![Sample our Urban Studies journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/110826/TOC-Subject-Sample-2021-Urban-Studies.jpg"})
Abstract
Álvarez-Ayuso I. C., Condeço-Melhorado A. M., Gutiérrez J. and Zofío J. L. Integrating network analysis with the production function approach to study the spillover effects of transport infrastructure, Regional Studies. The production function approach is used to analyse the role of transport infrastructure on regional gross domestic product (GDP) using new definitions and measures of road network capital stock that represent the real benefit obtained by regions when accessing markets. Improving the existing methodologies, the infrastructure stock is weighted with trade data so as to estimate the direct effects on production of a region's own infrastructure (what is termed here ‘internal stock’), as well as the spillover effects that it receives from using that of neighbouring regions (imported stock). The methodology is illustrated by using Spanish data for the 1980–2007 period and these internal and imported infrastructure stocks are calculated using geographical information system (GIS) network analysis based on generalized transportations costs. With this new data set, successive regressions are performed controlling for endogeneity and the obtained results are compared with those of previous research. The validity of this methodology and the existence of significant and rather large spillover effects that even outweigh the effect of the internally endowed capital stock on aggregate production are confirmed. On average, the relative magnitude of the spillover effects to that of the internal (own) stock effect increases with the level of territorial disaggregation, i.e., it is larger for provincial data than for regional data. Unfortunately, it is also found that spillover effects are asymmetric, exhibiting negative values for poorer regions, as they do not profit from the capital stock existing in their neighbouring areas as do their richer counterparts, thereby casting doubts on the cohesion effects attributed to transport infrastructure investments.
Álvarez-Ayuso I. C., Condeço-Melhorado A. M., Gutiérrez J. and Zofío J. L. 整合网络分析和生产函数法，以研究交通基础建设的外溢效应，区域研究。生产函数法，用来分析交通基础建设之于区域的地区生产总值 (GDP) 所扮演的角色，此方法运用路网资本存量的新定义和测量，呈现出区域准入市场时获得的真正效益。为了促进既有的方法，基础建设存量以贸易数据权衡之，以评估一个区域所拥有的基础建设所生产的直接效应（亦称之为“内在存量”），以及使用邻近区域的内在存量所获得的外溢效应（引进存量）。我们透过运用西班牙自 1980 年至 2007 年的数据来勾勒此一方法，并运用根据一般化运输成本的地理信息系统（GIS）网络分析，计算上述的内在及引进的基础建设存量。本研究用此一新数据集，控制内生性以进行逐步回归，获得的结果则将与过去的研究进行比较。此一方法的有效性，以及显着、且相当大幅——甚至比内在固有资本存量对集合生产的效应更大的外溢效应被确认。平均而言，外溢效应之于内在（拥有）存量效应的相对重要性，随着领域的解集程度而增加，例如其之于省层级的数据便较区域层级来得大。可惜的是，本研究同时发现，外溢效应是不对称的，并对较穷困的区域呈现出负面的价值，因其无法像较富裕的对照区域一般，从邻近地区的资本存量中获益，因此对推断为交通基础建设投资带来的凝聚效应提出了质疑。
Álvarez-Ayuso I. C., Condeço-Melhorado A. M., Gutiérrez J. et Zofío J. L. L'intégration de l'analyse des réseaux avec l'approche de la fonction de production pour étudier les retombées de l'infrastructure routière, Regional Studies. On emploie l'approche de la fonction de production pour analyser le rôle de l'infrastructure routière dans le produit intérieur brut (Pib) régional utilisant de nouvelles définitions et mesures du stock de capital relatives au réseau routier et qui représentent les avantages obtenus par les régions au moment où elles ont accès aux marchés. En développant les méthodologies actuelles, le stock d'infrastructure est pondéré par des données commerciales afin d'estimer les effets directs sur la production de l'infrastructure propre à une région (à savoir, ‘le stock interne’), aussi bien que les retombées qui proviennent de l'utilisation de celle des régions avoisinantes (à savoir, ‘le stock importé’). On présente la méthodologie à partir des données espagnoles de 1980 à 2007, et ces stocks d'infrastructure interne et importé sont calculés au moyen d'une analyse des réseaux à base d'un système d'information géographique fondé sur les frais de transport généralisés. Ce nouvel ensemble de données permet des régressions multiples qui tiennent compte de l'endogénéité. On compare les résultats obtenus à ceux des recherches antérieures. On confirme la validité de cette méthodologie et la présence d'importantes retombées de grande ampleur qui l'emportent même sur l'effet sur la production globale du stock de capital doté en interne. En moyenne, l'ampleur relative des retombées par rapport à l'effet du stock (propre) interne augmente avec le niveau de désagrégation territoriale; autrement dit, il s'avère plus important pour les données provinciales que pour les données régionales. Malheureusement, il s'avère aussi que les retombées sont asymétriques, présentant des valeurs négatives dans les régions plus pauvres, parce qu'elles ne bénéficient pas du stock de capital qui existe dans les zones avoisinantes comme le font leurs homologues plus riches, ce qui remet en cause les effets de cohésion attribués à l'investissement dans l'infrastructure de transport.
Álvarez-Ayuso I. C., Condeço-Melhorado A. M., Gutiérrez J. und Zofío J. L. Integration der Netzwerkanalyse in den Produktionsfunktionsansatz zur Untersuchung der Übertragungseffekte der Verkehrsinfrastruktur, Regional Studies. Mithilfe des Produktionsfunktionsansatzes analysieren wir die Rolle der Verkehrsinfrastruktur auf das regionale Bruttoinlandsprodukt unter Verwendung von neuen Definitionen und Maßstäben des Grundkapitals von Straßennetzen, die den wahren Nutzen für Regionen beim Zugang zu Märkten repräsentieren. Durch eine Verbesserung der bisherigen Methodologien wird der Infrastrukturbestand mit Handelsdaten gewichtet, um die direkten Auswirkungen der regionseigenen Infrastruktur (die hier als ‘interner Bestand’ bezeichnet wird) auf die Produktion sowie die Übertragungseffekte, die die Region durch eine Nutzung der Infrastruktur von benachbarten Regionen verzeichnet (importierter Bestand), abzuschätzen. Die Methodologie wird anhand von spanischen Daten für den Zeitraum von 1980 bis 2007 illustriert; die internen und importierten Infrastrukturbestände werden auf der Grundlage der generalisierten Verkehrskosten mithilfe einer Netzwerkanalyse des geografischen Informationssystems (GIS) berechnet. Mit diesem neuen Datensatz werden unter Berücksichtigung der Endogenität aufeinanderfolgende Regressionen durchgeführt; die Ergebnisse werden mit denen von früheren Studien verglichen. Es bestätigen sich die Gültigkeit dieser Methodologie sowie die Existenz von signifikanten und recht umfangreichen Übertragungseffekten, die sogar den Effekt des intern vorhandenen Grundkapitals auf die Gesamtproduktion ausgleichen. Im Durchschnitt nimmt die relative Größe der Übertragungseffekte im Vergleich zur Größe des Effekts des internen (eigenen) Bestands mit dem Umfang der territorialen Disaggregation zu, d. h., sie fällt für Daten auf Provinzebene höher aus als für Daten auf Regionalebene. Leider stellt sich auch heraus, dass Übertragungseffekte asymmetrisch sind und für ärmere Regionen negative Werte aufweisen, da diese nicht wie reichere Regionen von dem in ihren Nachbargebieten vorhandenen Grundkapital profitieren; dies lässt Zweifel an den Kohäsionseffekten aufkommen, die Investitionen in die Verkehrsinfrastruktur zugeschrieben werden.
Álvarez-Ayuso I. C., Condeço-Melhorado A. M., Gutiérrez J. y Zofío J. L. Integración del análisis de redes con el enfoque de la función de producción para estudiar los efectos indirectos de la infraestructura de transporte, Regional Studies. Mediante el enfoque de la función de producción analizamos el papel de la infraestructura de transporte con respecto al producto interno bruto (PIB) regional mediante nuevas definiciones y medidas de la reserva de capital para la red vial que representan el beneficio real obtenido por las regiones al acceder a los mercados. Con una mejora de las metodologías existentes, ponderamos las reservas de infraestructuras con los datos del comercio para calcular los efectos directos en la producción de la propia infraestructura de una región (lo que aquí se denomina ‘reservas internas’), así como los efectos indirectos que recibe al utilizar la infraestructura de las regiones vecinas (reservas importadas). Ilustramos la metodología mediante datos españoles entre 1980 y 2007 y calculamos estas reservas de infraestructuras internas e importadas mediante un análisis de redes basado en el sistema de información geográfica (SIG) a partir de los costes generalizados del transporte. Con este nuevo grupo de datos, llevamos a cabo regresiones sucesivas controlando la endogeneidad y comparamos los resultados obtenidos con los estudios anteriores. Se confirman la validez de esta metodología y la existencia de efectos indirectos significativos y bastante grandes que incluso compensan el efecto de las reservas internas de capital en la producción agregada. De promedio, la magnitud relativa de los efectos indirectos en comparación con el efecto de reservas internas (propias) aumenta con el nivel de la desagregación territorial, es decir, es mayor para los datos provinciales que para los datos regionales. Por desgracia, también observamos que los efectos indirectos son asimétricos y muestran valores negativos para las regiones más pobres, ya que no se aprovechan de las reservas de capital existentes en sus zonas vecinas tal como ocurre con sus homólogos más ricos; esto por tanto pone en cuestión los efectos de la cohesión atribuidos a las inversiones en la infraestructura del transporte.
Acknowledgements
The authors are grateful to Carlos Llano and the C-intereg project team for sharing the data on the Spanish interregional trade (http://www.c-intereg.es), as well as for their helpful comments and suggestions.
Supplemental data
Supplemental data for this article can be accessed at http://dx.doi.org/10.1080/00343404.2014.953472.
Notes
1. Many authors study the contribution of transport infrastructure to improve the access to markets (e.g., Spiekermann and Neubauer, Citation2002; Martín et al., Citation2004; Zofío et al., Citation2014).
2. Some studies have used a translog specification, which differs from the Cobb–Douglas function in that it does not imply various restrictions on the production structure, e.g., the inputs elasticities of substitution are, by definition, equal to 1 in the Cobb–Douglas function.
3. Absolute engineering unit prices (price per kilometre) corresponding to each road type are obtained from the Ministerio de Fomento (MFOM) (Citation2000). For example, in 2005, current prices for one kilometre of highway according to European Union standards in the flat, plateau topography, corresponding to the Spanish meseta, amounted €2 million/km (Toledo province), while in mountainous regions it amounts €6.3 million/km (Asturias province).
4. As it has an illustrative purpose and to save space, is truncated (i.e., not all 47 Spanish provinces are shown). The complete table is available from the authors upon request.
5. Additionally, the existence of unit roots would suggest spurious correlation problems in the estimations and can tested by performing the Harris–Tzavalis test for panel data. In the present case, the small time dimension in the data set, T, requires a unit-root test specification that assumes this dimension fixed, as in the case of Harris and Tzavalis (Citation1999). Unit roots in the dependent (GDP) and explicative variables are checked for: labour, private capital, and internal and imported capital stock, rejecting the null hypothesis on the existence of unit-roots in all cases, and, therefore, the possibility of spurious correlation in the estimation. These results support the estimation of the extended production functions with the internally and imported capital stocks by fixed effects, and using the within estimator incorporating individual dummies.
6. The test for exogeneity for the fixed effects model proposed by Wooldridge (Citation2002) on the internally used capital stock is performed, with statistic values (t-Student) of 0.11 (3.05) and 0.05 (3.59) for the regional and provincial levels, respectively. The null hypothesis is that this variable is exogenous, which is rejected at the 5% level, implying that the variable is endogenous.
7. Stock and Yogo (Citation2005) classify a group of instruments as weak, or ‘performing poorly’ if the bias of the IV estimator relative to the bias of the ordinary least squares (OLS) estimator exceeds a given threshold. The present study uses 10% and 15% maximal IV sizes.
8. The distance to Zaragoza as an IV for market access in a GDP regression was applied to Spanish data by López-Rodríguez et al. (Citation2011). The different spatial distribution of the regions and provinces determines the use of alternative IVs. In northern Spain there are many small regions (Comunidades Autónomas), while in the south the opposite can be observed. For this reason, at a regional level the geographical centre is situated in the north-west and, specifically, in Zaragoza. For the provincial disaggregation, since all provinces are of similar size, the centre moves to Madrid. Both kinds of clustering follow the argument posed by Marquez and Hewings (Citation2003) and Marquez et al. (Citation2003, Citation2006), who develop a competition structure for a multiregional economic system using co-integration/error correction modelling techniques that support its existence over both the short- and long-term.
9. Performing the test for exogeneity for the fixed-effects model for both the internally used and imported capital stocks, the following statistic values (t-Student) are obtained: 0.07 (1.87) and 0.09 (3.38) at the regional level and 0.03 (2.67) and 0.03 (1.77) at the provincial level. In both cases it is confirmed that these two variables are endogenous.
10. This is a common justification for the existence of spillover effects. Studies performed on a lower than the national scale obtain infrastructure elasticities smaller than those reported at the national level (García-Milá et al., Citation1993; Holtz-Eakin and Schwartz, Citation1995; Boarnet, Citation1998). Furthermore, Pereira and Roca-Sagales (Citation2003) obtain similar results confirming the relevance of the spillover effects from public capital in the case of Spanish regions. More recently, several studies have continued analysing the direct and indirect effects of public capital in regional growth. Solé-Ollé et al. (Citation2012) edited a special issue of Papers in Regional Science where several authors discuss the role of transport investment on growth and the localization of economic activity, and combining aspects related to productivity change and financing. This is the case, among others, with Rodriguez-Pose et al. (Citation2012) who study the case of Greek regions, highlighting the fact that public investment externalities seem to be more relevant for regional growth than direct public investment.
11. This reference as well as the associated Stata codes are available from Christian Hansen's webpage (http://faculty.chicagobooth.edu/christian.hansen/research/).
12. Considering the GDP in low-income provinces, it can be observe that it represents 32% of national GDP, while according to income per capita the sense of magnitude changes because in this case it represents 46%. In terms of population, 71% is affected if one takes into account GDP and 65% in low-income per capita provinces. Therefore, the negative spillover effect implies a higher effect in poor provinces when one translates those into per capita terms, although population percentage is less since this criterion. So, the correlation between them is 0.30.