Poverty in Mozambique : new evidence from recent household surveys

This paper has three primary objectives: (i) to investigate potential problems regarding Mozambique's most recent nationally representative household survey on poverty dynamics; (ii) to assess the robustness and reliability of official poverty statistics; and (iii) to provide alternative estimates of poverty and welfare indicators in light of the methodological and analytical issues raised in areas (i) and (ii). It is determined that at least two significant weaknesses affect the official poverty-rate estimates: measurement errors in consumption data and flaws in the methodology used to calculate poverty lines (the cost-of-basic-needs approach based on provincial food bundles with entropy correction). A number of observations appear to be affected by substantial measurement errors, which severely distort the official poverty statistics. The paper provides methods to correct the consumption distribution by recalculating poverty lines based on a single national food basket -- as opposed to the current estimates, which are based on province-specific food baskets. The revised poverty statistics differ considerably from the official estimates of poverty across provinces and are far more consistent with other poverty indicators. In addition, poverty appears to be highly concentrated in certain areas, with dramatically higher rates found in Central and Northern Mozambique, as well as in rural areas overall, compared with relatively low rates in Southern Mozambique and in the country's urban centers. These findings substantially contradict the government's official poverty figures, which appear to systematically overestimate poverty rates in Mozambique's Southern provinces and urban areas while simultaneously underestimating the prevalence of poverty in the country's Central and Northern regions and in rural areas nationwide.