www.decisionscillc.com
Industry
Update
Version 1.0, March 2025
The Impact of Medical Debt
Removal and Targeted Mail
for Financial Services
2025
Copyright © 2025 Decision Science LLC. All Rights Reserved.
Foreword
The end of Q4 2024 and Q1 2025 have posed new challenges to data
science rms working with nancial service providers. A CFPB regulation
removing medical debt from credit reporting was nalized in January,
however, credit bureaus began implementing these changes in the
leadup to this ruling in December. This change in data has broad impacts
across all reported attributes. In addition to medical-specic attributes,
those summarizing total debt or calculating ratios of other debt types
relative to total debt were also aected. These changes came with little
notice. For many users, these attributes have minimal impact as isolated
datapoints. However, rms like Decision Science incorporate these
attributes into larger predictive models, and as a result have experienced
signicant eects. Our models are built using hundreds of these credit
attributes, and as such the small changes are compounded, resulting in
model scores that dier greatly from what they would otherwise have
been. As a result, many nancial service providers engaged in direct
mail marketing have experienced lower marketing response rates and
reduced conversion metrics.
Upon receiving notice of the removal of medical debt, we quickly
transitioned to several of our lesser impacted legacy models. This has
ensured continued strong performance for our clients. Simultaneously,
we began incorporating new data and real-world performance insights to
rebuild our latest models, which are set to go live in early Q2.
We continuously rene our models and explore new modeling
techniques to keep pace with evolving economic conditions and shifting
consumer behavior. To maintain accuracy and eectiveness, we monitor
model performance closely and rebuild our models at least twice per
year - or more frequently when needed. This proactive approach allows
us to stay ahead of industry changes and deliver superior results for our
clients. Additionally, we have expanded our server capacity to enhance
model building and scoring eciency. This upgrade not only increases
processing speed but also adds extra redundancy, including o-site
hardware backups alongside our existing o-site data backups.
We look forward to continuing to serve you and are excited to showcase
our new models and new servers.
Since its decision to remove medical debt from consumer credit reports, the CFPB has remained largely silent, with no
recent news emerging from the agency over the past few months.
Overview:
CFPB Decision to Remove Medical Debt
In January 2025, the Consumer Financial Protection Bureau (CFPB) announced its intention
to remove all medical debt from consumer credit reports. This announcement, coming on
the wings of a ruling last year which removed medical debt in collections, marks a signicant
change in the way nancial institutions and industries assess consumer creditworthiness.
“The Consumer Financial Protection Bureau is issuing a nal rule amending
Regulation V, which implements the Fair Credit Reporting Act (FCRA),
concerning medical information. The FCRA prohibits creditors from considering
medical information in credit eligibility determinations. The CFPB is removing
a regulatory exception that had permitted creditors to obtain and use
information on medical debts notwithstanding this statutory limitation. The
nal rule also provides that a consumer reporting agency generally may not
furnish to a creditor a consumer report containing information on medical
debt that the creditor is prohibited from using.”
Brief Excerpt from the CFPB Final Ruling on Medical Debts
The stated goal of this change is to reduce the negative nancial
impact that medical debt has on consumers, particularly those
who may have otherwise strong credit proles but suer from
unavoidable medical expenses. However, this shift has led to
widespread disruptions in credit data models and selection criteria
for many industries that rely on credit bureau data.
01
02
03
Pre-removal of medical debts, a majority of aected
consumers had a FICO score of 581
75% of aected consumers had other derogatory
marks on their credit reports
Nearly half of these consumers experienced a credit
score increase between 17-20 points.
Summary of the CFPB’s Findings on
Medical Debt and Creditworthiness
The Consumer Financial Protection Bureau had previously raised
concerns about the accuracy and relevance of medical debt as a
factor in determining consumer creditworthiness. Their analysis
highlights multiple points where errors can arise in the reporting
of collections tradelines, leading to potential inaccuracies in
consumer credit reports.
Medical debt, in particular, is viewed as problematic because
its presence on a credit report does not necessarily reect a
consumer’s nancial responsibility in the same way as other forms
of debt. The CFPB points to the complexities of medical billing, the
delays caused by claims adjudication, and variations in provider
policies regarding debt collection as factors that can result in
medical debt appearing on credit reports under inconsistent and
sometimes misleading circumstances.
Additionally, the CFPB cites its own research, which found
that medical collections tradelines are less predictive of future
delinquency than non-medical collections. This has led to changes
in credit scoring models, such as FICO 9 and VantageScore 3.0,
which now dierentiate between medical and non-medical
collections, or remove certain paid collections from scoring
altogether. The CFPB supports such renements, arguing that they
produce a more accurate reection of consumer creditworthiness
and reduce the likelihood of medical debt unfairly lowering
consumer credit scores.
A Data Science Perspective:
The Broad Impact of Removing Medical
Debt from Credit Reports
While the CFPB’s approach aims to protect consumers from the adverse eects of medical
debt, its policy change has unintended consequences that extend far beyond individual credit
scores. From a data science and credit modeling perspective, the removal of medical debt
fundamentally disrupts the predictive accuracy of credit scoring models in ways that may not
have been fully considered.
Credit models, including those used by nancial institutions and lenders, are built on hun-
dreds - if not thousands - of attributes derived from consumer credit reports. These attri-
butes are interconnected, meaning that even if an individual data point seems minor, its
removal can cause signicant ripple eects across various predictive factors. While the direct
removal of medical debt from a single credit report may appear to have a negligible eect,
the aggregated impact on model performance is profound.
01
02
03
Alteration of Key Risk Indicators: Attributes summarizing total debt, debt-to-income
ratios, and past delinquencies are often inuenced by medical debt. With medical
debt removed, these attributes shift, sometimes in ways that make consumers
appear less risky than they actually are. This can lead to misclassication of
creditworthiness, with lenders potentially extending credit to consumers who may
have previously been considered high risk.
Compounding Eects in AI and Data Models: Credit scoring and risk assessment
models rely on historical data patterns to make accurate predictions. When a large
percentage of attributes - potentially up to one-third - are suddenly altered, the
historical basis for these models becomes unreliable. AI-driven models, which are
particularly sensitive to shifts in input data, may produce inaccurate risk assessments
as a result.
Reduced Predictive Power for Lenders: Lenders depend on credit models to make
informed decisions about extending credit, determining interest rates, and assessing
risk. With medical debt removed, many traditional risk indicators have been
weakened, potentially leading to an increase in mispriced loans and higher default
rates over time. This could result in lenders tightening credit criteria in response,
which may inadvertantly harm the very consumers the CFPB aims to protect.
The CFPB’s policy change benets some consumers in the short term, however the long-term
consequences of disrupting credit models at scale could outweigh these initial gains.
Decision Science:
How We Are Adapting
The consumers most impacted by this change closely align with the target audience of
many of our clients. Consequently, selection models that rely on historical credit data
have faced signicant accuracy challenges. Models trained on past data were not designed
to accommodate the abrupt removal of medical debt as a reporting factor, resulting in
unforeseen deviations from expected outcomes.
To mitigate the disruptions caused by the sudden disappearance of medical debt from
reporting data, we have implemented a multi-faceted approach that includes:
Comprehensive Data Analysis: Conducting an in-depth examination of
post-change credit data to assess the broader impact on credit attributes
and scoring models.
Model Retraining & Renement: Updating and retraining our selection
models to align with the new consumer credit landscape, ensuring
continued precision in identifying qualied consumers
Infrastructure Enhancements: Expanding our server capacity to support
increased data processing demands, providing additional redundancy
and improving system resilience.
Restoring Model Accuracy &
Enhancing Predictive Capabilities
The recalibration of our selection models has allowed us to restore targeting accuracy to
pre-medical-debt-removal levels. By leveraging real-world performance data, we have rened
our predictive frameworks to maintain consistency in consumer identication, ensuring our
clients continue to see strong campaign performance.
We are now preparing to introduce a new generation of selection models built entirely on
post-removal credit data. These models will be optimized for the evolving credit reporting
framework, oering even greater precision and insight into consumer nancial behavior.
Looking Ahead
As the credit industry continues to evolve in response to regulatory shifts, adaptability is key.
The ability to pivot in real time, reassess underlying data structures, and rene predictive
models will separate industry leaders from those struggling to keep pace.
At Decision Science, we remain committed to ensuring our clients
stay ahead of these changes. By continuously monitoring credit
data trends, enhancing our modeling techniques, and investing
in infrastructure improvements, we position both our company
and our clients for long-term success in a transformed credit
environment.
For more information on how these updates may impact your
selection strategies, please reach out to our team.
Our Recent Publications:
Newburyport, MA
Email: info@decisionscillc.com
website: www.decisionscillc.com
Decision Science stands as a leader in data-driven direct mail marketing solutions, equipped
with cutting-edge machine learning technology and Franklin, our industry-leading AI. Through
advanced modeling, rigorous data testing, and real-world client insights, we enable clients
to achieve enhanced targeting precision, higher response rates, and sustainable campaign
growth. Our commitment to innovation, transparency, and industry partnerships ensures our
clients have a strategic advantage in navigating the competitive landscape of direct mail
marketing.
The Decision Science Advantage
