top of page

​

Working Papers

 

Jump Starting the AI Engine: The Complementary Role of Data and Management Practices. 2022. Job Market Paper (Draft available upon request)

Abstract: Artificial intelligence is transforming business and society, but evidence that AI is boosting productivity is limited. To address this gap, I construct novel measures of AI investment for a large longitudinal sample of publicly-traded U.S. firms, as well as their data and management practices. I find that the average effects of AI investment on the firms’ productivity and stock market performance are noisy. However, instrumental variable regressions suggest a strong and sizeable positive causal impact. I also find significant heterogeneity across firms: The distribution of AI investment is skewed, and the impact of AI is only salient for a subgroup of distinctive firms. In particular, AI has a positive effect for firms with more intensive data or management practices, while the marginal effect of AI may not be statistically different from zero if the complementary practices are low. These findings highlight the complementary role of data and management in leveraging AI investment to boost firms’ productivity and market value.

​

​

​

​

​

​

​

​

​

​

​

​

​

Measuring the Effects of Firm Uncertainty on Economic Activity: New Evidence from One Million Documents. 2020. (w/ Kyle Handley) NBER Working Papers 27896, National Bureau of Economic Research, Inc., Invited to Revise and Resubmit at Review of Economic Studies Ungated version

Abstract: We construct a new measure of firm-level uncertainty from analyzing the text of mandatory reports filed with the U.S. Securities and Exchange Commission. Using firm and establishment level panel data on investment margins and employment dynamics, we find that periods of high firm uncertainty (1) reduce investment rates by 0.5% and attenuate the response to shocks by about half and (2) reduce employment growth rates by 1.4% and the responsiveness to shocks by 30%. Consistent with “wait and see” dynamics, uncertainty affects new investment activity, e.g. plant births and acquisitions, more than disinvestment margins. The fall in responsiveness is higher in recent business cycles, which helps explain declining business dynamism.

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Robot Hubs: The Skewed Distribution of Robots in U.S. Manufacturing. 2022. (w/ Erik Brynjolfsson, Catherine Buffington, Nathan Goldschlag, Javier Miranda, and Robert Seamans) Working Paper, Invited for publication in AEA Papers and Proceedings 2023

Abstract: In this paper we present results on the distribution of robots in U.S. manufacturing, using new establishment-level data collected by the U.S. Census Bureau. We use the data to present a number of facts about the location and use of robots. We find that the distribution of robots is surprisingly skewed across locations, even accounting for the different mix of industry and manufacturing employment across locations. Some locations - which we call “Robot Hubs” - have many more robots than one would expect if distribution of robots was uniform, after accounting for industry and manufacturing employment. We characterize these Robot Hubs along several industry, demographic, and institutional dimensions. Finally, we also characterize the establishments with robots.

​

​

AI Adoption in America: Who, What, and Where. 2022. (w/ Kristina McElheran, Erik Brynjolfsson, Zachary Kroff, Emin Dinlersoz, Lucia Foster, and Nikolas Zolas) [The author order is randomized.] Invited to Revise and Resubmit at Journal of Economics \& Management Strategy (Draft available upon request)

Abstract:​ Artificial Intelligence (AI) has the potential to transform business and society. However, a lack of detailed data on AI use by firms has left its adoption and implications poorly understood. We analyze data from the Census Bureau’s 2018 Annual Business Survey of over 850,000 firms to establish a number of stylized facts about early AI adoption in the U.S. While less than 6 percent of firms use any of the AI technologies we measure, adoption is prevalent in a subset of distinctive firms. At least some AI is used by most firms with over 5,000 employees. AI use is associated with owners who are more educated and experienced, yet also younger, and motivated by aspirations such as bringing new ideas to market or helping the community. Firms with early markers of high-growth entrepreneurship, that innovate, and that pursue growth-oriented strategies are also more likely to use AI. AI use is concentrated in a handful of “superstar” cities. In turn, AI is conditionally correlated with significant later-stage firm growth. The concentration and growth potential of AI's leading edge portend economic and social impacts far beyond this limited early diffusion, along with a potential “AI divide” if early patterns persist.

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

​

Gerontocracy in America: The Aging of Political, Artistic and Scientific Leaders. 2022. (w/ Seth Benzell and Erik Brynjolfsson) Working Paper (coming soon)

​

Selected Working in Progress

​

The Impact of Robot Adoption, Evidence from Microdata. (w/ Erik Brynjolfsson, Javier Miranda, and Robert Seamans)​

​

Technological Change and the Life Cycles of Skills: Identifying Skill Diffusion and Substitution from Job Postings. (w/ Wang Jin and Sebastian Steffen)

​

The Relationship between Competition and Innovation under Financial Constraints. (w/ Wang Jin and Georgios Petropoulos)

​

Supply Chain Uncertainty and Firm Responses (w/ Xuan Wei and Wenjian Xu)

​

USA_AI_Adoption_Percentile.png
marginal_olstfp.png
marginal_q.png
binsreg_top_bottom_noCTRLs_CIs.png
binsreg_invest_poly_noctrl.png
crux_figures_wp_decomposition.png
crux_figures_wp_delaycaution.png
bottom of page