Can we trust Africa's stats?

Shantayanan Devarajan, chief economist for the World Bank's Africa Region, argues that the continent is in the grips of a "statistical tragedy":

Consider economic growth, which is measured in terms of growth in GDP.GDP in turn is measured by national accounts. While there has been some progress, only half the countries (housing 68 percent of Africa’s population) use the 1993 UN System of National Accounts; the others use earlier systems, some dating back to the 1960s.

That this is not an arcane point is illustrated by the case of Ghana, which decided in 2010 to update its GDP to the 1993 system. The update also provided an opportunity to review in-depth the basic data underlying the country’s GDP estimates. The combination of improved basic data sources and adoption of the 1993 system led to Ghana’s GDP being 62 percent higher than previously thought. Ghana’s per capita GDP is now over $1000, making it a middle-income country. Their debt-to-GDP ratio is much lower.

Newspaper reports of Ghana’s remarkable “growth” in GDP prompted several other countries to consider revising their national accounts. Malawi, for example, recently revised its GDP estimates by over 30 percent (Young, 2010).

The “tragedy” is that the development community was happily publishing GDP statistics and growth figures for Ghana over the last decades, pointing out how well the country had been doing. Now we have to revise those figures. So in fact we did not know how well Ghana was doing.

There is a related problem with population statistics. Most of these are extrapolated from the last census. Since the standard for population census periodicity is 10 years, extrapolation is largely the rule. However, in Africa, only 32 countries representing 65 percent of the total population have had a census during the last 10 years. In Angola, for instance, the most recent census was in 1975. Ethiopia, Africa’s third most populous nation, had its first census covering the whole country only in 2007. In Nigeria, population head-counts are prone to inflation (because they affect fiscal transfers to states) and often controversial. In short, in presenting GDP per capita for many African countries, we cannot be sure of either the numerator or the denominator.

Devarajan argues that the state of affairs is even worse for poverty statistics. For instance, he writes that a widely touted state noting that the percentage of people in Africa living on $1.25 a day, declined from 58 percent in 1999 to 47.5 percent in 2008 was based on a sample of countries representing only 72 percent of the African population.

This is related to an issue I discussed in a post on the new Human Development Index rankings last week. Libya didn't jump up 23 spots on this year's index because the country made such remarkable strides in economic development last year, new statistics simply because of an updated IMF estimate of the country's GDP growth in 2011. 

Readers intuitively understand from domestic politics that leaders have an incentive to produce overly positive statistics in order to demonstrate to voters that they're better off than they were four years ago, or overly negative numbers to show outside funders that they still need aid. Nontheless, journalists, readers, and policymakers are probably a bit too quick to unquestioningly treat reported economic statistics as reality when the process of producing them is often just as political as the policies they're being used to evaluate. 

Sean Gallup/Getty Images