Per capita income disparities, particularly evident between developed and l
ess developed countries, are of noticeable importance even amongst the regi
ons of the European Union and do not seem to reduce their extent following
the predictions of the neoclassical model, despite the high degree of openn
ess typical of regional economies. The empirical analysis of convergence am
ong regions confronts two problems that, if not correctly addressed, can re
duce the validity of the results. The first concerns the definition of regi
on. Very often, in fact, empirical analyses within a regional context make
use of the available data for administrative regions without adequately con
sidering the consequent results of this choice. The second problem is relat
ed to the nature of the growth process. The methodology followed by the vas
t majority of researchers, i.e., cross-sectional regression, is based on th
e hypothesis of a growth process characterised by a smooth path towards a s
teady-state. As some authors have demonstrated empirically that this underl
ying hypothesis is invalid, it is interesting to develop an alternative met
hodology the results of which are not subject to this restrictive hypothesi
s on the growth process. The methodology suggested in this study focuses di
rectly on the cross-sectional distribution of per capita income, modelling
the growth process as a time homogeneous Markov chain. The methodology is t
hen applied to per capita income data for 122 EU functional regions over th
e period 1979-1990. (C) 1999 Elsevier Science B.V. All rights reserved.