Does Concentration Deliver? Evidence from Japan's Dual-Use R&D Program

• In Event: Innovation Policy, R&D Systems, and Science Funding

Abstract

This study asks whether the observed concentration of defense R&D funding aligns with innovation efficiency. While the Japanese government positions national research institutes as central drivers of innovation (7th Science, Technology, and Innovation Basic Plan), prior literature suggests that private-sector R&D may exhibit higher efficiency in translating inputs into measurable outputs, such as publications and patents. This raises an empirical question: does concentrated funding toward specific institutions generate superior innovation outcomes, or does it instead reflect institutional constraints on participation?

To address this question, this paper analyzes the policy outcomes of Japan's Security Technology Research Promotion Program, a competitive funding scheme launched in FY2015 by an agency of the Ministry of Defense to support advanced dual-use basic research across universities, national research institutes, and private firms. Japan provides a critical case for this analysis, as its science and technology system has historically been institutionally separated from defense activities, generating resistance within academia to defense-related funding. As a result, participation in the program — covering 239 funded projects by FY2025 — has been uneven across institutional types, with national research institutes exhibiting substantially higher participation rates than universities and private firms, producing a highly concentrated distribution of funded projects across a small number of institutions.

The analysis proceeds in three steps. First, funding concentration across institutional types is examined using the Herfindahl–Hirschman Index (HHI) and Gini coefficients. Second, Negative Binomial regression models are employed to estimate the association between project-level funding and research outputs. Third, Data Envelopment Analysis (DEA) is used to assess the relative efficiency of major recipient institutions. To account for heterogeneity, the analysis further examines whether these relationships vary across research fields. The paper constructs an original dataset covering all funded projects from FY2015 to FY2025 by systematically compiling administrative records and project-level outcome data from government reports, including information on publications and patent applications.

Preliminary findings reveal several notable patterns. While national research institutes receive the largest number of projects, private firms receive the largest share of total funding. Larger-scale projects are associated with disproportionately higher outputs in both publications and patents. Efficiency varies systematically by institutional type: private firms exhibit substantially higher patent output per unit of funding, whereas universities perform more efficiently in producing publications. National research institutes do not demonstrate a clear comparative advantage across these outcome measures. These patterns suggest a systematic misalignment between funding allocation and institutional comparative advantage. The program has generated measurable outputs, including 398 publications and 573 patent applications; however, none of the funded projects have to date been linked to defense procurement outcomes (Japan's Ministry of Finance) , raising questions about the program's downstream effectiveness.

These findings contribute to broader debates on the governance of public R&D and the role of state-led funding in shaping innovation outcomes in dual-use research contexts.

Author

• Makoto Tsujiguchi, Keidanren Policy Research Institute
  Presenting Author