Skip to main content
Eliminating bias with decision automation

The reporting of this study Study: Blacks, Latinos pay more for mortgages was distressing for me and others in this industry. After all the introduction of credit scoring back in the 50s was in part designed to eliminate bias by ensuring that only things that affect your creditworthiness should be considered. Several stories exist of the Fair Isaac founders (Bill and Earl) being asked to add race as a factor in scorecards early on and refusing on the very practical grounds that it had no bearing on creditworthiness. 

The key to consistent and fair decisions in sub-prime lending (the topic of the report) is to have very clear credit policy rules, based on solid analytics that are automated and followed without exceptions close to 100% of the time.  Automating this using an EDM approach to combine the analytics and the rules helps eliminate and residual personal bias. Lenders who enforce these policies rigorously and verify the customer claims of key variables such as income, stability of employment, total debt burden and value of the collateral, should get the same resulting loan rates irrespective of race or color.

The fact that some still are not makes me, like the study's authors, think that this is due to bias - conscious or sub-conscious. Clearly we have a ways to go.

By the way, if you're worried about your credit rating and what impact it might have on loans, you can get free credit reports at and check your scores (and find out about them) at (shameless plug for Fair Isaac's consumer website). The report was authored by the Center for Responsible Lending.

related posts