| Author Name | INUI Tomohiko (Faculty Fellow, RIETI) / KIM YoungGak (Senshu Univertity) / KWON Hyeog Ug (Faculty Fellow, RIETI) |
|---|---|
| Creation Date/NO. | January 2026 26-J-004 |
| Research Project | East Asian Industrial Productivity |
| Download / Links |
Abstract
This paper examines how artificial intelligence (AI) development is related to firm productivity and business dynamism in Japan. Using firm-level panel data from the Basic Survey of Japanese Business Structure and Activities matched with the Institute of Intellectual Property (IIP) patent database, we use AI-related patent applications as a proxy for AI development and analyze its association with productivity, firm organization, and industry-level reallocation. The empirical analysis combines fixed-effects estimation, event-study methods, and inverse probability weighting to address potential endogeneity.
We find that firms engaging in AI development tend to exhibit higher productivity in the medium to long run, while displaying distinctive dynamics around the timing of AI development. In particular, productivity temporarily declines around the onset of AI development and subsequently improves. This pattern is consistent with adjustment costs associated with the adoption of new general-purpose technologies, but it is also compatible with endogenous selection into AI development, and therefore, the analysis does not make strong causal claims.
Importantly, the association between AI development and productivity is heterogeneous across firms. The positive relationship is more pronounced among firms with higher initial productivity levels and larger firm size, while it is weaker among smaller and lower-productivity firms, highlighting the role of complementary assets such as skilled labor and organizational capabilities.
We further show that AI development is associated with changes in firm organization and labor composition. While total employment does not decline significantly, the share of high-skilled workers increases, and the number of subsidiaries tends to decrease. At the industry level, greater AI development is associated with stronger reallocation toward more productive firms, linking AI development to business dynamism in the Japanese economy.
