1252 Spatial Analysis of Economic Convergence
27.1 Introduction
The last two decades have witnessed a renaissance in the field of growth econo-
metrics. Defined as the set of statistical tools for the study of growth (Durlaufet al.,
2005), growth econometrics can be organized along two dimensions. In the first,
which we refer to as the regression approach, predictions from formal neoclassi-
cal (and other) growth theories have been tested using cross-sectional, time series
and panel data econometric specifications. The second group of methods departs
from the representative economy assumption underlying most of the regression
approaches and instead examines the entire distribution of incomes. Here the focus
shifts to how different parts of the distribution may behave over time, and to ques-
tions of whether there are changes in modality and shape of the distribution and
concerns with the intradistributional dynamics and mixing. These methods are
distinct not only in their focus, but also in the specific statistical techniques that
are employed.
A prominent trend in the growth literature has been a decidedly regional turn,
where the focus has shifted from cross-country analyses to examine the nature
of convergence as it may operate at the sub-national scale (cf. Barro and Sala-i-
Martin, 1992; Carlino and Mills, 1993; Neven and Gouyette, 1995; Sala-i-Martin,
1996; Rey and Montouri, 1999; Fingleton, 2004). Early in this regional turn, there
was some appreciation for the challenges that regional data posed for the appli-
cation of standard growth models. Given that regions typically display a greater
deal of openness than is the case for national economies, various forms of regional
interdependencies, such as labor and capital flows along with trade, take on
increased importance at the finer spatial scale. Yet there was also a perception that
there was a lack of econometric methods for modeling regional interdependen-
cies. When confronted with evidence of spatial dependence and heterogeneity in
regional growth sets, the strategy adopted by many researchers has been to remain
in the closed-economy model but to adjust the closed model for these spatial
effects.
Paradoxically, during this period of renewal and resurgence of interest in growth
econometrics, there was a similar burst of activity in the fields of spatial econo-
metrics and spatial statistics (for example, Anselin, 1988; Cressie, 1993; Anselin
et al., 2004). Indeed, as Arbia (2006, p. 3) has noted: “until a few years ago, spatial
econometric methods were well developed in the literature but the drama was that
no one used them in the mainstream applied economic analysis!”
Towards the close of the century this began to change, with a number of stud-
ies adopting methods and tools of spatial analysis to the question of regional, or
sub-national, economic convergence beginning to appear in the literature. The
intersection of these two literatures has generated a large, and growing, body of
empirical studies, as well as the identification of a number of challenging method-
ological issues and some advances in the modeling and analysis of spatial growth
and convergence. The growing recognition of the unique characteristics of spatial
data is reflected by Durlaufet al. (2005), who note: “The problem of spatial corre-
lation has been much studied in the regional science literature, and statisticians in
this field have developed spatial analogues of many time series concepts ...”