12 Comments
User's avatar
Javi Guiroy's avatar

Excellent review. What stands out most is that you avoid easy skepticism (the data are bad, therefore useless) and instead highlight something more uncomfortable: many sophisticated analyses rest on statistical foundations we rarely scrutinize.

The Tanzania example is particularly striking. This is not just a technical issue; it’s a reminder that macroeconomic series are institutional constructs, not neutral photographs of reality. When the base year changes, so can the entire development narrative.

Perhaps the key implication is not to abandon GDP, but to systematically complement it with independent sources, household consumption surveys, administrative data, satellite imagery (night lights), and digital transaction data. In high-informality contexts, triangulation should be standard practice, not an afterthought.

Thank you for bringing Jerven back into the discussion; his critique remains uncomfortably relevant.

Louis Becker's avatar

Alwyn Young’s “The African Growth Miracle”(https://personal.lse.ac.uk/YoungA/TheAfricanGrowthMiracle.pdf) is a super interesting attempt to re-estimate GDP growth using reliable (?) micro-data — recommend!!

Nadim (Abolish NDIS and EPBC)'s avatar

This is why the West needs to completely defund the aid industrial complex including the World Bank and IMF. It's not worth taking development of African countries seriously when they themselves don't take it seriously. What an absolute waste of time.

Allan Holst's avatar

I'm in charge of Mexico's balance of payments. How can we estimate a margin of error for, say, the current account as a share of GDP?

Toby Tremlett's avatar

Hey! This is really interesting- do you mind if I crosspost it in full onto the EA Forum? (https://forum.effectivealtruism.org/?tab=global-health-and-development) Under your account if you have one (or as a linkpost on mine if you don't).

Noah Birnbaum's avatar

Very interesting! Thanks!

Robert Inklaar's avatar

The reason for the large decline in the PWT series is that the figure shows the series based on multiple PPP benchmarks. PWT also has a growth series that matches countries' official series and for Tanzania this does not show a 1980s collapse in GDP per capita

Anssi's avatar

Thanks, I learned something new from your post.