«Cesare Imbriani* - Filippo Reganati** PRODUCTIVITY SPILLOVERS AND REGIONAL DIFFERENCES: SOME EVIDENCE ON THE ITALIAN MANUFACTURING SECTOR *Cesare ...»
Università degli Studi di Salerno
CENTRO DI ECONOMIA DEL LAVORO E DI POLITICA ECONOMICA
Cesare Imbriani* - Filippo Reganati**
PRODUCTIVITY SPILLOVERS AND REGIONAL
DIFFERENCES: SOME EVIDENCE ON THE ITALIAN
*Cesare Imbriani **Filippo Reganati
University of Roma, La Sapienza University of Cassino Istituto di Economia e Finanza Dipartimento Economia e Territorio P. le Aldo Moro Via Mazzaroppi 3 00100 Roma, Italy 03043 Cassino (FR), Italy E-mail: firstname.lastname@example.org E-mail: email@example.com Tel: +39 6 490642 Tel: +39 776 299438 Fax: +39 6 49910648 Fax: +39 776 312035 DISCUSSION PAPER 48 giugno 1999
CENTRO DI ECONOMIA DEL LAVORO E DI POLITICA ECONOMICA
Adalgiso Amendola, Guido Cella, Ugo Colombino, Cesare Imbriani, Giancarlo Marini, Pasquale Persico, Nicola Postiglione, Enrico Pugliese, Salvatore Vinci 2 Index
2. Productivity spillovers and multinational firms: some issues..... 9
3. Data, definitions of variables and statistical model................. 11
4. Statistical results
5. Concluding remarks
3 4 Abstract This work examines the main theoretical and empirical inter- pretations regarding the effects of foreign direct investment on productivity of local firms and, in particular, in which way produc- tivity spillovers are related to the existence of regional differences.
By taking into consideration the Italian manufacturing sector and using cross-section data, we find that although at a national level productivity levels are higher in the domestic sectors where multinational firms account for larger shares, productivity spillovers are concentrated only in the north-western area of Italy.
Key words: foreign direct investment, productivity spillovers.
JEL classification F23; 030 5 6 Introduction1 1.
The technological diffusion is a process by which innovations (i.e. new products, new processes or new management methods) spread within and across countries. In the economic theory, it is commonly agreed that the creation and the diffusion of new tech- nology is one of the major determinants of economic growth, inter- national competitiveness and trade performance. This view has been recently supported by the endogenous growth theory (Grossman and Helpman, 1991) which has emphasised the crucial role played in an international environment by both dynamic comparative advantages and international competition.
By exploiting comparative advantages embodied in host countries and/or overcoming market imperfections, foreign direct investment (FDI) and multinational enterprises (MNEs) have been traditionally considered as an important vehicle in the process of diffusion of technological and organisational innovations. In such one-way process, the multinational enterprises' (MNEs) home country has been essentially regarded as the centre from where 1 This paper has been presented at the Twelfth World Congress of the I.E.A.
Financial support from C.N.R. and MURST is gratefully acknowledged. Although
this paper has been jointly developed by the authors, it was written as follows:
section 1 and 2 C. Imbriani; section 3, 4 and 5 F. Reganati.
7 innovations flow to the various host countries.
However, the change in the dynamics of the international competition has pushed MNEs to transform themselves into learning organisations, which are interested in finding access to technology and innovations in a variety of host countries. As a result, FDI has been increasingly used as a means to tap innovation capabilities which are present in the host countries (Levitt and March, 1988; Kogut and Chang, 1991), thus preserving opportunities for future organisational learning and fostering technological spillovers on local firms. In such a picture, a new element, which has been recently emphasised in the literature (Amin and Tomaney, 1995), is the sub-national or regional dimension of the systems of innovation. Cantwell and Iammarino (1998, p. 384) have pointed out that “the MNE networks for innovation conform to a geographical hierarchy in different regional centres; accordingly, the technological specialisation of foreign firms in different regional locations depends upon the position of the region in the hierarchy, i.e. whether the regional system is at the top of the hierarchy (higher order location) or is a lower order regional centre”. This issue is very important not only to better understand the process of globalisation of technological activities but also to interpret more carefully the potential benefits of FDI for the host countries.
By taking into consideration the Italian manufacturing sector and using cross-sectional data, the purpose of this paper is twofold. First, we have tried to demonstrate whether and how technical knowledge has been transferred to domestic firms owing to the mere presence of MNEs; second, we have examined whether productivity spillovers are related to the existence of regional differences. This paper is organised as follows. In the next section, we give a short discussion on the nature of productivity spillovers;
section 3 describes the data set and the statistical model used; in section 4 the empirical results are presented and, finally, in section 5 we draw the main conclusions.
82. Productivity spillovers and multinational firms: some issues
The term “spillover” refers to the indirect effects generated by the presence of foreign firms both in the industrial structure of the host country and in the conduct and performance of local firms.
There are a number of spillover effects of FDI identified by the literature (Caves, 1974; Blomstrom, 1989; Blomstrom and Kokko, 1996). In particular, it is argued that the productivity of local firms may be mainly stimulated by three factors such as an increasing competition, the enhancing of human capital and the diffusion of new technologies. Since the degree of foreign penetration is likely to be higher in those sectors where the barriers to entry for new firms are high, the entry of foreign firms increases the degree of market competition and improves the allocative efficiency in the host country industrial structure. Also, another source of gain to the host economy could arise from the enhancing of human capital, which is due to the training of labour, and management that may later be employed by local firms. Moreover, foreign affiliates may speed up the cross-border transfer of technology and innovation causing them to disseminate faster than otherwise among domestic firms that competes with them. On the basis of such arguments, it has been postulated that: productivity levels are higher, ceteris paribus, in the domestic sectors where multinational firms account for larger shares. This proposition has been tested by some statistical analyses for different countries: Caves (1974) for Australia; Globerman (1979) for Canada; Blomstrom and Pearson (1983) and Kokko (1994) for Mexico; Haddad and Harrison, (1991,1993) for Morocco; Aitken and Harrison (1991) for Venezuela; Imbriani and Reganati (1997) for Italy, Kokko, Tansini and Zejan (1996) for Uruguay and Aslanoglu (1998) for Turkey.
Using cross-section data for three or four digit classification, Caves (1974), Globerman (1979) Blomstrom and Pearson (1983), and Imbriani and Reganati (1997) found that the foreign presence had a significant impact on the labour productivity of domestic firms and, therefore, spillovers were found significant. On the contrary, using panel data Haddad and Harrison (1993) concluded that although domestic firms exhibited higher levels of productivity in 9 sectors with a larger foreign presence, there was no significant relationship between larger foreign presence and higher productivity growth in domestic firms. Aslanoglu (1998) reached a similar result for the Turkish manufacturing sector. In particular, he found that while the presence of foreign firms increases competition in domestic industries, there was no significant contribution on the productivity of domestic firms.
An issue that has aroused some controversy in the theoretical literature concerned the relationship between spillovers and the size of the technology gap. In fact, we can distinguish at least two different positions. Some scholars (Koizumi and Kopecky, 1977;
Findlay, 1978; Wang, 1988, Wang and Blomstrom, 1992) have pointed out that spillovers grow with the size of the technology gap between domestic and foreign firms since the country's technical efficiency is an increasing function of the country's capital stock owned by foreign residents which are supposed to possess superior technical knowledge.
The second position concerns the idea that according to the "technological accumulation" literature (Cantwell, 1989) spillovers are more important in the industries where the technology gap is small. If foreign affiliates invest in a host country which represents itself a centre for innovation in the industry concerned, they are likely to have a positive impact on the host country economy because they contribute to strengthen and diversify local research and to stimulate the innovation of local competitors. Over time, FDI might set in motion a virtuous circle of increasing research intensive activity and a faster output growth.
The link between spillovers and the size of the technology gap has been tested for the case of the Mexican economy by Kokko (1994), for the Uruguayan economy by Kokko, Tansini and Zejan (1996) and for the Italian manufacturing sector by Imbriani and Reganati (1997).
Kokko (1994) pointed out the “enclave” characteristics of sectors in determining productivity spillovers. “Enclave” characteristics refer to industries where large technology gap and high foreign shares coincide. He found that in industries with enclave characteristics foreign firms take over and force local firms into narrow niches where the products and technologies of MNEs are not profitable. Accordingly, there is little scope for positive spillovers on 10 domestic industries. Imbriani and Reganati (1997) found that productivity levels are higher the lower the sizes of the technology gap between domestic and foreign firms. As a result, they concluded that if MNEs have chosen the Italian location because it represents itself a centre for innovation in the industry at a global level and, therefore, the presence of foreign firms is justified by the possibility to find an environment capable of increasing their technological advantage. Finally, Kokko, Tansini and Zejan (1996) found that a positive and statistically significant spillover effect only in plants with a moderate technology gap vis-à-vis foreign firms.
However, it is worth noting that all these studies have examined the effects of FDI on domestic firms by interpreting the manufacturing sector at a national level. In other terms, no tests have been performed on the existence of regional differences in productivity spillovers. In the Italian case, this seems to be a central point if we consider the particular productive structure of the country which shows a strong dualism between the North and the South areas as well as strong differences even within the two areas. Cantwell and Iammarino (1998) have found that the location of technological activities of foreign firms tends to be strongly agglomerated at a sub-national level. Lombardia and Piemonte can be considered higher order locations at the top of the scale, strongly attracting a broad range of foreign-owned technological activities due to their regional systems of innovation. Therefore, in this work we intend to use some unit of analysis that allows for investigation below the traditional country level.
3. Data, definitions of variables and statistical model
The empirical analyses are based on industrial data supplied by the Italian Central Institute of Statistics (I.S.T.A.T.) which we gratefully acknowledge. In particular, firstly we have selected a sample of 942 foreign firms, which undertake international production in the Italian manufacturing sector. In this work, we define domestically owned firms, as firms where the share of domestic ownership is above 50 per cent. Then we have collected information on the following variables: total number of employees, value added, gross output, investments, number of both manual and non-manual workers and concentration ratios. As a result, we obtained for the year 1992 a set of information covering 93 industries at a three-digit level of the manufacturing sector. However, when our analysis moved to the territorial distribution of foreign firms, we had to increase the degree of industrial aggregation, because the ISTAT by law cannot provide information on sectors where less than three firms operate. Therefore, in this paper we are able to analyse data for 35 industries.
Following other empirical studies conducted at a sector level (Caves, 1974; Globerman, 1979; Blomstrom and Pearson, 1983;
Kokko, 1994; Aslanoglu, 1998), we have analysed the presence of spillovers through a statistical model based on linear estimations of the labour productivity of domestic firms as a function of the degree of foreign penetration. If we find a positive significant relation between the domestic labour productivity and the foreign affiliates' market share, it follows that the foreign investment does raise the productivity in domestically-owned firms through spillovers.
However, to test extensively the spillover hypothesis, it is necessary to take into consideration other factors that can explain labour productivity, i.e. capital intensity, labour quality and concentration levels. Therefore, we hypothesise that the labour productivity of domestic firms can be estimated by the following function:
Val d = a1 + a2FP + a3KLd + a4LQd + a5CONC + e
In this equation the dependent variable (VALd) represents the labour productivity of domestic firms. This variable has been calculated by dividing the value added to the total number of employees in domestically owned firms. The degree of foreign penetration in each industry (FP) has been measured by the ratio of the foreign firms' employment to total employment. If spillover takes place, it is expected to have a significant positive effect on local labour productivity.