Credit pricing is changing. Traditionally, lenders priced consumer credit by using a small set of borrower and loan characteristics, sometimes with the assistance of loan officers. Today, lenders increasingly use big data and advanced prediction technologies, such as machine-learning, to set the terms of credit. These modern underwriting practices could increase prices for protected groups, potentially giving rise to violations of anti-discrimination laws.
What is not new is the concern that personalized credit pricing relies on characteristics or inputs that reflect preexisting discrimination or disparities. Fair lending law has traditionally addressed this concern through input scrutiny, either by limiting the consideration of protected characteristics or by attempting to isolate inputs that cause disparities.
But input scrutiny is no longer effective. Using data on past mortgages, I simulate algorithmic credit pricing and demonstrate that input scrutiny fails to address discrimination concerns. The ubiquity of correlations in big data combined with the flexibility and complexity of machine-learning means that one cannot rule out the consideration of a protected characteristic even when formally excluded. Similarly, in the machine-learning context, it may be impossible to determine which inputs drive disparate outcomes.
Despite these fundamental changes, prominent approaches to applying discrimination law in the algorithmic age continue to embrace the input-centered approach of traditional law. These approaches suggest that we exclude protected characteristics and their proxies, limit algorithms to pre-approved inputs, and use statistical methods to neutralize the effect of protected characteristics. Using my simulation exercise, I demonstrate that these approaches fail on their own terms, are likely unfeasible, and overlook the benefits of accurate prediction.
I argue that the shortcomings of current approaches mean that fair lending law must make the necessary, though uncomfortable, shift to outcome-focused analysis. When it is no longer possible to scrutinize inputs, outcome analysis provides a way to evaluate whether a pricing method leads to impermissible disparities. This is true not only under the legal doctrine of disparate impact, which has always cared about outcomes, but also, under the doctrine of disparate treatment, which historically has avoided examining disparate outcomes. Now, disparate treatment too can no longer rely on input scrutiny and must be considered through the lens of outcomes. I propose a new framework that regulatory agencies, such as the Consumer Financial Protection Bureau, can adopt to measure the disparities created when moving to an algorithmic world, enabling an explicit analysis of the trade-off between prediction accuracy and other policy goals.
The Impact of the Introduction of Positive Credit Reporting on the Australian Credit-seeking Population (Univewrsity of Sydney Business School, 2019) by Andrew Grant comments
This report examines the introduction of Australia’s comprehensive credit reporting (CCR) regime and its impact on the population of Australian credit applicants. Under the regulatory change, lenders will be able (and in some cases, are required) to share new sources of information about their borrowers with credit bureaus. In addition, credit scores, which under the previous regime incorporated only ‘negative’ information (such as loan defaults and credit enquiries) about a borrower, are now augmented with ‘positive’ information related to the granting of, and the servicing of these loans. Using data from one of the major credit bureaus in Australia, illion, we examine how the changes to the credit reporting regime help lenders gain greater visibility to the creditworthiness of consumers and how this translates to the consumer gaining access to different types of credit. Consistent with predictions from prior literature, the changes to the credit reporting environment result in greater dispersion of credit scores among the population, potentially leading to lower adverse-selection risks for lenders. More than two-thirds of the population experience a score increase following the implementation of positive reporting, lowering their perceived credit risk by an average of 25-35%. The balance of the population experiences a score decrease of a similar magnitude, apart from a small subset for whom CCR results in a large decrease in credit score.
We examine the proportion of the population that crosses three key credit-score thresholds (480, 600, and 720) with the advent of CCR. A borrower crossing the 600 threshold may be broadly interpreted as moving towards a ‘prime’ credit risk , potentially providing access to an average 4.5% reduction in the interest rate on personal loans, and also up to a threefold increase in credit card limits (conditional on meeting lender credit servicing requirements). In aggregate, we find that 8.02% of individuals cross the 600 credit-score threshold with the introduction of positive credit reporting, compared with 5.27% of individuals who fall below the threshold. We interpret the difference of 2.75% as a basic measure of the growth in the prime credit population following the regime change. A net difference of 11.13% is observed at the higher 720-score threshold, while a -3.03% difference at the lower 480 credit-score threshold.
Using demographic based information, we find that applicants who cross the 600 credit-score threshold are disproportionally younger, from higher-risk geographical areas, with lower estimated incomes and wealth, and from less established households. Arguably, this group represents borrowers that may have been traditionally underserved under the negative scoring regime. Following the introduction of positive reporting, they can demonstrate prudent credit behaviour and benefit from this. There is a lesser relationship between observed demographic characteristics and falling below the 600 credit-score threshold. Applicants from higher socioeconomic areas, with higher estimated wealth and income, and more established households stand to benefit from a greater likelihood of crossing of the 720-score threshold, enabling access to premium credit products (and even lower priced credit as risk based pricing becomes more commonplace). Overall, the introduction of CCR benefits the population on average by providing a net increase in credit access, with further benefits expected in terms of more efficiently priced lending products (especially at the higher 720 score threshold) and greater borrower discipline in order to maintain a premium credit record.
