How not to present poverty research results: The South African case

Because of their vital role in charting progress (or the lack thereof) in the pursuit of the povert y reduction, statistics are of obvious importance. In South Africa, these leave much to be desired.Disagreements among academics on the severity of poverty, the result of the failure of Statistics South Africa to conduct the appropriate surveys, are the inevitable result. Far from losing money (or sleep) as a result, some in the profession resort to further research, some of it quite highly paid, to squeeze new results out of old, often unreliable data. This could have serious consequences for the poor – policy failure caused by faulty monitoring can easily damage the vulnerable. Regardless of the reliability or otherwise of their findings, it is argued in the present paper that researchers would do well to offer them in a way that minimises the possibility of their being misinterpreted and/or misrepresented, and that maximises the likelihood that the non-specialist reader will be able to understand them. It is common practice to give poverty estimates in the form of the (FGT) ratios suggested by Foster, Greer and Thorbecke (1984), often without accompanying estimates of the absolute magnitudes involved. This, the present paper claims, allows overly optimistic conclusions to be drawn, making possible the concealment of rising misery behind a veil of aggregate improvement. Commencing with a glance in the abstract at the FGT ratios, the paper concludes that in order for poverty statistics not to convey a misleading impression of changes in the phenomenon they seek to represent, the ratios have to be augmented with sufficient information of concurrent changes in the income distribution. Most poverty studies look at changes in inequality. Often, however, the inequality results are not linked directly to the changes in poverty. As far as income poverty is concerned, the present piece of research suggests that doing so is the only appropriate way to present results. Having sketched a conceptual foundation, the paper looks at the regurgitation by government, without comment, of poverty statistics that directly contradict each other. After that, the strange case of an undeserved accolade government awards its anti-poverty policies, is found to be based upon a misinterpretation of their own findings by the authors of a recent poverty and inequality study (Bhorat and van der Westhuizen, 2008). A new set of poverty and inequality estimates (Leibbrandt et al, 2010), although it does not conform to the mode of presentation suggested above as necessary, points (as do the Bhorat and van der Westhuizen findings) to the strong likelihood that although the poverty headcount ratio may have fallen since the advent of democracy in the country, the poverty headcount is likely to have risen by several million between 1993 and 2008. An appendix at the end of the paper offers a little speculation on what poverty levels might have been had the AIDS epidemic not killed so many people.