The authors are members of the faculty of the Department of economics and the Centre for Asian Development Studies at Boston University, USA. An earlier, highly abbreviated, version of the same research was published in the Journal of Development economics, 23 (1986) , pp.55-65. Even earlier, PIDE published Development Strategy, Growth, Equity and the Political Process in Southern Asia (by G. F. Papanek) which included one Lecture on the same subject. These earlier versions used somewhat different methodologies, reported preliminary results and are far less complete. N.V. Jagannathan, I. A. S., did an excellent job of collecting the data on which this essay rests. It provides the framework for a more detailed study of five Asian countries supported by the Agency for International Development (AID) under grant AID/OTR-G-1 872. We are grateful for this support, and the research assistance of N. V. Jagannathan, but neither can be held responsible for our analysis or conclusions.
"Down" when the curve is U-shaped, because income shares are measured on the vertical axis. It is an inverted U if the Gini coefficient is on the vertical axis.
For an excellent discussion of the conflict between equity and efficiency or growth and the consequent relationship of efficiency and political repression, see Sheehan 1241.
In the 1963 - 1971 period the Gini coefficient is a reasonable .36 to .38 but the share of the poorest 40 percent in the same years ranges from 26 percent to 29A percent. For neighbouring and similar Bangladesh and India, with Ginis of .34 to A8, the share ranges are from 13.1 percent to 20.2 percent. At Pakistan's per capita income, the Bangladesh shares are rather typical.
The results including Pakistan are available from the authors.
With some loss of degrees of freedom one can construct a model to allow for systematic change in the differences between rural and urban income inequalities as per capita income in-creases. This is left for further work.
Note that most previous studies have been limited to data for countries as a whole and have simply ignored differences in definition. Taking account of such differences should increase the accuracy of the results.
Except for countries with extremely low levels of income.
See, however, Cline  who postulates the higher wage in the expanding modern sector and therefore concludes that the outcome is indeterminate.
Ahluwalia, in concludes that time-series data show no systematic deterioration in the share of the poorest 40 percent
Income distribution inevitably is highly egalitarian at per capita incomes below $100. “In countries with an average per capita income of $100, if the poorest 20 percent had less than 40 percent of total income, quite typical of countries with higher incomes, the average income of the poor would be less than $20 a year, probably not enough to live on, even if nearly all income goes for carbohydrates” egalitarian as $100 is approached.
Support comes from Bacha’s results He has to exclude both Pakistan and Sri Lanka from the analysis to obtain any significant relationship between per capita income and income distribution.
Note that regional differences are much less significant statistically when the dependent variable is the share of the poorest 40 percent (Table 3) than when it is the Gini coefficient (Table 2). These results suggest that the share of the poorest 40 percent is quite similar in different regions except in Sub-Saharan Africa.
In South and Central America the Gini coefficient is higher (less egalitarian than other regions), but for the share of the poorest 40 percent the region actually shows greater equality. A plausible explanation is that the share of the middle class is less, and of the rich greater, than elsewhere in the world. This conclusion runs counter to conventional wisdom. But when it is said that the middle class is more significant in middle-income countries, such as Latin America, reference is usually to the urban middle class. The usual middle peasantry is probably more important in Asia and in the developed countries than in Latin America. The weakness of the rural middle class in Latin America could be reflected in the results for the Gini coefficient.