![Sample our Global Development journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5929/TOC-Subject-Sample-2021-Global-Development.jpg"})
Abstract
County-level location patterns of INC5000 companies provide one map of American entrepreneurship and innovativeness, and understanding the local factors associated with these firms’ emergence is important for stimulating regional economic growth and innovation. We draw on the knowledge spillover theory of entrepreneurship to motivate our regression model, and augment this theory with additional regional features that have been found to be important in the firm location literature. Zero-inflated negative binomial regressions indicate that these firms exist in counties with larger average establishment size, higher educational attainment and more natural amenities. Income growth, a mix of higher paying industries, and more banks per capita are associated with a smaller presence of these types of firms, all else equal. We conclude that the local conditions favouring high-growth firms are likely to be different from those favouring new firms in general, and that these conditions differ significantly in urban and rural areas and by industrial sectors.
Acknowledgement
EMSI data is proprietary. ARIES is an industrial affiliates programme at Virginia Tech, supported by members that include companies in the energy sector. The research under ARIES is conducted by independent researchers in accordance with the policies on scientific integrity of their institutions. The views, opinions and recommendations expressed herein are solely those of the authors and do not imply any endorsement by ARIES employees, other ARIES-affiliated researchers or industrial members. Information about ARIES can be found at http://www.energy.vt.edu/ARIES. Neither ARIES nor any of its partners has seen or reviewed this work. The authors want to thank two anonymous reviewers for their constructive comments.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
Funding from USDA’s National Institute of Food and Agriculture (NIFA) under Grants 2012-70002-19385 (NARDeP) and 2014-51150-22094 (NERCRD) is gratefully acknowledged, as is funding from Penn State University’s Agricultural Experiment Station. Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the view of the U.S. Department of Agriculture, or of Penn State University.
Notes
1. Existing studies mostly focus on European countries such as the UK (Lee Citation2014), France (Lasch, Robert, and Le Roy Citation2013), Scotland (Mason and Brown Citation2013), Italy (Bonaccorsi et al. Citation2014), Germany (Stuetzer et al. Citation2014), Spain (Lopez-Garcia and Puente Citation2012), as well as a study of 184 cities in 20 European countries (García Citation2014). There is also a literature, beyond the scope of our work, that examines internal strategies and characteristics of high-growth firms (Smallbone, Leig, and North Citation1995; Delmar, Davidsson, and Gartner Citation2003). Moreno and Casillas (Citation2007) conduct a discriminant analysis to examine variables that separate high growth from other firms.
2. Starting in 1982, the magazine listed the 500 firms with highest revenue growth in the USA; in 2007, it expanded the list to 5,000 firms.
3. Examples of papers published since 2010 that use the same conceptual framework include Hanson and Rohlin (Citation2011), Manjón-Antolín and Arauzo-Carod (Citation2011), Frenkel (Citation2012), Arauzo-Carod and Manjón-Antolín (Citation2012), Arauzo-Carod (Citation2013), Alañón-Pardo and Arauzo-Carod (Citation2013), Basile, Benfratello, and Castellani (Citation2013), Mota and Brandão (Citation2013), Buczkowska and de Lapparent (Citation2014), Liviano and Arauzo-Carod (Citation2013, Citation2014).
4. At the same time, we note that factors contributing to firm emergence may not also ensure their long-term survival (see e.g. Brixy and Grotz Citation2007 for a sample of German firms).
5. See also Acs, Audretsch, and Lehmann (Citation2013). For a cautionary statement about this theory, see Knoben, Ponds, and van Oort (Citation2011).
6. The fact that the Tesla Company has bought large tracts of land in rural Nevada, both because of lower cost and to shield its research on batteries from competitors, is anecdotal evidence of the disadvantages of agglomeration.
7. Malecki (Citation1993, 123) discusses how, in the past, lack of information about market conditions elsewhere created disadvantages especially for smaller firms.
8. Data are for 2007, which is the only year in which our INC5000 data contains NAICS codes. In other years, the INC5000 industry definition is not comparable to the NAICS definition used by the US Census.
9. As another extension, Pergelova and Angulo-Ruiz (Citation2014) suggest that neither firm revenues nor profits are reliable guides for choosing firms to support with public financial resources. Their sample consists of new US firms.
10. Maps and analyses using 2007 and 2009 data, which we compiled from the website, are available from the authors upon request. We are grateful to INC Magazine for providing us with an electronic file containing the 2012 data.
11. The data are available here: http://www.ers.usda.gov/data-products/rural-urban-continuum-codes.aspx (accessed August 12, 2014).
12. The data are available here: http://aese.psu.edu/nercrd/community/tools/social-capital (accessed August 12, 2014).
13. EMSI uses data from multiple sources including the BEA, BLS, County Business Patterns, and the Quarterly Census of Employment and Wages to fill ‘suppression holes,’ in which publically available sources do not report disaggregated industry employment and wage data, especially for rural counties to protect confidentiality. EMSI has developed an algorithm to fill these holes, and their data are reported to be relatively accurate (Dorfman, Partridge, and Galloway Citation2011; Fallah, Partridge, and Olfert Citation2011).
14. Note that the use of four-digit EMSI employment data allows us to more precisely measure industry mix demand shifts than is typical in the literature, which usually relies on one-digit or at best two-digit data.
16. In a sensitivity analysis described below, we included only the youngest firms to consider this possibility.
20. Results using 2007 and 2009 data are not materially different, and available from the authors upon request. We also estimated models with interactions terms as part of our sensitivity analysis (see footnotes below).
21. In the zero inflation regressions, which are jointly estimated with the negative binomial regression, counties with more college graduates, banks per capita, INC5000 firms in 2009, and 2008 population as expected had statistically significant greater odds of also hosting one or more INC firms in 2012. Note that the inflation stage regression predicts the absence of firms.
22. These results also are robust to alternative measures of specialization, including the Hirschman–Herfindahl index and the Shannon Entropy index.
23. We do note that the coefficients’ estimates on the unemployment rate in all and metro-only counties are positive, and not that far from being statistically significant.
24. The difference in the coefficients of metro and non-metro counties mostly reflects scaling in that population density is much higher in metropolitan counties. The elasticity of high-growth firm location and population density is 0.12 in metro counties and 0.19 in non-metropolitan counties. The larger non-metropolitan response is intuitive as we would expect non-metropolitan high-growth firms to benefit a little more at the margin from higher density because they have so relatively little to begin with.
25. Results are available from the authors.
26. Additional analysis shows that the association between proprietor or self-employment rates and the presence of INC5000 firms follows the shape of an inverted U. Conceivably, the presence of too many self-employed eventually crowds out opportunities for establishing rapidly growing firms, although at least initially the self-employed form a pool from which such firms are likely to emerge.
27. An important exception is that in the sensitivity analysis, where we include interactions among and non-linear effects of other variables, the effect of labour market freedom is statistically significant and negative in the model containing all counties. This may suggest that a policy of reduced labour market freedom creates incentives to start a high-growth firm. This could be explored in future research.
28. Young, smaller firms have been shown to drive a disproportionate share of US job growth (Haltiwanger, Jarmin, and Miranda Citation2013). Subtle differences emerged by metro and non-metro status of firms when we separately considered newer (<11 years) and older INC5000 firms (results available on request). For example, newer non-metro firms, establishment size and prior (2005–2008) population growth each had a statistically significant and positive effect while neither had a statistical effect when all non-metro firms were considered together. Also, newer non-metro firms were less likely to emerge in counties with higher high school dropout rates, indicating that these firms require better-educated workers. When we considered only newer metro firms, a major difference was that industry specialization mattered, whereas it did not when all firms were considered together. Yet, the overriding theme was that the locational determinants between new and older fast growth firms were quite similar, suggesting that firm age is not a key intervening factor in their location. Another sensitivity test involving 2007 and 2009 data revealed largely consistent results. There is not enough variation in the dependent variable to permit estimating a panel data model.
29. Manufacturing is one sector for which the sample size is large enough to allow separate regressions for newer and older firms. Interestingly, for newer manufacturing firms, establishment size does not matter, and for newer metro-based manufacturing firms, the dropout rate is not statistically important.