Are Credit Scores Becoming Less Predictive? – Episode #73

October 29, 2008 by  
Filed under Episodes

In a recent article on, Tony Hughes posted data on how credit scores might not be assessing risk as well as in the past.  Matt Stichnoth summarized his findings in a SeekingAlpha article and my responses address his main points.

Hypothesis #1: People have learned to game the system. They accept line increases (which raise scores) when offered, for instance, even if they have no intention of using them. Or they might even pay to piggyback on another, high-score consumer’s card. The moves reflect the individual’s understanding of how credit scores are tallied, not changes in his actual creditworthiness.

Response: Piggybacking is an atypical event impacting a small part of the market that is shrinking. The FICO ’08 model allows authorized users, but Fair Isaac says it has found a way to spot nefarious practices around authorized users which should reduce the number of people trying to use this method to increase their credit scores.   Plus, credit limits are now on the decline, putting downward pressure on scores.

Hypothesis #2: People (especially those with high credit scores) are using their credit cards more than ever. Higher usage means a thicker credit file, and thick credit files tend to make for high scores.

Response: Using your credit cards more can lead to using more of your credit limits and that will hurt your credit utilization ratio and that lowers your credit score.

Hypothesis #3: Because of a quirk in how they’re calculated, credit scores haven’t adequately taken into account the broad rise that’s occurred in mortgage delinquencies. A consumer’s score tends to get dinged when the consumer’s payment pattern varies from the broad average, but when the broad average itself shifts, changes in scores might not adequately reflect broad changes in creditworthiness.

Response: Scoring models care more about number of delinquencies versus pinning them against average behavior.  The new FICO ’08 model will help people with occasional delinquencies but hurt people who are late more frequently.

Hypothesis #4: Consumer card balances are rising. In the past, high balances have correlated with strong consumer creditworthiness. Perhaps no longer, however. Consumer confidence is in the tank and the HELOC spigot shut off for many borrowers, so rising card balances could be a sign, if anything, of rising consumer desperation.

Response: This is an argument for lower scores not higher scores. High balances hurt your credit score because you are using more of your credit limits and that’s a signal for higher risk.

Finally, Matt points out that deliquencies are rising at the same time average credit scores are rising.  Using the “mean” instead of the “median” [which is what Fair Isaac reports] may not be the best measurement.  A mean average calculation can be impacted by shifts in high scores that impact the mean score.  A median shows the true middle score, and I suspect that it going down.

Similar Posts:


Ask a question, make a comment, share tips