Research

We study a core mechanism underlying state entrepreneurship: how bureaucratic promotion incentives shape public officials’ willingness to bear risk. We exploit a sharp institutional feature of the cadre system in China — city mayors become effectively ineligible for further promotion once they reach age 58 — to identify the causal effect of promotion incentives in a regression discontinuity design. We proxy risk preferences using revealed behaviour in government VC investments, focusing on investments that are typically riskier: early-stage deals, deals without private co-investors, and investments in new local industries. We find officials with promotion incentives undertake substantially more government VC investments, with the increase concentrated in these high-risk categories. These effects are robust to balance tests, permutation placebos, and alternative specifications. We then examine longer-run consequences using patent outcomes of government VC-backed firms over the subsequent five years. On average, greater risk-taking does not yield robust improvements in innovation. We further document heterogeneity in treatment effects. Longer-tenured mayors respond less strongly to promotion incentives and generate higher-quality innovation, while less capable mayors respond by taking greater risks, which worsens the average innovation performance of government VC–backed firms.

Presentation(s): AOM (2026 scheduled, best papers award), Oxford Applied Micro Seminar (2026), Oxford Development Workshop (2026)

We study the effects of a mechanical change in industrial policy instrument choice by exploiting China’s 2014 Budget Law, which compulsorily curtailed local subsidies and shifted support towards government venture capital (VC). Using an event-study design with a continuous treatment across prefecture-level cities that differed in pre-amendment subsidy intensity, we estimate the consequences of this substitution. We find the shock increased government VC investment, especially in late-stage deals. It also precipitated a decrease in firm entry—mainly in manufacturing—while entrant quality rose and innovation output was unaffected. The composition of post-shock government VC matters: a higher early-stage or stand-alone share increased entry and amplified innovation. We also find that government VC crowded in private VC, with the effect strongest in locations with a higher share of early-stage government VC investments and lower ex-ante private capital availability. Overall, replacing subsidy tool with equity tool tilted industrial policy towards selectivity and mobilisation of private finance.

Presentation(s): AEDC (2026, scheduled)*, CSAE (2025, scheduled), SALDRU-ERSA (2025), Oxford Development Therapy (2025)