Orphan drug policies target rare-disease indications, but firms make development decisions across drug portfolios that often span rare and common diseases. We study how these portfolio-level decisions shape the effectiveness and targeting of incentives under the U.S. Orphan Drug Designation (ODD) program. Using a novel dataset that tracks development histories and designation status across indications within each drug, we show that approximately 70 percent of orphan drugs are developed for both rare and common indications, with most mixed portfolios beginning with a common indication. At the drug level, ODD and broader development scope are associated with greater survival and entry into new indications, although the incremental ODD advantage declines as portfolios broaden. For rare indications that receive designation, however, the survival advantage does not diminish with portfolio scope, while the association between designation and entry is stronger in drugs with broader portfolios and larger markets. Rare indications in mixed portfolios also progress approximately 2.6 years faster than those in exclusively rare-disease portfolios, consistent with spillovers across indications. Motivated by these findings, we develop a dynamic model of multi-indication drug development to distinguish economies of scope from policy-induced investment and strategic portfolio choices. Structural estimation quantifies when orphan incentives induce otherwise unprofitable development, subsidize projects that would proceed without support, or remain insufficient to induce investment. Counterfactual analyses evaluate whether tailoring incentives to portfolio scope and commercial potential can improve support for rare-disease innovation. Our framework highlights the importance of accounting for spillovers and economies of scope in firms’ portfolio decisions when designing targeted incentive policies.

This paper studies how patent trading affects innovation in the U.S. pharmaceutical industry. I construct a novel dataset linking the timing of patent transfers to drug development stages and show that 82% of trades occur before launch, and that such trades significantly increase success rates. To interpret these patterns, I develop and estimate a dynamic model where firms differ in stage-specific expertise and can trade patents under search frictions and transaction costs. The model reveals that transferring patents to experienced firms raises success rates and innovation value, but trade frictions hinder efficient transfers. Counterfactuals show that reducing transaction costs boosts launch rates and innovation value, while targeted subsidies at specific stages outperform uniform ones in efficiency and cost-effectiveness.

This paper studies how patent trade affects subsequent innovation. Using U.S. pharmaceutical patent, reassignment, and firm-level data, I develop and estimate a dynamic model of patenting, buying, and selling that allows internally generated and acquired patents to affect future innovation differently. The estimates show that internally generated patent capital strongly supports subsequent innovation, while the effect of acquired patents varies across firms with different commercialization capabilities. Transaction costs significantly affect firms’ patent-trading decisions. A 50 percent reduction in transaction costs increases patent trading for all firms but raises innovation mainly among innovation-specialized firms, particularly smaller ones. At the industry level, the long-run patent stock increases by about 0.5 percent. The results show that patent-market frictions shape not only the reallocation of existing inventions, but also the location and amount of future innovation.

Artificial intelligence promises to improve drug discovery, yet its contribution to pharmaceutical R&D productivity remains uncertain. This paper examines how the employment of AI-skilled scientists relates to the scale, novelty, and progression of drug development, and how these relationships vary with firms’ scientific capabilities and resources. We construct a new dataset linking scientists’ employment histories and job postings from Revelio Labs to drug development pipelines from Cortellis, supplemented by firm financial and funding information. This linkage allows us to trace scientific AI expertise alongside drug development milestones and compare outcomes across firms and development stages.
Preliminary estimates indicate that greater employment of AI-skilled scientists is associated with larger pipelines and more novel drug candidates. The association with pipeline size is strongest among leading developers, while evidence of faster progression from discovery to preclinical development remains mixed. These patterns suggest that the relationship between AI expertise and research activity may differ from its relationship with subsequent development progress. Ongoing analysis addresses endogenous AI hiring, unobserved firm capabilities, and selective allocation of AI resources to distinguish AI’s contribution from underlying differences in firms’ research prospects. The study contributes evidence on the conditions under which AI may improve innovation productivity and whether its benefits broaden opportunities for biotech entrants or reinforce established developers’ advantages. The findings inform workforce investment and innovation policy in the biopharmaceutical industry.

Open Science Partnerships (OSPs)— mission-oriented, public-private collaborations that share research outputs with minimal IP claims—are attracting growing policy interest as mechanisms through which to advance biopharmaceutical drug development. This article argues that OSPs generate positive economic effects through three underappreciated mechanisms: they can reallocate early-stage scientific risk and change firms’ investment incentives; they generate longer-run spillovers through reusable platforms, human capital, and networks; and they build shared standards and infrastructures that shape participation in the innovation ecosystem. Because many effects emerge later and outside the original partnership, conventional output metrics systematically understate OSP value. To properly evaluate the impact of OSPs on drug development, the article calls for the development of linked evidence connecting partnership activity to downstream scientific, commercial, and clinical outcomes.