(with Zhao Chen, Zhikuo Liu, Juan Carlos Suarez-Serrato and Daniel Yi Xu), conditionally accepted, Review of Economic Studies
A universal fact of firm-level data is that investment is lumpy: firms either replace a considerable fraction of their existing capital (spike) or do not invest at all (inaction). This paper incorporates the lumpy nature of investment into the study of how tax policy affects investment behavior. We show that tax policy can directly impact the lumpiness of investment and that the effectiveness of tax incentives in stimulating investment depends crucially on interactions with investment frictions. We illustrate these results by studying one of the largest tax incentives for investment in recent history: China's 2009 VAT reform. Using administrative tax data and a difference-in-differences design, we document that the reform increased investment by 36% and that this effect is driven by additional investment spikes. We then simulate the fiscal cost of stimulating investment through different tax policies using a dynamic investment model that is consistent with the reduced-form effects of the reform. Policies that directly reduce the likelihood of firm inaction (e.g., investment tax credits) are more effective at stimulating investment than policies that only reduce the tax cost of investment (e.g., corporate income tax cuts).
Information and Communication Technology and Firm Geographic Expansion (Job Market Paper)
Information and communication technologies (ICT) can widen firms' geographic span of control by reducing internal communication costs. Combining comprehensive establishment-level datasets with ownership linkages, geographic locations, and ICT usages, I document that US manufacturers have expanded geographically over the past three decades and that firms with more advanced technologies (i.e., Intranet) have larger geographic coverages. To estimate the effects of ICT on firms' geographic span of control, I exploit a historic event in ICT developments in the United States: Internet privatization in the early 1990s. Results suggest that better access to ICT helped firms expand to more counties and that Internet privatization accounted for around 38% of the total increase in firms' geographic coverage during 1995-2007. Using a model where firms endogenously adopt advanced technologies and choose multiple production locations, I estimate that the Internet privatization reduced the manufacturing price by 1.76%. The counterfactual analysis highlights the importance of multi-unit production in evaluating the benefits of ICT improvements. Compared to a trade-only model, a model with multi-unit firms predicts that efficiency gains are larger and more geographically dispersed.
WORK IN PROGRESS
Knowledge Diffusion Through Multi-Unit Firms (with Rahul Gupta)