| Abstract: | The traditional empirical approaches to the analysis of economic growth,cross‐section and panel data regressions are substantially uninformative withrespect to the issue of convergence. Whether national or regional economies appear to converge in terms of per capita income or productivity levels (the so‐called β‐convergence) critically depends on the way in which the empirical model is specified. Traditional specifications witness a disproportionate presence of proxies for forces leading towards divergence among the conditioning variables. It is therefore hardly surprising that these analyses find a positive and statistically significant value for the estimate of the speed of convergence. A more constructive use of cross‐section and panel data regressions is in the analysis of the determinants of growth. The present paper therefore builds on recent work on the role of different growth determinants (Cheshire and Carbonaro 1996) and analyses the growth performance of 122 Functional Urban Regions (FURs)over the period 1978–1994. This model explicitly recognizes growth as amultivariate process. In this new formulation it incorporates a spatialized adaptation of Romer's endogenous growth model (Romer 1990), developing the work of Magrini (Magrini 1997). Magrini's model originated from the view that technological knowledge has a very important tacit component that has been neglected in formal theories of endogenous growth. This tacit component, being the non‐written personal heritage of individuals or groups, is naturally concentrated in space. As a result, technological change is profoundly influenced by the interaction between firms and their local environments. The present paper reports the results of the estimation of a fully specified model of regional growth in per capita income. Particular attention is played to the role of research and development (R&D) activities, and to the influence of factors such as Universities that shape the local environments and have important policy implications. These results are then applied to quantifying the scope for policy to influence the growth process. Several simulations are presented deriving alternative growth outcomes across European regions that would have been obtained if those variables over which policy might have control—including the contribution of human capital—had had alternative values reflecting the realistic scope of policy makers' influence. The implications for convergence/divergence in regional per capita income levels are then analyzed using a Markov chain approach (Quah 1993 and 1996; Magrini 1999). |
