Study Finds Wide Credit Score Gaps Across Bureaus Amid Tri-merge Debate

A study by Andrew Davidson & Co. found meaningful credit score discrepancies among the three major bureaus — a key data point for those arguing in favor of maintaining the tri-merge standard rather than shifting to a bi-merge or single-report model.

According to the paper, released Friday, 35% of the 245 million scored consumers in the dataset had at least one bureau score that differed from the traditional tri-merge result by 10 points or more. Another 18% had a variance of at least 20 points, while 7% saw differences of 40 points or more.

The three national credit bureaus — Equifax, Experian and TransUnion — provided Andrew Davidson & Co. with anonymized consumer data from October 2023, including VantageScore 4.0 credit scores. The sample represents the broader U.S. population and is not limited to mortgage borrowers or loans purchased by the government-sponsored enterprises (GSEs).

The research comes as a proposal to replace the longstanding tri-merge credit report with a single-file model has reignited debate over borrower costs, credit access and systemic risk.

“We are going through a modernization phase in the mortgage industry,” Sanjeeban Chatterjee, director of behavioral modeling at AD&Co, said in a statement. “At such times, it is important to understand the impact of the changes so that the stakeholders can make the right decisions. This study shows why knowing more is better from a risk management and affordability perspective.”

By credit score group

Score variances were most pronounced among lower-scoring borrowers. Among consumers with median scores between 600 and 639, 47% had at least one bureau score that differed from the tri-merge standard by 10 points or more. Another 26% saw differences of at least 20 points while 10.5% experienced gaps of 40 points or more. 

The study suggests that some borrowers could fall below minimum qualification thresholds. Roughly 30% of the 13 million scored consumers in the 620 to 639 range — including an estimated 240,000 GSE loans using VantageScore 4.0 within that band — could drop below a 620 cutoff under a single-score framework and potentially be denied a GSE mortgage.

For consumers in the 640 to 779 range, 40% had discrepancies of at least 10 points, 21% had differences of 20 points or more, and 8.1% had gaps of at least 40 points when compared with the tri-merge standard. 

Andrew Davidson & Co. noted that for a $350,000 GSE loan with a 90% loan-to-value ratio, moving between adjacent pricing tiers due to a score shift could raise or lower the combined cost of the mortgage and mortgage insurance by $3,000 to $5,000 in present value over the life of the loan.

Different proposals

Some proposals have suggested a 700 score cutoff to determine whether a tri-merge report would be required. But in the 700 to 779 range, the study found that 18% of the consumers still had at least one single-bureau score that differed from the tri-merge result by 20 points or more.

Under a hybrid model — in which the first pulled score of 700 or higher would serve as the representative score, while borrowers below 700 would still require a tri-merge — 4% of consumers in the 640 to 659 range and nearly 8% in the 660 to 679 range had a maximum bureau score of at least 700. That dynamic could shift some borrowers into different pricing or eligibility categories depending on which report is used, the study shows. 

Conversely, about 9% of all consumers — and 11% of those in the 640 to 779 range — could see their representative credit score increase by 20 points or more compared with the tri-merge standard, potentially improving pricing or eligibility outcomes.