Productivity Spillovers from Foreign Multinational Enterprises to Domestic Manufacturing Firms: To What Extent Does Spatial Proximity Matter?

Abstract

Mariotti S., Mutinelli M., Nicolini M. and Piscitello L. Productivity spillovers from foreign multinational enterprises to domestic manufacturing firms: to what extent does spatial proximity matter?, Regional Studies. This paper analyses the role of spatial proximity in productivity spillovers from foreign multinational enterprises (MNEs). Using a database of foreign affiliates located in Italy from 1999 to 2005, it is shown that local firms benefit from the presence of MNEs through both backward and forward linkages. It is found that: (1) spillover effects are more strongly associated with MNEs in knowledge-intensive business services and utilities; and (2) co-location positively affects productivity spillovers, and the effect is stronger for foreign affiliates that are distantly located. This is especially true in the service sector due to the role of temporary spatial proximity and other dimensions of proximity.

Mariotti S., Mutinelli M., Nicolini M. and Piscitello L. 外资跨国企业（MNEs）对国内製造厂商的生产力外溢：空间邻近性具有多大程度的影响力，区域研究。本文分析空间邻近性在外资跨国企业（MNEs）的生产力外溢中所扮演的角色。本文运用设立于意大利的外国子公司自 1999 年至 2005 年的数据库，显示地方企业同时藉由向后与向前连结，受益于跨国企业。本研究发现：（1）外溢效应在知识密集的服务业与事业中，与跨国企业的连结较强；（2）座落于同一区位对生产力外溢有正面的影响，但对于座落于一定距离之外的外国子公司的影响却较大。由于时空邻近性以及其他面向的邻近关係，上述的情形在服务业部门尤其如是。

Mariotti S., Mutinelli M., Nicolini M. et Piscitello L. Les retombées sur la productivité des entreprises multinationales étrangères au profit des entreprises manufacturières nationales: jusqu’à quel point la proximité spatiale importe-t-elle?, Regional Studies. Cet article analyse le rôle de la proximité spatiale pour ce qui est des retombées sur la productivité des entreprises multinationales étrangères. À partir d'une base de données auprès des filiales situées en Italie de 1999 jusqu’à 2005, on montre que les entreprises locales profitent de la présence des entreprises multinationales étrangères grâce aux relations en amont et en aval. Il s'avère que: (1) les effets de retombées sont plus fortement associées aux entreprises multinationales étrangères dans les services aux entreprises à forte intensité de connaissances et aux services de distribution; et (2) la co-localisation influence positivement les retombées sur la productivité, mais l'effet s'avère plus fort pour les filiales étrangères plus éloignées. C'est particulièrement évident dans le secteur des services en raison du rôle de la proximité spatiale temporaire et d'autres dimensions de proximité.

Mariotti S., Mutinelli M., Nicolini M. und Piscitello L. Produktivitätsübertragungen von ausländischen multinationalen Unternehmen an einheimische produzierende Firmen: Wie wichtig ist räumliche Nähe?, Regional Studies. In diesem Beitrag analysieren wir die Rolle der räumlichen Nähe bei Produktivitätsübertragungen von ausländischen multinationalen Unternehmen. Anhand einer Datenbank mit in Italien angesiedelten ausländischen Zweigfirmen im Zeitraum von 1999 bis 2005 zeigen wir, dass lokale Firmen durch Vorwärts- und Rückwärtsverknüpfungen von der Präsenz ausländischer multinationaler Unternehmen profitieren. Wir stellen fest, dass: (1) Übertragungseffekte in wissensintensiven Geschäftsdienstleistungen und Versorgungsbetrieben stärker mit multinationalen Unternehmen assoziiert sind und dass (2) sich ein gemeinsamer Standort positiv auf Produktivitätsübertragungen auswirkt, wobei der Effekt aber für entfernt angesiedelte ausländische Zweigfirmen stärker ausfällt. Dies gilt aufgrund der Rolle der temporären räumlichen Nähe und anderer Dimensionen der Nähe insbesondere für den Dienstleistungssektor.

Mariotti S., Mutinelli M., Nicolini M. y Piscitello L. Desbordamientos de productividad de empresas multinacionales extranjeras a empresas manufactureras nacionales: ¿es importante la proximidad espacial?, Regional Studies. En este artículo analizamos la importancia de la proximidad espacial en los desbordamientos de productividad de empresas multinacionales extranjeras. A partir de una base de datos entre 1999 a 2005 de filiales extranjeras ubicadas en Italia, demostramos que las empresas locales se benefician de la presencia de empresas multinacionales extranjeras a través de vínculos regresivos y progresivos. Observamos que: (1) los efectos de desbordamientos se asocian más con empresas multinacionales extranjeras en servicios comerciales con alto nivel de conocimientos y servicios públicos; y que (2) la ubicación conjunta afecta positivamente a los desbordamientos de productividad, pero el efecto es más fuerte en las filiales extranjeras que están situadas más lejos. Esto es particularmente cierto en el sector de servicios debido al papel de la proximidad espacial temporal y otras dimensiones de proximidad.

• Multinational firms
• Co-location
• Spatial proximity
• Productivity spillovers
• Customer–supplier interaction
• Vertical linkages

• 跨国企业
• 座落于同一区位
• 空间邻近性
• 生产力外溢
• 客户供应商互动
• 垂直连结

• Entreprises multinationales
• Co-localisation
• Proximité spatiale
• Retombées sur la productivité
• Interaction entre le client et le fournisseur
• Liens verticaux

• Multinationale Firmen
• Gemeinsamer Standort
• Räumliche Nähe
• Produktivitätsübertragungen
• Wechselwirkungen zwischen Kunden und Lieferanten
• Vertikale Verknüpfungen

• Sociedades multinacionales
• Ubicación conjunta
• Proximidad espacial
• Desbordamientos de productividad
• Interacción entre cliente y proveedor
• Vínculos verticales

JEL classifications:

• D24
• F23
• O19
• R15

Notes

1. Generally, MNEs also generate spillovers that do not directly change firms' productivity. For example, ‘market access spillovers’ occur when MNEs involuntarily transmit knowledge of foreign market conditions to local firms, thereby helping local firms enter into new businesses abroad (Blomström and Kokko, 1998). However, this paper is focused on spillovers that affect productivity.

2. Data refer to 2009.

3. NACE (Nomenclature statistique des activités économiques dans la Communauté européenne) is the industrial classification provided by EUROSTAT. In this paper, data are classified according to NACE revision 1.1.

4. Data on domestic companies refer to all firms operating in Italy, including foreign-owned firms.

5. There are several alternative ways of measuring TFP, including index numbers, data envelopment analysis, stochastic frontier analysis, instrumental variables estimation techniques and semi-parametric estimation techniques. For an exhaustive description of the advantages of using the Levinsohn and Petrin (2003) estimation procedure in the context of spillovers from FDI and linkages with MNEs, see Altomonte and Pennings (2009).

6. Quoting Griliches and Mairesse (1998), also reported by Levinsohn and Petrin (2003, p. 321):

The major innovation of Olley and Pakes is to bring in a new equation, the investment equation, as a proxy for ω (i.e. the productivity shock), the unobserved component of the error term. Trying to proxy for the unobserved ω (if it can be done correctly) has several advantages over the usual within estimators (or the more general Chamberlain and GMM [generalized method of moments] type estimators): it does not assume that ω reduces to a ‘fixed’ (over time) firm effect; it leaves more identifying variance in l (labour) and k (capital), and hence is a less costly solution to the omitted variable and/or simultaneity problem, and it should also be substantively more informative.

Additionally, unlike other alternative methods (e.g., the GMM instrumental variables approach), this approach also allows separate production functions across sectors to be specified easily, which is particularly relevant for a sectoral heterogeneous sample such as that considered in the present study.

7. Due to the small number of firms present in these few sectors, these industries were aggregated. Specifically, the food and tobacco industries (15 and 16), the paper products and printing and publishing industries (21 and 22), and the manufacturing not elsewhere classified (n.e.c.) and recycling industries (36 and 37) were aggregated. It is worth noting that these aggregations are those suggested by the NACE classification.

8. The industry and GDP deflators are available on the EUROSTAT website.

9. The authors thank an anonymous referee for stimulating them to clarify this issue.

10. Referring to the EUROSTAT scheme of territorial classification, Italian municipalities correspond to the NUTS-5 level (EUROSTAT, 2011).

11. This definition is consistent with standard definitions of urban areas in economic geography that define urban areas through commuting patterns. This definition is also consistent with the notion of a ‘functional region’, defined as ‘a territorial unit resulting from the organization of social and economic relations in that its boundaries do not reflect geographical particularities or historical events’ (Organisation for Economic Co-operation and Development (OECD), 2002). Italian LLAs also roughly follow the criteria used to define metropolitan statistical areas in the United States, travel-to-work areas in the UK, metropolitan areas and employment areas in France, and urban employment areas in Japan.

12. The correlation matrix is not reported for the sake of space. However, it is available from the authors upon request.

13. The full models, both backward and forward, were also estimated (the results are reported in Table A1 in Appendix A). However, as already observed in previous related empirical studies (e.g., Driffield et al., 2004), collinearity problems may weaken these estimates. In fact, satisfactory results for the manufacturing and utility services industries are obtained; for KIBS, a stronger impact from the forward channel is still observed, while the backward variable changes sign, albeit with a small coefficient. The authors thank an anonymous reviewer for stimulating them to investigate and clarify this issue.

14. More precisely, an increase of 1 SD of the relevant variable (i.e., KIBS_forw) causes a 0.597 increase in the standard variation of the dependent variable. Regarding the backward channel, local customers benefit more from foreign affiliates in utility services, as an increase of 1 SD of the relevant variable causes a 0.515 increase of the standard variation of the dependent variable.