Technically this paper doesn’t say “poor to rich” it says “subprime borrowers to super-prime borrowers”. Income-to-FICO score is only moderately correlated. Well, it says rich to poor in the abstract and conclusions, but not the actual writing.

The paper says high-income borrowers who run balances “lose” the most in this transfer - because they spend more in absolute terms, and banks are better able to capture that through balance increase.

To quote: “our findings are inconsistent with the reverse Robinhood hypothesis”.