The Invisible Continent. Counting the Missing Poor in Developing Countries

Because the poor face higher mortality, they are under-represented in the samples used to measure poverty. This paper quantifies this missing poor bias across 92 developing and least developed countries by comparing observed headcount poverty rates with counterfactual rates that would prevail if all individuals enjoyed the survival conditions of the top income class. Using income distributions and life expectancy by wealth quintile, we find the bias to be substantial and highly heterogeneous, from negligible in parts of Central Asia to 15 percentage points in Bolivia and 36 in Bhutan. Aggregated, the correction reveals an invisible continent of more than 116 million uncounted poor, large enough to reverse international poverty rankings. Instrumenting health system coverage with malaria ecology, we then show that universal health coverage causally reduces both the bias and the steepness of the income-mortality gradient through which it arises. A ten point gain in coverage lowers the bias by about one percentage point, three times the OLS estimate. Progress toward universal health coverage therefore yields a second dividend, correcting a measurement distortion concentrated where deprivation is deepest.