Early Joiners and Startup Performance 

Using employee-employer matched data on all US startups and information on the premature death of workers, we show that early joiners---non-founder employees in the first year of a startup---play a critical role in explaining firm performance. Losing an early joiner has a large negative effect on firm size that persists for at least ten years. 

Review of Economics and Statistics, Forthcoming (with Joonkyu Choi, John Haltiwanger, and J. Daniel Kim)

Press: WSJ

Slides: CAED 2019, Online Appendix 

Tracking Firm Use of AI in Real Time: A Snapshot from the Business Trends and Outlook Survey (2024) 

We provide new, real-time estimates of current and expected future use of AI for business purposes based on the Business Trends and Outlook Survey for September 2023 to February 2024. During this period, bi-weekly estimates of AI use rate rose from 3.7% to 5.4%, with an expected rate of about 6.6% by early Fall 2024. AI users often utilize AI to substitute for worker tasks and equipment/software, but few report reductions in employment due to AI use. 

(with Bonney, Breaux, Buffington, Dinlersoz, Foster, Haltiwanger, Kroff, Savage)

High-Growth Firms in the United States: Key Trends and New Data Opportunities (2024) 

Using administrative data from the U.S. Census Bureau, we introduce a new public-use database that tracks activities across firm growth distributions over time and by firm and establishment characteristics. With these new data, we uncover several key trends on high-growth firms—critical engines of innovation and economic growth. 

(with Joonkyu Choi, John Haltiwanger, and J. Daniel Kim)

The 2010 Census Confidentiality Protections Failed, Here's How and Why (2023) 

Using only 34 published tables, we reconstruct five variables (census block, sex, age, race, and ethnicity) in the confidential 2010 Census person records. Using the 38-bin age variable tabulated at the census block level, at most 20.1% of reconstructed records can differ from their confidential source on even a single value for these five variables. We perfectly reconstruct 2010 Census records for 97 million people, correctly infering race and ethnicity for 3.4 million vulnerable population uniques.

(with Abowd, Adams, Ashmead, Darais, Dey, Garfinkel, Kifer, Leclerc, Lew, Moore, Rodríguez, Tadros, & Vilhuber) 

An In-Depth Examination of Requirements for Disclosure Risk Assessment (2023) 

We argue that any proposal for quantifying disclosure risk should be based on pre-specified, objective criteria. We illustrate this approach, using simple desiderata, to evaluate the absolute disclosure risk framework, the counterfactual framework underlying differential privacy, and prior-to-posterior comparisons. We conclude that satisfying all the desiderata is impossible, but counterfactual comparisons satisfy the most while absolute disclosure risk satisfies the fewest. 

(with Jarmin, Abowd, Ashmead, Cumings-Menon, Hawes, Keller, Kifer, Leclerc, Reiter, Rodríguez, Schmutte, Velkoff, Zhuravlev)

On The Role of Trademarks: From Micro Evidence to Macro Outcomes (2023)

We study the effects of trademarks on the US economy. Motivated by evidence from firm-level data on trademark registrations and outcomes suggesting that trademarks protect firm value and are associated with higher firm growth and marketing activity, we introduce trademarks in a general equilibrium framework to quantify their aggregate effects. Analysis of the calibrated model indicates that trademarks generate higher product variety, quality, and welfare, along with higher concentration.

(with Emin Dinlersoz, Mehmet Yorukoglu, and Nikolas Zolas)

Where Have All the "Creative Talents" Gone?  Employment Dynamics of US Inventors (2023)

We study how the allocation of inventors across firms affects an economy's innovative capacity. We build a model where an inventor works for an entrant or incumbent and strategic considerations lead incumbents to poach the inventor, offering higher wages, and not implement the idea. Using the employment history of 760 thousand US inventors, we find evidence consistent with the model, showing that (i) inventors are increasingly concentrated in large incumbents and less likely to become entrepreneurs, and (ii) when an inventor is hired by an incumbent, compared to a young firm, their earnings increases by 12.6 percent and their innovative output declines by 6 to 11 percent.

(with Ufuk Akcigit)

Coverage: WSJ, NPR Planet Money, Becker Friedman Institute, Marginal Revolution

Measuring the Characteristics and Employment Dynamics of U.S. Inventors (2023) 

We introduce a new dataset linking patent inventors to survey, census, and administrative microdata at the U.S. Census Bureau. We use this data to provide a first look at the demographic characteristics, employer characteristics, earnings, and employment dynamics of US inventors. 

(with Ufuk Akcigit)

Slides: FSRDC 2022

Robot Hubs: The Skewed Distribution of Robots in U.S. Manufacturing (2023)

We use data from the Annual Survey of Manufactures to characterize the geographic concentration of robot use in US manufacturing. The top ten percent of CBSAs, by share of active robots, account for over 77 percent of all robots. We identify RobotHubs as geographies where robot use is far higher than one would expect given industrial composition alone. RobotHubs are much more likely to have robot integrators, firms that facilitate the installation of robots, and higher union membership. 

AEA Papers and Proceedings (with Erik Brynjolfsson,  Catherine Buffington,  J. Frank Li,  Javier Miranda, and  Robert Seamans)

The Characteristics and Geographic Distribution of Robot Hubs in U.S. Manufacturing Establishments (2023)

We use data from the Annual Survey of Manufactures to study the characteristics and geography of investments in robots across US manufacturing establishments. Robotics adoption and robot intensity (robots per employee) is more strongly related to establishment size than age, and the distribution of robots is highly skewed across establishments’ locations. Some locations, which we call Robot Hubs, have far more robots than one would expect even after accounting for industry and manufacturing employment, and these Robot Hubs tend to have robot integrators and higher levels of union membership. 

(with Erik Brynjolfsson,  Catherine Buffington,  J. Frank Li,  Javier Miranda, and  Robert Seamans)

Advanced Technology Adoption: Selection or Causal Effects? (2023)

We use data on the use of automation technologies from the 2019 Annual Business Survey combined with data on the historical business dynamics of firms from the LBD to explore the relationship between firm size and technology use. We find that large firms are much more likely to use automation technologies, even controlling for firm age and industry, but that firms using these technologies were already large and growing before the technologies became widely available, favoring the view that firms select into adopting technologies rather than technologies causing employment growth.

AEA Papers and Proceedings, Forthcoming (with Acemoglu, Anderson, Beede, Buffington, Childress, Dinlersoz, Foster, Haltiwanger, Kroff, Restrepo, and Zolas) 

Automation and the Workforce: A Firm-Level View from the 2019 Annual Business Survey (2022)

Using a new module introduced in the Census Bureau’s 2019 Annual Business Survey, we describe the use of automation technologies (AI, robotics, dedicated equipment, specialized software, and cloud computing) by US firms across all economic sectors. Adoption remains low (especially for AI and robotics), varies substantially across industries, concentrates among large and young firms, and its use is associated with a 15% higher labor productivity, accounting for 20–30% of the higher labor productivity achieved by the largest firms in an industry.

(with Acemoglu, Anderson, Beede, Buffington, Childress, Dinlersoz, Foster, Haltiwanger, Kroff, Restrepo, and Zolas) 

Redesigning in Longitudinal Business Database (2021) 

We describe the US Census Bureau's redesign and production implementation of the Longitudinal Business Database (LBD) first introduced by Jarmin and Miranda (2002). The LBD is used to create the Business Dynamics Statistics (BDS), tabulations describing the entry, exit, expansion, and contraction of businesses. We describe in detail how the LBD is created from curation of the input administrative data, longitudinal matching, retiming of economic census-year births and deaths, creation of vintage consistent industry codes and noise factors, and the creation and cleaning of each year of LBD data. 

NBER WP 28839 (with Chow, Fort, Goetz, Lawrence, Perlman, Stinson, and White) 

Research Experience as Human Capital in New Business Outcomes (2021) 

We develop several measures of human capital based upon experience working on federally funded research grants and experience in R&D labs, High Tech businesses, and universities, showing how these measures are negatively related to survival but positively related to employment, suggesting that high human capital individuals select into high-risk startups that exhibit "up or out" dynamics.

In Measuring and Accounting for Innovation in the 21st Century (with Ron Jarmin, Julia Lane, and Nikolas Zolas)

An Anatomy of U.S. Firms Seeking Trademark Registration (2021) 

We introduce a new dataset linking trademarks to firms-level microdata at the US Census Bureau, which provides measures of firm characteristics, performance, and dynamics of firms that own trademarks along with the incidence and timing of trademark registrations over the firm's life cycle. We show that (1) firms experience substantially higher employment and revenue following their first trademark filing, (2) there is significant overlap between firms with trademarks, R&D, and patenting, suggesting that trademark- based metrics may serve to improve measurement of innovation in the economy.

In Measuring and Accounting for Innovation in the 21st Century (with Emin Dinlersoz, Amanda Myers, and Nikolas Zolas)

Press: WIPR

Advanced Technologies Adoption and Use by U.S. Firms: Evidence from the Annual Business Survey (2020)

Using a new module introduced in the Census Bureau's 2018 Annual Business Survey, we describe the extent of digitization and the diffusion of advanced technologies including machine learning, robotics, and augmented reality. Digitization and cloud computing are widespread while advanced technologies are rare and generally skew towards larger and older firms. Adoption patterns are consistent with a hierarchy of increasing technological sophistication. While few firms are at the technology frontier, they tend to be large so technology exposure of the average worker is significantly higher.

NBER WP 28290 (with Zolas, Kroff, Brynjolfsson, McElheran, Beede, Buffington, Foster and Dinlersoz)

Business Dynamics of High Tech Industries (2020)

We compare methods of identifying High Tech industries, describe our preferred method using the concentration of STEM employment, and preview Business Dynamics Statistics of High Tech industries including startup rates and job creation rates.

Journal of Economics & Management Strategy (with Javier Miranda)

BDS-HT Tables, America Counts Blog, Code (STEM concentration)

Tracking the Technological Composition of Industries with Algorithmic Patent Concordances (2019)

We provide concordances between patent data (CPC, USPC) and industry and product classifications (NAICS, ISIC, HS), improving the utility of each and aiding researchers of innovation.

Economics of Innovation and New Technology (with Travis Lybbert and Nikolas Zolas)

Download Crosswalks

Proximity and Economic Activity: An Analysis of Vendor‐University Transactions (2019)

We use admin data on transactions associated with federally funded research grants (UMETRICS), finding: (1) expenditures on research inputs tend to be physically close to the universities, and (2) firms that supply research deepen ties with research institutions, opening new establishments near the university and supplying additional projects.

Journal of Regional Science (with Julia Lane, Bruce Weinberg, and Nikolas Zolas)

Squeezing More Out of Your Data: Business Record Linkage with Python (2018) 

We introduce the Multiple Algorithm Matching for Better Analytics (MAMBA) program, a flexible, scalable, and transparent software platform for business record linkage applications using Census microdata. MAMBA leverages multiple string comparators to assess the similarity of records using a machine learning algorithm to disambiguate matches.

(with John Cuffe) 

Is Federal Regulation to Blame for the Decline in American Entrepreneurship? (2018)

We use RegData, which uses the text of all federal regulations measure regulation by industry, to test whether regulation explains the secular decline in business dynamism, finding that changes in regulation are not associated with changes in business dynamism---highly regulated and lightly regulated industries saw declines in entry and job reallocation.  

Economic Policy. (with Alex Tabarrok)

Press: Data and code, Reason, Bloomberg, Washington Monthly, Bloomberg View, CEI, CLS Colombia Law blogs

Just Passing Through: Characterizing U.S. Pass-Through Business Owners (2017) 

We investigate the use of administrative data on the owners of partnerships and S-corporations to develop new statistics that characterize business owners. Income for these entities is "passed through" to owners and captured on Form K1 information returns, making it possible to link owners to businesses.

(with J. Daniel Kim and Kristin McCue)

Wrapping it up in a person: Examining employment and earnings outcomes for Ph.D. recipients (2015) 

We use admin data on federally funded research projects (UMETRICS) to study the outcomes of researchers, finding: (1) nearly 40% of supported doctorate recipients entered industry, tending to work for high-wage establishments in High Tech and professional services industries.

Science, 350(6266), 1367-1371. (with Zolas, Jarmin, Stephan, Owen-Smith, Rosen, Allen, Weinberg, & Lane)

Press: Nature, Business Insider, Vox, Science Daily, Phys, Times Higher Education

Is Entrepreneurship in Decline? (2015) 

We present a series of alternative perspectives on the secular decline in business dynamism, highlighting that the richest economies tend to have the largest firms, "entrepreneurship" that can occur within large firms, the negative effects of excess churn, and the international dynamism of domestic firms.

In Understanding the Growth Slowdown (pp. 169-187). Washington DC: Cato Institute Press. (with Alex Tabarrok)

Press: WIPR

Technical Notes & Data

"Imputing Establishment Robotics Data: I’m Afraid I Can’t Do That." 2022 . (with Jones, Miranda, and Smith)

“Building a Better Bridge: Improving Patent Assignee-Firm Links.” 2018. CES Technical Notes Series No. 2018-01. (with Dreisigmeyer, Krylova, Ouyang, and Perlman)

DISCLAIMER: This site is maintained by Nathan Goldschlag. Opinions expressed here do not reflect the official views of the U.S. Census Bureau or any other public or private organization.