Wealth and Population Data in Africa

Brian Sturgess - May 2012

Key Points
  • Most sub-Saharan African GDP data is of very poor quality, and in most cases probably significantly under-estimates the real level of economic activity.
  • Most African population data is of equally poor quality, and may over estimate population in many cases.
  • Due to the persistent underestimation of GDP and the likely overestimation of population , real income per capita in Africa may be significantly higher than previously assumed.
  • As a result almost all UN, World Bank , Aid Agency or similar data based on African income per capita statistics should be viewed as extremely unreliable.

Last year we published an article by Morton Jerven [1] of the University of British Colombia in Vancouver who demonstrated based on his own in-depth research and country visits to national statistical offices that faith in the accuracy of most sub-Saharan African country’s national accounts was little more than folly. Jerven’s interest stems from questions such as the appropriate level of aid and development targets, but the inaccuracies have implications for investors too. He comments:

“..what needs rethinking is the database for development analysis. What do we know about income and growth in Sub-Saharan Africa? The answer is: much less than we like to think. The data are unreliable and potentially seriously misleading…… If income and growth statistics in Africa do not mean anything, a great part of development analysis and policy targets are similarly meaningless. “

The size of the problem was demonstrated strongly with the recent case of Ghana. On the 5th of November, 2010, the Ghana Statistical Services office announced that its GDP estimate for the year 2010 was revised upwards to 44.8 billion cedi, as compared to the previously estimated 25.6 billion cedi. Overnightthe average Ghanaian had become 75% better off in terms of GDP per capita.  The inaccuracies in sub-Saharan African GDP data, the numerator of the GDP per capita ratio, arise as a result of a combination of a lack of local resources and international laxity.

An important issue concerns the time lag between regular GDP estimates and the base year chosen to estimate the structure of the economy.. It is recommended that a country rebases its GDP estimates regularly, optimally every 5 years because of structural changes  affecting the balance between economic sectors. However, few African countries have made these necessary adjustments and having surveyed 48 African countries, Jerven found that for 13 countries no official data was available on the base year while only 19 were using a base year less than a decade ago. Furthermore, even when local national statistics offices have provided no recent GDP estimates the World Bank still reports updated data almost miraculously. As Jerven notes: 

“The prevailing sentiment seems to be that data availability is more important than the quality of the data that are supplied. “

Ghana’s income was revised upwards as a result of rebasing the economy from 1993 to a new base year of 2006. This allowed a measurement to be taken of sector’s such as telecommunications which has shown spectacular growth while in 1993 the mobile phone had not yet arrived in Ghana. As Jerven notes with a 1993 base year the communications sector was accounted for solely through the numbers of home phones and receipts from the national telecommunication company.

There is little reason, therefore, to have much confidence in the numerator involved in the calculation of African GDP per capita estimates. It is also possible that confidence in the accuracy of African population data is equally misplaced which means that another source of error is introduced into ranking of countries by GDP per capita through the denominator of the ratio. In this issue we publish a controversial and challenging article by Deborah Potts of Kings College, University of London which looks at problems in the estimation of population data in sub-Saharan Africa and in particular with issues concerning the rate of urbanisation. She states that the most frequently used sources of urban population data the World Bank and the UN-Habitat reports are misleading and provide exaggerated estimates of urbanisation in Africa. Dr Potts contrasts the 2008 UN-Habitat report on urban populations with a report by the French Africapolis team which used satellite evidence to compare census and other data with what could be deduced from urban areas observable from space. The Africapolis evidence for Nigeria, the continent’s most populous state, estimated an urbanisation level of 30% while the UN Department of Economic and Social Affairs declared that Nigeria’s urbanisation level in 2006 was 49%. In terms of absolute numbers the UN figure implies an urban population of 69 million, while that of Africapolis implies an urban population of 42 million, an exaggeration by the UN of 64%.


Both of these articles suggest the reasons for rising inaccuracy and often the blame does not lie solely with local statistics offices, but it must be shared with international institutions who have replaced detailed analysis and regular investigations with Homeric assumptions. On population data, according to Potts, the UN and the World Bank have simply projected forwards current and future urban populations based on the rapid urbanisation that occurred in the 1950s to 1970s. This laxity has been encouraged by the lack of census data for many countries until the late 1990s. According to Potts:


“Politicians, civil servants, donors, urban planners, city authorities and academics persisted in using urban population data based on increasingly flawed assumptions about growth rates. In time, fictitious figures became facts by being constantly re-stated. Even when census data became available which provided a corrective, it could be many years before datasets and projections were amended accordingly.”


Country GDP per capita data are estimated by dividing the total, not the urban population, into GDP, but official estimates of total population are also subject to a wide margin of error. In Nigeria, for example, there was a gap between 1991 and 2006 in the taking of census data. But in Nigeria regional population numbers are highly political and Potts also notes that every census since 1952 has been contested. Furthermore, there is still no official breakdown of urban populations. The Democratic Republic of Congo, believed to have the third largest population in sub-Saharan Africa, has not conducted a census since 1984 and Pott’s notes that the “population figures – and projections – for Kinshasa, the capital city, are little more than guesswork.”


What does this all mean? First, it implies that ranking countries like Ghana which has rebased its economy with neighbours such Cote d’Ivoire and Nigeria who have not rebased the national accounts recently is meaningless. Second, if there are serious underestimates of GDP and overestimates of total population for many sub-Saharan African economies, then GDP per capita data will be doubly deflated from what it should be. This creates another source of measurement error separate from the rebasing issue since just as nobody knows how much all of the African economies have change structurally over the last two decades nobody also knows how far real population figures lie from those assumed or estimated from poor quality infrequent census data.


Oskar Morgenstern in an important, but too often ignored book, made a damning critique of the value of economic statistics by noting that unlike the natural sciences, no attempt is usually made to assess potential observation error. To do so partially can lead to shocking results that call into question much economic analysis. For example, taking Nigeria as an illustration of the underlying problem discussed above the country has not rebased its economy since 1990 while few believe the results of the 2006 population census. In 2010, Nigeria had an estimated GDP (PPP) of US$377.9 billion with an official estimated population of 162.5 million producing a GDP per capita figure of US$2,330. However, using the Ghana rebasing estimate and the urban population overestimate discussed above, Nigeria’s GDP per capita could have been as high as US$6,670 or triple the official figure.


Finally, in another paper in this Journal Freidrich Schneider,[2] a leading expert on the size and importance of shadow or grey economic activity across the world estimated that in the case of Nigeria in 1998 the shadow economy was a potential 48.8% of the size of the official one.  This would raise effective income to US$9,913 per capita.


 Far- fetched, perhaps, but the uncertainty around these economic figures should be embarrassing for official aid agencies or unwelcome for governments seeking funding. But for other users accurate and transparent statistics are essential indicators of economic potential and if African economies wish to continue to attract rising investment interest, the issue of data quality needs to be urgently addressed..



Africapolis (2008), Urbanization Trends 1950–2020: A Geo-statistical Approach, West Africa, Fact Sheets By Country (AFD-SEDET)


Jerven, M. (2010), Counting the Bottom Billion: Measuring the wealth and progress of African economies, World Economics, 12, 4, 35-52.


Morgenstern, O. (1950) On the accuracy of economic observations. Princeton, New Jersey: Princeton University Press


Schneider, F, (2010) The Shadow Economy Labour Force: What do we (not) know?, World Economics, 12, 4, 53-92.


[1] See Jerven (2010)

[2] Schneider (2010)