Does Your Underwriting Face a Systemic Risk?
Giving too much weight to FICO scores may pose major risks for lenders
By Jeffrey Feinstein, senior director of analytic strategy, LexisNexis | bio
Some mortgage-industry analysts have argued that there’s an inherent
flaw in lenders focusing on a single risk score when making their underwriting
decisions. Of course, these arguments often focus on problems that could arise
from a borrower’s FICO score being the
primary tool on which lenders rely. When the industry relies on a single
point of reference to judge a consumer’s
creditworthiness, lenders may expose themselves to systemic risk. Should this
single reference point prove inefficient or inaccurate, the entire underwriting
system could crumble.
If you take this line of reasoning a step further, the
systemic risk arguably doesn’t lie in an analytic solution like a FICO score,
but instead lies in the use of a single point of reference. Scores such as FICO
are based on credit-bureau data and use methodology that’s consistent with
those of other credit-bureau scores in the marketplace. These scores all share
a common data source. Consequently, the systemic risk that lenders face is
based on their use of a single data source; if the underlying credit-bureau data
fails, then all credit- bureau scores fail.
For example, some lenders may choose to suppress reporting
credit- limit information on a credit report for competitive reasons. The
thinking is that if credit card issuer No. 1 discloses a $10,000 limit on a credit
report, credit card issuer No. 2 may use that information to woo a consumer
away with an offer for a $12,000 limit.
To avoid this situation, credit card issuer No. 1 could suppress
reporting for its credit limit.
In the calculation of scores, however, credit limit is a key
component in the computation of credit utilization, which is a key component of
credit-bureau scores (visit myFICO.com
to learn more about this subject). The strategic decision to suppress credit
limits could adversely impact credit-bureau scores overnight, and in this case,
the effectiveness of any score by any vendor based on this data would be
It therefore makes sense to broaden the data used in your
credit decisions. By diversifying the
information in a decision, the impact and efficacy of any particular score
becomes less impactful to that decision.
There’s also a fair-lending implication to this argument. By
judging credit applicants by the same tools across many lenders, a consumer
could be consistently turned away from credit and could be essentially shut out
of the entire credit system. Given the increased scrutiny of fair-lending
practices, it behooves lenders to broaden their scope beyond this shared risk
and abandon the cookie-cutter approach of requiring a FICO score that exceeds a
certain cutoff. If borrowers are qualified and disqualified by the same score,
lenders will likely be unable to make refined decisions when it comes to giving
loans. A borrower who is rejected by one lender is likely to be rejected by all
lenders because the underlying credit data doesn’t change — in fact, that
underlying credit data indicates increasingly greater risk as inquiry after
inquiry piles up. The borrower’s only choices would be to seek out more
expensive credit or opt out of the credit system entirely.
Even more problematic is that overreliance on credit-bureau
data means that those without a credit footprint are essentially credit
invisible. That is, these potential borrowers are often unable to establish
credit because no mainstream lender is willing to underwrite somebody who can’t
be assessed by credit-bureau data alone. These credit-invisible consumers skew
toward underserved minorities such as black and Hispanic homebuyers.
The solution to these risks is for lenders to invite other
data sources into the decisionmaking process. For example, alternative data
such as public records, property ownership records and wealth may provide
additional information about a person’s creditworthiness. This data also adds
depth to decisions based on credit- bureau data by indicating high- and
low-risk behaviors outside of a
consumer’s wallet. Lenders could then
make more refined decisions based on the
combination of broader data elements in which credit bureau data is a piece of
the decision rather than the sole driver of the decision.
Alternative data also may serve as a replacement to credit-bureau data when such
data is unavailable. It may also help certain consumers — especially
underserved minorities — emerge from
their credit- invisible status. Many of these consumers could then open
affordable credit accounts, purchase cars and homes, or use a credit card with
In short, alternative data deserves to be a key part of
mortgage lenders’ decision-making strategies, as this data may provide
increased predictive value when used with a credit-bureau score, and enables
scoring of many consumers who can’t be scored using credit-bureau data alone.
This in turn may help reduce the systemic risk of underwriting campaigns across
Jeffrey Feinstein is senior director of analytic strategy at LexisNexis, where he’s responsible for developing the LexisNexis RiskView Credit Risk Score. He leads innovation efforts to bring new analytic solutions to the market leveraging LexisNexis data. Before joining LexisNexis in 2010, Feinstein developed expertise in credit data and received three U.S. Patents for his innovations in credit modeling. He holds a PhD from Ohio State University in experimental psychology and applied statistics. E-mail Jeffrey.Feinstein@LexisNexis.com.