Quixotic Monuments to False Prediction: Oliver Kim and Morten Jervens on GDP in developing countries

(And relevant, I suspect, to more developed ones).

This piece from Oliver Kim resonates with my struggle to get good (methodologically clear, sufficiently granular, consistent, publicly available) literacy statistics in Indonesia. Recommended.

[In 2013’s Poor Numbers, Morten] Jervens writes:

“In 2010, I returned to Zambia and found that the national accounts now were prepared by one man alone… Until very recently he had had one colleague, but that man was removed from the National Accounts Division to work on the 2010 population census. To make matters worse, lack of personnel in the section for industrial statistics and public finances meant that the only statistician left in the National Accounts Division was responsible for these data as well. (pg. x)”

…

Without the staff to collect and analyze survey data, statistical agencies are usually forced to improvise, guessing the size of the economy from population figures, which are themselves extrapolated from censuses that are decades-old.

To name one prominent example, Nigeria (thought to be the world’s sixth-largest country) has not conducted a national census since 2006. With almost twenty years since the last data collection, the population figure of 230,842,743 that you can currently pull from Wikipedia is a quixotic monument to false precision.

Indeed, it would not be a great exaggeration to say that we have never known how large Nigeria is.

The first census after independence, in 1962, was marred by regional infighting, because of its role in determining federal spending and assembly seats. The initial results, which reduced the population share of the politically dominant north, were rejected by the Northern People’s Congress.

A redo in 1963 was also rejected as fraudulent—this time largely by southern politicians.

The 1973 census was largely rejected as illegitimate; the 1980s and 1990s saw no censuses under military dictatorship; and the 2001 census produced an implausibly low number.

Even in the relatively successful 2006 attempt, census enumerators were attacked and killed, and the results were still disputed.

…

In the absence of good underlying data, the prevailing approach for GDP in developing Africa can be summarized as:

“Income estimates… derived by multiplying up per capita averages of doubtful accuracy by population estimates equally subject to error. (Jervons, p.39)”

Oliver Kim – How Much Should We Trust Developing Country GDP?

See also:

Statistics: The Galton Board and more
Statistical Rethinking: Richard McElreath on Bayes in the Garden of Forking Data
Building systems, questioning statistics, finding things out, decline and fall
Sampling bias; or, small n
Statistical dojo