Local business ownership and local economic performance: evidence from US counties

ABSTRACT

Local business ownership and local economic performance: evidence from US counties. Regional Studies. A rich US county-level dataset is utilized to obtain a statistical characterization of the relationship between the size of the local-based business sector and local economic performance. The study also disaggregates the local businesses sector by establishment size. The results do not provide conclusive evidence that the size of the local businesses sector is positively and significantly associated with local economic performance. The disaggregated results show some evidence of a significant and positive association of smaller local businesses with local economic performance compared with larger local businesses.

摘要

地方企业所有权和地方经济表现：来自美国各郡的证据. 区域研究。本研究利用美国一个丰富的郡层级数据集，以获取地方企业部门规模和地方经济表现之间的关联性之统计特性描述。本研究同时依照成立规模来区隔地方企业部门。研究结果，并未能提出地方企业部门的规模与地方经济表现正向且显着相关的总结性证据。区隔结果的部分证据显示，较小型的地方企业相较于较大型的地方企业而言，与地方经济表现有较为显着且正向的关联性。

RÉSUMÉ

La propriété des entreprises locales et la performance économique locale: des résultats provenant des comtés situées aux É-U. Regional Studies. On emploie un ensemble de données détaillées rassemblées au niveau des comtés aux É-U pour obtenir une caractérisation statistique du rapport entre la taille du milieu d'affaires local et la performance économique locale. En outre, l’étude ventile le milieu d'affaires local en fonction de la taille des établissements. Les résultats n'apportent pas de preuves concluantes que la taille du secteur des entreprises locales est associée de façon positive ou est étroitement liée à la performance économique locale. Les résultats ventilés fournissent certaines preuves d'une association significative et positive des petites entreprises locales avec la performance économique locale par rapport aux grandes entreprises locales.

ZUSAMMENFASSUNG

Lokales Firmeneigentum und lokale Wirtschaftsleistung: Belege aus Bezirken der USA. Regional Studies. In dieser Studie wird mithilfe eines erweiterten Datensatzes auf Bezirksebene in den USA eine statistische Charakterisierung der Beziehung zwischen der Größe des lokalen Unternehmenssektors und der lokalen Wirtschaftsleistung durchgeführt. Ebenso wird der lokale Unternehmenssektor nach Firmengröße aufgegliedert. Die Ergebnisse liefern keinen zwingenden Beweis für einen positiven und signifikanten Zusammenhang zwischen der Größe des lokalen Unternehmenssektors und der lokalen Wirtschaftsleistung. Die aufgegliederten Ergebnisse enthalten im Vergleich zu größeren lokalen Unternehmen einige Anzeichen für einen signifikanten und positiven Zusammenhang zwischen kleineren lokalen Unternehmen und der lokalen Wirtschaftsleistung.

RESUMEN

Propiedad empresarial local y rendimiento económico local: evidencia de condados estadounidenses. Regional Studies. En este artículo se utiliza un amplio conjunto de datos de condados estadounidenses para obtener una caracterización estadística de la relación entre el tamaño del sector comercial de ámbito local y el rendimiento económico local. También se desglosa el sector comercial local en función del tamaño de los establecimientos. Los resultados no ofrecen pruebas concluyentes de que el tamaño del sector comercial local esté relacionado de manera positiva y significativa con el rendimiento económico local. Los resultados desglosados demuestran cierta evidencia de una asociación significativa y positiva entre negocios locales más pequeños y el rendimiento económico local en comparación con negocios locales más grandes.

KEYWORDS:

• employment growth
• income growth
• local-based businesses
• poverty
• spatial analysis
• US counties

关键词:

• 就业成长
• 所得成长
• 地方企业
• 贫穷
• 空间分析
• 美国各郡

MOTS-CLÉS:

• croissance de l'emploi
• hausse des revenus
• entreprises locales
• pauvreté
• analyse spatiale
• comtés aux É-U

SCHLÜSSELWÖRTER:

• Beschäftigungswachstum
• einkommenswachstum
• lokale unternehmen
• armut
• raumanalyse
• bezirke der USA

PALABRAS CLAVES:

• crecimiento del empleo
• crecimiento de ingresos
• negocios de ámbito local
• pobreza
• análisis espacial
• condados estadounidenses

JEL:

• L25
• O18
• R11

ACKNOWLEDGEMENT

The views expressed here are those of the author and cannot be attributed to the Economic Research Service, US Department of Agriculture.

SUPPLEMENTAL DATA

Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/00343404.2015.1119264

Notes

1. See youreconomy.org/.

2. While it can be understood that there are other measures related to local business spillovers such as profits and other contributions by local businesses to local communities which can be used to study the relationship between local businesses and economic performance, the author did not have access to such information in the data through the Edward Lowe Foundation.

3. One main criticism of D&B data is that they are inconsistent in establishment and employment counts over time (Davis et al., 1996). Neumark, Zhang, and Wall (2005) conclude, after assessing the reliability of the NETS data along several aspects, that it is a reliable data source, despite its limitations. A recent study by Choi, Robertson, and Rupasingha (2013) for the US state of Georgia finds that the NETS data do not match comparable official government establishment-level statistics.

4. See note 1 above.

5. The data received from the Edward Lowe Foundation had the self-employed category included in stage 1.

6. For studies that use establishment-level data directly from the National Establishment Times Series (NETS), not county-level aggregates, see Neumark et al. (2005, 2011), and Kolko and Neumark (2008, 2010).

7. US Department of Agriculture (USDA) – Economic Research Service (ERS) rural urban continuum codes were used to classify counties as metro (codes 1–3) and non-metro (codes 4–9) counties.

8. One shortcoming of this measure, as observed by Beck, Demirguc-Kunt, and Levine (2005), is that it is a static measure. The data do not take into consideration the entry of new firms, firm deaths or growth of firms.

9. Due to space constraint, correlations are reported in Appendix A in the supplemental data online.

10. Higher order spatial weight matrices (up to two and three nearest neighbours) were also used as robustness checks for all the estimations discussed in this paper; it was found that the results of the variables of interest were largely unchanged in terms of the signs and significance. These results are not presented here for the sake of brevity but are available from the author upon request.

11. Each dependent and independent variable was standardized by subtracting its mean and dividing by its standard deviation to avoid badly conditioned variables and negative definite matrices.

12. The test of the null hypothesis that the difference between the coefficients of the shares of local and non-local employment is equal to zero is rejected at the less than 1% significance level, justifying the identification of commercial employment as local and non-local. This is also true for employment growth and change in poverty equations results, which are discussed below.

13. Unexpected results may be due to the high correlation (see Table A2 in Appendix A in the supplemental data online) between the shares of local and non-local sectors, and this issue is addressed further below.

14. Unlike the results discussed in Fleming and Goetz, the results presented here accounted for spatial dependence bias, which is important when using county-level data.

15. The results based on establishment size also seem to differ between the two studies for metro and non-metro areas, but a direct comparison of these results is not possible due to differences between the two studies in classifying counties as metro and non-metro.

16. The author is grateful to Tom Mickiewicz, Associate Editor of this journal, for suggesting this direction.

17. The author does not see the need to impose this structure on model 2 since the correlations among establishment size categories are not high (see Table A2 in Appendix A in the supplemental data online). Also, imposing this structure on model 2 is not possible due to many missing observations in the newly created ratio variables based on establishment size. This happens as a result of zero share of non-local establishments in many counties, especially when the establishment size gets larger. For example, in the non-metro sample, the ratio of local large to non-local large variable results in 1333 missing observations since that many non-metro counties had no reported large non-local establishments.

18. Table 4 only reports results related to the variables of interest and full results are available from the author upon request.

19. See LeSage and Pace (2009) and others (Fischer et al., 2009; LeSage & Fischer, 2008) for details regarding the specific calculations.

20. Again, because of space limitations, the discussion here is limited to a few examples of direct effects.

21. The author is grateful to an anonymous referee for making this point.

22. The author is grateful to Tom Mickiewicz for suggesting this alternative explanation.