The Great African Data Debate
24 September 2013
At the recent UNU-WIDER conference here in Helsinki on inclusive growth in Africa, the quality of national accounts was hotly debated. Part of the conference was focused on innovative methods, theories and empirical approaches to measuring poverty, inequality and social and economic mobility. The use of macro, sectoral, price, qualitative, and other data as a cross-check on traditional poverty survey data was also discussed.
In recent years there has been some debate about the poor condition of African statistics. One of the contributors, Morten Jerven, author of the critical book Poor Numbers), actually arrived fresh with a controversy in tow. Here is a useful overview of the subject by UNECA executive Carlos Lopes.
Recently the Chief Economist of the World Bank’s Middle East and North Africa Region, Shanta Devarajan, talked about a ‘statistical tragedy’ in a lecture (adapted for YouTube here and also developed in a new article for the Review of Income and Wealth). He gives one example, the rebasing of Ghana’s GDP numbers and role of the World Bank:
‘The tragedy here is that we were happily publishing estimates of Ghana’s GDP for the last twenty years, including extolling how Ghana was growing so rapidly and Ghana was actually reducing poverty at a very rapid clip, and how well it was doing—and we were taking quite a lot of credit for it, I might add’.
Devarajan was thus also reflecting upon the Bank’s own role in the ‘tragedy’; It seems like African growth has picked up since the mid-1990s and, thanks to that growth, poverty is declining.
‘The statistical tragedy is that we cannot be sure of either of these phenomena.’
We know less about the poor
Morten Jerven’s key message is that most African economies are richer than what GDP numbers indicate, but at the same time economic growth is slower than what the same evidence says. He also noted that there is a bias in the statistics towards the more well-off.
‘We know less about poor countries, and less about poor people in those countries.’
At the conference he underlined that his suggestion is not to throw all available data overboard, but to discuss ways to make it more reliable. Still, Erik Thorbecke noted in a response to an audience question that household survey data already is actually good quality, which also means that estimates of poverty and inequality should be reliable.
Poor numbers is being discussed in other places as well because of the great risk that both private and public investments decisions are based on flawed data. In the recent post-2015 report on Millennium Development Goals it was also suggested that something of a data revolution is needed improve the quality of statistics and information. The African Development Bank Group (AfDB) has responded to that challenge with The Open Data for Africa platform. At the conference Arjan de Haan also presented the six Indices of Social Development, a relatively new innovative database supported by the World Bank that was launched in March 2011.
During the last two decades there has been a wide range of initiatives to address the problem with bad statistics and there is some progress, as explained by Jeffery Round in his WIDER working paper. But the problems are, as Justin Sandefur and Amanda Glassman note in a in their draft conference paper, not easily solved due to a number of reasons such as political interference, limited human capacity, as well as perverse incentives linked to the collection of data.
Another problem is volatile and unpredictable aid funding as donors account for the larger part of activities of statistical offices in developing countries. This volatility makes improvements in data accuracy difficult to sustain.
For statistical offices in Africa the debate at our conference could actually be utilized as a good source of strong arguments for more sustainable funding and technical support.