In the context of cross-country patterns of growth, ‘convergence’ refers to currently poorer economies having the opportunity to catch up with those currently richer. The questions concern the behaviour over time of cross-section distributions of income (or output or welfare). Are the distributions collapsing, so that everyone shows a tendency to become equally well off? Or are the distributions instead increasing in dispersion, so that those relatively better off are becoming more so? Or, are the distributions tending towards clusters and subgroups, with the population is polarizing and then stratifying into distinct classes? Such questions are useful for appreciating patterns of cross-country growth, just as they are for understanding patterns of dynamics and mobility in individual incomes within societies.

In Discussion Paper No. 1355, Research Fellow Danny Quah provides a critical review of both traditional and more recent work in this area. The newer, more macro-oriented research differs from the earlier work on the dynamics of personal income distributions. The differences are manifest in two significant ways. Empirically in the range of data available and theoretically in the types of hypotheses used to explain particular patterns of inequality across economies. Despite the differences, it is apparent that the macro growth research and the micro social welfare income distribution work should each have insights worth sharing. Traditional analyses of growth and convergence focus on cross-section ‘convergence’ regressions, and hypothesize in terms of production function accounting. Such work rarely aspires to understand the dynamics of the entire distribution – and when it did, Quah argues, it typically got things wrong. Newer analyses draw their empirical basis from explicit models of distribution dynamics, and on the incentive structure underlying group and coalition formation in economic growth.

The difference across traditional and newer analyses is not simply cosmetic. As an example, compare the following. Key findings from traditional analyses are that convergence occurs at the rate of 2% per year, and that this implies physical capital’s factor share is larger than that reported in national income accounts. By contrast, key findings in the newer analyses include that the cross-section distribution of incomes across countries is polarizing into rich and poor, with the middle-income group eventually vanishing. Both findings derive from the same data. Both are technically correct. The two pictures of world income dynamics depicted differ dramatically, however. Furthermore, the different empirical findings motivate entirely different lines of theoretical research.

The traditional line of analysis takes itself to be a test of a particular theoretical growth model. The newer, distribution-dynamics approach also provides a test of the same growth model predictions, and simultaneously seeks to quantify dynamic patterns of growth and inequality across countries. The traditional line of research focuses on further refining production-function estimation and accounting. The newer approach seeks to understand the economics behind cross-country interaction and coalition formation, and provides insight into, among other things, how new technologies and patterns of production and communication imply alternative configurations of group dynamics. Lastly, the paper points to where subsequent empirical research is needed, and suggests alternative theoretical ideas to explore.

Twin Peaks: Growth and Convergence in Models of Distribution Dynamics
Danny T Quah