropriability of spillovers. Alexander P. Frankel Stanford Graduate School of Business 518 Memorial Way Stanford, CA 94305 afrankel@chicagobooth.edu Joshua L. Krieger Harvard Business School Rock 213 Boston, MA 02163 jkrieger@hbs.edu Danielle Li MIT Sloan School of Management 100 Main St, E62-484 Cambridge, MA 02142 and NBER d_li@mit.edu Dimitris Papanikolaou Kellogg School of Management Northwestern University 2211 Campus Drive, Office 4319 Evanston, IL 60208 and NBER d-papanikolaou@kellogg.northwestern.edu Scientific breakthroughs often build on earlier research efforts, including those that initially end in failure.1 Yet despite the importance of exploration and learning for innovation (Romer, 1990; Aghion and Howitt, 1992; Scotchmer, 1991; Furman and Stern, 2011), research and development projects are often evaluated only on their potential for direct success. For instance, traditional net present value (NPV) calculations multiply a project’s probability of success times its profits if successful, assuming zero benefits if the project does not succeed. In this paper, we examine how firms prioritize learning spillovers in their present R&D decisions.2 Our empirical work focuses on pharmaceutical R&D, a setting where breakthrough innovations can lead to enormous gains in welfare, where cumulative learning is important, and where failure is common. We first develop a new measure of cross-product knowledge spillovers. Using this measure, we show that novel drug candidates generate greater spillovers, and that much spillover value comes from drug candidates that themselves fail. Next, we use these facts to motivate a model of R&D investments. In our model, a firm must decide whether to invest in developing a drug candidate, which can either be incremental or novel. Firms are better able to evaluate the success likelihoods of incremental drugs because they are related to ideas that have previously been investigated, enabling them to screen out weak candidates. Firms know less about novel drugs, but by investigating a new area, they generate knowledge spillovers and learn about the success of future related drugs. Using this model as a guide, we provide empirical evidence that firms prioritize evaluation over learning, i.e. they are more reluctant to invest in novel relative to incremental drugs. Following Krieger, Li, and Papanikolaou (2021), we measure a drug candidate’s novelty by comparing its molecular structure with that of previously developed candidates. In this paper, we complement these backward-looking molecular linkages by defining new measures 1For example, in 2013, a large clinical trial for one of the most promising HIV vaccine candidates to date was halted due to lack of efficacy. Known as DNA/rAd5, the proposed vaccine sought to prime the immune system by injecting DNA plasmids that code for the production of protein structures also present in HIV. For this approach to be effective, the injected DNA must successfully enter a cell’s nucleus so that its instructions can be carried out. Yet, in postmortems, scientists worried that DNA plasmids were too easily destroyed before reaching the nucleus. These concerns suggested two avenues for follow on research: vaccines based on mRNA (which need only enter a cell’s cytoplasm to be effective) and delivery mechanisms that protect genetic material (Harris, 2021). Both these innovations are present in Moderna and Pfizer-BioNTech’s vaccines for SARS-CoV-2, which deliver mRNA wrapped in lipid nanoparticles. 2Risk-adjusted NPV (rNPV), a modern tool widely adopted by pharmaceutical firms, refines the traditional NPV calculation by adjusting for a drug candidate’s likelihood of progressing through each stage of R&D. Both the traditional the risk-adjusted approaches would assign zero value to the failed DNA/rAd5 vaccine of footnote 1, though, ignoring its potential to inform future vaccine development efforts.
Economics,Business