23
The Econometrics of Convergence
Steven N. Durlauf, Paul A. Johnson and Jonathan R.W. Temple
Abstract
The presence or absence of convergence between rich and poor countries represents one of the
most important questions in the new growth economics. New growth theories have been explicitly
designed to explain forms of divergence which do not appear in their neoclassical counterparts.
Despite substantial empirical work on convergence, there is no consensus as to whether it is present
in cross-country data. This chapter surveys the econometrics of convergence as well as the range
of empirical claims that have appeared. Particular attention is given to the relationship between
statistical versus economic notions of convergence. We argue that the disparities in claims across
empirical studies can to some extent be understood as reflecting inadequate attention to the
relationship between the statistical and economic notions.
23.1 Introduction 1087
23.2 β-convergence 1089
23.2.1 Convergence and the neoclassical growth model 1089
23.2.2 Cross-country regressions andβ-convergence 1091
23.2.3 Critiques 1092
23.3 Nonlinearities and multiple growth regimes 1096
23.4 σ-convergence 1098
23.5 Convergence and the cross-country distribution of per capita income 1099
23.5.1 Structural analysis 1103
23.6 Time series approaches to convergence 1103
23.6.1 Transitions versus steady-state dynamics 1105
23.7 The economics of convergence 1106
23.8 Conclusions 1109
23.1 Introduction
The question of whether nations converge or diverge has long been of interest to
historians and social scientists. One classic passage of Edward Gibbon’sThe Decline
and Fall of the Roman Empire^1 expresses the hope that all societies will exhibit
progress in the long run:
If, in the neighbourhood of the commercial and literary town of Glasgow, a race
of cannibals has really existed, we may contemplate, in the period of Scottish1087