Do agglomeration externalities affect firm survival?

ABSTRACT

Do agglomeration externalities affect firm survival? Regional Studies. This paper analyses the impact of spatial agglomeration externalities on Italian start-up firms’ survival. Italy represents a relevant case study given the well-known role of firm clusters in the country's economic development. Results obtained support the hypothesis that industry variety reduces the likelihood of firm exit. Specifically, related variety, which contributes to the generation and diffusion of new knowledge, has a positive effect on firm survival in manufacturing sectors, while unrelated variety, which may work as a portfolio strategy, plays a positive role in services sectors. Localization economies positively influence firm survival only in services sectors. Finally, urban density is not robust to the control for firm characteristics.

摘要

聚集的外部性是否会影响企业生存？区域研究。本文分析空间聚集的外部性对意大利新创企业的生存之影响。意大利呈现作为一个相关案例研究，因企业集群在该国经济发展中扮演的角色众所皆知。本研究获得的成果，支持产业多样性减少了企业退出的可能性之假说。特别是导致新知识生成和扩散的相关多样性，对于製造业部门的企业生存具有正面的效应，而不相关多样性，则可能作为组合策略，而在服务业部门中扮演正面的角色。在地化的经济，只有在服务业部门正向地影响了企业生存。最后，城市密度对于控制企业特徵而言并不强健。

RÉSUMÉ

Les externalités d'agglomération, est-ce qu'elles influent sur la survie des entreprises? Regional Studies. Cet article analyse l'impact des externalités d'agglomération spatiales sur la survie des entreprises italiennes en démarrage. L'Italie constitue une étude de cas appropriée étant donné le rôle bien connu des clusters d'entreprises dans le développement économique du pays. Les résultats obtenus soutiennent l'hypothèse selon laquelle la variété de l'industrie réduit la possibilité d’échec. En particulier, la variété connexe, qui contribue au développement et à la distribution de nouvelles connaissances, a un effet positif sur la survie des entreprises dans les secteurs industriels, tandis que la variété non-apparentée, qui peut se servir de stratégie de portefeuille, joue un rôle positif dans les secteurs des services. Les économies de localisation n'influent de façon positive sur la survie des entreprises que dans les secteurs des services. Pour finir, la densité urbaine ne tient pas bien compte des caractéristiques des entreprises.

ZUSAMMENFASSUNG

Wirken sich Agglomerationsexternalitäten auf das Überleben von Firmen aus? Regional Studies. In diesem Beitrag untersuchen wir die Auswirkungen der Externalitäten von räumlichen Agglomerationen auf das Überleben von italienischen Start-up-Firmen. Italien stellt angesichts der wohlbekannten Rolle von Firmenclustern für die Wirtschaftsentwicklung des Landes eine relevante Fallstudie dar. Die erhaltenen Ergebnisse stützen die Hypothese, dass Branchenvielfalt die Wahrscheinlichkeit einer Firmenschließung senkt. Insbesondere wirkt sich verbundene Vielfalt, die zur Erzeugung und Verbreitung von neuem Wissen beiträgt, positiv auf das Überleben von Firmen in produzierenden Sektoren aus, während unverbundene Vielfalt, die als Portfolio-Strategie fungieren kann, eine positive Rolle in Dienstleistungssektoren spielen kann. Lokalisierungswirtschaften wirken sich nur auf das Überleben von Firmen in Dienstleistungssektoren positiv aus. Die Stadtdichte schließlich ist gegenüber der Kontrolle auf Firmenmerkmale nicht robust.

RESUMEN

Afectan las externalidades de aglomeración a la supervivencia de las empresas? Regional Studies. En este artículo analizamos el impacto de las externalidades de aglomeración espacial sobre la supervivencia de las empresas incipientes italianas. Italia representa un estudio de caso relevante dado el papel bien conocido de las aglomeraciones de empresas en el desarrollo económico del país. Los resultados obtenidos avalan la hipótesis de que la variedad de la industria reduce la probabilidad del cierre de las empresas. En concreto, la variedad relacionada, que contribuye a la generación y difusión de conocimiento nuevo, tiene un efecto positivo en la supervivencia de las empresas en los sectores de manufactura, mientras que la variedad no relacionada, que podría funcionar como una estrategia de carteras, desempeña un papel positivo en los sectores de servicios. Las economías de localización influyen positivamente en la supervivencia de las empresas solamente en los sectores de servicios. Finalmente, la densidad urbana no tiene suficiente solidez cuando se tienen en cuenta las características de las empresas.

KEYWORDS:

• survival
• start-up firm
• agglomeration economies
• Italy

关键词:

• 生存
• 新创企业
• 聚集经济
• 意大利

MOTS-CLÉS:

• survie
• entreprise en démarrage
• économies d'agglomération
• Italie

SCHLÜSSELWÖRTER:

• überleben
• start-up-firma
• agglomerationsökonomien
• Italien

PALABRAS CLAVES:

• supervivencia
• empresa incipiente
• economías de aglomeración
• Italia

JEL:

• C81
• L11
• L25
• R12

Notes

1. Exceptions are De Silva and McComb (2011), Van Oort et al. (2012), Renski (2011) and, for the case of Italy, Carree et al. (2011) and Cainelli et al. (2014).

2. Source: Unioncamere, the Union of the Italian Chambers of Commerce. Exit rate is defined as the number of firms exiting the industry divided by the number of incumbent firms in the industry in the previous period.

3. See Duranton and Puga (2001) for a formal treatment of this mechanism.

4. In 2004, Italy recorded the highest entry rate since 1999 and a reduction in the exit rate (ISTAT, 2011).

5. Note that the subject of the analysis is the firm and not the plant. However, this is never considered as a shortcoming in the case of Italy, where the phenomenon of multi-plant firms is limited (about 6% according to ISTAT).

6. ‘Small firms' are those with a number of employees fewer than 50; ‘medium firms' are those with a number of employees higher than 50 and lower than 250; and ‘large firms' are those with more than 250 employees.

7. Other suitable and popular functional form for are: (1) log-time specification and (2) polynomial-in-time specification, where one can have linear, quadratic, cubic or other polynomials for the hazard function.

8. In biostatistics the unobserved heterogeneity is also referred to as ‘frailty'.

9. The most relevant exogenous variables affecting firm survival and considered in the dominant empirical literature are ‘size’ and ‘age’. This paper does not need to control for firm age, since only considered are cohorts of firms established between 2004 and 2006. Moreover, the paper avoids the inclusion in the model of firm-level variables, which are strongly suspected to be endogenous, such as measures of productivity and/or profitability.

10. All the data were obtained by consulting the Italian Statistical Atlas of Municipalities (Atlante Statistico dei Comuni).

11. An alternative absolute measure of localization economies, based on the ratio between the number of employees in the two-digit sector s in the LLS and the area (km²) of the LLS, was also used. However, the results of the analysis turned out to be robust to the choice of the variable.

12. In the literature, some authors (e.g., Renski, 2015) identify related sectors by using buyer–supplier linkages among industries and/or shared occupations as weights applied to regional industry employment. However, this paper uses the entropy approach because the aim was simply to distinguish Jacobs' externalities and not to identify the specific sources of Marshall's localization economies.

13. When not properly captured by a right-hand-side variable, this spatial spillover effect generates spatial autocorrelation in the residuals, with negative consequences for the estimates in terms of efficiency and/or consistency. See also Neffke et al. (2012) for a similar way of computing potential measures of agglomeration externalities.

14. Neffke et al. (2012) tested whether the effect of agglomeration externalities on firm survival differs between corporate and non-affiliated plants. Taking into account that multinational enterprises have a corporate structure, the present analysis assimilates multinationals with corporate firms.

15. The robustness of the results were assessed against alternative measurements of agglomeration variables. Specifically, two alternative models were estimated. In the first, the agglomeration variables were computed without including the weighted spatial lag (that is, the weighted values of the neighbours defined according to the inverse distance-decay criterion). In the second model, the agglomeration variables were computed as ‘potential measures' (that is, including the weighted spatial lag), but using the squared-inverse distance-decay instead of the inverse distance-decay criterion to identify the set of neighbours. The results (available from the authors upon request) strongly confirm those reported in the paper. The only relevant exception concerns the coefficient of related variety that turns out to be positive in the case of services, but this does not change the main message of the paper (related variety works in favour of the survival of manufacturing firms, while unrelated variety works in favour of services firms).

16. As discussed in the fourth section, the random effects cloglog model is based on the assumption of orthogonality between unobserved heterogeneity and the explanatory variables, so the estimated parameters cannot be fully interpreted as causal effects.

17. Due to data constraints, spatial and industrial fixed effects were controlled for in the estimations, but not for firms' fixed effects (only random effects were considered). Thus, the estimated parameters cannot be used to infer causal relations and the analysis cannot give conclusive results; rather the evidence can be used to define new theoretical hypotheses on the different effect of agglomeration economies in tradable and non-tradable sectors, which can be investigated in further studies.