Subsidy Design Without Capability Screening
Entry Composition in China's Battery Sector
Authors: Guankai Zhai (Stanford), Charles E. Eesley (Stanford)
Status: Working paper, 2026. In preparation for submission to Research Policy.
Keywords: Industrial policy · Subsidy design · Entrepreneurial entry · Entry composition · Battery manufacturing · China · Difference-in-differences
In plain English
China's local governments have deployed massive supply-side subsidies — equipment grants, tax breaks, land discounts, factory construction subsidies — to attract battery manufacturers. Across 300 prefecture-level cities from 2009 to 2025, these subsidies do attract more firms: about 1.3 additional battery firms per city-quarter, a result confirmed by 500 placebo permutations of fake treatment dates.
But the marginal entrants exhibit low industry specificity. They are disproportionately non-corporate — sole proprietorships rather than limited liability companies — exit at significantly elevated rates within twelve months, and produce no comparable short-run rise in city-level battery patenting. The entry surge is transitory: it peaks in the first two post-treatment years and reverses for the earliest-treated cohorts. Crucially, new-entrant survival deficits hold even in cities where the pre-existing battery cluster is thriving post-subsidy — a pattern incompatible with healthy ancillary cluster development.
The combined pattern fits one account of the marginal entrant: subsidies that pay for industry participation, without screening on productive capability, attract participation without productive commitment. The mechanism is concentrated in cities with pre-existing battery-sector density; in thin ecosystems the same policies have little detectable effect. Subsidies amplify existing clusters more readily than they create new ones.
The findings speak directly to ongoing debates in Research Policy on innovation and industrial subsidy design, and to the question of whether China's STI apparatus is producing genuine industrial capability or expanding headcount.
Abstract
Chinese local governments have deployed massive supply-side subsidies to attract battery manufacturers. The stated goal is to build production capacity. We ask what the entry response actually looks like, and what it tells us about subsidy design. Using the staggered adoption of city-level supply-side battery manufacturing subsidies across 300 Chinese prefectures between 2009 and 2025, we estimate the Callaway–Sant'Anna difference-in-differences estimator on a quarterly panel that recovers treatment timing at sub-annual resolution.
Subsidies increase battery-sector firm entry by approximately 1.3 firms per city-quarter, a result confirmed by randomization inference. The marginal entrants exhibit low industry specificity: they are disproportionately non-corporate and survive at significantly lower rates than comparable cohorts in non-subsidized cities. The entry surge is not matched by a comparable short-run increase in city-level battery patenting. The entry effect is transitory and is concentrated in cities with pre-existing battery-sector density; in thin ecosystems the same policies have little detectable effect.
Subsidies designed without capability screening attract participation without productive commitment. The compositional effect emerges from a legitimate market response to a broadly accessible supply-side payoff, without requiring political capture. The findings speak to ongoing debates in Research Policy on the design of innovation and industrial subsidies, and to the question of whether China's STI apparatus is producing genuine industrial capability or expanding headcount.
Data and methods
The treatment-identification procedure is modeled on the systematic policy search used by Banares-Sanchez et al. (2026) for Chinese solar industrial policy, drawing on the PKULaw repository of municipal legal and policy documents. Manual coding identified 56 treated cities and 244 never-treated controls, with sub-annual treatment timing recovered for every treated city. Firm-level data come from Chinese business-registration records covering the universe of firms registered under battery manufacturing classifications. Patent data are extracted from the China National Intellectual Property Administration under IPC class H01M.
Available on request
The current draft is available to academic colleagues, policymakers, and journalists on request. Please email cee@stanford.edu. An SSRN posting and final journal submission are planned for 2026.
Related work
This paper is part of a broader research program on industrial policy and the reallocation of entrepreneurial activity:
- When External Shocks Attract Capital: Export Controls and Venture Capital Allocation in China (with Yikai Cao, Wanru Deng, and Guankai Zhai)
- Industrial Policy Reshapes Venture Capital Allocation and Growth Trajectories in Climate Technologies (with Yikai Cao and coauthors)