Intraregional income convergence: Cross section and time series evidence from the USA

Abstract

The publication of the ground breaking work of Baumol (1986) was the spark that ignited an enormous interest to the issue of convergence in per capita income (e.g. Aghion and Howitt 1998; Baldwin et al. 2003; Capello 2006; Le Gallo 2004; Overman and Puga 2002; Ioannides and Overman 2004; Li and Haynes 2010). As perhaps anticipated, there is a growing number of attempts to assess regional convergence using extensive datasets, such as the regions of the European Union (e.g. Button and Pentecost 1995; Cuadrado-Roura et al. 1999; Rodríguez- Pose 2001; Rodríguez-Pose and Fratesi 2004; Lopez-Bazo et al. 2004; Alexiadis and Tsagdis 2010), the US states (e.g. Christopoulos and Tsionas 2007; Checherita 2009) and the regions of individual countries (e.g. Rodríguez-López et al. 2009; Hierro and Maza 2010). Most of the literature concerning convergence has been developed in terms of per-capita income using cross-section data. Nevertheless, convergence is by no means a mechanical phenomenon, which happens everywhere and always (Cuadrado-Roura 1996, p. 47). Regional convergence is characterised by rapid transformations and adjustments, properties that are difficult to be examined in a cross-section context. This has led to the development of alternative methodologies based on cointegration analysis generating a considerable amount of empirical literature (e.g. Bernard and Jones 1996; Carlino and Mills 1993; Sun et al. 2010).1 Still, the crucial question of the adjustment towards steady-state equilibrium, which lies at the heart of the convergence debate, remains unanswered. An approach to this issue can be provided through an Error-Correction-Model (hereafter ECM). Recent years have witnessed a growing number of attempts to implement this model to examine the evolution of regional employment and unemployment (e.g. Baddeley et al. 1998, 2000; Martin and Tyler 2000; Gray 2004; Hunt 2006; Alexiadis and Eleftheriou 2010).