Mortgage Notes: Getting Consumer Credit Scores Right

In this issue of The Institutional Risk Analyst, David Battany, Executive Vice President, Capital Markets, at Guild Mortgage Company talks about the disconnect between the desire to improve the use of credit scores in the world of mortgage finance and how consumer credit data is gathered and assembled into a score. Consumers and policy makers will probably be surprised to learn how imprecise the world of aggregating consumer data remains in 2026, almost 40 years since the first standardized consumer credit score, known as the FICO Score, was published in the United States in 1989. The fact that several new scores have been published by the aggregators of consumer data has not improved the situation appreciably. How can the consumer finance industry and the regulatory agencies get this right?

Getting Consumer Credit Scores Right

By David Battany

July 1, 2026 | The correct credit score for a borrower is obtained by gathering the data from all of their reported credit accounts and running all of this data through the most current credit score model. There is only one correct answer to the question of a borrower’s credit score, regardless of which credit model is used.

Obtaining three different credit score reports, each created based upon three different and incomplete credit account histories, will produce three different and three wrong credit scores. Averaging these three wrong scores will not create the correct credit score to predict a person’s probability of default.  

Assume a person has a history of 10 total credit accounts, and eight have perfect pay histories, one has a few late pays, and one has several late pays or no pays.  Late payments are not the only factor that determines if a particular credit history has a “good” or “bad” impact to a person’s overall credit score.  

The length of time an account has been open, the account balance and many other factors determine how a specific credit history has a good or bad impact on a person’s overall total credit score. Let’s assume for the sake of this discussion that a sample person has eight “good” and two “bad” accounts with respect to all factors evaluated by a credit score model. Assume the creditors for all 10 accounts each report to at least one credit bureau. 

Today when a lender orders what is known as a “tri-merge” credit score report, it is highly likely that not all three bureaus are getting the data from all 10 reported accounts. Tri-merge credit report rules are primarily set by the Federal Housing Finance Agency (FHFA) for conventional loans, and the Department of Housing and Urban Development (HUD) for FHA and VA loans. These government entities mandate that lenders pull data from all three major bureaus to ensure standardized underwriting.

If all three bureaus had identical data, we would know this because they would all produce nearly identical credit scores. The current tri-merge rules give no incentive for any bureau to get all data from all 10 accounts and arguably provides a disincentive to do so. If each bureau had a complete data set, each would produce effectively the same and correct credit score, so there would no longer be a need for a tri-merge report.

Each bureau today uses slightly different versions of FICO Classic, so there could be single digit score deltas between the three scores produced by the three bureaus. Assume in this hypothetical scenario that bureau A gets data on nine of the 10 accounts, bureau B gets data on eight of the 10 accounts and bureau C gets data on seven of the 10 accounts held by a consumer.  

---

If each bureau had a complete data set, each would produce effectively the same and correct credit score, so there would no longer be a need for a tri-merge report.

---

As each bureau is running a credit score on a prospective borrower, if they happen to pick up the two accounts with bad histories, that means the consumer is not getting any positive offset from their other 1- 3 accounts with positive histories that are not picked up by the bureau. 

By not including their 1-3 accounts which had perfect pay histories, this borrower’s score is derived from this incomplete history and is incorrectly lower than their true correct credit score. On the flip side, if any of the bureaus miss one or both or the two bad accounts, then they are reporting credit scores incorrectly higher than a borrower’s true correct credit score, by not including their 1-2 accounts which had bad pay histories.

When we see a tri-merge report with three very different scores, we know for a fact that at least two of the three scores are wrong, and possibly all three are wrong.  However, we have no idea which ones are wrong and whether they are wrong to the high side or low side.  

The average of three wrong scores does not bring us to the correct score, and the average of the three wrong scores could be materially off the borrower’s correct score.  What is the “correct score?” The score we would see if all 10 of their reported accounts were run through the same credit score model.  

To generate an accurate credit score, we need all 10 of the borrower’s accounts – their complete reported credit history – to all be run through the credit score model in order to know their correct credit score and their most accurate prediction of their probability of default. 

The evidence of the above is plainly visible when we pull a tri-merge report and see a 40-80 point range in the scores from the three bureaus for the same borrower on the same day.  Roughly speaking, every 40 points of FICO score has historically translated to about a doubling of expected future default rates.  

If we see a score range of 80 points, one bureau is telling us a borrower is about  4X as likely to default as another. This is a material gap, and averaging the scores does does not fix the math error of three incorrectly calculated credit scores if some or all are based upon incomplete reported credit history data. 

Simply put, any time a credit bureau is running a credit score utilizing a person’s incomplete credit history, meaning they are not looking at all of their reported credit accounts, the credit histories they are missing would likely have had a positive or negative impact on the person’s score. In the cases of 40-80 point score variance between the three credit data bureaus, the impact of not running a score on the complete reported history is material to the borrower, the mortgage lender, and the credit risk holders behind the mortgage.  

Our industry and home buyers rely upon credit scores to be accurate predictors of default. A score determines the possibility of a person being approved for a loan, the type of products they are eligible for, and the interest rate and points they must pay for the loan.

The credit score also impacts the value of a mortgage servicing right or MSR, in some cases the value of the MBS, and the amount of capital a GSE or private mortgage insurer or MI would need to hold to offset their future expected credit losses.  There is no excuse to not get credit scores right.  Using partial data to run a score does not produce the correct score.

The solution to the above is straightforward.  We should modify our current process of having the three credit bureaus, each running three partially complete sets of consumer data, knowing each have some amount of overlap, through a model which is highly sensitive to the data from every single credit history, and then averaging the three poorly calculated credit scores, incorrectly thinking this average is the correct score.

Instead, we should gather all of the reported data from all three bureaus, eliminating duplicates, giving us one complete reported credit history data set, and then run this single complete data set through the credit score model, one time. The resulting single credit score will be the person’s correct credit score.  It will be much more predictive of a person’s default than the current process of averaging three incorrect credit scores derived from three incomplete data sets.

As our industry moves to more modern credit score models such as FICO 10T and Vantage 4.0, for us to benefit from the improved ability to predict default risk, we need to change how we feed the data into these new models. A borrower’s credit score should always be determined by analysis of all of their reported credit histories, not by the random good or bad luck that happens in today’s tri-merge process where each bureau is very possibly, and perhaps likely, calculating scores on an incomplete credit histories. Incomplete consumer data is unfair to borrowers and results in incorrect credit scores, which understate or overstate a person’s true probability of default.

Today the mortgage industry is playing a stupid game of Russian Roulette, but this situation is so easy to fix.  As an industry that serves the needs of consumers, we should want a complete data set of all of a person’s reported credit history and for it to be assessed as one complete credit file through the best available credit score model. It does not matter which credit score a lender uses if the data is not complete.  We have a legal and moral obligations to get this right. 

The Institutional Risk Analyst (ISSN 2692-1812) is published by Whalen Global Advisors LLC and is provided for general informational purposes only and is not intended for trading purposes or financial advice. By making use of The Institutional Risk Analyst web site and content, the recipient thereof acknowledges and agrees to our copyright and the matters set forth below in this disclaimer. Whalen Global Advisors LLC makes no representation or warranty (express or implied) regarding the adequacy, accuracy or completeness of any information in The Institutional Risk Analyst. Information contained herein is obtained from public and private sources deemed reliable. Any analysis or statements contained in The Institutional Risk Analyst are preliminary and are not intended to be complete, and such information is qualified in its entirety. Any opinions or estimates contained in The Institutional Risk Analyst represent the judgment of Whalen Global Advisors LLC at this time, and is subject to change without notice. The Institutional Risk Analyst is not an offer to sell, or a solicitation of an offer to buy, any securities or instruments named or described herein. The Institutional Risk Analyst is not intended to provide, and must not be relied on for, accounting, legal, regulatory, tax, business, financial or related advice or investment recommendations. Whalen Global Advisors LLC is not acting as fiduciary or advisor with respect to the information contained herein. You must consult with your own advisors as to the legal, regulatory, tax, business, financial, investment and other aspects of the subjects addressed in The Institutional Risk Analyst. Interested parties are advised to contact Whalen Global Advisors LLC for more information.