Mesurer les inégalités : que captent réellement les enquêtes ? Analyse de deux enquêtes ivoirienne et malgache

Apart from the fact that methodological choices influence the measurement of standard of living and its distribution, it is well known that several measurement errors bias also household survey data: errors of sampling, unit non- response, items non-response and selective observation. The aim of this work is to detect carefully all these bias in two household survey data (EPM93 Malagasy survey an ENV98 Ivorian survey) and to try to correct them. The analysis of the measurement errors has been done, firstly by checking the consistency between consumption and income information, and secondly by confronting household data with others sources of data like National Accounts and the Balance of Payments data. Several adjustments are proposed. It seems that standard of living are under estimated up to 15 % and 50 % and that inequality levels in both countries are nearer to level of countries like Brazil than it is said by standard evaluation.