Inequality measures in Latin America

I have arrived to the Kellogg Institute (a fantastic place to work with really helpful people around) for a year on a project on the interactions between political and economic inequality in Latin America.  One of the first problems I am dealing with?  Indicators!  In the case of political inequality, the problem is that there is little agreement on how to measure it and a lack of relevant data.  The case of income inequality is more straight forward, well-known but also frustrating.

We often say that "Latin America is the most unequal region in the world", forgetting the diversity of distributional outcomes within the region.  But which countries are doing best and worst on income distribution?  The answer is that it depends on who is measuring it.  There are two different cross-country databases on income distribution: one from the Economic Commission of Latin America and the Caribbean and one from the World Bank together with the Centre for Distributive, Labor and Social Studies in Argentina.    Both use the same sources (country-level household surveys), but make different adjustments.  The result?  The magnitude of inequality and the order of countries varies (sometimes a lot) depending on which of the two we use.

This is evident when considering the so-called Palma Index (which compares the income of the top 10% with that of the bottom 40%) in both cases:

Note two things: the Palma index as measured with ECLAC data is larger in several countries like Honduras, Guatemala but also Chile and Peru.  Also, the comparative levels of inequality change: Colombia is the most unequal country when using SCEDLAS data but not there are other countries with more inequality when using ECLAC data.

There are several reasons to explain these differences but I want to concentrate on the implications here.  First, much of our econometric results may be driven by data issues... that are seldom fully studied.  In our graph, Honduras is a total outlier that may be eliminated from some regressions in one case or one of many in another.  Second, regional studies that treat the data carefully and are based on descriptive statistics may be more valuable than commonly recognised (more on this at a later stage). Third, at the end, we may need to always work with stylised facts when discussing inequality and triangulate as much as we can.  There is little doubt that Colombia and Honduras are very unequal and that Uruguay is probably the least unequal country in the region... yet what happens in the middle is less clear and needs careful consideration and a lot of triangulation. 

More on all these topics in upcoming weeks; I hope to use the sabbatical to write about inequality in this blog more often.  Feedback most welcome (as it will feed directly in the new project).