Research
Job market paper
Endogenous Exit and China’s State-Sector Transformation
Abstract. Between 1998 and 2007, China’s GDP grew by nearly 10 percent annually. Song, Storesletten and Zilibotti (2011) emphasize resource reallocation from state-owned enterprises (SOEs) toward more productive private firms as a central driver of this growth. Yet Hsieh and Song (2015) highlight substantial productivity growth within the shrinking state sector. We argue that exit should be evaluated against the productivity growth that would have occurred had exiting firms continued operating. Conventional productivity decompositions and the analysis of Hsieh and Song (2015), however, overlook the subsequent productivity deterioration these firms might have experienced, potentially understating the contribution of exit. We develop a framework that recovers this unobserved potential productivity distribution and find that selective exit explains 58 percent of SOE productivity growth. Our year-by-year analysis shows that selection dominates during early “grasp the large, let go of the small” reforms, while incumbent improvements become more important following later governance reforms. We then estimate a structural model showing that reducing government financial support for struggling SOEs promotes exit and reallocates resources toward more productive private firms, whereas governance reform improves incumbents but can impede reallocation. Our findings suggest that state-sector productivity growth reinforces, rather than challenges, the importance of reallocation.
Presentations: Midwest Macro Meeting (Cleveland, 2025); Women in Macro Meeting (Chicago, 2026); North American Summer Meeting of the Econometric Society (Atlanta, 2026); HHEI Conference (2026).
Working paper
A Friction-Robust Estimator of Labor-Capital Substitution
Abstract. The elasticity of substitution between capital and labor is important for long-run economic growth. Conventional estimators infer this elasticity from firms’ input responses to relative factor prices. Capital frictions can distort these responses and bias estimates even with exogenous instruments. We develop a framework that exploits firms’ optimal labor choices to jointly identify capital–labor substitution and demand elasticity without imposing a frictionless capital optimality condition. Monte Carlo experiments demonstrate the performance of our estimators in economies with capital frictions and policy distortions. Applying the framework to China’s 2004 Industrial Census yields greater capital–labor substitutability and lower markups than conventional methods, highlighting the importance of accounting for capital frictions in estimation.