Portal de Periódicos da CAPES

Industrial renewal in the 21st century: evidence from US cities

2015; Routledge; Volume: 51; Issue: 3 Linguagem: Inglês

10.1080/00343404.2015.1100288

1360-0591

Thor Berger, Carl Benedikt Frey,

Regional resilience and development

ABSTRACTIndustrial renewal in the 21st century: evidence from US cities. Regional Studies. Where and why do new industries emerge? Using revisions of official industrial classifications, this paper documents the appearance of new industries in the US economy between 2000 and 2010 stemming directly from technological advances. Examining differences in new industry creation across cities, this paper shows that new industries mainly emerge in human capital abundant places and cities that specialize in industries that demand similar skills. Instrumental variables estimates that exploit the location of 19th-century land-grant colleges as an instrument for contemporary differences in human capital assigns a causal interpretation to these results.摘要二十一世纪的工业復兴:来自美国城市的证据. 区域研究。新兴产业在何处、以及为何兴起?本文运用修订后的官方产业分类,纪录从 2000 年到 2010 年之间,美国经济中直接源自于技术革新的新兴产业。本文检视各城市创造新兴产业的差异,显示新兴产业主要是从专司需要类似技能的产业、且人力资本充沛的地方及城市中兴起。将十九世纪赠地大学区位视为促成当代人力资源差异的工具变量评估,对此般结果提出了因果性的诠释。RÉSUMÉLe renouveau industriel au cours du 21e siècle: des résultats provenant des grandes villes situées aux É-U. Regional Studies. D'où et pourquoi les nouvelles industries est-ce qu'elles apparaissent? À partir des modifications des nomenclatures industrielles officielles, cet article documente la naissance des nouvelles industries dans l'économie aux É-U entre l'an 2000 et 2010 qui est directement le résultat des avances technologiques. En examinant à travers les grandes villes les différences de la création de nouvelles industries, cet article montre que les nouvelles industries apparaissent principalement dans des lieux et des grandes villes abondants en capital humain et qui se spécialisent dans les industries qui exigent des compétences similaires. Des estimations des variables instrumentales, qui exploitent l'emplacement des land-grant colleges du 19e siècle (universités aux É-U construites à partir d'une concession foncière faite par le gouvernement fédéral) comme outil des différences contemporaines du capital humain, attribuent une interprétation causale à ces résultats.ZUSAMMENFASSUNGDie Entstehung neuer Branchen im 21. Jahrhundert: Belege aus Städten der USA. Regional Studies. Wo und warum entstehen neue Branchen? Anhand einer Überprüfung der offiziellen Branchenklassifikationen wird in diesem Beitrag das Entstehen von neuen Branchen in der Wirtschaft der USA als unmittelbares Resultat technischer Neuerungen im Zeitraum von 2000 bis 2010 untersucht. Durch eine Untersuchung der Unterschiede bei der Entstehung neuer Branchen in verschiedenen Städten zeigen wir, dass neue Branchen vor allem in Orten mit reichlich Humankapital sowie in Städten entstehen, die auf Branchen spezialisiert sind, für die ähnliche Qualifikationen erforderlich sind. Zur kausalen Interpretation dieser Ergebnisse nutzen wir Instrumentalvariablenschätzungen, in denen der Standort von Land-Grant-Colleges des 19. Jahrhunderts als Instrument für aktuelle Unterschiede beim Humankapital genutzt wird.RESUMENRenovación industrial en el siglo XXI: ejemplo de ciudades estadounidenses. Regional Studies. ¿Dónde y por qué surgen nuevas industrias? A partir de un examen de las clasificaciones industriales oficiales, en este artículo documentamos la aparición de nuevas industrias en la economía estadounidense entre 2000 y 2010 que proceden directamente de los progresos tecnológicos. Examinando las diferencias entre la creación de nuevas industrias en varias ciudades, mostramos que las nuevas industrias surgen principalmente en lugares con abundante capital humano y en ciudades especializadas en industrias que exigen calificaciones similares. Para hacer una interpretación causal de estos resultados, calculamos las variables instrumentales utilizando la ubicación de universidades del siglo XIX por cesión de terreno como un instrumento para las diferencias contemporáneas en el capital humano.KEYWORDS: citiesnew industrieshuman capitaltechnological change关键词: 城市新兴产业人力资本技术转变MOTS-CLÉS: grandes villesnouvelles industriescapital humainmutation technologiqueSCHLÜSSELWÖRTER: StädteNeue BranchenHumankapitalTechnischer WandelPALABRAS CLAVES: ciudadesindustrias nuevascapital humanocambio tecnológicoJEL: J24O31O33R11View correction statement:Erratum ACKNOWLEDGEMENTSThe authors are grateful to the editor, two anonymous referees, Liana Christin Landivar at the US Census Bureau, and Kerstin Enflo, Martin Henning and Michael Wolf at the Bureau of Labor Statistics for advice, comments and suggestions that substantially improved the paper. The usual disclaimer applies.DISCLOSURE STATEMENTNo potential conflict of interest was reported by the authors.FUNDINGThor Berger gratefully acknowledges funding from the Swedish Research Council [grant number 2014-1491] and the Crafoord Foundation [grant number 2013–0812]; Carl Frey gratefully acknowledges funding from Citi.SUPPLEMENTAL DATASupplemental data for this article can be accessed at http://10.1080/00343404.2015.1100288Notes1. For brevity, this paper interchangeably refers to new 'technology-related' industries simply as 'new industries' throughout.2. Similarly, while flea markets are now for the first time sufficiently popular to constitute an industry, the first flea markets in the United States date back to the Canton Monday Trade Days beginning in 1873.3. Urban status is defined based on whether a CZ intersects a metropolitan statistical area. The robustness checks reported in Appendix A in the supplemental data online shows that the main results are similar, though smaller in magnitude, in a sample that also includes the rural CZs.4. Controls are based on the 2000 Census. While income growth between 2000 and 2010 may be endogenous to new industry creation, this control variable is thought of as capturing a wide range of productivity shocks that may be correlated with the creation of new industries.5. Educational attainment is captured by a dummy for a bachelor's degree, a master's degree, a professional degree, a PhD and whether the degree is in a science, technology, engineering and mathematics (STEM) field respectively.6. While this approach addresses spatial sorting based on observable characteristics, sorting on unobservable ability may still bias the estimates upwards if more able workers select cities with a larger fraction of skilled workers.7. Moretti (Citation2004) uses the location of the land-grant institutions to study the extent of human capital spillovers.8. Kantor and Whalley (Citation2014) examine the long-run impact of US agricultural experiment stations, established following the Hatch Act of 1887 that provided federal funding to establish experiment stations at the land-grant colleges. However, they show that the productivity effects of these stations in most instances had faded already by the 1920s, with little impact on productivity differences today.9. More precisely, it requires that the land-grant institutions do not differ in important dimensions from other universities. In support of this assumption, Moretti (Citation2004) shows that workers in cities with a land-grant college differ little from workers in cities without a land-grant institution in terms of demographic and racial characteristics as well as Armed Forces Qualification Test (AFQT) scores.10. The share of workers with a college education remains a statistically significant predictor of new industry creation also when focusing only on the 51 new industries that are associated with digital technology, holding each city's share of workers employed in IT industries constant. A 1 SD (standard deviation) increase in the share of college-educated workers is associated with roughly 0.67 SD increase in new digital industry employment, suggesting that general human capital is important over and above related skills.11. Patents are assigned to counties based on the location of residence of the first-named inventor, which may differ from the location of work, although this should induce negligible bias in the estimations. Publicly available county-level patent counts from the USPTO are matched with CZs, using crosswalks available from the US Department of Agriculture (USDA).12. More precisely, it includes controls for the number of title IV degree-offering institutions per capita and the income-adjusted college graduation rate based on Chetty, Hendren, Kline, and Saez (Citation2014).13. The corresponding Stock and Yogo (Citation2005) critical value for a maximum 10% IV bias is 16.38 for one endogenous regressor and one instrument, meaning that one can clearly reject the null of weak instruments.14. An additional threat to the validity of these estimates is that employment in IT clusters or patenting activity may be higher in cities in proximity to the land-grant institutions. However, controlling for the share of workers employed in the IT industry and the average number of patents per university only slightly reduces the coefficient on the share of college-educated workers respectively to 0.013 (p = 0.019) and 0.011 (p = 0.046) in the second stage.