Industrial Location at the Intra-Metropolitan Level: The Role of Agglomeration Economies

Abstract

Arauzo-Carod J.-M. and Viladecans-Marsal E. Industrial location at the intra-metropolitan level: the role of agglomeration economies, Regional Studies. The objective of this paper is to analyse the influence of agglomeration economies on location decisions taken by new firms inside metropolitan areas. As it is assumed that these economies differ according to firms' level of technology, the sample comprises new firms from high-, intermediate- and low-technology industries. Of particularly interest is in analysing the effects of agglomeration economies that are felt over very short distances (inside the metropolitan areas). Introduced in the estimation is the effect of the central city as a determinant for the location of new firms in the rest of the metropolitan area.

Arauzo-Carod J.-M. et Viladecans-Marsal E. Sites industriels au niveau intramétropolitain: le rôle des économies d'agglomération, Regional Studies. L'objectif de cet article est d'analyser l'influence des économies d'agglomération sur les décisions prises par de nouvelles entreprises concernant l'installation de leur site dans des zones métropolitaines. Partant de l'hypothèse que ces économies diffèrent en fonction du niveau technologique des entreprises, notre échantillon comprend des entreprises nouvelles d'un niveau technologique élevé, intermédiaire et bas. Nous nous intéressons particulièrement à l'analyse des effets des économies d'agglomération qui sont ressenties sur de très courtes distances (à l'intérieur des zones métropolitaines). Nous introduisons dans notre estimation l'effet de la ville centrale comme déterminant dans le choix des sites des nouvelles entreprises dans le reste de la zone métropolitaine.

Sites industriels Ville Zone métropolitaine Économies d'urbanisation

Arauzo-Carod J.-M. und Viladecans-Marsal E. Industriestandorte auf intrametropolitaner Ebene: die Rolle der Agglomerationswirtschaften, Regional Studies. Mit diesem Beitrag soll der Einfluss von Agglomerationswirtschaften auf die Standortentscheidungen von neuen Firmen in metropolitanen Gebieten untersucht werden. Da wir annehmen, dass diese Wirtschaften je nach dem technologischen Niveau der Firmen unterschiedlich ausfallen, enthält unsere Stichprobe neue Firmen aus Branchen mit hohem, mittlerem und niedrigerem technologischen Niveau. Insbesondere interessieren wir uns für die Analyse der Auswirkungen von Agglomerationswirtschaften, die über sehr kurze Entfernungen hinweg (innerhalb der metropolitanen Gebiete) spürbar werden. In unserer Schätzung berücksichtigen wir die Auswirkung der zentralen Stadt als Determinanten für den Standort neuer Firmen im übrigen metropolitanen Gebiet.

Industriestandort Städte Metropolitane Gebiete Agglomerationswirtschaften

Arauzo-Carod J.-M. y Viladecans-Marsal E. Localización industrial en un nivel intrametropolitano: el papel de las economías de aglomeración, Regional Studies. El objetivo de este artículo es analizar qué influencia ejercen las economías de aglomeración cuando las nuevas empresas eligen su localización en áreas metropolitanas. Suponiendo que estas economías difieren en función del nivel de tecnología de las empresas, en nuestra muestra abarcamos nuevas empresas de industrias con tecnologíaalta, intermedia y baja. Nos interesa en particular analizar los efectos de las economías de aglomeración que se hacen palpables sobre distancias muy cortas (dentro de áreas metropolitanas). En nuestro cálculo introducimos el efecto de la ciudad central como determinante para ubicar las nuevas empresas en el resto del área metropolitana.

Ubicación industrial Ciudades Áreas metropolitanas Economías de aglomeración

Keywords:

• Industrial location
• Cities
• Metropolitan areas
• Agglomeration economies

JEL classifications:

• L60
• R12
• R30

Notes

In Spain, the smallest political and administrative units are the cities, understood as municipalities. There are more than 9000 municipalities of very differing sizes. The next level up in political terms is that of the provinces (of which there are 50 in the Spanish case) and this level is equivalent to the Nomenclature of Territorial Units for Statistics (NUTS) III.

The third section provides a detailed description of the characteristics of the Spanish metropolitan areas used in this paper.

The authors expected to obtain different location patterns according to the broad characteristics of each industrial sector (measured in terms of technological level), so it was not necessary to use data from all manufacturing activities.

The nine municipalities that did not receive new industrial firms were: San Juan Bautista, Tiana, Cañete de las Torres, Espejo, Hoyo de Manzanares, Teverga, Gorliz, Lekeitio and Plentzia. These are small municipalities with a mean population of 3759 inhabitants.

In those models the dependent variable is a count variable (here, the number of times that an industrial establishment locates in a municipality).

Obviously, working at a local level involves more observations than at the regional or national levels. The ‘problem’ is the in-built restrictions of the econometric software concerning the maximum number of alternatives that can be estimated using a conditional logit model. Therefore, having more observations constitutes a major ‘problem’.

One problem with this argument is how to choose the samples. Because an undetermined number of firms were not able to locate, they were not counted. All these are counted as zero.

For a detailed discussion as to how zero observations contribute to the likelihood function, see Cameron and Trivedi (1998).

The goodness-of-fit test divides the range of the data into intervals. The number of points within each interval is then compared with the expected number of points for that interval according to the hypothesized distribution of the data (here, the hypothesis should have a Poisson distribution). Here the deviance statistic (Cameron and Trivedi, 1998) is used.

Alpha determines the degree of dispersion. Specifically, if α = 0, the negative binomial distribution is equivalent to a Poisson distribution (and there is no over-dispersion).

Measuring urbanization economies at a local level is very common in the empirical literature (for example, Arauzo, 2005; Arauzo and Manjón, 2004; Barrios et al., 2006; Coughlin and Segev, 2000; Duranton and Puga, 2000; Figueiredo et al., 2002; Guimarães et al., 2000; Viladecans, 2004).

N_ij is a count variable in which there are zero and non-negative values.

For a more detailed analysis, see Arauzo (2005).

Nevertheless, empirical work about the incidence of the qualifications of the labour force usually shows ambiguous results. For example, Holl (2004b) finds both a (mainly) positive and a negative effect on firm location depending on the industry analysed.