Implications of Recent Credit Score Migration on CECL Models

Home News & Insights Implications of Recent Credit Score Migration on CECL Models

Credit score migration acts as an early indicator of credit stress, and recently the story is getting more interesting. Across the country and across nearly every retail product type we are seeing more movement than usual. Some scores are drifting up, but what stands out is the magnitude of the downward moves.

Under normal conditions, migrations cluster in a narrow band. Scores within a given tier might shift up 5 to 10 points here, down 5 to 10 points there. Recently the downside swings are much wider, with consumers dropping multiple tiers at a pace we have not observed in years.

Why does this matter?
Because losses are not linear.

A 20 point improvement in score does not offset a 20 point decline. The math of credit risk is convex.

As scores fall, the marginal increase in expected loss accelerates.
As scores rise, the marginal decrease plateaus.

So when the negative migrations become larger and more common than the positive ones, the reserve impact becomes asymmetrical. Even if ups and downs look balanced in count, the financial effect is not.

This is exactly what we are observing in CECL models across institutions right now. Downward migrations are pulling more loans into higher risk tiers, and the resulting increase in expected losses is outweighing the benefit from any upward drift in the portfolio.

What to watch going forward:

  • Whether the spread between positive and negative movements continues to widen
  • Whether the downward migrations correlate with pockets of delinquency already emerging
  • Whether consumer level liquidity indicators such as savings rates or revolving balances confirm the trend
  • How much of this is persistent versus temporary noise driven by seasonal spending patterns

Score migration usually drifts quietly. When it speeds up and becomes uneven, it tells you where the stress is building. For institutions projecting risk and calibrating CECL reserves, this deserves close attention. Ongoing asymmetry in score migration is rarely a positive sign.

Recent Insights