The Gini Needs a Deviation Measure

No person interested in development economics has not heard of the immensely useful Gini Index and its broad use in telling about the degree of income inequality in a country. I find in readings that several assumptions are made about what it implies in addition to its utility in policy. In the Slate Magazine here, Mark Gimein highlights some of the methodological limitations by comparing the respective Gini measures for the US and India and reaches the conclusion that on its own, the Gini coefficient is not very useful in telling much more than the aggregate. The implication is that many pundits use the Gini Coefficient in a way that leads to the suspicion that it is being incorrectly substituted for poverty figures. Conclusions and proposals for policy leading from these comparisons are therefore understandably error-prone.

I accept that as a broad measure of inequality, the Gini is a fantastic innovation but it is demonstrably incomplete. To my mind, it is comparable to the statistical mean which is not too useful without mention of the standard deviation since the mean on its own tells nothing about the dispersion of the data from which it is derived. Perhaps what needs to be developed for use alongside Gini figures is the equivalent of the standard deviation so that one can make reasonable inferences about multiple countries with similar Gini scores but whose average incomes vary substantially. It cannot be correct to state that the level of inequality is the same in two countries on account of equivalent Gini scores when the respective per capita incomes are US$ 1000 and US$ 20,000.