Abstract

This dissertation studies how the allocation of scarce research talent across heterogeneous firms shapes innovation and aggregate productivity growth. Although advanced economies have devoted a rising share of resources to R&D, productivity growth has stagnated. We argue this trend reflects not a shortage of research effort but a misallocation of research talent, which operates through the quality composition of innovation or the diffusion of knowledge rather than the scale of R&D investment. The two chapters develop complementary Schumpeterian growth models that formalize distinct channels of this misallocation and discipline them with Korean firm-patent matched data. Chapter 1 studies how the allocation of scarce research talent across firms shapes the quality composition of innovation. We build a Schumpeterian growth model with heterogeneous innovation in which the probability that an innovation becomes a radical breakthrough is determined endogenously by firms' allocation of scarce skilled researchers. R&D subsidies sustain low-capability firms in R&D activities, raise the skill premium, and crowd out breakthrough investment by more productive firms. This mechanism operates through the quality composition of innovation rather than its frequency, and is amplified when subsidies are size-dependent, as in Korea. Calibrating the model to Korean firm-patent data, we find that eliminating R&D tax credits has a negligible growth effect as quantity losses are offset by quality gains, while equalizing tax credits across firm sizes raises growth by 0.04 percentage points entirely through the quality channel. Expanding the supply of high-skilled researchers by 25% generates a 0.35 percentage point growth gain, with 80% operating through quality improvements that are absent in models with exogenous radical innovation probability. Chapter 2 studies a distinct margin of misallocation that arises from defensive hiring by market leaders, who employ researchers to block knowledge diffusion rather than to innovate. In a step-by-step innovation model with a sector-specific researcher market, leaders' blocking demand simultaneously reduces the knowledge diffusion rate and crowds out followers' productive R&D. Calibrated to Korean firm-patent matched data, the model implies that leaders absorb 87.6% of sector-specific researchers for blocking, reducing equilibrium diffusion by 19%, with roughly one-fifth of aggregate R&D expenditure generating no innovation. Expanding research resources alone cannot mitigate this distortion—only policies that reallocate researchers away from defensive activities can. Together, the two chapters identify distinct but complementary channels through which the allocation of research talent—across firms of different innovative capability, and between productive and defensive uses within sectors—shapes the quality and diffusion of innovation. They share a unifying message: aggregate growth depends not only on the supply of R&D resources but on how research talent is allocated across firms and activities.

Committee Chair

Yongseok Shin

Committee Members

Francisco Buera; Mariagiovanna Baccara; Oksana Leukhina; Rodolfo Manuelli

Degree

Doctor of Philosophy (PhD)

Author's Department

Economics

Author's School

Graduate School of Arts and Sciences

Document Type

Dissertation

Date of Award

4-27-2026

Language

English (en)

Download

Available for download on Sunday, April 23, 2028

Included in

Economics Commons

Share

COinS