Lower the per capita.
Liberia just ran its first census in many years. The tentative population number: about 3.5 million.
For our impact evaluations in the countryside, the census bureau shared its village population estimates. It looks like they may have missed a few zeroes. The current count for of Voinjama, a bustling city with (I would hazard) 10 or 20 thousand people? 4500.
More than ten thousand census enumerators fanned out in a single week of enumaration earlier this year. It was a spectacular enterprise, and a testament to the capacity of the post-war government. But it may have been a little ambitious to expect an accurate count. I wouldn’t be surprised if national population estimates were off by 20 or 30 percent.
I wonder: if the accuracy of developing country census bureaus improve over time, could we be manufacturing slowdowns with the data? Income growth would be biased up the first decade or two, and biased down later on.
Cheaper, easier, and more accurate would have been a sample-based approach. But nations rich and poor alike cling to the census. A badge of honor? A misplaced principle? A statistical misunderstanding?
In the meantime, wait for the growth boom in Liberian growth statistics.