The Mortgage Bankers Association gathered in Chicago last month, where its president, Bob Broeksmit, declared he is “increasingly confident that tri-merge won’t be around for long.” The Federal Housing Finance Agency, expected to follow with a two-bureau announcement, has since said only that it is “exploring” the idea.
Prophecy is a hard business.
The FHFA announced a move to two credit reports in 2022, confirmed it in 2024, and scrapped it in July 2025 under its current director. The standard has survived one death sentence already, and for a reason: lenders found they could not do without three reports, one each from Equifax, Experian, and TransUnion, merged into one file.
I spent my banking career at Los Alamos National Bank and Wells Fargo, and the lesson that stuck is simple: a mortgage rarely goes wrong at the closing table. It goes wrong in an underwriting file that was missing something.
Three files catch what one misses. Not every creditor reports to every bureau, and rent, utilities, and small-bank tradelines land unevenly across the three, as Patrick Brenner wrote in The Washington Times. The merge catches the debt one bureau never saw, which flatters a borrower, and the on-time history one never recorded, which shortchanges her.
The gaps are not small. A recent brief cited Andrew Davidson & Co.’s study of 245 million consumers: 35 percent had a bureau score at least 10 points off their three-score median. A mortgage is priced in tiers; in one modeled $350,000 loan, a one-tier move was worth $3,000 to $5,000. Under tri-merge, Fannie Mae takes the middle score, so no single bureau’s number sets the price on its own.
MBA wants the option to stop at one report when the first pull shows 700 or better. The prize: a tri-merge runs between $85 to $100, so skipping a bureau saves roughly $30 at closing. If the rate moved an eighth of a point, that would cost about $10,000 on a $350,000, 30-year loan held to maturity. Whether fewer reports would move the rate at all is the question FHFA has not answered, and $30 is a thin reason not to ask it.
What worries me most is not the lost data but what replaces it: choice. Once an originator can see several results and deliver the most flattering, the score reflects selection as much as credit history. Even Mr. Broeksmit acknowledges this: without enforceable guardrails, he wrote in April, his own reform could “trade one inefficiency for a new form of gaming.” His answer is a rulebook, no picking the highest score, deliver every report pulled, policed across thousands of lenders and millions of pulls. That is a new problem, invented to replace a standard that does not have it: three numbers, the middle wins, and no one has to be policed.
My colleagues and I have argued that buyers cannot compare homes they never see. Neither can a lender price a risk it never sees.
State capitols have worked the same problem from the consumer’s side. Albany, Springfield and Sacramento have removed medical debt from credit reports, and Boston plans to follow, deliberate edits made one category at a time, not an option to set two of three files aside. Oklahoma City and Harrisburg are debating whether credit information belongs in insurance pricing at all; a different question, same conviction that credit data deserves careful handling. And Sacramento tells its own residents: “Not all creditors provide information to all the bureaus.” Not a mortgage rule, but the same lesson: the files differ, and the differences land on real people.
None of this is a quarrel with the bureaus. Three independent files are why the merge works; the disagreement among them is the signal, not the defect. And the bureaus are making the file richer while the lobby argues for reading less of it. TransUnion pioneered trended data in mortgage underwriting a decade ago, has added rental and utility tradelines, and now prices its VantageScore 4.0 mortgage score at 99 cents through 2028, a score price, not a report price, but real. If the lobby wants to fight about cost, itemize the bundle: resellers assemble the merge, score providers supply the models, and lenders say report-and-score prices rose 40 to 50 percent this year. Find out whose line that is first.
Keep tri-merge as the default until FHFA publishes matched-applicant evidence that fewer reports produce the same approvals, pricing, and performance. If there must be a pilot, pick the bureau before anyone sees a score, deliver every report pulled, and publish results for first-time and thin-file buyers, the ones a single file could miss.
Mr. Broeksmit has been confident before; in February he called three reports “an anachronism.” Tri-merge is still here, convicted of costing $30 and never tried on what it saves. Before anyone carries out the sentence, let the evidence be heard. Lenders should not get to choose which version of a borrower’s credit history the loan is priced on.