![Sample our Built Environment journals, sign in here to start your access, Latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5926/TOC-Subject-Sample-2021-Built-Environment.jpg"})
Abstract
Problem: Many policymakers and planners believe that entrepreneurship is key to rejuvenating American communities amid continued pressures of globalization and technological change; yet, despite its present popularity, we still know relatively little about the influence of different types of locations on new business creation and performance.
Purpose: This study examines how new firms' entry, survival, and growth differ among central cities, suburbs, small cities, and rural areas inside and outside metropolitan areas, to allow policymakers and planners to develop more informed development strategies.
Methods: I use a unique establishment-level longitudinal database to examine the formation, survival, and growth of new manufacturing and advanced services firms in the continental United States using a mix of descriptive and inferential statistical methods.
Results and conclusions: This study shows that entrepreneurial performance differs across a continuum of locations from nonmetropolitan rural to urban core. New firms in central cities have higher failure rates but faster rates of employment growth in advanced services. Nonmetropolitan rural places are undersupplied with new high-tech manufacturing and both high-tech and conventional advanced services firms, and have lower growth rates in both high-tech and conventional manufacturing and advanced services. Intermediate places in the urban hierarchy (suburbs, small cities, and rural segments of metropolitan areas) often have relatively high rates of new firm entry, survival, and growth. No environment favors entrepreneurship across all sectors and performance measures.
Takeaway for practice: Policymakers seeking to develop effective entrepreneurial strategies should understand the relative advantages and disadvantages facing entrepreneurs in different types of areas. Central cities' lower new firm entry and survival rates may be due to higher costs and regulatory barriers, or may be a result of their attracting riskier endeavors that also have potential for more rapid growth. The growth of firms started by rural entrepreneurs, by contrast, is constrained by limited local markets and relative isolation, and likely to benefit from strategies that expand the reach of new firms into broader markets.
Research support: This research was supported by the Ewing J. Kaufman Foundation's Dissertation Fellowship program and the National Science Foundation's Doctoral Dissertation Research Fellowship program. All research using the confidential Longitudinal Database (LDB) was conducted on site at the Bureau of Labor Statistics offices in Washington, DC, and has been cleared for public distribution. The author is solely responsible for the content of this article.
Notes
Note: a. The location quotient is calculated as the area type's share of new firms, divided by its share of population.
*p 0.05 (two-tailed)
**p 0.01 (two-tailed)
***p 0.001 (two-tailed)
†p 0.10 (two-tailed)
*p 0.05 (two-tailed)
**p 0.01 (two-tailed)
***p 0.001 (two-tailed).
†p 0.10 (two-tailed)
*p 0.05 (two-tailed)
**p 0.01 (two-tailed)
***p 0.001 (two-tailed)
†p 0.10 (two-tailed)
*p 0.05 (two-tailed)
**p 0.01 (two-tailed)
***p 0.001 (two-tailed)
1. I calculated this average of annual private sector gross employment due to openings using the Bureau of Labor Statistics, Business Employment Dynamics series from 1997 to 2007.
2. I also do not include establishments whose size at birth is 250 employees or larger. Thorough inspection of individual records suggests that most establishments I rejected solely because they were above the 250-employee threshold were probably miscoded, and were either not new or were not independent of pre-existing businesses. In any case, only a handful of establishments were eliminated by this criterion.
3. I used the Standard Industrial Classification (SIC) system (which grouped data on firms according to their products) because it was in effect during my study period. It has since been replaced by the North American Industry Classification System (NAICS; which groups data on firms according to their methods of production). In both systems, fewer digits represent more general categories. Manufacturing is an SIC sector, or single-digit grouping of one or more SIC industries.
4. I dropped one industry (tobacco manufacturing, SIC 21) from my analysis because it had very few entrants during my study period.
5. I did not include firms in predominantly local-serving industries such as those in retail trade and personal services, although they often make up a large share of new businesses. Establishments in these industries locate close to the resident population that is their market and depend for success on wealth derived from sectors that produce goods and services. Because they have less independent potential to generate increases in long-term employment and wages, such industries are of less importance from an economic policy perspective.
6. I calculated this using the Bureau of the Labor Statistics, Current Employment Estimates series for the period 1997 to 2007.
7. Please refer to CitationHecker (1999) for a more detailed discussion of the classification method and a complete list of three-digit SICs classified as high-tech and conventional.
8. Researchers familiar with state ES-202 files (the originating source for the Federal QCEW) may question the accuracy of the data. I have considerable confidence in the quality of the data used in this study. During the early 1990s, the BLS began requiring establishment-level reporting, uniform reporting standards, and additional quality controls among the states. The Federal BLS also conducts its own quality control after the state records have been submitted. In addition to these institutional procedures, I conducted my own quality control checks, developing programs to flag suspicious establishments, such as those showing explosive growth between quarters or other dramatic changes. I inspected the suspicious records, eliminating those that could not be corrected or verified.
9. I deliberately avoided relying on the earliest years of the LDB (1989–1992), which are known to have reporting errors (CitationFeser & Sweeney, 2006).
10. I used the LDB's unique identifier that allows identifying new establishments and tracking them over time. Predecessor and successor identification numbers make it possible to track establishments even when they change owners. A more detailed explanation of the business tracking method is available upon request from the author.
11. I use year-specific dummy variables to control for macroeconomic conditions that affect all new firms, regardless of location. Many additional years of data, preferably spanning multiple cycles, would be necessary to explore how business cycles influence performance at the level of the establishment. This topic is left for future research.
12. The Census distinguishes urbanized areas, urban clusters, and rural areas based on population density, geographic contiguity, and population thresholds, without respect to overlap with metropolitan areas and central cities. The OMB defines metropolitan areas according to the economic integration of counties and, in the case of New England, minor civil divisions. The 1999 OMB definitions of metropolitan areas were also used in the 2000 Census of Population and Housing.
13. I compiled population data by area type from the STF3 geographic area summary files of the 2000 Census of Population and Housing.
14. Proper measurement of the relationship between age and growth must also account for the selective attrition of small firms (CitationDunne, Roberts, & Samuelson, 1989; CitationEvans, 1987; CitationHall, 1987). Smaller firms are more likely to fail in their first few years of life (CitationAudretsch & Mahmood, 1995; CitationAudretsch, Santarelli, & Vivarelli, 1999; CitationEvans, 1987). Because small firms drop out early and large firms remain, average establishment size increases even if the remaining firms have not actually grown. To account for this, I limit the analysis to entrants that eventually survive the entire study period.
15. I credit an anonymous referee for providing this interpretation.