Why Should We Analyse Convergence Through the Distribution Dynamics Approach?

The convergence hypothesis has stimulated heated debate within the growth literature. The present paper compares the two most commonly adopted empirical approaches – the regression approach and the distribution dynamics approach – and argues that the former fails to uncover important features of the dynamics that might characterise the convergence process. In particular, the empirical section highlights the interpretational advantages stemming from the use of stochastic kernels to capture the evolution of the entire cross-sectional income distribution. Incidentally, comparison between the results obtained from alternative sets of Italian regions suggests that the use of administrative regions may lead to ambiguous results.