In this article, I found this graph of the percentage of people earning less than $7.40 a day worldwide, excluding China. That number has barely shifted from 1981 to 2015 - it's down by 3 percentage points from 62% to 59%. We can contrast this with the drop in people earning less than $1.90 a day: even excluding China, this has dropped by 17 percentage points (from 29% to 12%) in the same time. I played around with the data source for a while, and the $7.40 threshold doesn't seem particularly cherrypicked (nor are the dates). Excluding China is not totally standard but seems reasonable since otherwise the data from this one country would make it difficult to see more general trends.
The original article uses this data (and other arguments, like critiques of neoliberalism in the 80s and 90s, and criticism of Rosling's data sources) to argue that books like Rosling's Factfulness and Pinker's Enlightenment Now are misleading. To be fair, it makes sense that when you use a higher threshold, the percentage of people crossing that threshold in a given time decreases. But going from 62% to 59% is a very small difference over 34 years! So I'm tempted to conclude that the decline in extreme poverty is a less robust indicator of beneficial long-term economic growth than I thought, and that the two books I mentioned are missing important trends in global poverty. Is this reasonable?
Does it really explain anything to say that poverty rates are stagnant outside of China because those countries have a lack of economic growth? I guess it suggests that changes in foreign aid or in income inequality aren't the culprit, but this explanation still feels like a dormitive principle.