Spatial dynamics in warehousing activities and - ART-Dev

Spatial dynamics in warehousing activities and agglomeration economies: lessons from the French case

Sophie Masson, Romain Petiot

Document de travail ART-Dev 2015-2 Avril 2015 Version 1

Spatial dynamics in warehousing activities and agglomeration economies: lessons from the French case Sophie Masson1, Romain Petiot2 1

Univ. de Perpignan Via Domitia, UMR ART-Dev F-66860, Perpignan, France, 2 Associate Professor, Univ. Perpignan Via Domitia, UMR ART-Dev F-66860, Perpignan, France Abstract The geographical dynamic of logistics infrastructures location is characterised by concentration of activities. This paper puts forward the hypothesis that the spatial concentration of warehousing activities can be explained by the presence of agglomeration economies. The econometric estimation of the link between labour productivity and density of warehousing activities conducted on French data reveals the role of intersectoral and intrasectoral externalities. We show that agglomeration economies have a minor impact on urbanisation and localisation diseconomies for warehousing activities. The analysis of productivity shows that the areas with the highest job densities are not systematically those with the highest productivity. Keywords: Logistics, Spatial concentration, Productivity, Agglomeration economies, Econometric estimation Titre Dynamiques spatiales des activités d’entreposage et économies d’agglomération : enseignements du cas français Résumé Les infrastructures logistiques sont des outils majeurs de production de la logistique et marquent fortement le territoire. Leurs dynamiques géographiques d’implantations se caractérisent par une concentration des activités en quelques points du territoire. Ce papier pose l’hypothèse que la concentration spatiale des activités logistiques d’entreposage peut s’expliquer par la présence d’économies d’agglomération comprenant les bénéfices nets d’économies de localisation et d’urbanisation. L’analyse porte sur des données françaises. Diverses mesures de la concentration de ces activités ainsi qu’une mesure de la spécialisation des espaces dans cette activité sont proposées confirmant la concentration spatiale logistique. L’estimation économétrique de la relation entre la valeur de la productivité locale au coût du travail et la densité des activités d’entreposage permet d’appréhender les externalités intersectorielles (ou économies d’urbanisation) et le rôle des externalités intrasectorielles (économies de localisation), approchées par la spécialisation sectorielle locale. Les résultats montrent globalement un très faible effet des économies d’agglomération et, en outre existence de déséconomies d’urbanisation et de déséconomies de localisation s’agissant des activités logistiques d’entreposage. Ainsi, l’analyse des différences de productivité entre espaces géographiques montre que les productivités les plus fortes ne se trouvent pas systématiquement dans les zones d’emploi les plus denses. Mots-clés : Activités logistiques d’entreposage, concentration spatiale, productivité, économies d’agglomération, estimation économétrique

Pour citer ce document :

Masson, S., Petiot, R. 2015. Spatial dynamics in warehousing activities and agglomeration economies: lessons from the French case. Document de travail ARTDev 2015-2. Auteur correspondant : [email protected]

1

Introduction

There has been a considerable increase in third-party logistics activities since the 1990s. Between 1993 and 2009 there was a 67% increase in the number of warehousing jobs in France, compared with 21% for jobs of all types during the same period. Logistics infrastructure consists of large systems that produce logistics activities (Abrahamson et al., 2003; Fender and Pimor, 2013), which create the added value resulting from the geographical distribution of goods in order to supply production units or commercial units. The functions of logistics infrastructure, traditionally referred to as warehouses or logistics hubs, are changing. They perform an increasing number of high added value operations, due to the intensification of corporate vertical disintegration. This process of outsourcing encourages the emergence of a market for logistics service providers. Their operational functions aside, logistics activities have an impact on local areas. Decisions about where to locate warehouses have a marked impact on freight transport demand, and strongly influence mode choice for freight transport (Bowen, 2008). It is particularly useful to attempt to understand the locational decisions of logistics firms as the demand for logistics space is expected to increase considerably in the advanced economies (Van den Heuvel, et al., 2013a). A better understanding of the spatial dynamics that affect logistics activities is therefore of benefit both to public decision-makers (in order to understand what attracts logistics activities) and firms (property developers, logistics service providers, etc.). Nevertheless, the academic literature that deals specifically with the locational decisions of logistics firms is meagre (Masson and Petiot, 2013; Van den Heuvel, et al., 2013a). In addition, Hesse and Rodrigue (2004) stress that research into freight transport and logistics is generally under-represented in regional science. While there is an abundant literature that analyses the links between transport and regions, few studies deal with logistics from the spatial standpoint and consider its geographical impacts even though, as part of the process of globalisation, it impacts strongly on the organization of regions. However, in recent years the topic has attracted more academic interest. Thus, Bowen (2008), Cidell (2010; 2011), Hesse (2007), Hesse and Rodrigue (2004, 2006), McKinnon (2009), Wagner (2010), Dablanc and Andriankaja (2011) and Dablanc and Rakotonarivo (2010) have investigated changes in the geography of logistics activities, highlighting the dual process by which logistics activities are becoming concentrated geographically at a few principal points in a country or continent while at the same time spreading outwards into urban peripheries, and stressing the vast amounts of land they require. Hesse and Rodrigue (2004, 2006), Jakubicek and Woudsma (2011), Masson and Petiot (2011, 2012b), Mérenne-Schoumaker (2007), Raimbault, et al. (2013), Savy (2006), Strale (2013a; 2013b) and Wackermann (2005) have analysed the drivers of the spatial dynamics of logistics activities. More specifically, Dablanc and Ross (2012), Masson (2013), Masson and Petiot (2013) and Raimbault, et al. (2013) have examined the strategies of stakeholders in the production process of logistics spaces, and considered the governance of these activities at local and regional level. Bhatnagar and Sohal (2005), Melo, et al. (2009) have conducted operations research analyses of the location of logistics firms, but their findings have been subjected to little empirical testing. Last, Sheffi (2012), Van den Heuvel, et al. (2013a; 2013b) have studied the understanding of the spatial concentration of logistics activities and questioned the public authorities about the effects of the colocation of these activities. This paper deals with the spatial dynamic of logistics warehousing activities and particular analyses their geographical concentration. This process, while it has been described a great many times, has rarely been measured. The aim is to better understand the principles which govern the location of logistics activities and, in particular, their tendency to become spatially 3

concentrated. We shall examine how agglomeration economies contribute to the concentration process. In addition to attempting to minimise traditional costs, the policy which is pursued with regard to the location of hubs also sets out to achieve economies of scale, as the organisation of logistics depends on the creation of circuits that consolidate flows. What advantages do these activities derive from grouping together? Do any agglomeration economies result from their spatial concentration? The aim of the paper is thus to examine how external economies of scale affect the spatial dynamic of logistics warehousing activities. First, a literature review will describe the spatial dynamic of warehousing activities and its determinants. We shall present an account of the geographical distribution and a measure of the spatial concentration of warehousing activities in France. Second, we shall describe the theoretical framework of the determinants of the spatial concentration of economic activities and the estimation strategy applied in the literature to measure agglomeration effects resulting from the concentration of economic activities. The role of intersectoral externalities (urbanisation economies) and intrasectoral externalities (localisation economies) in generating spatial concentration in warehousing activities is then measured. 2

The spatial dynamic of warehousing activities

The location of warehousing activities is receiving an increasing amount of attention in the literature. This highlights the existence of the spatial dynamic characterised by the two processes of concentration and sprawl that is driven by global and local determinants (2.1). Based on this brief literature review we shall describe the geographical distribution of warehousing activities in France and propose a measure of their geographical concentration (2.2). 2.1 The case of warehousing activities: major trends and determinants The literature highlights the major trends exhibited by the spatial dynamics of logistics activities. The first is the lasting trend to become concentrated in metropolitan areas. We can observe that logistics establishments are clustering together within zones that are limited to a few hundred hectares. This tendency for logistical activities to become spatially concentrated is apparent in all developed economies, be it in the United States (Dablanc and Ross, 2012; Cidell, 2010), the United Kingdom (McKinnon, 2009), the Netherlands (Van den Heuvel, et al., 2013), Belgium (Strale, 2013a; 2013b) or France (Masson and Petiot, 2012b; Raimbault, et al., 2013; Wemelbeke and Mariotte, 2007). This concentration has rarely been precisely measured. Researchers propose a number of different explanations for this tendency to concentrate in the metropolitan areas. Savy (2006) highlights the fact that logistics activities cannot be delocalised: the logistics activities that are related to local production and consumption must be necessarily produced locally. Thus, proximity to the markets of production and consumption is a decisive factor in the metropolitization of logistics activities. In addition to this, Hesse and Rodrigue (2004), Bowen (2008) and Strale (2010) stress that the geography of logistics activities is also affected by the importance of accessibility and integration within global transport networks. This concentration of logistics activities is also due to a process of centralisation which leads to the construction of fewer but larger warehouses (McKinnon, 2009). It is driven by the desire to minimise stocks (which leads to a reduction in the number of warehouses) and the consolidation of flows (leading to an increase in the size of warehouses). It is furthermore assisted by the fact that transport costs have remained durably low.

4

Another explanation for the concentration of logistics is forces of agglomeration other than market proximity, for example the attempt to achieve agglomeration economies (Masson and Petiot, 2012b; 2013; Van den Heuvel et al., 2013a; 2013b). In this paper we shall explore the way agglomeration economies affect the spatial dynamic of logistics activities. Furthermore, the trend towards the concentration of logistics is combined with one of logistics sprawl, i.e. an outward movement into the periphery of metropolitan areas. Cidell (2010) has performed an empirical study of this process in the metropolitan regions of the United States and shown there to be a trend for logistics to move towards the outskirts of the central city. The same trend has also been measured in the Paris region by Dablanc and Andriankaja (2011), and it has also been observed, in particular, in the Atlanta Megaregion (Dablanc and Ross, 2012). This process is partly explained by the greater availability and lower price of land in peripheral areas, which makes it possible to build ever larger warehouses. Bowen (2008) takes the view that the growing preference shown by logistics activities for the outskirts of metropolitan areas is the outcome of better access to transport networks, which has become a vital factor in the globalised economy, and the important role of trade between distant economies. Last, the move into peripheral areas is explained by the specific nature of logistical urban planning which comes into conflict with residential urban planning (Bahoken and Raimbault, 2012): the NIMBY effect leads local governments to move polluting logistics sites into the outskirts. 2.2 An evaluation of the concentration of warehousing activities in France The concentration of warehousing activities in France can be measured from a number of data sources. The Sitadel1 database shows the extraordinary expansion that has taken place since the end of the 1970s. Thus, in 2011, France had approximately 100 million m² of logistics space, 35 to 40% of which was in warehouses of more than 5000m² (data from Sitadel 2011). While recently there has been an overall reduction in the number of warehouses built annually, those built now are larger. In less than 10 years, the mean size of warehouses has increased by a factor of more than four (DTZ, 2011). The largest warehouses (with a surface area of over 35,000m²) account for a tenth of the total number of warehouses but almost 40% of the total surface area (Commissariat général au développement durable, 2012). Furthermore, the downward trend in stock levels that results from just-in-time management tends to be accompanied by a reduction in the surface areas that are dedicated exclusively to warehousing, this reduction is in reality less than proportional to the reduction in storage needs (Petiot and Masson, 2011). Furthermore, conventional warehousing space is being partially replaced by areas used for consolidation (crossdocking) and adding value to goods (co-manufacturing, co-packing) (Becker, 2003). Warehouses are gradually taking on new roles which are required by changes in corporate strategy resulting from the implementation of Supply Chain Management. Recently created warehouses are mainly located in large urban centres. It is very clear that logistics activities are very much concentrated in certain parts of the country. Three regions of France (Ile-de-France, Rhône-Alpes and Nord-Pas-de-Calais) contain more than a third of the country’s total surface area of warehouses (see Map 1 and Appendix 1). The Ile-de-France and Rhône-Alpes contain 28% of the nation’s logistics jobs and five regions (Ile-de-France, Rhône-Alpes, Nord-Pas-de-Calais, Centre and Picardie) account for 54% of the nation’s logistics jobs. Warehouse space is also concentrated. Six regions contain more than 50% of the total warehouse surface area, although they only account for 32% of Metropolitan 1

Base Sitadel: Information and automated processing system for basic data on dwellings and premises.

5

France’s surface area. This concentration of logistics therefore leads to a high degree of spatial differentiation as regards logistics services, with logistics activities concentrated in metropolitan areas (the metropolitan areas of the North of France and the Paris, Lyon and Marseille regions), transit corridors (the Rhône valley, the Lorraine corridor), infrastructure junction points and the main seaports, while other areas have a very low logistics density (Savy, 2006). Map 1. Distribution of warehousing surface area per region in 2010 (Data from Sitadel)

2

Unit: m Sources: Masson and Petiot (2012a)

We shall use a measure of logistical concentration based on the Unistatis jobs database held by the Pôle emploi Directorate for Studies and Statistics2. This provides information on establishments and their workforce according to the location (département-municipality), principal economic activity (APE code 3, 2008 NAF4 code rev. 2) and also gives the size of the establishment. We collected the jobs data between 1993 and 2009, for logistics activities for both the regions and the départements of France. We have examined in particular the warehousing activities that come under NAF 2008, rev. 2 category 52.1 "Warehousing and storage". 2

Pôle emploi is the French State body responsible for employment. Its data do not cover all members of the workforce and exclude individuals, in particular, self–employed workers and civil servants, who do not pay contributions to ASSEDIC (ASSociation pour l’Emploi dans l’Industrie et le Commerce) which pays out unemployment benefit to jobseekers. Within its limits it is an extremely reliable source of data. 3 APE (Principal Activity Exercised): the INSEE (the French National Institute of Statistics and Economic Studies) provides all companies and each of their establishments with a code characterising their principal activity on the basis of the French classification of activities (NAF rev. 2). 4 NAF (French Classification of Activities): a code assigned by the INSEE to each economic sector. It enables the Institute to assign an APE Code to every French firm and each of its establishments. NAF 2008 rev. 2 was introduced in January 2008 as part of the process of harmonising European classifications in the framework of NACE (Statistical Classification of Economic Activities in the European Community).

6

Table 1 shows that four regions of France contained 49.9%, 52.9% and 46.6% of warehousing jobs in 1993, 2000 and 2009 respectively. While the Ile-de-France, Rhône-Alpes and Nord-Pas-de-Calais regions are tending to maintain their lead over other regions with regard to warehousing activities, it is noteworthy that the Centre region is becoming increasingly differentiated from the others, having displaced Provence-Alpes Côte-d’Azur to join the "top 4" most important warehousing regions from the year 2000 onwards. For the last 10 years, Centre has been France’s fourth most important region for the creation of warehouses. It is geographically near to Ile-de-France, and has the benefits of a central location without its disadvantages (saturated infrastructure and land scarcity). Thus, the spatial dynamic of establishments seems to be marked by a transfer to the outer ring of the Paris basin (Centre, Picardie). Table 1. Percentage of warehousing activities in the regions of France (1993, 2000, 2009) % of total no. of Variation warehousing jobs 1993 2000 2009 1993-2009 Ile-de-France 19,00% 19.3% 17.1% -1.9% Rhône-Alpes 12.2% 11.3% 11.1% -1.2% Nord-Pas-de-Calais 9.5% 10.9% 9.5% 0,00% Provence-Alpes-Côte d'Azur 9.2% 6.1% 6,00% -3.2% Centre 6.9% 11.4% 8.9% 2,00% Picardie 4.86% 4.9% 7.4% 2.5% Total 61.7% 63.7% 60,00% -1.67% Sources: Masson and Petiot (2012a). Unistatis employment data

Rank 1993 1 2 3 4 5 7

2000 1 3 4 5 2 6

2009 1 2 3 6 4 5

Geographical concentration is defined as the distribution of the importance of the regions in a specific economic sector. A given sector is “concentrated” if a large proportion of its production is conducted in a limited number of regions. Several indicators are available for measuring the spatial concentration of economic activities (Krugman, 1991a; 1991b; Ellison and Glaeser, 1997; Combes, et al., 2006; Maurel and Sédillot, 1999; Brülhart, 2001; Albert, et al., 2011; Houdebine, 1999; Hallet, 2000; Goschin, et al., 2009; Aiginger and Leitner, 2002; Barlet, et al., 2010; Mouhoud, 2010). We have chosen to use the Hirschman-Herfindahl index and the Gini index (see Appendix 2). This decision was based on the performance of these indices in relation to the data are available to us. It is possible to supplement the measurement of geographical concentration by a measurement of regional specialisation which corresponds to the concentration of the activities of one or more sectors in an area. The index of specialisation we selected is Hoover’s index (also known as the Hoover-Balassa index) (see Appendix 3). Our analysis of logistical specialisation covers the period 1993-2009 and is based on the computation of two Hoover-Balassa indices, the first for regions and the second for départements. On this basis, we have constructed several maps showing specialisation. Maps 2 and 3 show the change in the regional specialisation of warehousing activities between 1993 and 2009.

7

Map 2. Degree of specialisation of the French regions for jobs in warehousing activities (1993)

Sources: Masson and Petiot (2012a). UNISTATIS employment data

Map 3. Degree of specialisation of the French regions for jobs in warehousing activities (2009)

Sources: Masson and Petiot (2012a). UNISTATIS employment data

It can be seen that in 1993, the Haute-Normandie region had a very high level of specialisation, as did, though to a lesser degree, Nord-Pas-de-Calais, Provence-Alpes-Côte d’Azur, Alsace and Rhône-Alpes. In 2009, the same specialisation structure can be observed 8

except for the Rhône-Alpes region which no longer seems to be more specialised than the rest of France. However, the specialisation index of the Picardie and Centre regions has increased sharply. This relative specialisation is explained by a number of factors: the presence of major seaports (Le Havre and Rouen in Haute-Normandie and Marseille in Provence-Alpes-Côte d’Azur) which encourage trade and require supporting logistics activities, location on a border or a major trade corridor (Nord-Pas-de-Calais, Alsace and Rhône-Alpes), and closeness to a zone of very dense population and economic activity (Ile-de-France). Map 4. Degree of specialisation of the French départements for warehousing jobs (1993)

Sources: Masson et Petiot (2012a). UNISTATIS employment data.

9

Map 5. Degree of specialisation of the French départements for warehousing jobs (2009)

Sources: Masson and Petiot (2012a). UNISTATIS employment data.

Observing specialisation in warehousing activities at département level reveals a number of interesting spatial dynamics. In 1993, the highly specialised départements were those in the Haute-Normandie region (départements 27 and 76 on Map 5) as a result of the activity of the port of Le Havre, the Départements in the outer ring of the Paris basin (60, 77, 78, 91), in particular Loiret (45), and the Départements in the Rhône Valley (13, 26, 84, 69). This high degree of specialisation may be explained by the presence of a large amount of transport infrastructure, a high concentration of population and activities (the effect of markets and size) and location on a trade and transit corridor. In 2009, we can observe an increase in the specialisation of the départements in the outermost ring of the Paris basin and a confirmation of concentration in the Rhône Valley départements. We have analysed the geographical concentration of three categories of logistics activities over the period 1993-2009 initially on the basis of regions, then on the basis of départements. The first category is so-called “extended logistics activities” and covers all transport and logistics activities (i.e. section H "Transport warehousing" in NAF 2008 rev.2). The second category is "extended hub activities" which covers activities associated with passage through a logistics hub (this combines all the activities in category 52 of NAF 2008 rev. 2 "Warehousing and auxiliary transport services"). Last, the category that we have already considered above contains only “warehousing activities” Table 2 sets out the results we obtained with the absolute (Herfindhal) and relative (Gini) indexes of concentration and the regional breakdown. Table 3 shows those obtained with the breakdown into départements.

10

Table 2. Concentration of logistics activities - for data on regions (1993, 2000, 2009) Annual variation Activity

Index

1993

2000

2009

1993-2009 (2000-2009)

Extended logistics activities

Extended hub activities

Warehousing activities

Gini

0.11

0.1

0.06

Herfindhal

0.09

0.08

0.1

Gini

0.18

0.16

0.15

Herfindhal

0.11

0.11

0.1

Gini

0.25

0.27

0.24

0.1

0.1

0.08

Herfindhal

-3.66% (-5.02%)

0.71% -1.44%

-1.24% (-1.11%)

-0.46% (-0.98%)

-0.29% (-1.13%)

-0.76% (-1.65%)

Sources: Masson and Petiot (2012a). Données emplois Unistatis.

From Table 2 we can draw some preliminary conclusions about the concentration of logistics activities in the regions of France. The tendency for concentration as measured by the Herfindhal index increased slightly over the period in the case of extended logistics activities. The relative index changed in a distinct manner, and shows a reduction in the concentration of logistics activities. While these activities have tended to become more concentrated (rising absolute indices), their geographical distribution has, however, drawn away from the mean for other sectors (reduction in the relative index). This reduction is very marked in the case of the extended logistics activities, which shows that transport and logistics activities tended to spread over the whole national territory during the period 1993-2009 (and to an even more marked degree in the period 2000-2009). This followed the application of deregulation policies in the European transport sector in the 1980s and increased outsourcing of transport and logistics activities. However, the reduction in concentration is much lower for extended hub activities and much higher for warehousing activities which are more concentrated at regional level than logistics and transport activities in general. Table 3. Concentration of logistics activities - for data on départements (1993, 2000, 2009) Activity

Index

1993

Extended logistics Gini activities Herfindhal Extended hub activities

2009

Annual variation 1993-2009

0.2

0.22

0.85%

0.02

0.02

0.36%

Gini

0.33

0.32

-0.31%

Herfindhal

0.03

0.02

-0.57%

Warehousing Gini 0.42 0.43 activities Herfindhal 0.03 0.02 Sources: Masson and Petiot (2012a). Unistatis job data.

0.18% -1.34%

In contrast, when concentration is measured on the basis of the breakdown into départements, it tends to become more stable, or even increase (see Table 3). It can also be 11

seen that the concentration at département level is proportionally greater than that observed at regional level. Last, warehousing activities are relatively concentrated, and there is a slight tendency for them to become more so. 3

The role of agglomeration economies in the spatial dynamic of logistics warehousing activities: theoretical framework and econometric analysis

Does the concentration of logistical activities increase their productivity? This question echoes one which has attracted much interest in the field of regional economics, namely whether spatial concentration increases efficiency, and if so, under what circumstances. A major line of research focuses on the role of agglomeration economies with regard to the concentration dynamic of economic activities. Its goal is to verify the influence of agglomeration economies (3.1). We shall present the data and identify some relevant variables with a view to conducting empirical tests to confirm the role of agglomeration economies on the concentration of logistics warehousing activities (3.2.). Based on specifications derived from the empirical literature, we shall then estimate the role of urbanisation economies and localisation economies (3.3). 3.1 Spatial concentration and agglomeration economies: the theoretical framework Generally speaking, the geographical location of a sector can be explained by three types of factors, the relative importance of which depends on the type of activity (Mouhoud, 2010):  the existence of tangible and intangible benefits that are to the advantage of the sector in question; 

the existence of positive externalities;



the ratio between firms and clients.

The concentration of logistics activities is the outcome, in particular, of the need to benefit from externalities5. Agglomeration economies are comparative productivity advantages which a set of firms receives in one region rather than another, because of its size and structure. The hypothesis that the geographical agglomeration of economic activities generates productivity gains is based on Marshall’s conception of externalities (1890) (see Catin, 1991, 1995, 1997; Fujita and Krugman, 2004; Fujita and Thisse, 1996; 2002; Glaeser et al., 1992; Rosenthal and Strange, 2004). In terms of productivity level, the effectiveness of concentration is interpreted as an economy of scale that is external to the firm and internal to the region in question. Traditionally, agglomeration economies are divided into two types: urbanisation economies and localisation economies. Urbanisation economies are due to population concentration and the presence of a large number of firms from different sectors in a location: these economies are external to the firm and the industry to which it belongs, that is to say they are intersectoral agglomeration economies. The presence of a large number of sectors of activity in a given zone increases the frequency of technological transfers from one sector to another. Industrial diversity can foster the exchange of ideas and technology which generates innovation and higher productivity. 5

Baudewyns (2005), in a brief overview of the determinants of the spatial agglomeration of economic activities, mentions (i) the interaction between transport costs and economies of scale (Krugman, 1991b), (ii) technological externalities, and (iii) pecuniary externalities resulting from interactions with the market and strategic factors.

12

These economies are also referred to as Jacobs externalities (Jacobs 1969). Little research has analysed the effect of urbanisation economies on a sector-by-sector basis. Nevertheless, there are exceptions, for example the work by Brülhart and Mathys (2008) and Foster and Stehrer (2009). This research reveals that urbanisation had significantly positive effects in all the sectors that were considered apart from agriculture (Foster and Stehrer, 2009), for which density within a region has a negative impact on productivity. Location economies are due to the presence of a large number of firms belonging to the same industry in the same geographical area. These are intrasectoral agglomeration economies. The benefits associated with the proximity of other producers depend on the sharing of facilities and public goods, access to a larger range of suppliers and a specialised pool of labour as well is the sharing of risks. Spatial proximity facilitates immediate access to competitors in the same sector and allows firms to keep abreast of market information for their negotiations with clients and suppliers. MAR (Marshall-Arrow-Romer) externalities also exist, which can also be known as specialisation economies, which involve the way sectoral specialisation affects the innovation brought about by the spatial concentration of firms belonging to the same sector of activity, in which a network of ties forms between backward and forward of a given sector (Arrow, 1962; Marshall, 1890; Romer, 1986). This encourages spillover of intrasectoral knowledge created by the interaction between economic agents which share information and common knowledge. In the case of warehousing activities, which require a relatively unskilled workforce and whose technology is relatively standardised, externalities of this type are doubtless low, not to say non-existent. Ultimately, it is considered that the pressures of competition which oblige firms in a given sector to innovate or disappear also lead to an increase in productivity. The last type are Porter’s externalities (1990) based on the theory of local innovation clusters which states that growth is driven by local intrasectoral competition between firms in a cluster. Brulhart and Mathys (2008) have shown that most of the time the effects of the sector itself are negative, which suggests that there are congestion diseconomies (except in financial services where density in the sector has strong positive impacts on productivity). The literature also contains a wide range of analytical and statistical work that attempts to identify the links between productivity and the economic size of regions. One important line of research attempts to verify the influence of agglomeration economies and their variety on, usually industrial, productivity: Catin (1991, 1995, 1997) on the French regions, Ciccone (2002) on the European regions, Ciccone and Hall (1996) on the American counties, Combes (2000), Combes, et al. (2006), Combes, et al. (2008; 2011) on industries and services in France at the employment area6 level, Rosenthal and Strange (2001) on American industries. Melo, et al. (2009) have performed a meta-analysis of this work. As yet, no study has dealt with the specific sector of logistics activities. 3.2 The situation and the selected variables 3.2.1 The sample Our data were sourced from the most recent annual survey of transport firms which was conducted in 2007 by the Department for Observation and Statistics (SOeS) which is part of the Commission for Sustainable Development at the French Ministry for the Environment, Energy, Sustainable Development and Regional Planning. The data related to French firms operating in metropolitan France whose head offices were also in metropolitan France. They come under section H "Transport and warehousing" of NAF rev. 2008. The database contains 6

An employment area is a geographical area within which most of the labour force lives and works, and in which establishments can find the main part of the labour force necessary to occupy the offered jobs.

13

13,396 observations. As our study only concerns warehousing activities, we have restricted ourselves to category 52.1 "Warehousing and storage" which contains the sub-categories 52.10A "Refrigerated warehousing and storage" and 52.10B" Non-refrigerated warehousing and storage". Our basic sample thus contained 682 observations, i.e. 112 refrigerated warehousing firms and 570 non-refrigerated warehousing firms. Database clean-up was performed to remove duplicates, observations with no data or without data on the variables which were of interest for our study, namely production and production factors, and left us with just 495 observations. Finally, as our aim was to explore the link between the productivity of the firm and its spatial location we were forced to eliminate 50 observations that related to multi-establishment firms. The database we finally worked with was made up of 445 clearly localised single-establishment firms (73 refrigerated warehousing firms and 372 non-refrigerated warehousing firms, or respectively 16 and 84% of the sample we analysed). 3.2.2 The selected variables Analysis of the productivity of a firm is performed on the basis of the ratio between the output (the amount of production or added value) and the inputs (all the factors of production that are mobilised to create this production or added value). The inputs are of two types: labour or capital.  Output The database we obtained from the national survey of firms contained information about the level of production, expressed in thousands of Euros. This is the value of the goods and services created and invoiced by the firm during the financial year, in other terms its turnover during the financial year. We have confined ourselves to this last production variable in order to estimate the firm’s output.  Inputs Labour Information concerning labour may be either physical (total full-time workforce in each firm) or monetary (payments made for each observation). The shortage of reliable information concerning workforces has led us to use the net cost of labour, namely the sum of salaries and social security contributions. Capital In order to measure inputs, we considered both tangible assets (land, warehouses, technical facilities, transport equipment, etc.) and intangible assets (goodwill, customer portfolio, right to renew a lease, etc.). The database provides the level of each type of asset at the beginning and end of the financial year. We then used these two items of information to construct a mean variable for the financial year. Amongst other factors, we took account of the intermediate consumption that enabled the firm to achieve its production during the financial year. This consumption includes, in particular, rent paid by the firm in order to exercise its activity, which allowed us to take account of the fact that firms often rent their warehouses.  The dependent variable Our goal was to test for agglomeration economies (urbanisation and localisation economies) due to the location of logistics warehousing activities. We have postulated that if such 14

agglomeration economies exist, they express the advantage to firms of concentrating in the same area, which can be measured by an impact on the performance of firms which we measure on the basis of productivity gains. It is reasonable to test these agglomeration economies with regard to 3 productivity indicators: the productivity of labour, the productivity of capital and the total productivity of the factors of production. In the case of the productivity of labour, we have constructed a productivity indicator that takes account of the cost of labour. This corresponds to the amount of production per unit of salary. The productivity of capital includes all the resources used for production with the exception of labour, namely assets (tangible and intangible) and intermediate consumption. Measuring the productivity of capital however poses some problems resulting from the quality of the information in the database. As a corollary, the total productivity of the factors which expresses the production resulting from all the production factors at play, namely labour and capital, cannot be explained by our analysis in view of the lack of reliable data about the productivity of capital. We shall therefore estimate the impact of location on the productivity of warehousing firms on the basis of single-factor productivity relating to the cost of labour.  The explanatory variables Urbanisation economies Urbanisation economies were estimated by evaluating the effect of job density in the employment zone in which the establishment is located. This is an indicator of the economic activity of all the sectors present in the zone. Localisation economies Localisation economies were estimated by evaluating the effect of the job density in a given sector in the employment zone. As a measure it can be refined by testing the effects of intrasectoral job density in the broad sense and job density for section H "Transport and warehousing". It allows us to test the effect of the proximity of all the activities in the transport and logistics sector. In addition, we have tested the effects of intrasectoral job density in the strict sense by limiting the analysis of job density to category 52.1 "Warehousing and storage" in the employment zone. Furthermore, the effects of the employment area’s specialisation in jobs in section H and in jobs in category 52.1 compared to jobs in section H were estimated in order to test the effect of the intensity of jobs in the sector on the productivity of firms. 3.2.3 Statistical description of the data The descriptive analysis of the data that make up our sample (see Table 4) highlights the high degree of dispersion between the observations, whether they relate to production, full-time equivalent workforce, or pay. Our survey contained a large variety of economic situations ranging from very small firms (37% of the firms had fewer than 9 employees) to large firms (2% of the firms had more than 250 employees). Very small and small firms are prevalent (83% of firms had fewer than 49 employees). This variety within the sample is also present in the cost of labour productivity, although to a less marked degree. Density variables exhibited a high degree of dispersion, in particular job density for sector H. This variable describes a wide variety of economic situations from employment zones with few jobs in the sector to employment zones where the sector has a strong presence. The situation is less straightforward in the case of job density in warehousing and storage activities which seems low in the employment areas in our sample. Nevertheless, the transport sector in the employment areas in question seems, on average, to account for a higher 15

proportion of warehousing and storage jobs than the transport sector does of jobs in all the other sectors. Table 4. Descriptive statistics of the variables in the employment areas N=445 Data Production (thousand Euros) Workforce (Full-time equivalent) Payments (thousand Euros) Dependent variable Cost of labour productivity Explanatory variables Job density Job density for Section H Job density group 52.1 Job specialisation in Section H Job specialisation in group 52.1 2 Size of employment zone (km )

Minimum Maximum

1st Quartile

Median

3rd Quartile

Mean

11 1 3

60,185 501 13,268

719 7 217

1,781 15 515

4,378 37 1,276

0.49

74.23

2.34

3.39

5.07

4.76

8.03 0.16 0.00 0.04 0.02

1,831.43 297.87 1.47 20.52 38.79

43.41 1.98 0.02 0.78 0.63

78.97 3.88 0.06 0.91 1.04

250.04 15.55 0.20 1.04 2.29

183.87 13.14 0.15 1.29 1.82

122.6

8,752

1,121.30

2,320.40

4,413.40

Variance (n-1)

3,761.17 34,674,581.26 32.58 2,892.11 1,081.88 2,762,261.65

Stand. dev. Coefficient (n-1) of variation 5,888.51 53.78 1,662.01

1.56 1.65 1.53

35.84

5.99

1.26

69,766.10 1,207.15 0.06 1.47 10.07

264.13 34.74 0.24 1.21 3.17

1.44 2.64 1.57 1.07 1.74

2,841.82 4,461,816.21

2,112.30

0.74

Analysis of the geographical distribution of the dependent variable, i.e. cost of labour productivity, gives some counter-intuitive findings. Maps 6 and 7 show that the productivity of warehousing is not at its highest in the regions with the greatest density of warehouses. For example, the mean productivity of the warehouses in Ile-de-France is not the highest. While the distribution of productivity according to size of urban area (Graph 1) shows a slight tendency for productivity to increase with size of urban area (except for group 05 municipality in an urban area with 35,000 to 49,999 inhabitants), it also appears to show that above a certain critical size (municipality in an urban area with at least 500,000 inhabitants), productivity falls as the size of the urban area increases. We can therefore make the hypothesis that congestion effects have an impact when an urban area exceeds a certain size. The distribution of the cost of labour productivity according to the size of the employment zone (Graph 2) does not reveal any really significant trend.

16

Map 6. Regional distribution of single establishment firms in category 52.1 "Warehousing and storage"

Map 7. Regional distribution of the productivity of labour in single establishment firms in category 52.1 "Warehousing and storage"

Graph 1. Distribution of productivity at the labour cost according to size of urban area

Graph 2. Distribution of productivity at the labour cost according to size of employment area 6

9

Productivité moyenne au coût du travail

Productivité moyenne au coût du travail

8 7

6 5 4 3 2

5

4

3

2

1

1 0

0 00

01

02

03

04

05

06

07

08

09

1

10

2

3

4

5

6

7

8

9

10

Code taille Zone d'Emploi

Code taille Aire Urbaine

Key: Urban Area Size Code (ad hoc classification) 1 – EZ with less than 20,000 jobs; 2 – EZ with between 20,000 and 34,999 jobs; 3 – EZ with between 35,000 and 49,999 jobs; 4 – EZ with between 50,000 and 99,999 jobs; 5 – EZ with between 100,000 and 149,999 jobss; 6 – EZ with between 150,000 and 249,999 jobs; 7 – EZ with between 250,000 and 449,999 jobs; 8 – EZ with between 500,000 and 849,999 jobs; 9 – EZ of Lyon; 10 – EZ of Paris.

Key: Urban Area Size Code (INSEE classification) 00 - Municipality not in an urban area; 01 - Municipality in an urban urea with less than 15,000 inhabitants; 02 - Municipality in an urban area with between 15,000 and 19,999 inhabitants; 03 - Municipality in an urban area with between 1 20,000 and 24,999 inhabitants; 04 - Municipality in an urban area with between 25,000 and 34,999 inhabitants.; 05 - Municipality in an urban area with between 35,000 and 49,999 inhabitants; 06 - Municipality in an urban area with between 50,000 and 99,999 hab.; 07 - Municipality in an urban area with between 100,000 and 199,999 hab.; 08 - Municipality in an urban area with between 200,000 and 499,999 hab.; 09 - Municipality in an urban area with between 500,000 and 9,999,999 hab.; 10 - Municipality in Paris urban area

Sources: graphs anf maps constructed using data from EAE 2006

3.3 Estimation method and empirical results The most practical estimation approach for quantifying the productivity gains associated with an increase in the geographical concentration of activities is the reduced form with log-linear

17

specifications (Combes et al., 2006; Combes and Lafourcade, 2012)7. The specifications we have adopted make it possible to differentiate between the effects of urbanisation externalities and those of locational externalities. We have used a Cobb-Douglas function to model each of the relationships. This allows us to estimate the impact on cost of labour productivity as an elasticity using the ordinary least squares method. 3.3.1 Testing the effects of urbanisation economies In the literature, it is the density of economic activities rather than the overall size of the market which is seen as decisive for productivity. The reason for this is that the overall size of the market is more sensitive to the spatial division which is applied while density is a more continuous variable (Combes and Lafourcade, 2012). The first paper that considered the impact of job density on the productivity of firms was that by Ciccone and Hall (1996). Following their example, and like much other research, we have used a log-linear relationship to measure urbanisation economies, based on the link between job density in the employment area and productivity at the cost of labour8 : (1) where denotes the productivity at the labour cost for firm located in employment area and denotes the job density in employment zone . The term is a constant, is the coefficient of elasticity to be estimated and an error term that expresses everything that cannot be explained by the density. The use of logarithms means we can estimate directly the elasticity of the dependent variable, i.e. the percentage variation in productivity brought about by a 1% increase in density. Specification (1) therefore allows us to evaluate the proportion of the differences in productivity that can be attributed to density differences between zones, or the potential impact on productivity of an increase or reduction in the density in a zone. Although the correlation between the logs of job density and the surface area of the employment areas is fairly low (- 0.23), the densest employment areas nevertheless tend to be the smallest. Consequently, as the dispersion of the surface areas of employment areas is quite high, we have, following Barbesol and Briant (2008), included the surface area of the employment area in specification (1) in order to estimate the influence of the size of the employment area and the effect of density in the case of a given surface area. We have thus tested a measure after adding the surface area of the employment area as a variable: (2) where denotes the surface area of employment area . 7

Other types of approach have recently been adopted to test the effects of agglomeration economies (Combes and Lafourcade, 2012) such as so-called structural forms based on the equilibrium relationships predicted by the models of the New Economic Geography. These link productivity, or more precisely salaries, to the geographical concentration of activities. In most cases they have been derived from monopolistic competition models like the Dixit-Stiglitz model (Dixit and Stiglitz, 1977), or the seminal models developed by Krugman (1991b) and Krugman and Venables (1995). 8 The theoretical foundations of this specification have been described by Combes et al. (2008). The study in question assumed that firms in all regions and sectors produce at constant marginal costs. These costs take account of salaries and the cost of the other inputs. It is assumed that the technology and the efficiency of work are specific to each firm. The firm’s profit is the sum of its sales on the markets minus the cost of the production factors and trade-related costs (including transport) incurred in order to supply external markets. The firm’s sales correspond to its volume of production and the profit derived from the sale of each unit, which we could call, perhaps mistakenly, the "price" of the product (for the firm). This price represents the mean benefit the firm derives from each unit sold on the destination markets, net of transport costs and intermediate consumption. This “price” is higher when the firm is positioned in or near a large market as such proximity reduces its supply costs. On the other hand, it is lower when competition is high, as is the case on large markets. This effect counteracts the first, and acts as a factor of dispersion. It can be shown that the first order condition for the optimum use of labour by firms, namely when each employee’s nominal salary is equal to their marginal productivity, gives a salary equation that can be used to estimate agglomeration effects. The nominal salary given by this equation is directly proportional to the worker’s individual skills and increases with the technological level of the firm and its “price” and decreases with the cost of the inputs other than work. The density of a zone can therefore increase the productivity of the firms located there, either because it improves their technology or the efficiency of their workforce, or because it increases the price of their goods, or alternatively because it reduces the price of their inputs because of the transport cost savings due to their greater closeness to purchasers (consumers or sub-contractors). On the other hand, density can also reduce the productivity of firms by creating conditions that are conducive to the appearance of congestion on the market for some inputs, leading to an increase in their cost.

18

Tables 5 and 6 set out the results of the regressions. The data for the employment area of Paris can interfere with the interpretation of the results because of its very high job density. We have therefore conducted tests of two different types, first with the full database and second with the full database minus the Paris employment area. The correlation between the logarithms of the job density and the surface area of the employment areas is in this case very strong (- 0,45). Table 5. Estimations of the effects of urbanisation according to specification (1) Dependent variable: ln PTiz N Constant a

Full Database

(0.128)

(< 0.0001)

ln DEz

-0.0619**

-0.0835**

(0.024)

(0.00565)

445 1.564

Table 6. Estimations of the effects of urbanisation according to specification (2)

Database minus the Paris ZE

415 1.646***

Dependent variable: ln PTiz N Constant a

Full Database

(< 0.0001)

(< 0.0001)

ln DEz

-0.0556**

-0.0813**

(0.04810)

(0.016)

lnSURF z

R² Adjusted R²

Jarque-Bera Statistic White's statistic Durbin-Watson Statistic dL of D-W dU of D-W

0.012 0.009

0.018 0.016

136.375***

126.013***

(< 0.0001)

(< 0.0001)

0.6444 1.843

1.3111 1.888

1.83987 1.848875

1.83404 1.8437

R² Adjusted R² F-statistic Jarque-Bera Statistic White's statistic Durbin-Watson dL of D-W dU of D-W

445 1.273***

Database minus the Paris ZE

415 1.591***

0.034

0.006

(0.328)

(0.888)

0.0136 0.0092 3.056

0.018 0.0137 3.872

(0.048)

(0.022)

138.764***

126.725***

(< 0.0001)

(< 0.0001)

3.207 1.8453

1.4893 1.8881

1.83535 1.85341

1.829195 1.84856

The figures in brackets are the critical probabilities for each statistic *, **, *** indicate that the null hypothesis has been rejected at a significance level of, respectively, 10, 5 and 1%.

The first striking result is the weakness of the relationship between job density in the employment area and the productivity at the labour cost in the case of warehousing activities (see Tables 5 and 6). Nevertheless, this relationship is significantly negative. Thus, the elasticity of the productivity at the labour cost and job density is 6.19% for the full database and 5.56% for a given employment area surface (as the impact of the latter on productivity is not significant). The model gains a little explanatory power when only the observations outside Paris are tested. The elasticity of the productivity at the labour cost to job density thus stands at 8.35% for the entire database and 8.13% for a given employment area surface. We can therefore suggest that the productivity at the labour cost for single-establishment firms falls slightly as job density rises. A doubling of job density leads to a reduction of approximately 6 to 8% in the productivity at the labour cost depending on the sample and the specification. Consequently, the location of warehousing firms in dense zones does not improve their productivity. On the contrary, the results show that the intensity of economic activities in the zone of location is significantly counter-productive. Strong labour demand in an employment zone therefore reduces the performance of the warehousing firms located in it. This is because the dynamic nature of local activities creates competition that increases salaries. The impact of the cost of labour productivity is naturally negative. Consequently, competition seems to have a greater effect than urbanisation economies because of a congestion effect on the labour market.

19

3.3.2 Testing the effects of localisation economies The role of locational externalities depends on the hypothesis that firms are more productive in zones where their industry is more highly concentrated. In this case it is not the density of the local economy which is responsible for productivity gains, but more specifically that of the economic sector in question. Initially, we shall examine localisation economies by applying the same approach we used in order to investigate urbanisation economies, namely exploring the link between the productivity at the labour cost and intrasectoral job density. However, the variable which is the most commonly used to capture intrasectoral externalities is the extent to which the employment area is specialised in the sector of activity in question, which we can measure on the basis of the proportion of local economy that is in the sector in question (Glaeser, et al., 1992).  The role of intrasectoral job density In the case of density, the models estimate the intrasectoral effect in the strict sense (firms belonging to category 52.1 "Warehousing and storage") and in the broad sense (firms that come under section H "Transport and warehousing"). Measurement of the effect of localisation economies in the broad sense is based on the link between the job density for sector H in the employment area and the productivity at the labour cost: (3) where denotes the job density for sector H in employment area . Likewise, measurement of the effect of localisation economies in the strict sense is based on the link between the job density for category 52.1 in the employment area and the productivity at the labour cost: (4) where denotes the job density for category 52.1 in employment area . Tables 7 and 8 summarise the results of the regressions. The correlation between the logs of the job density for section H and the surface area of the employment areas is slightly higher (- 0.25) in the case of the full database, and slightly lower for the database that excludes the observations for the employment area of Paris (- 0.41) than it was for the results of the previous analysis that related to urbanisation economies.

20

Table 7. Estimations of the effects of location according to specification (3) Dependent variable: ln PTiz N Constant a

Full Database

(< 0.0001)

(< 0.0001)

lnDEH z

-0.0629**

-0.081***

(0.0108)

(0.00361)

lnSURF z R² Adjusted R² F-statistic Jarque-Bera Statistic White's statistic Durbin-Watson Statistic dL of D-W dU of D-W

Table 8. Estimations of the effects of location according to specification (4) Dependent variable: ln PTiz N Constant a

Database minus the Paris ZE

445 1.172***

415 1.388***

0.028

0.001

(0.427)

(0.971)

0.0194 0.015 4.374

0.0248 0.0201 5.236

(0.013)

(0.00568)

130.937***

121.248***

(< 0.0001)

(< 0.0001)

1.414 1.85175

2.44251 1.061

1.83535 1.85341

1.829195 1.84856

lnDEE z lnSURF z R² Adjusted R² F-statistic Jarque-Bera Statistic White's statistic Durbin-Watson dL of D-W dU of D-W

Full Database

Database minus the Paris ZE

445 4.498***

415 1.026***

(< 0.0001)

(0.00029)

-0.026

-0.050*

(0.2697)

(0.030)

0.022

0.014

(0.5789)

(0.720)

0.0047 0.0002 1.047

0.0158 0.0110 3.309

(0.352)

(0.0375)

134.865***

129.668***

(< 0.0001)

(< 0.0001)

4.634 1.7792

10.308 1.8502

1.83535 1.85341

1.829195 1.84856

The figures in brackets are the critical probabilities for each statistic *, **, *** indicate that the null hypothesis has been rejected at a significance level of, respectively, 10, 5 and 1%.

Once again, the treatments reveal a strikingly weak link between intrasectoral job density, whether in the broad or strict sense, and the productivity at the labour cost for warehousing activities (see Tables 7 and 8). The effect of job density in sector H on the productivity at the labour cost is nevertheless very significantly negative. Thus, the elasticity of the productivity at the labour cost to the job density for sector H stands at 6.29% for the full database and 8.1% when only observations from outside the employment area of Paris are tested. The effect of job density in category 52.1 is however not significant in the case of the full database, and only slightly significant for the database without the Paris employment area. Thus, while there is no conclusive effect with regard to localisation economies for the sector in the strict sense, the impact of co-location is significantly negative for localisation economies in the broad sense. In addition to a congestion effect for jobs that exerts upward pressure on salaries, in particular from the intrasectoral standpoint, there is also an effect due to the nature of the activities. Sector H includes, amongst other types of firm, freight transport firms and providers of services that are ancillary to transport, in other words firms that have taken on logistics functions, so-called 3PL firms, which are both customers and competitors of the warehousing firms. The latter are always “price-takers”, in particular because compared to 3PLs they can only offer a very small range of support services. Warehousing firms are therefore subject to the market and its changing nature, which generates pressure on the prices of their services. As in our case production is measured by turnover, it is negatively affected by an intensification of competition.  The role of job specialisation The effect of localisation economies can also be tested by considering the link between job specialisation in the employment area and the productivity at the labour cost: (5)

21

where denotes the job specialisation for sector H in employment zone . We shall not report the tests conducted on the effect of job specialisation for category 52.1 "Warehousing and storage” in employment zone as no conclusive result was obtained. Table 9. Estimations of the role of job specialisation for sector H in the location effects according to specification (5) Dependent variable: lnPT iz N Constant a lnSPECH z R² Adjusted R² Jarque-Bera Statistic

Full Database

Database minus the Paris ZE

445 1.278*

415 1.248***

(0.031)

(< 0.0001)

-0.1496**

-0.1475**

(0.021)

(0.02354)

0.012 0.0097 136.473***

0.0124 0.01 131.755***

(< 0.0001)

(< 0.0001)

1.2408 1.84816

1.2248 1.837

White's statistic Durbin-Watson Statistic

dL of D-W 1.83987 1,83404 dU of D-W 1,848875 1,8437 The figures in brackets are the critical probabilities for each statistic *, **, *** indicate that the null hypothesis has been rejected at a significance level of, respectively, 10, 5 and 1%.

As previously, the link between the productivity at the labour cost and the employment area’s relative specialisation in sector H jobs is low (see Table 9). The elasticity of the productivity at the labour cost to job specialisation for sector H is nevertheless almost 15%, confirming the significant existence of localisation diseconomies as described by Combes and Lafourcade (2012) who mention that the geographical concentration of a sector of activity can increase competition between firms in the sector. If such localisation diseconomies are strong enough, they may outweigh localisation economies. 3.3.3 Testing the effects of agglomeration economies The last model allows us to test the impacts of agglomeration effects: (6) The results for this specification (6) are summarised in table 10.

22

Table 10. Estimations of agglomeration effects according to specification (6) Dependent variable: lnPT iz N Constant a

Full Database

lnDE z lnSPECH z lnSURF z R² Adjusted R² F-statistic Jarque-Bera Statistic White's statistic Durbin-Watson Statistic

Database minus the Paris ZE

445 1.289***

415 1.523***

(< 0.0001)

(0.00014)

-0.049

-0.070*

(0.084)

(0.0405)

-0.125

-0.114

(0.057)

(0.0853)

0.028

0.005

(0.430)

(0.893)

0.0217 0.0151 3.267

0.0255 0.0184 3.585

(0.02124)

(0.01391)

130.924

121.242

(< 0.0001)

(< 0.0001)

4.2066 1.856

4.7683 1.854

dL of D-W

1.8308 1.8243 1.85795 1.8534 dU of D-W The figures in brackets are the critical probabilities for each statistic *, **, *** indicate that the null hypothesis has been rejected at a significance level of, respectively, 10, 5 and 1%

The results of these final treatments confirm the fragility of the tested model (see Table 10). They nevertheless suggest that, overall, agglomeration diseconomies exist, as a result of location effects outweighing urbanisation effects. All things being equal, our results tend to suggest that intrasectoral concentration in the broad sense increases competition, which in turn increases labour costs and adversely affects the efficiency of warehousing activities. In addition, as warehousing activities are in a situation of economic servitude and all the firms provide much the same services, warehouse are in the position of a price-taker. Thus, intrasectoral concentration creates pressure on the price of their service. In other words, for this sector, the effects of competition seem to outweigh the opportunities provided by proximity. The small size of the sample, the fact that the findings are not statistically conclusive and disagree with the empirical literature that analyses the totality of economic activities leads us to examine the behaviour of residuals. In the case of all the models we tested, with the exception of specification (4) for the full database, the Durbin-Watson test either leads one to conclude that there is an overall absence of autocorrelation, or does not enable one to conclude that no such autocorrelation exists. While the White test leads us never to reject the hypothesis of homoscedasticity, the Jarque-Bera test raises an issue which is generally significant with regard to the normality of residuals. All the results of our analysis need to be qualified in view of the variable robustness of the estimations revealed by the analysis of residuals.

23

4

Conclusion

The originality of the research presented in this paper lies in the fact that it provides a measure of the spatial dynamics of logistics activities, warehousing in particular. For the first time, it provides the opportunity to test how agglomeration economies contribute to these dynamics. The analysis, performed on data from different sources, confirms, in the case of France, the process of geographical concentration highlighted in the literature. Moreover, the concentration measurements show that warehousing activities are more concentrated than the rest of the activities in the "Transport and storage" sector of the economy, and that this trend towards concentration is a lasting one. The New Economic Geography explains the advantages economic activities derive from grouping together because of the presence of agglomeration economies due to the combination of the urbanisation (or extrasectoral) economies that result from the concentration of the economic fabric in the area, and localisation or specialization (or intrasectoral) economies, related to the concentration of activities in the sector in question. We have sought to test the hypothesis that the productivity of logistics warehousing firms is positively correlated with both the total job density in the zone (urbanization economies) and the level of specialization in the "Transport and storage" sector (localization economies). Taken as a whole, the estimation results show that the chosen specifications explain very little of the variance in the productivity of warehousing firms (adjusted R² values of between 0.015 and 0.020 - the work of Barbesol and Briant (2008) reports adjusted R² values ranging from 0.127 to 0.205). However, the elasticities between the productivity of warehousing firms in a given area and its job density appear to be significantly negative. Thus, contrary to our original hypothesis, the effect of urbanization economies on logistics warehousing operations in France is very low, and to say the least, negative. This result contradicts the findings of research on economic activities in general which has shown that, until now, a high local density of economic activities increases the productivity of firms and workers. In the case of warehousing activities in France, the results of our study suggest that, although the presence of these activities is likely to increase the attractiveness of the area for all other activities, the density of the economic fabric tends to increase pressure on the prices of warehousing services - especially as these all very much resemble one another - and thus reinforce their position of "servitude" in the supply chain. This testifies to the fact that firms of this type are in the position of price-takers. Furthermore, our results also show the absence of local externalities of the MAR type and, conversely, the presence of localisation diseconomies for warehousing activities in France. This finding is less in contradiction with other research on the sector. The warehousing sector, which is characterised by the strong presence of low-skilled workers, is not affected by knowledge externalities, for example. In addition, and less intuitively, the results suggest that the proximity of complementary services, which could potentially foster inter-firm collaboration and help the development of more complex, tailor-made, logistics services, does not have a positive net impact. The results of this study also tend to cast doubt on the role of logistics clusters on the performance of firms, in spite of their marked development in recent years. Therefore, the article leads to the conclusion that the concentration of logistics warehousing operations probably tends to be more a source of agglomeration diseconomies, both intersectoral (urbanisation diseconomies) and intrasectoral (localisation diseconomies). Ultimately, geographical concentration does not appear to provide productivity gains for warehousing companies. However, the above conclusions should be qualified. The research we have described is open to both internal and external criticism. With regard to internal criticism, the main issue is the small size of the sample. In addition, as the fundamental 24

principle of our analysis was the location of the observed entities, observations on multiestablishment firms, which provide only general information, had to be removed. Although this was necessary in order to ensure our analysis was sound, it reduces our ability to observe the spatial dynamics of the major groups which operate in this area of activity. With regard to the dependent variable, for reasons of data availability, it was only possible to consider the productivity at the labour cost. The fact that total factor productivity was not analysed means that, for example, we were unable to consider the role of property pressure that inhibits this type of activity’s natural need for space. Finally, the proposed specifications are incomplete in that they do not incorporate all the factors that could explain the presence of agglomeration economies. Thus, the market potential9 is not taken into account while it should help to address the effect of external markets (or market accessibility). The inclusion of this variable could allow us to take account of the impact of the performance of transport infrastructure. The external limitations of our work are due first to the fact that it only considered warehousing that is conducted for a third party, and second to the fact that the analysis is conducted on a sectoral basis rather than on the basis of supply chains. However, location in a technology park (automotive, aerospace, etc.) can probably improve a warehousing firm’s productivity. There are also a number of possibilities for future research. We would like to confirm our findings by referring to a consolidated database or by including observations that cover a number of years for the "Warehousing and storage" activity category or by extending our observations to cover the entire "Storage and transportation support" category? This would be appropriate insofar as many warehousing services are now carried out by road transport undertakings or providers of services that are ancillary to transport. Furthermore, if the presence of agglomeration diseconomies is confirmed, we would like to better identify what drives the spatial concentration of logistics activities. Indeed, how can we explain the spatial concentration of warehousing activities when it reduces their productivity? The idea would be to test the hypothesis that despite a net negative effect, the spatial concentration of warehousing activities is strongly conditioned, as stated in the literature, by the configuration of supply chains (Petiot and Masson 2011). Finally, we would like to estimate how the effect of the market and accessibility impact the spatial concentration of logistics activities.

9

Calculated, for example, by summing all the activities on markets, discounted by the cost of transport to access them.

25

Appendixes Appendix 1. Economic features of the French administrative regions (INSEE, 2012) Administrative region Alsace Aquitaine Auvergne Basse-Normandie Bourgogne Bretagne Centre Champagne-Ardenne Corse Franche-Comté Haute-Normandie Ile-de-France Languedoc-Roussillon Limousin Lorraine Midi-Pyrénées Nord-Pas-de-Calais Pays de la Loire Picardie Poitou-Charentes Provence-Alpes-Côte d'Azur Rhône-Alpes

Urban units with more than 100,000 inhabitants (rank) Strasbourg (13), Mulhouse (26) Bordeaux (7), Bayonne (29), Pau (34) Clermont-Ferrand (24) Caen (35) Dijon (28) Rennes (20), Brest (33), Lorient (57) Tours (18), Orléans (23) Reims (31), Troyes (50) Besançon (49), Montbéliart (59) Rouen (12), Le Havre (27) Paris (1) Montpellier (15), Perpignan (36), Nimes (39) Limoges (37) Metz (21), Nancy (22), Thionville (51) Toulouse (6) Lille (4), Douai-Lens (10), Béthune (17), Valenciennes (19), Dunkerque (40), Maubeuge (58) Nantes (8), Angers (30), Le Mans (32), Saint-Nazaire (47) Amiens (43), Creil (56) Poitiers (53), La Rochelle (55), Angoulème (60) Marseille - Aix-en-Provence (3), Nice (5), Toulon (9), Avignon (14) Lyon (2), Grenoble (11), Saint-Etienne (16), Chambéry (38), Annecy (46), Valence (54) Metropolitan France

GDP (million Euros)

53 632 90 796 33 756 36 370 42 731 83 407 67 122 37 113 8 173 28 593 49 815 612 323 63 944 17 307 56 346 79 855 103 226 101 229 45 681 45 016 142 358 196 995 1 995 786

GDP/inhab. Rank 12 6 19 18 16 7 9 17 22 20 13 1 10 21 11 8 4 5 14 15 3 2

(Euros)

28 849 27 583 24 920 24 597 25 996 25 666 26 126 27 813 25 523 24 295 26 984 51 250 23 566 23 354 23 968 27 198 25 487 27 775 23 751 25 166 28 861 30 943 31 420

Rank 4 7 16 17 11 12 10 5 13 18 9 1 21 22 19 8 14 6 20 15 3 2

Sources: (INSEE, 2012)

Appendix 2. Concentration indices There are several concentration indices, of varying degrees of effectiveness. We have selected two based not only on their performance, but also on their compatibility with the data at our disposal. These are the Herfindahl-Hirschman index and the Gini index.  The Hirschman-Herfindahl index of absolute geographic concentration (Hk) The Hirschman-Herfindahl index is the most straightforward to compute and corresponds to the sum of squares of the sector's share in the area. Its value varies between 1/n, where n is the number of zones, and 1. It takes on a value of 1 when all the establishments in the sector are concentrated in a single zone and its minimum value when they are evenly distributed between the zones.

 Gini index of relative geographical concentration (Gk) Gini indices of geographical concentration are widely used in the empirical literature to measure the inequality of the distribution of a variable (Koenig, 2005). In the case of our study, for a given region or département, the Gini index of relative geographical concentration is referred to as relative because it evaluates the extent to which jobs in logistics activities are more geographically concentrated than economic activities in general. If the two distributions are very different, jobs in logistics activities are said geographically concentrated (the value of the index is close to 1). The use of relative indices makes it possible to make a 26

comparison between the importance of a sector in a zone and its importance in all the studied zones.

Appendix 3. The index of specialization: Hoover’s index Hoover’s index is the ratio between the proportion of jobs an activity accounts for in a region and the proportion of jobs it accounts for in the country as a whole. It is expressed using the same formula Isard used in 1960 to give what was termed the location quotient.

denotes the number of jobs in sector k in zone i and stands for the number of employees in zone i. Denotes the number of jobs in all the zones and expresses the total number of jobs in the economy as a whole. This indicator estimates whether a given category of activities is over-represented in a zone (when its value is greater than 1) or when it is under-represented (when its value is less than 1) compared to all urban areas.

27

References Abrahamson M, Aldin N, Staher F (2003) Logistics platforms for improved strategic flexibility. International Journal of Logistics Research and Application 6(3): 85-106. Aiginger K, Leitner W (2002) Regional concentration in the United States and Europe: Who Follows Whom? Weltwirtschaftliches Archiv 138: 652–679. Albert JM, Casanova MR, Orts V (2011) Spatial location patterns of Spanish manufacturing firms. Papers in Regional Science 91(1): 107-136. Arrow K (1962) The economic implications of learning by doing. Review of Economic Studies 29: 155-173. Bahoken F, Raimbault N (2012) La périurbanisation singulière de l’immobilier logistique du bassin parisien. M@ppemonde 106: 1-24. Barbesol Y, Briant A (2008) Economies d’agglomération et productivité des entreprises : estimation sur données individuelles françaises. Economie et Statistique 419-420: 41-54. Barlet M, Crusson L, Dupuch S, Puech F (2010) Des services échangés aux services échangeables : une application sur données françaises. Economie et Statistiques 435-436: 105-125. Baudewyns D (2005) Structure économique et croissance locale : étude économétrique des arrondissements belges, 1991-1997. Revue d’Economie Régionale et Urbaine 5: 635-656. Becker D (2003) Le développement des implantations logistiques en France et ses enjeux pour les politiques d'aménagement. Paris : Conseil Général des Ponts et Chaussées, Ministère des Transports, de l'Equipement, du Tourisme et de la Mer. Bhatnagar R, Sohal AS (2005) Supply chain competitiveness: measuring the impact of location factors, uncertainty and manufacturing pratices. Technovation 25(5): 443-456. Bowen J (2008) Moving places: the geography of warehousing in the US. Journal of Transport Geography 16: 379–387. Brülhart M (2001) Evolving geographical specialisation of European manufacturing industries. Weltwirschaftliches Archiv 137(2): 215-243. Brülhart M, Mathys NA (2008) Sectoral agglomeration economies in a panel of European regions. Regional Science and Urban Economics 38(4): 348-362. Catin M (1991) Économies d'agglomération et gains de productivité. Revue d'Economie Régionale et Urbaine 2: 261-278. Catin M (1995) Productivité, économies d'agglomération et métropolisation. Revue d'Economie Régionale et Urbaine 4: 663-682. Catin M (1997) Disparités spatiales de productivité, accumulation du capital et économies d’agglomération. Revue économique 48(3): 579-589. The Ciccone A (2002) Agglomeration effects in Europe. European Economic Review 6(2): 213-227. Ciccone A, Hall, RE (1996) Productivity and the density of economic activity. American Economic Review 86(1): 54-70. Cidell J (2010) Concentration and Decentralisation: the New Geography of Freight Distribution in US Metropolitan Areas. Journal of Transport Geography 18: 363-371.

28

Cidell J (2011) Distribution Centers among the Rooftops: The Global Logistics Network Meets the Suburban Spatial Imaginary. International Journal of Urban and Regional Research 35(4): 832– 851. Combes PP (2000) Economic structure and local growth: France 1984-1993. Journal of Urban Economics 47(3): 329-255. Combes PP, Duranton G, Gobillon L (2008) Spatial wage disparities: Sorting matters! Journal of Urban Economics 63: 723-742. Combes, PP, Duranton G, Gobillon L (2011) The identification of agglomeration economies. Journal of Economic Geography 11: 253-266. Combes PP, Lafourcade M (2012) Revue de la littérature académique quantifiant les effets d’agglomération sur la productivité et l’emploi. Rapport réalisé pour la Société du Grand Paris. Mission d’études des éléments de l’évaluation socio-économique du réseau de transport du Grand Paris. Combes PP, Mayer T, Thisse JF (2006) Economie géographique, Economica, Paris. Commissariat général au développement durable (2012) Les entrepôts et leur activité. Chiffres et Statistiques. 2012, 334 p. Dablanc L, Andriankaja D (2011) Desserrement logistique en Ile-de-France : la fuite silencieuse en banlieue des terminaux de fret. Flux 85-86: 72-88. Dablanc L, Rakotonarivo D (2010) The impacts of logistics sprawl: How does the location of parcel transport terminals affect the energy efficiency of goods’ movements in Paris and what can we do about it? Procedia – Social and Behavorial Sciences 2(3): 6087-6096. Dablanc L, Ross C (2012) Atlanta: a mega logistics center in the Piedmont Atlantic Megaregion (PAM). Journal of Transport Geography 24: 432-442. Dixit AK, Stiglitz JE (1977) Monopolistic competition and optimum product diversity. American Economic Review 67(3): 297-308. DTZ. (2011) 1998-2008-2011. Evolution de l'immobilier logistique. Ellison G, Glaeser E (1997) Geographic Concentration in U.S. Manufacturing Industries: a Dartboard Approach. Journal of Political Economy 105(5): 889-927. Fender M, Pimor Y (2013) Logistique et Supply chain. Dunod, Paris. Foster N, Stehrer R (2009) Sectoral productivity, density and agglomeration in the Wider Europe. Spatial Economic Analysis 4(4): 427-446. Fujita M, Krugman P (2004) The new economic geography: past, present and the future. Papers in Regional Science 83: 139-164. Fujita M, Thisse JF (1996) Economics of agglomeration. Journal of the Japanese and International Economics 10: 339-379. Fujita M, Thisse JF (2002) Economics of agglomeration, Cities, Industrial location and Globalization. Cambridge University Press, New-York. Glaeser EL, Kallal H, Scheinkman JA, Schleifer A (1992) Growth in cities. Journal of Political Economy 100(6): 1126-1152. Goschin Z, Constantin DL, Roman M, Ileanu BV (2009) Specialisation and concentration patterns in the Romanian economy. Journal of Applied Quantitative Methods 4(1): 95-111.

29

Hallet M (2000) Regional specialisation and concentration in the EU. Economic Papers 141, European Commission, March 2000. Hesse M (2007) The System of Flows and the Restructuring of Space, Elements of Geography of Distribution. Erdkunde, Jan-Mar. Hesse M, Rodrigue JP (2004) The transport geography of logistics and freight distribution. Journal of Transport Geography 12: 171-184. Hesse M, Rodrigue JP (2006) Global Production Networks and the Role of Logistics and Transportation. Growth and Change 37(4): 499-509. Houdebine M (1999) Concentration géographique des activités et spécialisation des départements français. Economie et Statistiques 326-327: 189-204. Jacobs J (1969) The Economy of Cities. Random House, New York. Jakubicek P, Woudsma C (2011) Proximity, land, labor and planning? Logistics industry perspectives on facility location. Transportation Letters: The International Journal of Transport Research 3(3): 161-173. Koenig P (2005) Concentration spatiale et décision d’exportation. Revue économique 56(6): 12271248. Krugman PR (1991a) Geography and trade. Leuwen University Press and the MIT Press, Cambridge. Krugman PR (1991b) Increasing returns and economic geography. Journal of Political Economy 99(3): 484-499. Krugman PR, Venables AJ (1995) Globalization and the inequality of nations. Quarterly Journal of Economics 110(4): 857-880. Marshall A (1890) Principles of Economics. Macmillan, London. Masson S, Petiot R (2011) On the Relationship between Space and Logistics, Sustainable Development, Territories and Firms Location Decision: toward a sustainable Attractivity? Bordeaux, LAREFI, 15-16 Septembre 2011. Masson S, Petiot R (2012a) Attractivité territoriale, infrastructures logistiques et développement durable. Les Cahiers Scientifiques du Transport 61: 63-90. Masson S, Petiot R (2012b) Localisation des activités logistiques en France : faits stylisés et déterminants, 49ème Colloque de l’Association de Science Régionale De Langue Française, Belfort, 9-11 juillet. Masson S, Petiot R (2013) Logistique et territoire : multiplicité des interactions et forces de régulation. Géographie, Economie, Société 15: 385-412. Maurel F, Sédillot B (1999) A measure of the geographic concentration in French manufacturing industries. Regional Science and Urban Economics 29: 575-604. McKinnon A (2009) The present and future land requirements of logistical activities. Land Use Policy 26(sup 1): S293-S301. Melo P, Graham P, Noland R (2009) A meta-analysis of estimates of urban agglomeration economics. Regional Science and Urban Economics 39(3): 332-342. Melo MT, Nickel S, Saldanha-da-Gama F (2009) Facility location and supply chain Management - A review. European Journal Of Operational Research 196: 401-412.

30

Mérenne-Schoumaker B (2007) La localisation des grandes zones de logistique. Bulletin de la Société géographique de Liège 49: 31-40. Mouhoud EM (dir.) (2010) Economie des services et développement des territoires. La Documentation Française, DATAR, Paris. O’Connor K (2010) Global city regions and the location of logistics activity, Journal of Transport Geography 18: 354-362. Petiot R, Masson S (2011) Dynamiques territoriales des infrastructures logistiques : Quels enseignemenrs retenir de l’étude du cas français ? Quelles perspectives pour la recherche ? 10ème séminaire francophone est-ouest de socio-économie des transports, MEDDTL, PREDIT, Ecole polytechnique de Montréal, Montréal, 1-5 juin 2011. Porter ME (1990) The Competitive Advantage of Nations. Free Press, New York. Raimbault N, Douet M, Frémont A (2013) Les implantations logistiques entre réseaux et territoires, L'Espace géographique 42(1): 32-43. Romer P (1986) Increasing returns and long-run growth, The Journal of Political Economy 94: 10021037. Rosenthal S, Strange W (2004) Evidence on the nature and sources of agglomeration economies. In: Henderson JV, Thisse JF (eds), Handbook of Regional and Urban Economics, Vol. 4, Cities and Geography, Elsevier, Amsterdam Rosenthal S, Strange W (2001) The determinants of agglomeration. Journal of Urban Economics 50: 191-229. Savy M (2006) Logistique et territoire. La Documentation française, Paris. Sheffi Y (2012) Logistics clusters, delivering value and driving growth. MIT Press, Cambridge, MA. Strale M (2010) La localisation des entreprises logistiques et le positionnement des régions urbaines nord-ouest européennes, Belgeo http://belgeo.revues.org/6841 Strale M (2013a) La logistique : localisation des activités et impacts territoriaux. Bruxelles : Thèse de Doctorat de Géographie, Faculté des Sciences, Université Libre de Bruxelles, 2013a. Strale M (2013b) Une typologie empirique des plates-formes logistiques et de leurs activités. Revue d'Economie Régionale et Urbaine 1: 131-162. Van den Heuvel FP, De Langen PW, Van Donselaar KH, Fransoo JC (2013a) Spatial concentration and location dynamics in logistics: the case of a Dutch province. Journal of Transport Geography 28: 39-48. Van den Heuvel FP, De Langen PW, Van Donselaar KH, Fransoo JC (2013b) Regional logistics land allocation policies: Stimulating spatial concentration of logistics firms. Transport Policy 30: 275282. Wackermann G (2005) La logistique mondiale. Transport et communication. Paris : Ellipse, 2005. Wagner T (2010) Regional traffic impacts of logistics-related land use. Transport Policy 17: 224-229. Wemelbeke G, Mariotte H (2007) L’essor de grandes zones logistiques accompagne la massification des flux routiers. SESP en bref 15.

31

Documents de Travail Art-Dev : 2012-01 Sourisseau JM, Bosc PM, Fréguin-Gresh S, Bélières JF, Bonnal P, Le Coq JF, Anseeuw W, Dury S, 2012. Représenter la diversité des formes familiales de la production agricole. Approches théoriques et empiriques. 2012-02 Michel, S., Randriamanampisoa H. La pauvreté multidimensionnelle au prisme du microcrédit. 2012-03 Ricci, F. Traps due to negative externalities arising from the uneven spatial distribution of innovative activities. 2012-04 Chevalier, P. Quels effets des reglementations nationales dans la programmation LEADER dans l’Union Européenne ? 2012-05 Meuriot, V, Analyse critique de l’économétrie des séries temporelles moderne. 2013-01 Giordano, T., Multilevel integrated planning and greening of public infrastructure in South Africa 2013-02 Meuriot, V, Diallo A.S., A comment on “Liberalization and food price distribution: ARCH-M evidence from Madagascar” (Barrett, 1997) 2013-03 Ghiotti, S., Riachi, R., La gestion de l’eau au Liban : une réforme confisquée ? 2013-04 Malizard, J., Is There Military Keynesianism? An Evaluation of the Case of France Based on Disaggregated Data. 2013-05 Poncet, C., Risque et flexibilité dans la gestion des opérateurs en capital-risque : Réflexions autour des critères d’intervention. 2013-06 Poncet, C., Le développement des opérateurs en capital-risque : le poids du contexte institutionnel. 2014-01 Bourgeois, R., The State of Foresight in Food and Agriculture: Challenges for Impact and Participation 2014-02 Bourgeois, R., Food (In)security: the New Challenges Ahead 2014-03 Bourgeois, R., Farmers Moving out of Poverty: What are the Challenges? 2014-04 Bourgeois, R., Constructive Destruction: What has to be Changed?.

32

Documents de Travail Art-Dev : 2014-05 Charlier, D., Efficacité énergétique dans le bâtiment et paradoxe énergétique : quelles conséquences pour la transition énergétique ? 2014-06 Charlier, D., Energy-Efficient Investments in the Housing Sector: Potential Energy Savings vs. Investment Profitability. An Empirical Analysis 2014-07 Charlier, D., Split Incentives and Energy Efficiency: Empirical Analysis and Policy Options 2014-08 Pesche D., Le Coq J.F., Costa Rican Payment for Environmental Services : between international and national influence, a feedback analysis 2014-09 Malizard, J, Droff, J., Economic versus Strategic Constraints: the Asymmetric Behaviour of Defense Spending in France 2014-10 Dachary-Bernard, J., Rivaud, A., Evaluation des préférences des touristes en matière d’aménagement des zones côtières : regard sur et par l’ostréiculture 2014-11 Le Coq, J-F., Sáenz-Segura , F., Salazar, J.A.V., Agriculture familiale dans le canton de Guatuso, Costa Rica 2014-12 Cortes, G., Fréguin-Gresh, S., Guétat-Bernard, H., Sourisseau, J-M., Les Systèmes Familiaux Multi-localisés : un modèle d’analyse original des ruralités aux Suds 2014-13 Bélières , J-F., Agriculture familiale et politiques publiques au Mali 2014-14 Le Coq JF., Pesche D., Legrand T., Saenz F., Politique publique et changement climatique: une lecture par les coalitions de cause du Programme de Paiement pour Services environementaux au Costa Rica. 2104-15 J-F. Le Coq, G. Serpantié, F. Andriamahefazafy, F. Saenz-Segura, R. Mora-Vega, R. Pierre, Les écolabels fournissent-ils des services environnementaux? Enseignements de filières agricoles au Costa Rica et Madagascar. 2014-16 L. Apithy, S. Guyard, S. Bouard, M. Passouant, JM. Sourisseau, JF. Bélières. 2014. Réaliser une enquête par sondage : une application au fonctionnement des unités socio-économiques de base du monde rural Kanak (Nouvelle-Calédonie). 2015-1 S. Masson, R. Petiot. 2015. Logistic facilities and territory: Spatial dynamics and policy issues.

33