Abstract
This paper investigates the relationship between market potential and the spatial variation in the number and the average size of firms. We adapt the canonical model of the New Economic Geography to demonstrate this relationship and to derive an empirical specification suitable for estimation through dynamic panel techniques. The model is tested against municipal data on the number of firms per adult in Chilean comunas for 2005–2010. Our results confirm that market potential along with place-specific fixed costs play an important role in determining the spatial variation in the number of firms per capita.
Résumé
la présente communication se penche sur les rapports entre le potentiel du marché et les variations spatiales dans le nombre et la taille moyenne des entreprises. Nous adaptons le modèle canonique de la nouvelle géographie économique afin de démontrer ce rapport, et dériver une spécification empirique appropriée pour l'estimation par le biais de techniques des panneaux dynamiques. On met le modèle à l’épreuve avec des données municipales sur le nombre d'entreprises par adulte dans des municipalités chiliennes pour 2005–2010. Nos résultats confirment que le potentiel du marché, ainsi que les coûts fixes spécifiques au lieu, jouent un rôle important dans la détermination de la variation spatiale du nombre d'entreprises par personne.
Resumen
este estudio investiga la relación entre el potencial del mercado y la variación espacial del número y el tamaño medio de las firmas. Adaptamos el modelo canónico de la Nueva geografía económica para demostrar esta relación y obtener una especificación empírica idónea para la estimación a través de técnicas para un panel dinámico. El modelo ha sido probado frente a los datos municipales del número de firmas por adulto en los municipios chilenos entre 2005 y 2010. Nuestros resultados confirman que el potencial del mercado, junto con los costes fijos específicos de la localidad, desempeñan un papel importante en la determinación de la variación espacial del número de firmas per capita.
摘
要,本文研究了市场潜力与在企业数量和平均规模方面的空间差异之间的关系。我们采用了新经济地理学的正则模型来证明这一关系,并通过动态面板技术取得适用 于估算的实证规模。该模型是根据 2005–2010 年期间关于智利城市中人均企业数量的市政数据进行测试的。研究结果证实,市场潜力与每个地方特有的固定成本在确定人均企业数量的空间差异方面均起到重要作 用。
We are grateful to Geoffrey Hewings, Rodrigo Paillacar, Damian Clarke, Miguel Atienza, Andrea Bentancor, Gianni Romani and two anonymous referees for helpful comments and suggestions. Any errors are of our own responsibility. Modrego wishes to thank the financial support of project Fondecyt 1130356 and of the Rural Territorial Dynamics Program funded by the International Development Research Centre of Canada (IDRC).
We are grateful to Geoffrey Hewings, Rodrigo Paillacar, Damian Clarke, Miguel Atienza, Andrea Bentancor, Gianni Romani and two anonymous referees for helpful comments and suggestions. Any errors are of our own responsibility. Modrego wishes to thank the financial support of project Fondecyt 1130356 and of the Rural Territorial Dynamics Program funded by the International Development Research Centre of Canada (IDRC).
Notes
1. We acknowledge recent developments in the related discipline of urban economics that have delivered microeconomic models of entrepreneurship that share many features with the NEG (Glaeser et al., Citation2010a, Citation2010b).
2. We thank an anonymous referee for suggesting this interpretation.
3. While neither NEG nor entrepreneurship models have explicitly identified this relationship before between the region's number of firms per capita and the regional market potential there is a theoretical literature within the field of industrial organisation, primarily associated with the work of John Sutton (Citation1991, Citation1998) which demonstrates that in a non-spatial Cournot setting, the number of firms and the levels of competition increase in a non-linear manner as the size of the market increases.
4. Isla de Pascua (Easter Island) Juan Fernández and Antártica.
5. http://www.sii.cl/estadisticas/empresas.htm (version with date of extraction: 17/08/2011)
6. There is a small number of firms without location in the SII data-set comprising 0.2% of the total number of firms in the period.
8. According to Gustavo Maillat, former Regional Ministerial Secretary of Economy in the Region of Coquimbo, while the costs of setting up a business would be of around US$700 in Santiago, they would add up to US$1,000 outside the capital city (http://diarioeldia.cl/articulo/casi-99-se-reduce-costo-crear-una-empresa)
9. 1 UF is approximately US$48.
10. The CASEN surveys have been taken every two to three years since 1987.
12. Mion (Citation2004) points to a risk of bias in Hanson's approach due to the inhomogeneous scale of the RHS and LHS variables and proposes a linearization of the NEG wage equation. Here, unlike Hanson (Citation2005), the dependent variable is not part of the market potential function, so the problem is likely smaller. On the other hand, and as the author points, linearization also comes at a cost of potential biases. Without a conclusive assessment, we decided to stick to Hanson's procedure.
13. According to SII criterion, small firms are those with levels of annual sales below UF 25,000 (USD 1 MM aprox.), medium firms up to UF 100,000 (USD 4 MM aprox.) and large firms above that.
14. Classification in the tertiary sector was based on the sectorial classification used by SII, closely resembling the 1-digit ISIC.
15. Windmeijer (Citation2005) proposes a finite sample correction for linear two-step GMM estimators, performance of which has not been well-established in the general non-linear case (Windmeijer, Citation2008).
16. This observation comes from a comment of an anonymous referee to whom we are grateful.
17. The estimated shares of expenditures on non-tradables (housing), is in line with estimates based on expenditures surveys in the country, ranging from 15% to 30% depending on what is included in the item ‘housing’ (INE, Citation2008).
18. We owe this idea to an anonymous referee.
19. Which are available from the authors upon request.
20. When these comunas are included the J test strongly rejects the null hypothesis, which indicates a specification error.
21. Transitory cash transfer of around US$80 given in 2010 to around 4 million poor people to cope with additional expenses during the month of March.
22. Geographically targeted programme supplying four cubic metres of wood to poor households of the Region of Aysén in Southern Chile. It is estimated by the government as a US$200 save in household's annual incomes.