> A first indication of this is seen in Figure 1a which shows a kernel density estimate of the distribution of productive assets pooling all wealth classes. The

> A first indication of this is seen in Figure 1a which shows a kernel density estimate of the distribution of productive assets pooling all wealth classes. The distribution is bimodal, with a mass of households around 0.25 and 6.5, and hardly anyone in between. This is an important result in understanding how to get people out of poverty in undeveloped countries, but I suspect it won't translate well to industrialized countries like many in the comments are suggesting. The key result seems to be dependent on the bimodal wealth distribution. (Giving a cow to someone in the lower bucket was enough to push her into the next bucket.) Income distributions from industrialized countries look like the cleanest log normal distributions you've ever seen, so there just isn't evidence of a great repulser in higher income countries that you could push poor people over like they do in this study. Caveat: They imply that income and consumption are bimodal, but never plot them, only wealth. Wealth is notoriously hard to measure, and I can only find good histograms of income for the USA. Income and wealth should track each other, at least over a log scale, but I am not comparing exactly the same thing.