Convergence Clubs and Spatial Externalities

A major concern for regional economists is whether regional per-capita incomes tend to converge or diverge over the long-run, and whether such trends apply to all or only limited groups of economies. This latter possibility is known as ‘club convergence’ and provides a realistic and detailed picture about regional growth (Fischer and Stirböck 2006). The notion of club convergence was originally introduced by Baumol (1986) in recognition of convergence within a subset of national economies. As Baumol and Wolff (1988, p. 1159) subsequently noted, however, “just how countries achieve membership in the convergence club, and on what basis they are sometimes ejected” is a difficult question to answer.

The study of regional growth has been dominated by two broad and contrasting theoretical approaches regarding regional convergence. According to the first, market forces will lead to a general convergence of per-capita incomes across an integrated space economy over time. This approach is labelled as ‘neoclassical regional growth theory’ and its premises are based upon the standard growth model, as outlined by the pioneering work of Solow (1956) and Swan (1956). Using a general equilibrium framework these models predict that disparities in per-capita incomes across regions are unlikely to occur or, at least, to be persistent, thus creating a pattern of convergence towards a unique level of per-capita income. By contrast, there is a large body of theoretical and empirical work, known as the ‘post-Keynesian approach’, which supports the argument that regional disparities in per-capita incomes are permanent and self-perpetuating and therefore divergence in per-capita incomes is the most likely outcome. Representative models can be found in the work of Myrdal (1957), Perroux (1950, 1955) and Kaldor (1967, 1970 and 1972). This chapter outlines the major approaches to regional growth, as put forward by the neoclassical and post-Keynesian schools of thought. Throughout this and subsequent chapters more emphasis is placed upon the neoclassical model, for two reasons. First, the neoclassical model offers both a theoretical explanation and testable predictions concerning the possibility of convergence in per-capita incomes across regions. Indeed, most of the conceptual definitions of regional convergence used in empirical studies derive directly from the neoclassical model. Second, the vast majority of empirical literature has in fact tested the neoclassical model rather than alternative models.

The 1980s and 1990s have seen the earlier neoclassical and Post-Keynesian models augmented by a new generation of growth theories, notably Endogenous Growth Theory in which technical progress develops within the economic system. Hammond and Rodriguez-Clare (1995) summarise the contribution of the endogenous growth models as follows:

Although the concept of ‘club convergence’ emerged from empirical evidence, its theoretical underpinnings can be found in neoclassical and endogenous growth models, outlined in Chaps. 2 and 3. Indeed, a prediction of several Endogenous Growth models, such as the ‘Technological Diffusion-Gap’ model, is that economies do not converge towards a common equilibrium. ‘New Economic Geography’ implies a polarisation of regions into different ‘clusters’, poor or ‘peripheral’ regions and rich or ‘central-core’ regions, with growing disparities and divergence among clusters. It is the purpose of this chapter to examine the theoretical framework for club convergence. Firstly, the notion of ‘club convergence’, as this has emerged from empirical studies, is introduced in Sect. 4.2. Section 4.3 outlines two theoretical approaches to multiple equilibria and club convergence proposed by Galor (1996) and Azariadis and Drazen (1990), which are, essentially, a reformulation of the neoclassical model. Section 4.4 describes the club convergence pattern within the framework of Endogenous Growth Theory, in which club convergence is attributed to the diffusion of technological innovations from leading economies. This process, however, appears to be exogenous and very little is said about how is determined. Diffusion of technology is not a simple and automatic process. Instead, it requires that lagging economies (countries or regions) should have the appropriate infrastructure or conditions to adopt or absorb the technological innovations. A simple model is developed in Sect. 4.5 in which club convergence is attributed to differences in the absorptive abilities of regions. Finally, Sect. 4.6 provides some conclusions.

The theoretical analysis of convergence presented thus far has examined the circumstances in which an economy converges towards an equilibrium level of output per-worker or a steady-state rate of growth. The possibility that groups of economies are likely to converge towards the same steady-state (absolute convergence) or towards different steady-states (conditional convergence) has also been examined. In discussing club convergence, the possibility of convergence towards a leading economy has also been addressed. In reality economies are not in equilibrium and are subject to all manner of shocks at different points in time. Therefore, an important question arises: ‘how is possible to test for convergence, when the steady-state is never achieved?’ The approach, in practice, is to direct empirical measures of convergence towards the process of convergence, rather than the equilibrium outcome.

Regional growth may be convergent or divergent, as discussed in Chaps. 2 and 3. Convergence may also be an exclusive property of a specific set of regional economies, which are likely to share similar characteristics. It is the purpose of this chapter to provide an assessment of whether or not absolute convergence is apparent across the regions of the EU-27, and whether this applies only to a selected club.

The previous chapter has examined club convergence in the context of the EU-27 regions, thus providing an alternative perspective on the issue of regional convergence in an enlarged Europe. While previous studies on European regions claim that convergence is slow, the empirical tests reported on Chap. 6 establish that convergence is a property that characterises the regions of the ‘old’ member-states of the European Union together with a selected set of regions located in new member-states.

This study is placed within a wide literature that is concerned with whether levels of per-capita income or labour productivity across economies converge or diverge in the long-run and using the NUTS-2 regions of EU-27 as an empirical context. It has addressed the issue of regional convergence using a data set that covers the period 1995–2006. An attempt is made to provide a detailed and thorough view of the issue of convergence across the regions of the EU-27 by considering various empirical approaches.