Chapter 10 offers a useful entry into the recent literature on growth and convergence. Although selective, the chapter provides a rich perspective on the issues at hand.

It starts with the standard neoclassical model whose prediction regarding growth and convergence lies at the heart of the ‘conditional convergence’ analysis of Barro (1991) and Sala-i-Martin (1995) and to which Mankiw, Romer and Weil (1992) have given a compact formulation. The critical assumption behind the empirical tests of conditional convergence offered by these authors is that countries are viewed as realisations of independent experiments of independent countries. The weakness is that countries do interact. If they were so well integrated as to be subject to the same intertemporal prices then one might equalise instantaneously the capital–labour ratio of all economies at a level that would perpetuate indefinitely any initial diversity of per capita income. Per capita output, on the other hand, would be instantaneously equalised, and the empirical implication (that we would want to see, but which is likely to be falsified) would be that the pattern of convergence of income per capita is slower than the patterns of convergence of per capita output (and all the more so when the economies are financially integrated). The irony, of course, is indeed that integration is bad for convergence of welfare.