Faces of poverty:: sensitivity and specificity of economic classifications in rural Vietnam

Aims: Poverty concepts and measurements have occupied philosophers for centuries and are subject to debate by researchers. A wide range of possible measures have been developed and used. Most research is country specific and different methods produce different pictures of poverty. This study aimed to compare measures of poverty within an epidemiological field laboratory in Bavi District, northern Vietnam (FilaBavi) and specifically to find out whether the official economic classification made by the local authority matched other measurements of socioeconomic status. Methods: Structured questionnaires were used to collect socioeconomic information in 11,547 households. In addition, the official classification for individual households was recorded. Five economic indicators were constructed: income, expenditure, household assets, housing conditions, and local authority's estimation. Results: Official economic classification and housing score were symmetrically distributed, while assets score and particularly income were highly skewed. Design effects were high because of high intra-cluster correlations. No indicator was closely correlated with any other. Sensitivity and positive predictive value for poverty were generally low for all indicators. Discussion: The authors' findings do not suggest that any of the indicators used is substantially better than the other or better than the Official Economic Classification made by local authority. The results also show that no indicator is particularly useful to predict the values of any other indicator and different poverty indicators may classify different socioeconomic groups as poor.