Panel evidence on the speed of convergence
Abstract: Estimates of the speed of convergence vary widely and depend on the methodology employed. While cross-sectional regressions typically find slow convergence, time series estimates suggest that incomes converge rapidly. This paper uses panel methods to combine cross-sectional and time series information. Our methods improve on prior research in three ways. Relative to univariate time series estimators, panel methods increase the precision of convergence estimates. Relative to earlier panel studies, this paper tests the key homogeneity assumption needed to bring cross-sectional information to bear on convergence estimates, and it uses panel versions of the methods of Andrews (1993) and Stock (1991) to adjust for small sample bias and construct confidence bands. Our estimates are closer to the results of cross-sectional studies and suggest that convergence occurs slowly.
Provider: Federal Reserve Bank of San Francisco
Part of Series: Working Papers in Applied Economic Theory
Publication Date: 1997