ITALY’S INDUSTRIAL CHANGES IN THE nineties: A COMPARISON BETWEEN LARGE ENTERPRISE AREAS AND INDUSTRIAL DISTRICTS
(First draft )
Please, do not quote
Gabi Dei Ottati / Laura GrassiniDipartimento di Scienze Economiche
Università di Firenze
/ Dipartimento di Statistica G. Parenti
Università di Firenze
Abstract This paper examines the uneven economic development of the Italian local labor systems (LLS) in the Nineties. The study focuses on the changes in employment, using the 1991 and 2001 Economic Census data. The chief aim of the paper is to compare the employment performances of different types of LLS, and particularly of industrial districts (geographically concentrated areas of specialized SME) and local systems characterized by the prevalence of large enterprises. The originality of the study consists in the territorial unit chosen for the empirical analysis, that is the 784 Local Labor Systems (LLS) of Italy identified by ISTAT on the basis of 1991 census data. Among those LLS there are 199 manufacturing LLS identified by ISTAT as Industrial Districts (ID), among the remaining manufacturing LLS we identified 72 LLS characterized by the prevalence of LES, the residual 513 LLS are considered Residual Systems (RS). In this paper, after an examination of the employment changes by sector and type of LLS in Italy, the local area competitive effect derived from the shift and share analysis is discussed and modeled on a set of covariates to investigate any different performances of the types of LLS. The data here analyzed show that also in the nineties the Industrial Districts proved to be the most dynamic local systems of Italy, performing better than the national average and particularly of Large Enterprise Systems. Moreover, the shift and share analysis demonstrates that in the nineties the local competitive effect in ID was high enough to more than compensate the negative industry mix, due to the drop of employment in the production of the goods for the person and the home. Conversely, in LES the local effect was so negative to determine a total negative effect, despite the positive industry mix.
Key words: economic development, manufacturing industry, industrial districts, local labor systems, made in Italy, shift-share, industry-mix effect, local area competitive effect.
1. Introduction
Considering the overall economic performance of Italy during the second half of the XX century, it looks like an undeniable success: to give just an idea, being 100 the level of real per-capita GDP of USA, the Italian level passed from 45.9 in 1938 to 75.1 in 1992, so exceeding the per-capita GDP of UK, which in 1938 ranked just after the USA (Maddison, 1995). In spite of that remarkable growth, due to the industrialization occurred since the second World War, the Italian economy, and particularly its industrial structure, persists markedly different from the other developed countries.
In particular, the Italian economy is characterized by small and medium sized firms, whereas large companies are very few. This is so, not only in comparison with USA and UK, but also with France or Germany. Moreover, the Italian manufacture continues to be specialized in sectors such as textiles, clothing, footwear, or furniture, instead of moving towards high tech and capital intensive sectors. Hence, several economists following the mainstream approach have recurrently indicated those features (the small size of firms and the specialization in so called traditional sectors) as responsible for the relative decline in the Italian GDP growth in the ’90s, and in more recent years.
Our approach is different and the related interpretation is more complex.
First of all, we believe that the prevailing small size of the Italian manufacturing firms and their economic activity specialization are two closely interlinked features (Becattini, 1996). Moreover, we think that those features cannot be properly understood without taking into account the territorial organization of the Italian economy: i.e. the importance of local systems and, especially, of those local systems characterized by many specialized small and medium sized enterprises (SME) that conform to the industrial district (ID) model[1].
According to that interpretation, any meaningful empirical study of the Italian economic development has to focus on the many different local systems that make up the mosaic of the country. Moreover, at this point, many studies have demonstrated that ID have played a remarkable role in the economic performance of Italy, at least since the ‘70s, especially as far as exports are concerned[2]. Various scholars[3], and among them Michael Porter (1990), have by now proven the substantial correspondence between the development of industrial districts and the success of the Italian exports of goods for the person (textiles, clothing, leather goods, jewels…) and for the home (furniture, lighting equipment, tiles…) (Personal and Household Goods) , together with the related machine tools (Light Mechanics).
Considering both the type of manufacturing specialization of Italy and the importance of ID for the Italian exports and economic development in general, we have chosen to focus the empirical analysis on local systems, instead of sectors at national level.
Being convinced that the two main “engines” of the Italian economy are industrial districts (ID) and local areas characterized by one or more large enterprises (LES), our principal aim is to compare their relative performances during the ‘90s. In particular, first we examine Italy’s employment changes, by sector and by type of local system with a special focus on manufacturing. Then, we apply the shift and share analysis to the employment data, by sector and local system, in order to separate the industry-mix effect from the local competitive effect in the economic performance of the area.
The paper proceeds as follows. In the next section the methodology and the data we use are specified. The economic performances of the different types of local systems are considered in section 3. Section 4 presents the shift and share analysis and the results of its applications. Some conclusions are drawn in section 5.
2. Data and methodology
According to the approach outlined in the previous section, since it was not possible to make a direct comparison between the performances of large enterprises, a phenomenon that exists essentially at company level, with those of industrial districts, which is principally territorial, we had to choose a territorial unit of analysis suitable for the comparison. The best-known territorial definition of the Italian ID is the Istat-Sforzi (1997) one, which is based on the 1991 census “local labor systems”[4]. Specifically, out of the 784 LLS of Italy, there were identified 199 local labor systems having some features of the ID model[5]
Therefore, we decided to use the 199 ID identified by Istat-Sforzi. Furthermore, always on the basis of 1991 Census data, we found those non ID LLS characterized by a manufacturing industry with the prevalence of large enterprises, through the following procedure. Non ID LLS were selected according to the degree of manufacturing employment. The LLS with a degree of manufacturing employment over the national average were considered “large enterprises systems” (LES) if the share of employment was concentrated in large establishments (establishments with more than 249 workers). Residual systems (RS) are those which fall neither into the class of ID, nor in that of LES. Among the non ID LLS, we identified a total of 72 LES[6].
It is necessary to observe that, at the end of 2005, a new map of LLS based on 2001 Population Census data was released by ISTAT (ISTAT; 2005). In this paper, we used the map at 1991 because other statistical information has been produced at this geographical levels (for example: the time series 1996-2002 of added values and employment for each 1991 LLS), that allow future development of the study of the transformations occurred in Italy at the end of the century. Moreover, the new map of LLS is only partially different from the one of 1991. The number of LLS was in 1991 784 with 199 ID, whereas in 2001 there are 686 LLS with 156 ID. If we refer to the municipalities (the common unit in the two classifications), we have that:
- 1856 of 2474 (83.8%) municipalities that in 1991 were in ID still belong in 2001 to ID;
- 5268 of 5627 (89.5%) municipalities that in 1991 were not in ID still belong in 2001 to non-ID LLS;
- a total of 7124 of 8101 (87.9%) municipalities do not change type of LLS between 1991 and 2001.
Table 2.1 and Figure 2.1 give a picture of the geographical location of the three types of LLS. We adopted the usual classification into the Istat four Italian macro-regions. We can see that the North East and Center regions concentrate the largest number of ID, corresponding to the so called Third Italy (Bagnasco 1977); Southern areas, instead, are characterized by a large prevalence of RS, whereas, in the North-west of Italy there is the highest proportion of LES.
Table 2.1: Geographical location of the different types of LLS (column percentages)
LLS / North west / North east / Center / South and isles / TotalN. / % / N. / % / N. / % / N. / %
ID / 59 / 42.2 / 65 / 45.5 / 60 / 44.1 / 15 / 4.1 / 199
LES / 30 / 21.4 / 16 / 11.2 / 13 / 9.6 / 13 / 3.6 / 72
RS / 51 / 36.4 / 62 / 43.4 / 63 / 46.3 / 337 / 92.3 / 513
Total / 140 / 100.0 / 143 / 100.0 / 136 / 100.0 / 365 / 100.0 / 784
Classification based on 1991 Economic Census data.
Figure 2.1: map of the LLS based on 1991 Population Census data
Having accounted for the reason we chose the LLS as the territorial unit of analysis, we want now to specify the classification of the official ISTAT economic activities here used. Before focusing on manufacturing industry, we considered all the economic activities of firms, subdivided into the following sectors (Table 2.2).
Table 2.2: Classification of economic activities
Code / DetailAB / Products of agriculture, hunting, forestry (A); fish and other fishing products (B);
CE / Products from mining and quarrying (C); electrical energy, gas and water (E);
D / Manufactured products
F / Construction work
G / Wholesale and retail trade services; repair services for motor vehicles, motorcycles and personal and households goods
H / Hotel and restaurant services
I / Transport, storage and communication services
J / Financial intermediation services
K / Real estate, renting and business services
MNO / Education services (M), health and social work services (N), other community, social and personal services (O)
Some sectors that are considered less important for the purposes of this paper (for example: CE, AB) have been aggregated to obtain a synthetic picture of the whole economy and in order to render workable the shift and share analysis[7].
Having devoted a specific investigation to manufacturing industries, we have now to explain why we grouped the 14 sectors of the ISTAT official manufacturing activities into four clusters of products. This choice is mainly justified by the purpose to subdivide the Italian manufacture in a way that can reveal, instead of concealing, the products in which Italy has gained a competitive advantage, and their peculiar industrial organization. Therefore, we grouped together the goods for the person and for the home (PHG)[8], which constitute the core of the Italian exports. Then, we classified as light mechanics (LM) the production of metal goods and machinery most of which are used to manufacture PHG, and whose share of the Italian exports is also significant. Actually, PHG and light mechanical engineering are so important for the Italian balance of trade, that they are often labeled as “made in Italy”. In particular, a large part of PHG manufactured in Italy have a high stylistic, cultural and even emotional value, as they remind the Italian art, landscape, climate and, more generally, way of living. Hence, PHG are clearly distinct from the usually undifferentiated mass products of most Italian heavy industry (Heavy), such as basic chemicals, or metals. The fourth cluster of manufactured goods is food and beverages (Food), which includes also some of the distinguishing products of the Mediterranean diet, such as olive oil or good reputed wines.
As far as producer services (K sector) are concerned, we are particularly interested in the subcategories: Computer and related services (K72) and Other business services (K74)[9]. The reason for a deeper investigation of these services is their striking growth in the nineties, and by the fact that these services are also directly connected with the evolution of industrial firms in developed countries.
Since the main aim of this paper is to compare the employment performances of ID and LES, it is necessary to isolate, as far as possible, if any, the ‘district effect’[10]. This is quite problematic, since local development variables not directly related to the districts and or institutional variables, like the legal structure of firms, may interact with the ID variable. For example, district LLS may exhibit higher performances just because ID are mainly located in regions with a higher quality of infrastructures (see table 2.1).