top of page

Review: How Costly is Noise? Data and Disparities in Consumer Credit

by Ruth Lee, CMB




The paper "How Costly is Noise? Data and Disparities in Consumer Credit" by Laura Blattner and Scott Nelson examines the role of credit scores in lending decisions and the potential for disparities in credit outcomes across social groups. The authors find that widely used credit scores are statistically noisier indicators of default risk for traditionally underserved groups, such as minorities and low-income borrowers. This noise emerges primarily through the explanatory power of the underlying credit report data (e.g., thin credit files), not through issues with model fit.


The authors then use a structural model of the US mortgage market to quantify the gains from addressing these information disparities. They find that equalizing the precision of credit scores can shrink differences in efficiency and in loan approval rates for disadvantaged groups by approximately 50%.


The authors argue that their findings provide a stark illustrative example of how promoting disadvantaged groups' ability to show their quality as borrowers can advance the goals of efficient and equitable credit access.


Here are some of the key takeaways from the paper:


Credit scores are statistically noisier indicators of default risk for traditionally underserved groups.

  • This noise emerges primarily through the explanatory power of the underlying credit report data (e.g., thin credit files), not through issues with model fit.

  • Addressing these information disparities can lead to significant gains in efficiency and equity in credit markets.

  • The paper's findings have important implications for policymakers and businesses that are interested in promoting fair and equitable access to credit. By understanding the sources of noise in credit scores, policymakers can develop policies that help to reduce these disparities. Businesses can also use this information to make more informed lending decisions and to create products that are more accessible to disadvantaged groups.

I hope this summary is helpful.

We show that lenders face more uncertainty when assessing the default risk of historically under-served groups in US credit markets and that this information disparity is a quantitatively important driver of inefficient and unequal credit market outcomes.

Article on Data Bias in Consumer Lending
.pdf
Download PDF • 7.08MB


30 views0 comments

Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page