/ 27 November 2007

Poverty shock! (Never mind the facts)

The press recently had a field day with reports that poverty in South Africa had doubled since 1994. The story seems to have originated with a press release from the South African Institute of Race Relations (SAIRR) that was picked up by major media. This is shocking news. But is it true?

The SAIRR report says the percentage of South Africans living on less than $1 a day doubled from about 4,5% to 8,8% between 1996 and 2005. The number of people in this category is said to have expanded from 1,9-million to about 4,2million.

South Africa’s poverty figures are hotly debated partly because no single official source provides reliable and regular trend data. In 2006 the HSRC brought together the country’s top poverty experts to exchange views. Despite deep differences, not one of them would argue that poverty has got worse since 2000.

For example, Prof Servaas van der Berg of the University of Stellenbosch has found that the number of people living on less than R3 000 a year fell from 16,2million to 15,4million between 1993 and 2004. While the population grew 16,3% (or 7,3-million people), there were fewer people living in poverty, both absolutely and relatively.

The All Media and Products Survey data shows that the poorest income group (LSM 1) fell from 10,5% to 4,8% of the population between 2001 and 2006/7. This is roughly the group referred to by SAIRR. The bottom three income groups (LSM 1 to 3) shrank from 38,8% to 27,7% of the population in this period. The incomes of people in the poorest groups grew by about 38% between 1993 and 2004, mostly as a result of the expansion of social grants.

What explains the discrepancy in the SAIRR report? Most researchers use a combination of mostly official data sources to arrive at a view on poverty levels and trends. Examples include the Income and Expenditure Surveys, the General Household Survey, the October Household Survey, the Labour Force Survey and other data sets available from Statistics South Africa. While not official, the AMPS provides the only annual data set on income trends dating back to the 1990s and is seen to be a reliable source.

In contrast, the SAIRR accessed a database purchased from a company called Global Insight. As the data set is not open to scrutiny, I called to investigate. The problem arises from two potential sources.

First, Global Insight draws together data benchmarked to the 2000 Income and Expenditure Survey (IES). If the 1995 and 2000 IES are compared, they show rising poverty. However, due to changes in methodology and sampling, it is generally accepted that they are not directly comparable. Moreover, much has happened since 2000 to reduce poverty, particularly the introduction in social grants. Global Insight notes that it did not take this into account when it constructed its trend.

Second, the dollar rate used by Global Insight uses purchasing power parity as its measure, so that the exchange rate was R4,80 and R3,50 to the dollar in 2005 and 1996 respectively. This meant that a $1 a day translated into R1 277 to R1 752 a year. It is precisely this income group that shrank the most. Complexities in this exchange rate translation might have thrown off their model in complex ways.

The SAIRR has issued a report meant to highlight business and employment trends, but has relied on someone else’s conclusions. It seems satisfied to issue dramatic statements without checking official sources or easily accessible analysis.

The poverty debate cannot rely on such shoddy work. A major news agency called me last week to ask for comment on the SAIRR report. I quoted the AMPS data showing how the poorest income groups got smaller in the past six years. They ran the story anyway as reported by SAIRR, taking a quote from one of my papers out of context to support this report.

The real problem of poverty is big enough. It is not helped by arbitrary inflation, sloppy research and poorly researched journalism.

Miriam Altman is executive director of the Employment Growth and Development Initiative at the Human Sciences Research Council