PMC peer-reviewed — excluding alternative data led to "a significant decline in model performance" (PMC11108212)

Claim. Peer-reviewed analysis (PMC11108212) finds that excluding alternative data from default-prediction models leads to a significant decline in model performance — the information is not redundant.

Quote.

"Led to a significant decline in model performance."

Source. PMC11108212 (accessed 2026-06-21).

Confidence. Verified. Peer-reviewed; ablation-study design isolates the marginal contribution of alt-data.

Caveats. "Significant" is statistical-significance language, not magnitude — read the underlying paper for the specific lift before quoting a number. Multiple studies show direction; specific lift varies by dataset.

Implication / use. Use to support the direction claim ("alt-data improves default prediction") without quoting a single number. Pair with NBER WP 29840 (Di Maggio, Ratnadiwakara, Carmichael, 2022) — "Invisible Primes: Fintech Lending with Alternative Data" for both mechanism and magnitude examples.