Local Production Systems in Global Value Chains: the Case of Italian Industrial Districts Giancarlo Corò and Mario Volpe University of Venice Ca’ Foscari, Department of Economics and Tedis-VIU October 2006 Draft Address for correspondence:
[email protected]
Abstract The focus of the paper is the internationalization process followed by Italian industrial districts. Starting from this perspective, we address the more general issue of changing and upgrading of clusters and local production systems. We therefore look whether the opening process of production organization is associated with an increase in innovation and competition and with significant effects on the local labor market. We try to measure the process and the effects from a systematic, macro-economic perspective: we therefore look at national accounts at the most available local and sector detail. The very special nature of clusters, dominated by small and medium enterprises and often specialized in mature and traditional sectors, poses theoretical and empirical problems. From a theoretical point of view, we refer to the literature on industrial districts and the literature on fragmentation and international integration through trade: studying the upgrading of clusters form local to global chain we end-up in a framework very similar to the GVC literature. From an empirical point of view, we propose a methodology based on international integration of correlated flows of intermediate goods, more suited for the case under investigation than the most used methodologies (FDI, for example). Obtaining some preliminary results, we are able to sketch some comments on empirical results, focusing on the issue of changing in specialization of clusters (e.g., form final to intermediate goods), variety of strategies and positioning of apparently similar clusters and increase in intersector linkages (final good and machinery). Keywords: fragmentation of production, outsourcing, international integration of production, local production systems. JEL-code: F15, L32, R12.
1. Introduction1 The aim of this paper is to analyse international re-organization processes of small and medium enterprise (SME) systems from the perspective of fragmentation of production. Our analysis extends from two starting points. The first, based on economic and social literature on industrial districts (Becattini, Rullani 1993; Brusco 1994; Porter 1998), is to recognize the specificity in the patterns of division of labour in SME production systems. These are systems where local market 1
The paper is the outcome of a research on the international extension of productive network of Italian traditional sectors. We have benefited from very helpful comments, suggestions and criticism by different scholars during workshops and seminars: we want to thank in particular Gary Gereffi, Charles Sabel, Luigi Signorini, Giuseppe Tattara and Antonello Zanfei. The authors are responsible for any omission and imprecision.
1
plays a crucial role in organization of the supply chain, in integration of different productive stages and in good matches between supply and demand of intermediate goods. The second is Feenstra’s insight on growing process of integration of the global economy through trade (Feenstra 1998), that has been developed also by “international fragmentation” theory (Arndt, Kierzowsky 2001) and by “global value chain” approaches (Gereffi, Humphrey, Sturgeon 2005). Our idea is that these two different standpoints may help us to look at international re-organization processes of Italian SMEs without relying on the multinational firm as the only model for the global economy. Indeed, if we analyse only foreign direct investment (FDI) flows for the Italian economy – and, especially, for those regions characterized by a wide presence of SMEs – the conclusion is that, despite a high export bent, the degree of international integration of production is very low. But if we analyse the growing intra- and inter-industry trade between several Italian clusters and the emerging economic regions in the world economy (Central-Eastern Europe, Mediterranean Basin, Far East and Central Asian countries) we may infer the existence of international productive networks. Our hypothesis is that the main way SMEs develop international linkages reflects the local division of labour patterns: from this perspective, to analyse international production re-organization processes of SME systems, we have to look how the local value chain is slicing up in the global economy (Krugman, 1995). This theoretical perspective also has a corollary in understanding the economic systems’ competitive position in international division of labour. By looking at “kaleidoscope comparative advantage” described by Bhagwati and Dehejia (1994), the economic position is not only decided by the industry in which the local or national systems have specialised, but also by the specific segments of the value chain marked in local or national activities. In other words, in accordance with Hummels, Rapoport and Yi (1997), we prefer to speak of “vertical specialization” rather than traditional comparative advantage based on structural analysis. This paper proceeds in the next chapter to sketch a macroeconomic framework to analyse the role of local SME systems in international productive processes. In chapter 3 we try to construct a measurement of international integration of production (the number of workers involved abroad) based on bilateral trade value in intermediate and final goods inside each supply chain. To seek a confirmation of this measure, in chapter 4 we propose different statistical tests, computing several historical series on trade data with co-integration models. In chapter 5 we describe the main results produced at the national and local levels by applying the measurement of international integration of production through trade. In chapter 6 we summarize some outlines for further investigations.
2
2. A macroeconomic framework to analyse the role of SME clusters in the process of productive internationalization. When measuring internationalization of production, the standard approach is to look at the extension of individual firms’ productive activities abroad. The economic literature tackles the problem from a microeconomic perspective, analysing the choices of firms in terms of commercial versus productive internationalization, based on direct surveys or data about firms’ investments abroad. These are firms able to expand their productive capacity abroad due to available resources: knowledge, organization, management and financial. In other words, these are multi-national or meta-national firms, which organize their production worldwide. Therefore, they are able to actively fragment their production into different multi-national sites, depending on various factors well analysed in the literature: cost reduction, access to specific knowledge, access to intermediate and final markets. The most evident form of this process is the building of new productive capacities abroad, through either the acquisition or the construction of plants. The degree of international integration of the firm is determined by size of the outgoing flow of investments. The appropriate measure could be the flow of investment in a given period, the size of the stocks or the number of employees, depending on the scope of the analysis. By summing up the international integration of the individual country-based firms, it would be possible to obtain a measurement of the international integration of production of an entire national economic system, as is extensively done in empirical research. On the other hand, there is a consistent share of economic activities, quite deeply involved in the international extension of productive activity, which is scarcely captured by this analytical and empirical analysis. It is that made by small and medium-sized enterprises usually clustered in local system of production. As far as the Italian economy is concerned, SMEs have exhibited quite a better performance than bigger firms in recent decades. The explanation of this success has been extensively investigated in the literature. What is of interest here are two main points captured by the analysis of the industrial organization of local SME networks: 1. The structure of SMEs does not fit well with the general framework outlined above, the one used to measure the degree of internationalization of an economic system based on multiand meta-national firms. Some of these firms, though of medium size, are playing a global competitive game. But as a whole this set of firms lacks sufficient resources to directly control the international extension of production. 2. Industrial districts based their success, in economic and social performance, on an early domestic fragmentation of the value chain, confined within the geographical boundaries of the district. The technical feasibility of breaking down the supply chain, together with variability and sufficient sizes of intermediate markets for local firms positioned along the supply chain, led to the emergence of a highly specialized local system. This has been due to the exploitation of locally based externalities. They became leaders in mature markets (the so-called “Made in Italy” sectors) due to the local fragmentation of production; more precisely, they were able to be innovators and competitors through the establishment of internal markets. Market transactions were not strictly confined within the local systems, but the emergence of these markets (dedicated machinery for textile or footwear makers) was the way to be dynamically competitive too, spreading innovation and new knowledge along the supply chain. Within the local systems of production, each firm plays a different role in 3
the supply chain. The role is two-fold: on the one hand, firms are complementary to each other in a vertical integration scheme; on the other hand, the many small firms compete with each other in their production. It is precisely this combination of coordination and competition that allows a dynamic performance of the local system, in terms of innovation, new products, knowledge growth and new employment. The crucial point for our reasoning is that the organization of production is made through market transaction. In the new landscape of global competition, these clusters are forced to change their internal organization. They have to face a stronger international competition and have to try to gain new competitive advantages. In other words, SME clusters have to take advantages of greater availability of resources abroad in two complementary ways: through increased efficiency and through access to new knowledge. In other words, they have to participate to the international reorganization of the value chain within the general framework of productive fragmentation (Gereffi 2004). This reasoning poses two important sets of research questions for industrial districts and clusters in general. First, given the point 1) above (the different structure of the cluster with respect to multi- or metanational firms), what are the instruments and the patterns of their internationalization of production? And, consequently, how can we measure it? Second, due the already existing domestic fragmentation of production, is the productive process constrained by the local organization of production? Is there any advantage in replicating the model of fragmentation already experienced in the local arena? Is this pattern of international integration producing different effects from one induced by an increase of FDI abroad, more related to the local economic system than to individual firm strategy? Our point is that in this area of research a macroeconomic approach is needed to capture the process regarding the system of firms, the productive districts, and the different forms of the internationalization of productive processes here analysed. As in the internal organization, in the international integration of production it is likely that districts use less resource-intensive tools and less formalized than FDI, such as productive agreements and market transactions. Given that these relationships are with foreign firms, they involve trade flows of intermediate goods. Therefore our analysis extends from two considerations: 1) The growing literature on FDI is not suitable for analysis of SME clusters. Firms embedded in productive districts, even dynamic and network related ones, cannot usually afford the large amount of sunken costs required to establish new productive capacities abroad, signalled by direct investments. Thus, FDI are the proper indicator for large, multi- and meta-national corporations, but not for the international extension of production organization involving SMEs and clusters of firms. It is a theoretical difficulty, as well as an empirical one. Surveys of the framework of Italian production (Rossi 2004) show the mismatch between the sector detail of outgoing FDI and export specialization pattern of the Italian economy. 2) International trade flows pertaining to the districts, both imports and exports, should be the appropriate tool to assess the scope of the process.
4
The point is relevant but somehow neglected in the economic literature. An interesting line of study is the one on industrial fragmentation (Arndt, Kierzkowsky, 2001; Zysman, Schwartz, 1998), stating that if the internationalization process is made within a fragmentation framework, the analysis should refer to disaggregated trade flows. It is exactly the breakdown of the supply chain in technical strokes that corresponds to the local organization of production in “Made in Italy” manufacturing districts. The focus is therefore on bilateral trade flows of intermediate goods between districts and foreign areas, both related to a specific value chain, as a signal of the existence of a district-governed international network of production. This process can follow different patterns, based on different outsourcing schemes: horizontal or vertical integration, collaborative agreements or pure market transactions. Along this line of research some scholars have used the source of Outward Production Trade (OPT) as an indicator of productive internationalization (Baldone, Sdogati and Tajoli, 2002). OPT, or the opposite, Inward Production Trade (IPT) are a clear extension abroad of the supply chain; they are specific trade regimes, due to particular bilateral agreements for custom reduction. Though this pattern of internationalization has theoretical relevance, nowadays its use is limited to a very specific vertical integration pattern and its importance is dramatically reduced due to the new institutional framework. Indeed, this trade regime was used mostly with Easter European countries (EECs), now within the enlarged Europe.
3. An index of productive internationalization for SME clusters If trade flows are used to extend the supply chain internationally, we could use this source as a proxy of the phenomenon. Following the above suggested macroeconomic approach, we refer to intermediate trade flows generated by specialized local systems of production. Domestic -within the district- market transactions are replaced by foreign transactions. Depending on the technical position of the production steps involved, there will be a different type of integration with the foreign areas. We will see vertical integration, if the role of the foreign supply is along the supply chain; we will see horizontal integration, if the foreign supply is positioned at the end of the supply chain. A third pattern is inter-sectoral integration: an important issue for the Italian case, where dedicated machinery is growing due to internationalization of the related final sectors. To pursue our analysis, as first step we have to describe statistically (more precisely, in terms of national accounting) the supply chains relevant to industrial districts. The sector breakdown must consider, in a comparable way, domestic activity and trade flows: availability of sector classification of statistical data is clearly limiting the analysis2. In the case of Italy, the focus is on “Made in Italy”(MII) value chains: • Textile and Apparel (T-A) • Leather and Shoes (L-S) • Wood and Furniture (W-F) We consider as the units of analysis the local Italian systems specialized into these value chains; local systems, namely provinces, are used as proxies for industrial districts (see note 1). 2
The statistical source we are using is Coeweb, the Italian national statistical office (ISTAT) web site for foreign trade. We are using the most detailed data available in terms of sector breakdown, Italian geographical area, foreign area: two digits classification for good and services, provinces for domestic area, single country for foreign area. As an example, the one we use in the statistical analysis: exports of textile, from Vicenza to Romania.
5
We statistically describe each cluster with the best available disaggregation. For T-A, for example, the cluster is broken down into Textile (code Ateco: DB172) Apparel, the final good (code Ateco: DB182) Dedicated machinery (code Ateco: DK295) As a second step we have to identify the geographical area involved in the productive internationalization. Considering all the potential countries involved into the internationalization of production by Italian district, we select the subset for the vertical international process as the countries with 1. Similarity in trade specialization, assessed through specialization indexes (see section 4). 2. Low-medium quality of final production of the relevant sector production. 3. Statistical correlation between export toward Italian districts of forward (final) goods than the value of imports from Italian district of backward (intermediate) goods 4. Presence of outward Italian FDI in the sector of specialization. The subset of horizontal integration, on the other hand, is made by countries with the same characteristics, with the exception of point 3) above, substituted by: 3bis. Bilateral flows in the specialized sectors such that there is not an evident correlation between import of intermediate goods from Italian districts and related export of semi-finite goods. What holds as statistical correlation is instead the one between exports, in the specialized sector, to the Italian districts and exports of Italian districts to the final markets of highly developed countries. Empirically, we proceed on this point observing the pattern of international trade flows of the national economy. During the last 15 years, there has been a strong change in the geographical orientation of both exports and imports. The dominant position of European partners has sharply decreased in favour of other areas: East European countries, North Mediterranean countries, India and China. As far as exports of Italian districts are considered, Italian districts’ products compete in high quality markets; as far as imports, the average quality of imported products is not well suited for the Italian final demand. That means the magnitude of international trade flows directed or originated by Italian districts contains more than pure commercial internationalization: it is likely that some degree of productive internationalization is happening through these flows. When we could identify a strong correlation between backward exports and forward imports in the pattern of trade flows between a district and a specific geographical area, within the same supply chain, we could infer the existence of strong vertical integration between the areas. The statistical analysis of this correlation is presented in the next section. Once identified, we proceed to measure and index international integration, for each district and for each specialized supply chain3.
3
A third pattern of integration is important: the inter-sectoral one. Exports of dedicated machinery play a significant role. Mechanics and machinery have always played a key role in Italy’s prosperous districts: it is precisely the contiguousness between final good production and machinery production which generated continuous process and product innovation. The same kind of inter-sector linkages is propagating in the international extension of productive networks. This role is so relevant, that many districts are in fact repositioning themselves from final good production to machinery production. This plays a major role in the productive internationalization with India and China.
6
We propose to calculate an indicator that we call “foreign involved employees” (hereafter FIE) to measure the intensity of international production for each industrial district. It consists of the number of foreign workers involved in the supply chain governed by Italian local system of production; the term “involved” comprises the direct and indirect linkages. The indicator considers the two main kinds of productive integration: the vertical, where some phases of the production cycle are done abroad (also in the extreme form of temporary exports) and the horizontal, where imports substitute the domestic production4. For each Italian district, we consider the international bilateral flows with foreign countries in the sector of specialization, as explained above. In the following with the term “imports” and “exports” we refer to goods pertaining to the supply chain governed by the specific MII district. We calculate the international vertical integration using the following steps: 1) Identification of the foreign countries involved in the international vertical integration, by using trade similarity indexes. The process is validated on a statistical basis, as explained in section 4). 2) Valuation of integration in terms of foreign production, calculating the difference, for each district, between the value of imports of a forward step, and the value of export of a backward step on the same supply chain. 3) Computation of employees by dividing the value of production by an average value of worker productivity, taken by official Italian data on FDI (Reprint 2003). In symbols, the indicator of “foreign induced employees” due to vertical integration is computed, for each supply chain φ of the local production system L, as: φLAdEVt
= ∑i ∑k (φLMvi(k+1)(t+lag) – φLXvikt)/φΠikt
Where: Mv(k+1) = Import due to vertical integration (forward step in the supply chain φ) Xvk = Export due to vertical integration (backward step in the supply chain φ) Π = Average foreign worker productivity (Output/workers of foreign plants) k = production stroke in supply chain i = Country-area target of vertical integration t = quarter lag = temporal lag required by foreign production, in quarters The horizontal integration, meanwhile, is calculated: 1) As in step one above, with the difference that there is lack of statistical validation for vertical integration; instead, the statistical validation points to correlation between imports of the Italian district from these countries and exports to the country with high-valued final markets; 2) Valuating the foreign integration of production as the difference between the value of bilateral imports of forward (finite or semi-finite) goods of Italian districts and the value of bilateral imports of the same good at the beginning of the time period under analysis. 3) Computation of the “foreign induced employees” index as stated above. In symbols, the indicator of “foreign induced employees” due to horizontal integration is computed, for each supply chain φ of the district L, as: 4
The distinction between horizontal and vertical integration may be ambiguous: it depends on how one defines the supply chain. Furthermore, we speak about horizontal integration instead of production displacement, because the link with productive specialization of local cluster: imports are governed by the domestic final production.
7
φLAdEOt
= ∑j ∑k (φLMojkt–φLMojk(t0))/φΠjkt
Where, in addition to how written above: Mo = Import due to horizontal integration j = Country-area target of horizontal integration (t0) = base year (first year of the integration process) Total amount of “foreign induced employees” for each local system L and industry φ is the sum of vertical and horizontal integration effects: φLAdEt = φLAdEVt
+ φLAdEOt
For the Italy as a whole, the total amount of FEI will therefore be given by: φNAdEt
= ∑L φLAdEt
Where the number L of local production systems considered are less than n as total amount of local production systems in the national economy. In fact, we consider only local systems specialized in the supply chain φ (industrial districts).
4. Empirical analysis: the Italian case 4.1. Measurement and data sources In the analysis of the international productive integration between Italian economic systems and foreign economic areas we have tried to study this phenomenon from different standpoints and to different extents through the analysis of time series of flows of goods between countries. The analyses have been focused on the traditional manufacturing sectors that Italy is specialized in, that is Textile and Apparel (T-A) and Leather and Footwear (L-F). Furthermore, the analyses have been restricted to two specific economic areas, EECs on the one hand and China and India on the other. The hypothesis of international productive integration has been tested at two different levels of analysis. First, the survey has been carried out at the local level. It has examined the integration between some Italian productive districts (provinces, the smallest administrative areas have been used instead) and the above-mentioned set of countries. Then the analysis was extended to a regional level. For the first step of the survey, the provinces of Veneto containing productive districts have been chosen. For the T-A sector the selected provinces were: Treviso, Padova and Vicenza. For the L-F sector, the analysed provinces were: Verona, Treviso and Venezia. In the second step of investigation, the analyses were carried out at the regional level. For each sector, the five Italian regions with the highest concentration of productive districts have been chosen. These are: Piemonte, Lombardia, Veneto, Emilia Romagna and Toscana for the T-A; Lombardia, Veneto, Toscana, Marche and Puglia for the L-F. According to the international production fragmentation theory, different phases of the same supply chain base their production in different countries. Exchanges of goods between these countries can be seen as a proxy of these ongoing integration processes. The patterns of international integration vary according to the features of the production processes themselves, as well as to the features of the foreign countries involved. As stated in the previous chapters, productive integration is vertical when the flows of goods between two economic regions involve two consecutive phases of the same supply chain (i.e. export of textiles and import of apparel between Italy and Romania). In this 8
case a phase of the productive process is de-localized abroad and the analysed flows of goods are inputs and outputs of that phase. There is horizontal productive integration when a firm enters the productive process of a foreign firm providing it with goods that will be exported afterwards to different countries (i.e. Italy imports textiles from China. Italy exports those goods to other final markets). The level of similarity between two countries with regard to a specific productive sector can be determined by using the intra-industrial trade index (IIT) for that specific sector. Two countries showing a high value of the IIT are likely to be integrated since the flows of goods of the same supply chain between each other are similar in value (or size). On the other hand, a low IIT means either that the two countries do not interact with each or that the flows of goods are one-way rather than two-way flows. Table 1 records the value of the IIT index between Italy and different economic areas of the world for the two productive sectors studied. From the results of Table 1 it can be inferred that there exists vertical integration between Italy and the EECs, whereas between Italy and China-India the low level of infra-industrial trade could be a proxy of horizontal integration. To investigate the horizontal productive integration, exports of clothing and footwear towards the typical final markets for the Italian manufacturing output, that is EU15, USA and Japan, have been regressed on the imports of the same goods from China and India separately. Table 1. Values of the IIT
Textiles and Apparel
Leather and Footwear
1992
2002
1992
2002
EU 15
0.520
0.476
0.229
0.344
EECs
0.958
0.871
0.618
0.925
North America
0.243
0.079
0.184
0.104
China and India
0.100
0.186
0.079
0.390
World
0.630
0.669
0.687
0.757
Source: Our elaboration of data from IST-AT datawarehouse
Among the EECs the analyses have been restricted to the four showing the highest volume of trade with Italy in the analysed sectors. These are Romania, Hungary, Bulgaria and Croatia. To these we have added Turkey. The four EECs accounted in 2004 for as much as 65% of all Italian textile exports to the EECs and for as much as 77% of apparel imports from the same region. As for the leather and footwear sector, these countries accounted for as much as 60% of leather exports and 72% of footwear imports. The research has tried to test the hypothesis of a long-run relationship, that is, cointegration, between imports and exports of goods between Italian provinces and regions and foreign economic areas. According to this hypothesis the trade flows of manufacturing between economic regions have not only commercial purposes. They represent a proxy of productive integration as well. In the case of vertical productive integration, the static OLS model can be represented as:
M n ,t = α + β X ( n −1),t −1 + u t Whereas for the horizontal integration the model can be expressed as:
X n ,t = α + βM n ,t −1 + u t
9
where M and X are flows of imports and exports, respectively, n indicates the nth phase of the productive process and ut is the error term which satisfies the general assumptions. Cointegration between series has been tested using the Error Correction Model (ECM), either in its restricted or unrestricted version. The ECM restricted can be represented as follows (vertical integration):
∆M n ,t = α + ∆β X ( n −1),t + ϕECM t −1 + u t where ECM is the residuals of the static OLS model. Due to its features this ECM has been applied in those cases where the two time series are both non-stationary, and integrated of the same order [I(1)]. The hypothesis of cointegration of the import and export time series has been tested with a ttest on the ECM term. For the cases where the included series have different order of integration, the unrestricted ECM model proposed by Pesaran et al. (1999) has been used. This can be applied irrespective of the order of integration of the series, and can be written as (vertical integration):
∆M n ,t = α + ∆βX n ,t + π 1 M n ,t −1 + π 2 X ( n −1),(t −1) + u t In this case the hypothesis of cointegration is tested through an F-test whose critical values have been tabulated by Pesaran et al. (1997). The data used in the analyses come from the data warehouse of the IST-AT. These are quarterly data on the value of the trade flows of goods in specific phases of the analysed value chains. The time series cover a period between 1991 and 2004. Due to the huge difference of scale between import of goods from China and India and export of goods towards the final markets, in the analysis of the horizontal integration the data have been log-transformed. 4.2. Empirical results Before analysing the long-run relation between series, a preliminary analysis on the order of integration of the series themselves has been done. It has resulted that all the series are either stationary or integrated of order one [I(1)], therefore a regular OLS model to verify the relation between variables cannot be used. The restricted ECM model has been used for those series both integrated of order one. In the other cases, the unrestricted ECM developed by Pesaran at al (1999) has been adopted. The results of the analyses of the vertical productive integration between the provinces of Veneto and the EECs plus Turkey are listed in table 2. The t and F-tests provide sufficient evidence of long-run relation between variables, since the null hypothesis of no cointegration is rejected all six times at the 0.05 level. The coefficients φ and π1 give the speed of adjustment, and they are all negative and smaller than one, there is a convergent path towards the long-run equilibrium adjusting the short-run deviation from the steady state. The coefficient β gives the short-run effect of the export on import, whereas the coefficient -(π2/ π1) represents the long-run effect. Looking at the results of the restricted ECM model for the provinces it seems that the export of intermediate goods towards the EECs determines a big effect on the import of final goods, either apparel or footwear, from those countries. The short-term effect of export of textile is not even significant for Vicenza, whilst its long-run effect is equal to 1.4. For both Padova and Treviso an increase in textile exports to EECs leads to a more than proportional increase in imports of apparel from the same economic area. The same can be said for Treviso and Verona for what concerns the L-F sector. 10
Table 2. Test for cointegration between provinces of Veneto and EECs plus Turkey. ECM Restricted ECM Unrestricted PROVINCE
α
Β
Φ
α
Β
π1
π2
-(π2/ π1)
F-test
Textiles and Apparel Vicenza
coef. P>|t| coef. P>|t| coef. P>|t|
Padova Treviso
1.364 0.00** 1.139 0.00**
0.09 0.513
-0.468 0.00**
0.657 0.00**
1.4
11.81
0.641 0.00**
-0.461 0.00**
0.864 0.00**
1.87
9.7
-0.794 0.00** -0.209 0.03**
Leather and Footwear Venezia
coef. P>|t| coef. P>|t| coef. P>|t|
Treviso Verona
2.174 0.00** 1.619
-0.126 0.04** -0.618
0.00**
0.00**
*significant at 0.1 level, **significant at 0.05 level. Critical value for the F-test at 0.05 level: 8.199
Table 3. Test for cointegration between regions and EECs plus Turkey. ECM restricted ECM Unrestricted α α Β Φ β π1 π2 REGION
-( π2/ π1)
F-test
Textiles and Apparel Veneto Lombardia Toscana Piemonte Emilia
coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t|
2574833
0.740
-0.533
0.01**
0.00**
0.00 **
0.728
-0.493
0.00 **
0.00 **
0.639 0.026**
-0.485 0.00**
0.840 0.00**
1.782
11.72**
3941995
-0.176
-0.378
0.1713
0.548
4.1
0.029**
0.65**
0.01**
0.05**
2.14
2.42
0.925
2.27
0.939
-0.283
0.606
0.00**
0.04**
0.03**
Leather and Footwear Veneto Lombardia Toscana Marche Puglia
coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t|
1.336
-0.222
0.00**
0.029**
0.132
-0.173
0.1066
0.026**
1.79
-0.094
0.00**
0.1098
0.482
-0.341
0.00**
0.001**
790121
-0.05
-0.232
0.174
0.05**
0.8
0.1*
0.282
*significant at 0.1 level, **significant at 0.05 level. Critical value for the F-test at 0.05 level: 8.199
11
The analysis of the productive integration at the regional level provides different results. The hypothesis of the long-run relationship is rejected for Toscana and Emilia Romagna in the T-A, and for Lombardia and Marche for the L-F sector. For all the other cases the hypothesis of cointegration is not rejected at 0.05 level and all the coefficients of long-run, short-run effects and speed of adjustment are significant and in accordance with the economic theory. The case of Veneto and Marche are particularly relevant in the L-F industry. The important productive districts of Vicenza, as far as leather production, and Verona, Venezia and Treviso, for footwear, seem to create an integrated system with the EECs. Even Marche, with the productive district of Fermo, seems to be productively integrated with the EECs and Turkey. Firms based in these productive districts are the most internationalized and they have moved in the EECs some of the phases of the productive processes. Table 4. Test for cointegration between Italian regions and China.
ECM Restricted Α
REGION
β
ECM Unrestricted Φ
α
β
π1
π2
-( π2/ π1)
F-test
Textiles and Apparel Veneto Lombardia Toscana Piemonte Emilia
coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t|
0.685
-0.306
0.00**
0.04**
13.55 0.00**
0.428 0.00**
-0.882 0.00**
0.219 0.00**
0.248
20.1
8.249 0.00**
0.541 0.00**
-0.506 0.00**
0.101 0.04**
0.199
7.32
7.48
0.461
-0.663
0.307
0.463
12.7
0.00**
0.00**
0.00**
0.00**
23.9
0.311
-1.424
0.149
0.104
48
0.00**
0.01**
0.00**
0.04**
Leather and Footwear Veneto Lombardia Toscana Marche Puglia
coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t|
0.221
-0.206
0.00**
0.01**
0.179
-0.179
0.00**
0.00** 9.145
0.203
-0.469
-0.001
0.00**
0.00**
0.00**
0.9
21.863
0.189
-1.21
0.104
0.00**
0.00**
0.0**
0.05**
4.665
-0.01
-0.25
-0.003
0.01**
0.87
0.015**
0.74
9.4 0.08
39.1 2,5
*significant at 0.1 level, **significant at 0.05 level. Critical value for the F-test at 0.05 level: 8.199
The hypothesis of productive integration between Italy (Italian districts) and China and India is different from the one we tested for the EECs. For China and India there supposedly exists a horizontal productive integration, which implies that from those countries Italian firms import goods that had not previously exported. The productive cycle ends partially with the domestic consumption of the final goods and partially with the export of these goods to final markets. The hypothesis of long-run equilibrium between imports and exports is therefore tested by analysing the cointegration between imports of apparel and footwear from China (and India) and export of the same goods to the most relevant final Italian markets is not rejected eight out of ten times for the two analysed sectors. In regards to trade with China, for both the sectors, the long-run relationships suggest that a 10% increase in the import of goods from China leads to an increase of exports to the final markets 12
between 1 and 5%. This is a reasonable result since the volume of import of goods from China is smaller than that of going to final markets. China is a big market for Italy’s imports of both apparel and footwear, but up to the beginning of 2005 there existed an agreement which prevented Chinese exports to massively enter the EU market. Speeds of adjustments are always negative and smaller than one for all the cases but Piemonte and Marche in the L-F sector. This suggests that the short-term shocks are absorbed, and imports and exports of apparel and footwear tend towards the steady state in the long-run. Table 5. Test for cointegration between Italian regions and India.
ECM Restricted Α
REGION
β
ECM Unrestricted Φ
α
β
Π1
π2
-( π2/ π1)
F-test
Textiles and Apparel Veneto Lombardia Toscana Piemonte Emilia
coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t|
0.311
-0.515
0.00**
0.00**
0.354 0.00**
-0.636 0.00**
0.188
-0.838
0.00**
0.00**
0.469
-0.416
0.00**
0.00**
0.578
-0.815
0.00**
0.00**
Leather and footwear Veneto Lombardia Toscana Marche Puglia
coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t| coef. P>|t|
0.743
-0.379
0.00**
0.02**
0.169
-0.602
0.8431
0.00**
0.527
-1,692
0.00**
0.00**
0.04
-0.426
0.01**
0.00**
9.118
0.496
-0.700
0.299
0.00**
0.00**
0.00**
0.01**
0.427
13.88
*significant at 0.1 level, **significant at 0.05 level. Critical value for the F-test at 0.05 level: 8.199
The analysis of the long-run relationship between Italian regions and India produces results similar to those for China. In the case of India the restricted ECM model has been used all but one time due to the fact that each pair of analysed time series have the same order of integration [I(1)]. Imports and exports of both apparel and footwear are always cointegrated. The results of the short-run elasticity of imports suggest that a 10% increase of imports from India leads to a less than proportional increase of exports of the same goods (apparel and footwear) to final markets in Italy.
5. The degree of international integration of production in Italian districts Analyses of international productive integration processes between Italian firms and foreign economies may be approached by looking at how the spatial organization of involved supply chains is evolving. Until the end of the 1980s most of the productive processes of the manufacturing firms took 13
place within the domestic economy. In the 1990s some firms began internationalizing the value chains of their production. The international fragmentation of productive processes has generated global flows of intermediate goods, measured also by the growing relevance of intra-industry trade. Therefore, flows of goods of the same supply chain between countries suggest the presence of productive integration processes. Thus, a small or medium-sized enterprise can activate foreign firms based in the host countries to work on “delocalized” phases also through market relationships (trade), not just with FDI. In this way, the small firm in an Italian industrial district introduces several foreign workers into its supply chain (network strategy) without its own foreign branch (property strategy). In some cases this implies that the internationalization of the supply chain produces “substitution effect” between workers employed in manufacturing. But this effect does not always appear, because often the delocalization processes respond either to an outside growth strategy of the firm, or to international outsourcing which strengthens the enterprise’s competitiveness. However, the change in the domestic (Italian) local labour market and the increase in the number of workers based abroad is a phenomenon that our analysis tries to recognize. But because the productivity level in the new host countries is usually lower than in Italy (this is true in general, but particularly in “Made in Italy” industries), therefore – given the value of production – the number of workers employed should end up being bigger. To determine the number of “foreign induced employees” for every industry, the index described in the two preceding sections has been used.. As explained above, the number of foreign induced employees is a function of both the flows of goods between Italian and foreign firms and the level of productivity of the foreign workers involved in the industry studied. Table 65 shows the evolution of the foreign induced employees in the three macro geo-economic regions considered, for the three industries studied, in 1996, 2001 and 2003. At the beginning of the survey’s period, foreign induced employees value was about 141,000. EECs hosted 76,000 of them, whereas MED and ASIA had 26,000 and 38,000 respectively. Looking at the same data for 2003 it appears that the phenomenon of delocalization increased very sharply. Indeed EECs, MED and ASIA hosted around 195,000, 51,000 and 98,000 workers respectively. T-C is the industry employing the highest number of workers, followed by L-F and W-F. Deepening the analysis it appears that in 2003 EECs accounted for as much as 56% of the total Fie of the T-C. For the L-F the share of Fie in the EECs was about 46% and for the W-F the portion was 68%.
5
Data contained in Table 6 and Figures 1-12 are original elaborations of data from the ISTAT datawarehouse
14
Table 6. Induced Foreign Employees in the Italian Districts Induced foreign employees T-A EEC
Mediterra nean Basin
China and India
16,792 14,987 9,227
69,802 50,005 24,975
117,054 103,889 38,612
2003 2001 1996
NI 96=100
Tot
203,648 168,880 72,814
% on Tot. Empl.
280 232 100
NI Xtot 96=100
114 125 100
28% 11%
Induced foreign employees L-F EEC
Mediterra nean Basin
China and India
24,877 24,933 11,225
18,359 15,866 7,672
37,745 34,713 15,169
2003 2001 1996
NI 96=100
Tot
80,981 75,512 34,066
% on Tot. Empl.
238 222 100
NI Xtot 96=100
108 128 100
37% 15%
Induced foreign employees W-F EEC
Mediterra nean Basin
China and India
9,301 10,421 5,533
9,556 7,673 5,848
40,522 36,734 22,812
2003 2001 1996
NI 96=100
Tot
59,379 54,828 34,193
% on Tot. Empl.
174 160 100
NI Xtot 96=100
120 134 100
23% 14%
Figure 1 shows the values of the Delocalization Index (DI) of the T-C sector for each of the local industrial systems studied, in 1996, 2001 and 2003. 1. Delocalization Index of the T-A supply chain 2,500
2,000
1,500
1,000
0,500
1996
2001
Brindisi
Pescara
Bari
Benevento
Lecce
Prato
Ferrara
Rovigo
Novara
Pesaro Urbino
Perugia
Ancona
Rimini
Modena
Napoli
Teramo
Padova
Mantova
Brescia
Firenze
Arezzo
Vicenza
Vercelli
Varese
Pistoia
Verona
Venezia
Como
Reggio nell'Emilia
Biella
Bergamo
Lecco
Treviso
0,000
2003
The same applies for Figures 2 and 3 for the L-F and W-F sectors respectively.
15
2. Delocalization Index of L-F supply chain 2 1,8 1,6 1,4 1,2 1 0,8 0,6 0,4 0,2
o lin el
er ta as
en C
Av
a
a rli '-C Fo
As
1996
es
ne
zi
ol i Ve
a
ap
ri
vi
N
Pa
Ba
sa Pi
at
M
ac
er
ic e
a
no
zo ez
ia to
co
li P
Ar
e
Pi s
za
nz Fi re
e cc
en Vi c
o
Le
m Te
ra
va do
a Pa
cc Lu
is o ev
ro Ve
Tr
na
0
2001
2003
3. Delocalization Index of the W-F supply chain 0 ,6 0 ,5 0 ,4 0 ,3 0 ,2 0 ,1
1996
2001
Pistoia
Pesaro-Urbino
Bari
Matera
Como
Verona
Milano
Pisa
Forlì-Cesena
Perugia
Padova
Pordenone
Treviso
Udine
Venezia
Vicenza
0
2003
The DI is the ratio between the Fie and the number of domestic workers employed in the productive phases involved in the international outsourcing process. Values of DI larger than one mean that for the entire production process, firms of those local industrial systems do rely more on workers abroad than they do on employees in Italy (keeping in mind differences in productivity). 16
For both the T-A and L-F industries, the first few local industrial systems show a DI greater than one, at least since 2001. What is most evident for all three industries and for all the clusters is the increased value of the index between 1996 and 2003. Treviso, which is the leader of the T-A industry, grew its DI from 0.55 to more than two. This means that in 2003 for every “domestic” worker employed in this industry there were a couple of workers employed in the three economic regions abroad. Several other clusters followed the same path, although with lower results. In the LF industry the results are similar. Verona and Treviso, the first two local industrial systems, had DI close to two in 2003. In 1996 they had very low DI, especially Treviso, with DI less than 0.4. Due to its features, the W-F industry shows values of DI much lower for all the provinces studied. Only Vicenza has an index higher than 0.6 in 2003. Yet, between 1996 and 2001, all the provinces increased their DI. Figures 4-6 are constructed using data about the number of Fie employed in the EECs for each of the three industries. By looking at Fig. 4 it is easy to recognize the role of Treviso, and its firms, in the delocalization process of the firms producing textiles and apparel. In 1996, firms from Treviso employed slightly more than 5,000 workers in the EECs. In 2003 this number surpassed 22,000. As shown in Fig. 5 Treviso is the leading province in terms of total Fie in EECs, even for the L-F industry. Verona ranks second with fewer Fie, whereas the other provinces present a much lower productive integration with EECs. In 2003, Treviso and Verona totalled almost 16,000 workers in the countries of Eastern Europe, and as in the case of the T-A sector they sharply reinforced the delocalization process between 1996 and 2003.
5. EECs-Delocalized employees of the L-F supply chain 9000 8000 7000 6000 5000 4000 3000 2000 1000
1996
2001
Avellino
Caserta
Napoli
Teramo
Pavia
Forlì-Cesena
Pistoia
Vicenza
Arezzo
Venezia
Firenze
Bari
Pisa
Lucca
Macerata
Padova
Lecce
Ascoli Piceno
Verona
Treviso
0
2003
17
Fig. 6 shows the number of EECs Fie for the W-F sector. For this industry the productive district of Udine province (world leader in chair production) ranks first, while Treviso is second. As for the DI, for the number of Fie too, this sector shows much lower levels than the previous two sectors. Indeed, in 2003 Udine had 5,300 Fie and Treviso 4,100. The third province, Vicenza, had less than 3,000 Fie, although it had the higher DI for this sector.
6. EECs-Delocalized employees of the W-F supply chain 6 .0 0 0 5 .0 0 0 4 .0 0 0 3 .0 0 0 2 .0 0 0 1 .0 0 0
1996
2001
Pistoia
Forlì Cesena
Matera
Pisa
Como
Perugia
Pesaro Urbino
Bari
Verona
Padova
Venezia
Milano
Pordenone
Vicenza
Treviso
Udine
0
2003
Figures 7-12 report the number of Fie in 1996 and 2003 for each of the three productive sectors. These figures divide the Fie by area of economic integration. Therefore, Fig. 7 contains a graph on the number of Fie in 1996 for the T-A sector in the EECs, MED and ASIA. Fig. 8 has the same graph, referring to 2003.
7. T-A supply chain; number of employees by economic area of productive integration in 1996 6 .0 0 0 5 .0 0 0 4 .0 0 0 3 .0 0 0 2 .0 0 0 1 .0 0 0
Peco
Benevento
Brindisi
Pescara
Prato
Pesaro Urbino
Bari
Napoli
Ancona
Vercelli
Biella
Ferrara
Rimini
Rovigo
Lecce
B a c in o M e d ite rra n e o
Perugia
Lecco
Novara
Firenze
Mantova
Como
Teramo
Venezia
Modena
Varese
Padova
Verona
Pistoia
Arezzo
Brescia
Vicenza
Reggio nell'Emilia
Bergamo
Treviso
0
C in a e In d ia
18
8. T-A supply chain; number of employees by economic area of productive integration in 2003 25.000
20.000
15.000
10.000
5.000
Peco
Brindisi
Benevento
Vercelli
Prato
Pescara
Rimini
Ferrara
Lecce
Lecco
Rovigo
Novara
Pesaro Urbino
Bari
Biella
Bacino M editerraneo
Napoli
Ancona
Pistoia
Perugia
Firenze
Teramo
Venezia
Como
Mantova
Varese
Arezzo
Modena
Padova
Verona
Brescia
Vicenza
Reggio nell'Emilia
Treviso
Bergamo
0
Cina e India
By looking at these two figures it is easy to recognize the role of the EECs in the internationalization processes involving the Italian provinces. In both 1996 and 2003 the Fie based in Eastern Europe represented the vast majority of the total Fie for almost all the provinces. Furthermore, the polarization of Fie within the EECs increased, due to the reduction of relative weight of the other two economic regions. It was only for Firenze and Napoli that China and India were the most important productive partners. Figures 9 and 10 record the same information for the L-F productive sector. In this case too, the EECs have always been the most important productive partner. Unlike the T-A industry, MED is the second most important area of productive integration for L-F. Yet, between 1996 and 2003 there has been a substitution effect which increased the weight of the EECs and reduced those of MED and ASIA in the international productive integration processes. 9. L-F supply chain; number of employees by economic area of productive integration in 1996 4500 4000 3500 3000 2500 2000 1500 1000 500
Peco
B a c in o M e d ite rra n e o
C in a e In d ia
Avellino
Napoli
Caserta
Pavia
Forlì-Cesena
Teramo
Pistoia
Vicenza
Arezzo
Venezia
Bari
Lucca
Firenze
Pisa
Macerata
Ascoli Piceno
Padova
Lecce
Treviso
Verona
0
19
10. L-F supply chain; number of employees by economic area of productive integration in 2003 9000 8000 7000 6000 5000 4000 3000 2000 1000
Peco
B acino M editerraneo
Avellino
Caserta
Napoli
Teramo
Pavia
Forlì-Cesena
Pistoia
Vicenza
Arezzo
Venezia
Firenze
Bari
Pisa
Lucca
Macerata
Padova
Lecce
Ascoli Piceno
Verona
Treviso
0
C ina e India
The W-F industry (Figures 11 and 12) presents a slightly different situation.
11. W-F supply chain; number of employees by economic area of productive integration in 1996
4.500 4.000 3.500 3.000 2.500 2.000 1.500 1.000 500
UE
PECO
AFRICA
Matera
Pistoia
Perugia
Forlì Cesena
Bari
Pesaro Urbino
Pisa
Como
Verona
Padova
Venezia
Milano
Vicenza
Pordenone
Treviso
Udine
0
ASIA
20
12. W-F supply chain; number of employees by economic area of productive integration in 2003
6 .0 0 0 5 .0 0 0 4 .0 0 0 3 .0 0 0 2 .0 0 0 1 .0 0 0
UE
PECO
A F R IC A
Pistoia
Forlì Cesena
Matera
Pisa
Como
Perugia
Pesaro Urbino
Bari
Verona
Padova
Venezia
Milano
Pordenone
Vicenza
Treviso
Udine
0
A S IA
The areas and the countries that have mainly traded with the Italian firms are different from those of the two previous industries. Though EECs have held primacy in terms of Fie, firms based in Western European countries still have a role in the productive integration processes of Italian firms. Besides the EECs and the Western European countries, the other economic areas productively integrated with the Italian provinces are Africa (not just MED) and all of Asia (not just China and India). These differences are due to the features of the industry as well as to the geographical location of the productive sites. Despite these differences, the EECs increased their relevance in terms of Fie between 1996 and 2003. In almost all the provinces the number of Fie based in the EECs relatively increased with respect to those based in the other areas.
6. Conclusion and remarks The empirical results here described should have some consequences for the policy debate: in particular, attention should be directed toward issues such as how to govern a spontaneous tendency toward internationalization. Even if SMEs are playing an active role in the general process of fragmentation of production, specific policy measures are useful and necessary. In the political debate, meanwhile, the repositioning of domestic supply in the international framework is seen as an outsourcing process, dangerous for maintaining employment, promoting economic opportunities and ensuring capacity to maintain some competitive advantage in international markets. Therefore, at the centre of the economic policy debate is the question of whether the internationalization of production is crowding out national employment, and consequently national income and capacity to sustain dynamic competition. From a preliminary analysis (see also the results in Rossetti and Schiattarella 2003), it is likely that there does exist a positive association between productive internationalization and economic performance. The higher the degree of supply internationalization, the lower the decrease in 21
manufacturing employment, the higher the total income of the clusters, the higher the level, and consequently the wages, of the employment demanded. We want here to mention again the important inter-sectoral linkages between “Made in Italy” clusters and the mechanics sector. As we have seen in the analysis of international trade flows, it is precisely the interdependence with the mechanized and machinery sectors of these clusters which has permitted a high degree of innovation (contrary to the idea that innovation requires large firm size) and high quality.
References Amin A. (1993), “The globalization of the economy: an erosion of regional networks?”, in Grabher G. (a cura di), The embedded firm. On the socioeconomics of industrial networks, Routledge, London Arndt S.W., Kierzkowsky H (eds.), 2001, Fragmentation. New Production Patterns in the World Economy, Oxford, Oxford University Press Arora A., Fosfuri A., Gambardella A., 2001, Markets for Tecnology: The Economics of Innovation and Corporate Strategy, Boston, MIT Press Barney J. (1991), “Firms Resources and Sustained Competitive Advantage”, Journal of Management, Vol. 17, n.1 Baldone S., Sdogati F., Tajoli L., 2002, “La posizione dell’Italia nella frammentazione internazionale dei processi produttivi”, in Ice-Istat, L’Italia nell’economia internazionale 2001, Roma Barba Navaretti G., Castellani D., 2002, Does investing abroad affect performance at home? Comparing Italian multinational and national enterprises, Milano, Working paper Centro Studi Luca D’Agliano Becattini G. (1998), Distretti industriali e Made in Italy, Bollati Boringhieri, Torino Becattini G., Rullani E. (1993), “Sistema locale e mercato globale”, Economia e politica industriale, 80 Bhagwaty J., Deheljia V.H., “Free Trade and Wages of the Unskilled”, in Bhagwati J., Koesters M. H. (eds.), Trade and Wages, Washington D.C., The American Enterprise Institute Press, 1994 Brown J.S., Duguid P. (2001), “Knowledge and Organization: a Social-Practice Perspective”, Organization Science, Vol. 12, n. 2, March-April Brusco S. (1994), “Sistemi globali e sistemi locali”, Economia e politica industriale, 84 Cooke P., Morgan K. (1994), “The creative milieu”, in Dodgson M., Rothwell R. (a cura di), The Handbook of Industrial Innovation, Edward Elgar, Cheltenham Corò G., Grandinetti R. (1999), “Evolutionary patterns of Italian industrial districts”, Human Systems Management, 18 Corò G., Volpe M. (2003), “Frammentazione produttiva e integrazione internazionale nei sistemi di piccola e media impresa”, Economia e società regionale, n. 1 Dierickx I., Cool K. (1990), “Asset Stock Accumulation and Sustainability of Competitive Advantage”, Management Science, Vol. 35, n.2 Dunning J.H. (1981), International production and the multinational enterprise, Allen&Unwin, London 22
Feenstra R.C. (1998), “Integration of Trade and Disintegration of Production in the Global Economy”, Journal of Economic Perspective, Vol. 12, Number 4 Gereffi G. (2004), “The Global Economy: Organization, Governance, and Development”, in Neil Smelser N. and R. Swedberg (editors), Handbook of Economic Sociology, 2nd edition, Princeton University Press and the Russell Sage Foundation, Hummels D., Rapoport D., Yi K. (1997), “Globalization and the Changing Nature of World Trade”, Chicago, University of Chicago paper Kogut B. (1985), “Designing global strategies: comparative and competitive value-added chains”, Sloan Management Review, summer Krugman P. (1995), “Growing World Trade: Causes and Consequences”, Brooking Paper on Economic Activity, 1 Levitt T. (1983), “The globalization of markets”, Harvard Business Review, May-June Onida F. (2001), “IDE: su cosa fare ricerca?”, Economia e politica industriale, n. 111 Pesaran, M.H., Y. Shin and R.J. Smith (1999), “Bounds Testing Approaches to the Analysis of Long-Run Relationship”, DAE Working Paper No. 9907, Department of Applied Economics, University of Cambridge. Pesaran, M.H. and B. Pesaran, (1997), Microfit 4.0: Interactive Econometric Analysis, Oxford, Oxford University Press. Porter M. (1990), The Competitive Advantage of Nations, The Free Press, New York Porter M. (1998) “Cluster and the new economics of competition”, Harvard Business Review, November-December Reprint (2003), L’Italia Multinazionale. Rossi A. (2004), La Governance dell’internazionalizzazione produttiva, Formaz, Roma. Savona M. and R. Schiattarella (2004), “International relocation of production and the growth of services: the case of the “Made in Italy” industries”, Transnational Corporations (Vol.13, No.2) Saxenian A. (1994), Regional Advantage, Harvard University Press, Cambridge (Mas) Schiattarella R. (1999), “La delocalizzazione internazionale: problemi di definizione e di misurazione. Un’analisi per il settore “Made in Italy”“, Economia e politica industriale, n. 103 Tattara G. (a cura di) (2001), Il piccolo che nasce dal grande. Le molteplici facce dei distretti industriali veneti, FrancoAngeli, Milano Zanfei A. (2000), “L’impatto delle multinazionali sui Paesi ospiti: il problema della creazione di legami a monte e a valle con le economie locali”, Economia e politica industriale, n. 105
23
