The Dynamic Evolution of the Proximity Effect in the Textile Industry

Abstract

This paper tests the evolution of the relationship between localization and proximity and their effect on firm performance for a sample of 10,490 Spanish textile firms in the period 2001–2006. Using coefficients of specialization two levels of geographical disaggregation are distinguished: (i) the provincial cluster and (ii) the industrial district. The results obtained show a positive and significant impact of the specialization level on productivity, both for industrial districts and provincial clusters (in this case only when the levels of specialization are high enough). In general, the effect is higher for industrial districts than for provincial clusters, although in both cases it decreases with time. Given that in most EU countries the textile industry has a tendency to be geographically concentrated in industrial districts, these results carry important implications for the textile industry.

Acknowledgements

The authors thank the editor and reviewers for valuable comments on the revision of this paper. They also thank A. Madhok and F. X. Molina-Morales for their helpful comments on earlier drafts. This research was partially funded by the Research Project UV-AE-10-24464 of the University of Valencia (Spain). An earlier version of this paper was presented at the 9th EURAM Conference, Liverpool.

Notes

Textile trade among WTO (World Trade Organization) members is regulated by the Agreement on Textiles and Clothing, which replaced the Multifiber Agreement when the WTO replaced the GATT in 1995. This agreement implemented the progressive reduction of import quotas imposed by the EU and North America to a number of low-wage countries. The schedule is as follows: 16% of products in 1995, 17% in 1998 (mostly clothing articles), 18% in 2002 (again clothing articles) and the remaining 49% in 2005 (affecting more evenly different types of textiles). Due to the spectacular growth of imports from Asia in the beginning of 2005, the EU was lobbied to activate a safeguard clause and the US followed suit, so that in practice some quotas were reactivated and there is not as yet a complete liberalization in the sector.

In this paper, however, we focus on the location of firms and their level of agglomeration. In establishing empirically the distinction between cluster and industrial district, we use both a finer geographical disaggregation and the contiguity criterion. Other authors such as Markusen (1996) and Paniccia (2007) have also used dimensions such as number and size of firms and the relationships between them.

The CE is a statistic that measures the presence of an activity (in this case the textile-apparel sector) in a territory with respect to the presence of that sector in the whole reference sample. It is defined as: , where E _ij is employment in sector i in location j, E _j is total employment in location j, E _i is total employment in sector i, E _n is total employment in the country (in this case Spain). A CE above 1 for a particular area indicates that area is more specialized than the sample average (in this case the Spanish average).

The SABI database contains information on over 550,000 Spanish firms and 65,000 Portuguese firms of various sizes (small to large), while others such as the Cabsa database contains data for the 5000 largest Spanish firms. The use of SABI brings additional advantages, such as the level of disaggregation at both the geographical and sectoral levels. Finally, the SABI database has been widely used on research about Spanish firms (Escribá-Esteve et al., 2009; Puig et al., 2009; Tomás-Miquel et al., 2009).

As census data are taken every 10 years, 2001 is the latest year for which employment data are available at the required level of disaggregation.

This area is located in two different provinces of the Comunidad Valenciana region (Alicante and Valencia) with CEs of 2.14 and 1.64, respectively. The Alcoi-Ontinyent industrial district had previously been identified in the literature (Ybarra, 1999; Cluster Competitividad, 1999; Puig and Marques, 2010; Tomás-Miquel et al., 2009). In 2008, this district employed over 15,000 people (8.5% of total employment in the Spanish textile-apparel sector and 9.5% of the sample firms).

Productivity is logged as customary in the literature in order to preserve statistical properties (log-normality) which are necessary for regression analysis.

Experience is measured as the number of years the firm has been in activity.

As in Puig et al. (2009), we distinguish the following five NACE subsectors: yarn (codes 17.1 and 17.2); (2) finished products (17.3); (3) home-technical (17.4 and 17.5); (4) knitted textiles (17.6 and 17.7) and (5) clothing (18.2).