more closely on the strand of literature originating from the work of Evans
and Karras (1996). However, we exploit some of the recent innovations in
the literature on panel unit root tests in order to incorporate two main im-
provements with respect to the their methodology. Firstly, we account for the
potential panel heterogeneity arising from the different economic structure of
each region. Secondly, we consider the possibility that each region might be
characterised by a different growth path. Finally, we incorporate the poten-
tial cross-sectional dependence due to common shocks hitting different sets
of regions at the same time.
Bearing all this in mind, the remainder of the paper is organised as follows.
The next section provides a further discussion on regional convergence in
Italy. Section 3 presents the steps of the econometric methodology of the
paper. Section 4 presents the empirical implementation and discusses the
results. The final section concludes.
2 Regional Convergence in Italy
The literature on the empirical estimation of convergence in Italy developed
after the work of Barro and Sala-i-Martin (BSiM, 1991). These authors
estimated absolute convergence at a rate of 2 percent for the period 1950-
1985. The stark contrast of this result with the dualistic nature of growth in
Italy subsequently led many researchers to question the robustness of their
analysis. Indeed, later studies have highlighted how this result depends on the
particular time period taken under consideration. There is now a widespread
agreement that during the 60s and the first part of the 70s the process of
convergence reached its apex, whilst the later decades are characterised by a
tendency for regional economies to diverge.2
However, these studies focus mostly on labour productivity and per capita
2See, among the others, Di Liberto (1994), Mauro and Podrecca (1994), Paci and
Pigliaru (1995), Cellini and Scorcu (1997), Paci and Saba (1998), and Margani and Ricciuti
(2001)