When the remedy is worse than the disease: Adjusting survey income data for price differentials, with special reference to Mozambique

In using survey data for money metric analysis of poverty and well-being, it is customary to adjust either the data or the poverty line for spatial prices differentials where data exist to make such adjustment. In developing countries where recorded price differentials between regions or provinces are large, using the remedy of adjusting for price differentials may sometimes lead to very wrong conclusions about the spatial distribution of poverty. This may have severe consequences for policy and may be detrimental to the poor. The paper deals with a specific situation, that of Mozambique, where large price differentials are said to exist between the capital (Maputo City) on the one hand, and the rest of the country. Official data that adjust for this may heavily over-estimate poverty in Maputo City, with consequences for the targeting of poverty. We use an asset index based on Multiple Correspondence Analysis (MCA) to show that the spatial poverty profile derived from the price-adjusted income data exaggerates poverty in Maputo City, and undertake further empirical analysis to show that not adjusting for the estimated spatial price differentials may have given more reliable estimates of well-being, judging by asset holdings.