That one looks right to me (or, “right” meaning consistently using period LEB) - it’s a little hard to compare because of the difference in granularity but it shows about a 20-year drop for WW2 which is what I would expect.

I edited my comment above; I think what’s happening is that the OP article is mixing different metrics for different parts of the chart. I think this one you’re sending is consistently using period LEB which is why the size of the dips is different.