The two major current approaches to explain economic growth are the neoclassical growth model and models of endogenous growth. In Discussion Paper No. 1075, Research Fellow Xavier Sala-i-Martin adds a new study to an academic discourse that has remained on the agenda for more than a decade. After defending the usefulness of the theoretical concepts of b - and a -convergence and explaining their interrelatedness, the paper extends the empirical evidence on regional growth and convergence across the US, Japan, and five European countries. It reconfirms that the remarkable Barro/Sala-i-Martin 2 per cent-rule (of b -convergence) is correct and applies to all examined countries. It is also shown that the inter-regional distribution of income in all countries has shrunk over time (a -convergence).

Statistical artefacts, regional price dispersion, migration, perfect capital mobility, government cohesion policies, and Quah's unit root hypothesis are all dismissed as unlikely to account for these findings. One-sector models of endogenous growth are also ruled out as plausible explanations of b - and a -convergence. The two approaches mentioned at the outset remain as plausible explanations of the phenomenon: the neoclassical model, modified so that capital includes human capital elements, and models of endogenous growth with technological diffusion. To take account of the relatively slow rate of convergence, a notion of gradual technology adaptation is introduced to the latter, theoretically justifiable through imitation and implementation costs. The study cannot assess which of the two approaches and their underlying hypotheses is more adequate, and suggests further research.

Regional Cohesion: Evidence and Theories of Regional Growth and Convergence
Xavier Sala-i-Martin

Discussion Paper No. 1075, November 1994 (IM)