Have you read this report? Could you summarize it for us, in a paragraph or two? 
@Lel wrote:Have you read this report? Could you summarize it for us, in a paragraph or two?
One of the provisions in various consumer credit laws passed by Congress a few years ago called for the Fed to study the effects of credit scoring on various demographic groups. In particular, Congress wanted to know whether credit scoring tended to discriminate against any particular demographic group (such as minorities, people from certain neighborhoods, etc.), by overpredicting their credit risk compared to other groups of people. While credit scoring algorithms don't include race, age, or zipcode directly there was concern that factors they did include might in fact be proxies for demographic factors that may not be included.
The Fed study concluded that for race, religion, national origin, etc., there did not appear to be any discriminatory effect. They only such effect they found was that how credit scoring treats age of accounts tends to discriminate somewhat against older people and recent immigrants, by making them look somewhat higher risk. But within each age group the age of accounts does have a very strong predictive value, such that removing age of accounts from the scoring models makes them less predictive overall, so they concluded that age of accounts needs to remain an allowed factor. In the case of recent immigrants, the problem is pretty obvious: their credit files look too much like those of younger people who are higher risk. So for recent immigrants, allowing the inclusion of age as an explicit factor in credit history scoring (which the law doesn't allow) would improve their scores by providing a means for scoring algorithms to distinguish them from younger people.
The problem with older people is more subtle: older people already get higher scores because their age of accounts tends to be longer, they tend to have more tradelines, and so forth. But younger people are statistically even higher risk than their credit scores would predict, so allowing the inclusion of age as an explicit factor in credit history scoring (which the law doesn't allow) would improve the scores of older people even more than the ways in which allowed factors tend to improve the scores of older people. In other words, a middle-aged or older person with a high credit score is even safer than that credit score says, while a young adult with a high credit score is not quite as safe a credit risk as that credit score says.
Another second-order effect was in how medical debts are handled: if medical debts were entirely removed from credit scoring, this would have a greater effect on older people than on younger people simply because older people are more likely to have substantial medical debts.
However, to me the description in the full paper of how the study authors did their analysis was much more informative and interesting than any brief summary of its conclusions can be. While fairly long, it is reasonably well written and will reward close reading.