Income poverty in 2004: a second engagement with the recent van der Berg et al figures

Differing analyses of the poverty reduction rate in South Africa

By: Meth C, SDS Working Paper, no 47. 2006
Via: Eldis

Recent poverty estimates, made by acclaimed South African academics, have become accepted as truth by a government in need of some good news. This paper makes use of the Labour Force Surveys (LFS) for 2001 and 2004 to argue that the previous poverty estimates, commonly assumed to be accurate, are in fact too low. Poverty reduction, the paper argues, has not been as dramatic as the statistics imply. The paper examines numerous problems associated with data analysis. These include the treatment of missing income estimates, the plausibility of the zero-income estimates, under-reporting of income and expenditure, and the inadequacy of social grants data. Following this, the author estimates poverty reduction between 2001 and 2004, and presents a critique of the prevailing consensus on the rate of poverty reduction.

The study finds that, using the LFS data, the number of people living in poverty had reduced by just over a million. This result contrasts strongly with those previously provided, which suggest a decline in the poverty headcount of over 3 million. The main conclusion, therefore, lies in the need for a more multi-dimensional analysis of existing data, in the absence of fully accountable statistics. For example, the authors conclusions are sensitive to the assumptions made about income under-reporting, about migrant remittances and about social grants. The paper ends by stressing the importance of any reduction in poverty, regardless of how large or small it may be. The author supports many welfare initiatives, particularly the social grants system, but maintains that there is a long way to go before the income gap narrows to a significant extent.