CFPB ’ s Concerns With Black-Box Credit Models
The Consumer Financial Protection Bureau ( CFPB ) issued a
circular on May 26 , 2022 , detailing their stance on black-box credit models . Although these models are fast , effective and usually helpful , the models may also contain unseen biases or other discriminatory tendencies due to the way data is mined and the history of bias in the American financial world .
“ Federal consumer financial protection laws and adverse action requirements should be enforced , regardless of the technology used by creditors , and that creditors cannot justify noncompliance with the Equal Credit Opportunity Act ( ECOA ) based on the mere fact that the technology they use to evaluate credit applications is too complicated , too opaque in its decision-making , or too new .” CFPB
In recent years , the explosion of digital financial services has been a double-edged sword for both consumers and lenders . The sudden influx of new customers as online banking gained traction meant financial institutions everywhere needed a fast , reliable and theoretically sound model to get customers through intake procedures and into their systems . But how was that to be done with so many consumers from so many areas , all wanting something different ?
The Answer : Black-Box Credit Models and Algorithms
Black-box credit models were designed to be effective and efficient , by taking enormous amounts of consumer data and history , and basing decisions off the predictive analysis of that data . Given algorithms are designed to make decisions faster and better than human beings , financial institutions faced with situations where a lot of data needed to be sorted through resorted to complex algorithms .
Typically speaking , a sufficient and adequate machine learning system will take as much data as it can , start making predications or analyses , and then will
6