The impact of the tertiarization process in Spanish economic growth from a regional perspective

Abstract

The article focuses on the impact that the growth of the service sector has had on the economic growth of the Spanish regions. Thus, the study tries to define regional groupings in order to identify different modalities of regional growth. As a result of the cluster analysis, we propose the groupings and analyse the beta-convergence in per capita GDP and in terms of the degree of tertiarization in the Spanish regions. Taking these groups as our starting point, a data panel model with fixed effects is applied and the study concludes that mainly capital stock and capital productivity in market services explain to a great extent the growth in per capita GDP in the Spanish regions.

Keywords:

• services
• growth
• regions
• convergence

Notes

These studies are based on two questions: (i) the incorporation of new arguments in the aggregate functions of production with the objective of explaining the differences in the levels of growth between countries and regions and (ii) the importance of the factors linked to territory to understand the divergences in economic growth observed in each territory.

According to Léo and Philippe (2005, p. 143), the analysis of sector-related development is ‘essentially an empirical approach starting with the observation of important changes of weights among the different activity sectors’.

For this we use the information provided by the census. Nevertheless, and given that for the rest of the variables biannual information is used, we opted to use the data obtained from the tendency generated by a polynomic function (of the third or fourth order, depending on the regions). Given the fact that what we wish to do is to establish the characteristics of the different Spanish regions in terms of their economic dynamics and, subsequently, try to explain to what extent the structural change in their economies towards the service sector has determined their growth, in this study we have used a set of variables that could be grouped into two blocks: macroeconomic variables and variables related to the service sector. Regarding the former (expressed in per capita terms), they are related to GDP that plays a fundamental role both in the calculation of labour productivity and of disposable family income. Nevertheless, it is true that the correlation between all of these variables is high given the regional disparities in productivity (due to their different productive structure) and disposable income (as a consequence of the important redistributive efforts made by Spanish Public Administrations), their inclusion in the cluster analysis could introduce some enriching nuances without generating any distortions. With reference to variables in the service sector, they have been used in relative terms, i.e. in terms of the national economy, and their inclusion is aimed at grouping Spanish regions in terms of their similarities in the process of expansion in the service sector. This analysis was carried out both in terms of employment and production in services in general and in the catering and hotel industry in particular. The inclusion of this sector is due to the need to introduce into our analysis the differential effect that the development of the tourist industry has had on the tertiarization process in some Spanish regions, a factor that was particularly important during the first 20 years of the period analysed.

The series used are those given by the BBVA Foundation (Fundación BBVA) in Pérez, Goerlich and Más (1996), as this study contains homogenous data on the variables, on a biannual basis (only for those years ending in uneven numbers), for a relatively long period of time.

These variables are the following: GDP and disposable family income, all expressed in per capita terms; labour productivity; the relative importance of the GVA of the hotel and catering industry and, finally, the degree of tertiarization (in production and employment).

In Navarro and Hernández (2004), we can find an overview of recent studies on the process of income distribution in Spain.

Nevertheless, according to Mas, Maudos, Pérez, and Uriel (1994, p. 131) ‘This evidence is compatible with the presence of perturbances in determined sub-periods of the sample, which temporarily increase disparity or simply decelerate its reduction’.

Although the adjustment in regression is good, the convergence coefficient (i.e. the slope of the regression line) suggests that the levelling process of income is very slow. A value of 1.207 indicates that a ‘typical region’ eliminates each year only 1.207% of the income differential with the interregional average.

See, for example, Martín (1992, 1996), Mas, Maudos, Pérez, and Uriel (1994) and De la Fuente, Vives, and Esteban (1994).

Employment in service activities has been widely used to study the impact of the growth of services in several aspects of the regional economy. García-Milá and McGuire (1998) found evidence that suggested that the shift towards services in the US economy may had benefited regional employment growth, in particular, through the knowledge spillovers from services to manufacturing.

For a more detailed analysis of the process of regional growth that has taken place since 1955 see the classic study of Pérez et al. (1996), or the more recent work of Martínez Chacón (2002), Villaverde (2001), Alcaide (2001) and Goerlich and Más (2001).

In this classification, business services are included in the branch of other market services.

The calculation of labour productivity is particularly difficult for services in comparison to other activities. According to Wölfl (2003), the problems that arise when measuring productivity in services refer to the definition of output of some services, the calculation of value added at constant price (especially when this has taken place at the same time as changes in the quality or mix of services) and the underestimation of productivity growth in services that are intermediate inputs for other industries (such as finance, transport, etc.). Along with the difficulty of measuring productivity in non-market services (Scarpetta et al., 2000, p. 48), this leads us to consider only capital productivity in market services.