TransUnion

On Oct. 1, Bloomberg reported, citing an unnamed source, that the Federal Housing Finance Agency is moving toward a bi-merge requirement, under which government-conforming mortgages would need only two credit files, from the current three. TransUnion, Equifax, and Fair Isaac shares fell after hours.

Why it matters: We think whether the bi-merge gains traction will turn on the details. Lenders currently use the median of the three credit scores. Under a bi-merge, it is unclear whether the average or the lower of the two scores would be used. We think this methodology matters and will change behavior. In other words, if policymakers do not adopt the lower of the two, they could invite "bureau gaming." As with VantageScore versus FICO, we think the greater risk of lower volumes is at the super-prime end. For example, if a borrower seeking a GSE loan has a TransUnion VantageScore of 803 and an Experian VantageScore of 807, pulling an Equifax VantageScore or a FICO score from either bureau will likely not produce a better outcome, because the top band starts at 780.

The bottom line: We plan to lower our fair value estimates for Fair Isaac, Equifax, and TransUnion by a single-digit percentage. Mortgage-related credit revenue is roughly 40% of Fair Isaac's total, versus less than 15% for Equifax and TransUnion. We see Fair Isaac, with its recently approved direct license model, as better-positioned to hold its economics versus the bureaus, and we expect it to introduce a pricing model that is agnostic to the number of bureaus used. That said, it has greater mortgage exposure than the bureaus, and transitioning to new fee models may not be seamless.

Bears say: The Rocket Mortgage announcement of a preferred credit score indicates that some lenders may adopt a waterfall in which they pull only FICO if they do not like the VantageScore result. Lenders may do the same with the bureaus, pulling files from two bureaus first and then deciding whether to pull a third.

FHFA Director Bill Pulte had long indicated lowering the number of bureaus required. The Bloomberg article said the bi-merge could be announced as soon as Oct. 12, when Pulte is scheduled to speak at a mortgage industry conference in Chicago.

For 2027, we expect Fair Isaac to shift to emphasize a performance-based pricing model that is agnostic to the number of scores pulled and the number of bureaus used. We also think the firm may charge different amounts for different loan sizes. Against its current price of $10 per score, we think the likely range of outcomes is an increase of 0% to 10%, or $10 to $11 per score. Because less than 5% of Experian's revenue is mortgage-related, we see Experian as relatively unaffected.

On Sept. 30, Pulte said in a post on X that "FICO Direct is approved on our end." The direct license program is Fair Isaac's preferred route to a performance-based model, under which it would charge a higher fee on scores for closed loans and a lower fee or no fee on scores for loans that do not close. We believe a performance-based model should be more attractive to lenders, as it better aligns economics with successful loan outcomes and allows a greater portion of the cost to be passed through to consumers at closing. That said, adoption is not guaranteed, particularly if lenders view the model as adding complexity. Our understanding is that under the DLP, a lender must pay for the credit file only once if it wants both FICO and VantageScore.

Aside from TransUnion extending its $0.99 VantageScore pricing through the end of 2028, neither Fair Isaac nor the bureaus have released additional pricing details for credit files or scores. We think it will be hard for the bureaus to raise VantageScore prices (to make up for a bureau volumes shortfall), as each bureau prices its VantageScore independently.

Regarding Rocket's Sept. 29 announcement, we would be cautious about extrapolating too much from its press release. First, we think it is entirely possible that the non-GSE market finds the better-of VantageScore and FICO framework confusing and therefore does not adopt it. Second, bank lenders may be more risk-averse about moving away from FICO for other purposes, such as evaluating mortgage servicing rights. Overall, we feel comfortable with our model, which assumes that Fair Isaac's market share declines to 85% by fiscal 2030. The biggest risk we see is in the top 25% to 40% of the market (super-prime borrowers). Because these borrowers already receive a top score, an additional score would add little from their perspective. However, moving to a waterfall creates a fragmented workflow and generally requires lenders to purchase the credit file twice.

We still think that in many instances it will make sense for a lender to pull both scores (and, depending on the details of the bi-merge, all three bureaus), given the savings from landing in a better band. To illustrate this, we used NerdWallet's tool to obtain mortgage quotes for a $500,000 home purchase from a single lender, holding the points fee constant. Being in the 740-759 FICO band rather than the 720-739 band would save a consumer roughly $2,700 over five years (on a 30-year fixed-rate mortgage with a 20% down payment). The stakes are higher for mortgages with a down payment below 20%, which requires mortgage insurance. 

While we think “score gaming” is real, we are starting to object to the “score gaming” moniker. We believe the rules are the rules, and following the GSE rules as they are intended are not “gaming.” For example, the IRS lets married couples pick between married filing jointly and married filing separately and pick the one with the lowest tax bill. By this logic, couples who do this are “tax gaming.” Fair Isaac and others in the industry have characterized lender choice as score gaming while VantageScore and the credit bureaus have sought to minimize score gaming, with VantageScore referring to it as a “myth.” We view VantageScore’s position as illogical, and while Fair Isaac and others have a point with gaming, we think at some point the rules are the rules. We believe over time, it is reasonable to assume that the GSEs will simply raise loan-level price adjustment matrices to adjust for this. Said differently, we think that over time a 700 FICO under the new lender choice system will be seen as less favorably (a higher credit risk) than a 700 FICO under the prior FICO-only system because the 700 FICO in the new system is conditional on it having a worse VantageScore.

We are increasing our fair value estimate to $86 from $85, primarily due to the time value of money and minor model adjustments. Our discounted cash flow-based fair value estimate equates to about 18/16 times our 2026/2027 adjusted earnings per share estimates when excluding stock-based compensation, and around 20/17 times our 2026/2027 adjusted EPS estimates when stock-based compensation is included.

We expect TransUnion's revenue to compound at an organic annual growth rate of about 8% over the next five years. Mortgage volumes are about 13% of current revenue, and with home purchase volumes near historical lows, we expect a modest rebound after 2027. We model TransUnion’s mortgage revenue (excluding FICO royalties and any upside from VantageScore) growing on average by about 10% per year over the next five years. Compared with Equifax, TransUnion generates a smaller portion of its revenue from mortgage-related inquiries, which can be lumpy. We expect nonmortgage US financial services to grow by about 5% per year from growing volume, pricing, and increased use cases in areas such as fraud and identity solutions. We expect emerging verticals to likewise grow in the 5% to 7% range, in line with historical averages. TransUnion’s US consumer business has underperformed peers such as Experian, and we take a cautious view and thus only model 1% growth. Excluding the impact of consolidating TransUnion de Mexico, we model about mid- to high-single-digit growth per year in the firm’s international business, a bit below the average of 12 % from 2018 to 2025. We expect growth to reaccelerate in India, given the size of the market opportunity, but expect slower growth in the firm’s developed markets.

We don’t expect much margin expansion. We forecast adjusted EBITDA margins of 35.9% to 35.9% in 2030 versus 36.0% in 2025. However, part of this is due to how we model pass-through FICO royalties in mortgage. Excluding the impact of FICO royalties on US mortgages, we forecast 2030 adjusted EBITDA margins of 39.4% compared with 37.5% in 2025. We peg TransUnion’s weighted average cost of capital at 8.7%.

We view TransUnion as a narrow-moat business with an intangible moat source due to its proprietary data. While we see TransUnion’s data as highly proprietary, we do not see the firm’s return profile as high enough to warrant a wide economic moat rating.

TransUnion’s credit bureau operations provide data to lenders and other institutions, including credit history, current credit status, payment history, address, and other identity information. TransUnion’s data is critical to customers' underwriting decisions and other workflows such as identity and fraud. The price of its services is small in comparison with the dollar amount of lending volume. For example, approximately $1.3 trillion in revolving (that is, credit card) debt is outstanding, and we estimate that US card revenue from all three bureaus is less than $1 billion, which equates to single-digit basis points. Because the accuracy and completeness of data are critical to credit decision-making, lenders often pull from more than one credit bureau (particularly in scenarios where they have incomplete information), and we do not believe pricing is the primary factor for choosing a credit bureau.

We believe the barriers to entering the credit bureau business are high, as replicating a database of millions of customers would be incredibly difficult. Because credit bureaus’ data is based on the voluntary reporting of thousands of financial institutions, it’s unlikely that a startup, particularly given heightened data security concerns, could convince banks to share consumer information.

Mortgage-related revenue makes up just over 10% of TransUnion’s revenue. Government agencies such as Fannie Mae and Freddie Mac require a credit report from all three bureaus, which is known as a “tri-merge” report. Currently, the majority of mortgages originated in the US are conforming and thus using TransUnion (along with EFX and EXPN) is essentially a requirement. While there is some talk about them going from a mandatory tri-merge to an optional bi-merge, this has been delayed indefinitely and has seen opposition from GOP senators. In addition, there is some speculation about Fannie Mae and Freddie Mac exiting conservatorship, however, the tri-merge requirement predates conservatorship. That said, we cannot rule out the risk entirely and in such a case TransUnion would be forced to compete more aggressively (including on price) with Expeiran and TransUnion.

About 45% of TransUnion’s US Markets revenue is from the core use cases of auto, mortgage, card, banking, and consumer lending. Over the years, TransUnion has grown its emerging verticals unit, which is now 35% of US revenue and includes other use cases for credit bureau data such as insurance, tenant/employment screening, telecom companies, marketing, and so on. And just under 20% of TransUnion’s US business is the consumer business, which is primarily through indirect channel partners such as the use of TransUnion’s credit bureau data provided to Credit Karma customers. We view the consumer segment as a natural extension of its traditional credit bureau operations and another path to monetize the moat surrounding legacy operations.

About one-quarter of the firm’s revenue comes from operations outside the US. While the competitive position of the company’s international operations varies a bit based on its relative size, international markets tend to echo the oligopolistic structure seen in the US, with generally only a few viable competitors in each country.

We also see little risk of artificial intelligence disruption. The core data sets of TransUnion are proprietary, and while TransUnion hasn’t given a statistic on how proprietary its data is, competitor Equifax views 90% of its revenue as being generated from proprietary data.

From a quantitative perspective, TransUnion’s returns on invested capital over the last five years (2021 through 2025) were 6.2% including goodwill and 15.0% excluding goodwill by our calculations. These ROICs were not strong enough, in our view, to warrant a wide rating. We peg TransUnion’s weighted average cost of capital at 8.7%.

After guidance misses and other issues, TransUnion's shares are cheap relative to peers and thus offer more room for upside.

Although the ramp-up could take many years, TransUnion has a leading position in India, the one emerging market that we believe has long-term revenue potential in line with the US.

TransUnion’s growth of its emerging verticals segment (insurance, public sector, and so on) means less sensitivity to the macro environment versus the Great Recession.

The majority of TransUnion's revenue is transaction-based and so growth is primarily driven by volume. Mortgage volumes have disappointed in recent years. Regarding mortgage, the move to a bi-merge from a tri-merge in US mortgages could result in lower volumes and more price competition.

Management’s predilection for M&A could ultimately dilute the economic moat surrounding the core business.

The firm's relatively mature offshore footprint could limit incremental margin expansion from further geographic workforce optimization.