Research

“More Mobile, More Concentrated: Inventor Flows, Firm Networks, and the New Geography of Innovation” (w/ Xian Jiang and William Kerr)

Abstract: We use U.S. patent records and metropolitan house-price data to document four changes in the geography of invention between 1980 and 2015. First, patents and inventors shifted sharply toward high-house-price metropolitan areas, while those areas’ population share remained broadly stable. Second, mobility among active inventors became more frequent, but the network narrowed: a growing share of moves ended in the highest-price cities or connected to such cities, and fewer metropolitan areas accounted for half of inventor flows. Third, patenting firms became active in more metropolitan locations. Approximately one-third of observed intercity inventor transitions occurred within assignees, and internal transferees became more common at new firm–MSA patenting locations. Fourth, technologies whose initial leading cities experienced stronger relative house-price appreciation subsequently became more geographically concentrated in patent production. House prices reflect both equilibrium indicators of successful clusters and potential costs of accessing them. The findings describe a U.S. innovation system that became simultaneously more mobile, more organizationally connected, and more spatially concentrated over the last five decades.

“Does Goliath help David? Anchor firms and startup clusters” Journal of Urban Economics 156 (November).

Abstract: Does attracting a “Goliath” firm to a county anchor the growth of regional industry clusters? Matching a hand-collected dataset of major corporate site selection contests to restricted-use U.S. Census microdata, I show that winning a Million Dollar Plant (MDP) generates 2,515 net supply chain jobs in the first four years after announcement, declining to 1,622 in the subsequent four years. Over the last four years of the panel between four and seven years after the MDP is announced, startups remain the largest contributor to job gains at 1,694 jobs, more than offsetting a decline among other incumbents (-61% of the net total), with local incumbents (41%), and affiliated firms (15%) also contributing positively. Post-announcement startups are more numerous but smaller and slower-growing than comparable cohorts in runner-up counties. The two main sources of job creation draw on distinct segments of the local labor market: relative to those in finalist counties, startups in winning counties rely disproportionately on immigrant workers whereas local incumbents employ more local workers, and neither shows evidence of worker flows from the MDP’s corporate network.

“Firms for Funding: The effect of Million Dollar Plants on school finances and student achievement” (w/ Viviana Rodriguez) Journal of Urban Economics 149 (September).

Abstract: We study the impact of large firm entry on local public education by comparing school districts in counties that win a Million Dollar Plant (MDP) to runner-up counties. Winning an MDP increases school district revenues by approximately 3%, primarily through higher local property tax collections. While total per pupil revenue rises modestly, we find gains in instructional spending and small improvements in test scores. Effects vary by firm type: manufacturing MDPs are associated with greater capital outlays but limited achievement gains, whereas high-tech MDPs see increased instructional spending and improved proficiency. Districts with MDPs in highly educated industries exhibit larger increases in instructional spending and student outcomes, even when overall revenue gains are similar.

“Who Benefits from Million Dollar Plants? The Missing Local Beneficiaries”” (w/ Saheel Chodavadia, William Kerr, and Stephen Ross)

Abstract: We study who receives jobs in the industries catalyzed by Million Dollar Plants (MDPs) during 2000-2015. Comparing winning versus runner-up counties, people working the winning county are on average 1.2% more likely to work in the MDP’s four-digit industry compared to the runner-up. This effect takes about nine years post MDP announcement to materialize. Surprisingly, however, there is no employment effect for the initial residents of the winning county compared to runner-up counties; nor do we observe any other difference in economic benefits. These results suggest MDPs generate the promised jobs but source their workforce from outside the winning county.

“Regional Resilience, Startups, and the Great Recession”

Abstract: This paper investigates how employment levels across geographic regions in the U.S. recovered after the Great Recession. Using a Bartik (1991) framework to leverage variation across regions according to their industry composition, this paper highlights a new pattern during the recovery from the Great Recession. Regions that experienced the most dramatic downturns do not recover as quickly as expected, and this deficit is largely explained by a diminished role for startups in job creation. Common perceptions about local labor market quality and credit availability to startups do not explain the startup deficit post-crisis.