Mortgage Assumption in Divorce: CA, MD & VA Laws

Every discussion of divorce and mortgage assumption runs into the same wall, and almost nobody says so out loud. No source in the United States records why an assumption occurred. Not FHA, not VA, not USDA, not Fannie Mae, not Freddie Mac, not Ginnie Mae, and not the Home Mortgage Disclosure Act. There is no reason code. A divorce-triggered assumption and a sale-triggered assumption are indistinguishable in every dataset that exists.

The gap is not an oversight at the margins. It is structural, and it has three separate causes.

Ginnie Mae cannot count them either, for a reason that is almost elegant: an assumed loan never leaves the pool, so no removal-reason code is ever generated. And the enterprises publish loan purchase and performance data without publishing assumption volume at all.

The industry has said this plainly. In December 2022, the Mortgage Bankers Association wrote to FHA and VA that “data on volume of assumptions remains scarce.” Nearly four years later, it still is.

That last row is the one to read twice. The words release, liability and novation appear in none of the three statutes. Each compels the availability of an assumption. None discharges anyone from the note.

Two features of the California statute deserve attention, and neither is widely reported.

It excludes the loans that motivated it. The Senate Banking analysis describes the bill as a response to mortgage rates rising from under three percent in 2021 to roughly seven percent in 2024. The enacted text then reaches only loans originated on or after January 1, 2027. Every legacy low-rate loan, the entire population the bill was written about, sits permanently outside it.

It has no enforcement mechanism, and the Legislature knew. The Senate Banking Committee analysis states that the bill “does not provide an enforcement mechanism” and suggested the author consider working with the Judiciary Committee to craft a narrow enforcement provision. No such provision was added.

Maryland is the only one of the three that reaches loans already in existence, which makes it the only one that touches a low-rate loan written before the statute. It is also the only one with an internal tension in its own applicability clause.

Section 2 of the Act contains two sentences that do not sit comfortably together. One deems conventional home mortgages exceeding the FHFA lending limits, entered into before the effective date, to include an assumption provision. The next states that the operative subsections “shall be construed to apply retroactively and shall be applied to and interpreted to affect any conventional home mortgage loan entered into prior to the effective date of this Act.” One sentence is limited to loans above the conforming limit; the next is general. A practitioner relying on Maryland retroactivity for a conforming-balance loan should read both sentences before relying on either.

Maryland's disclosure duty also runs earlier than Virginia's: the provision must be disclosed in writing to the applicant before completion of the loan application, not after.

Virginia's is the tidiest drafting of the three, and it was added to a section that has existed since 1982. Subsections (A) through (D) already gave a Virginia homeowner the right to demand, in writing, whether a holder will permit assumption, and to receive an answer within ten business days. The 2026 amendment converts a disclosure regime into a mandate, for loans secured going forward.

Note also that all three statutes define “conventional home mortgage loan” to exclude loans insured or guaranteed by the federal government. FHA, VA and USDA loans are governed by their own assumption rules, covered next, and are untouched by any of this.

Most of what practitioners believe about divorce and mortgage assumption comes from Garn-St. Germain, and most of what they believe about Garn-St. Germain is wider than the statute.

What Garn-St. Germain actually does is narrower and still valuable: its subsection (d) exemptions bar a lender from accelerating on certain transfers, including a transfer to a spouse or child, and a transfer resulting from a decree of dissolution, legal separation or incidental property settlement where the spouse becomes an owner. That protects the transfer of title. It does not compel an assumption, and it does not release anyone.

Then there is the fact that reframes the entire discussion for conventional loans.

This is precisely the gap the three new statutes address, and it explains their shape: they operate by requiring the assumption provision to be written into the loan in the first place, going forward, rather than by compelling a servicer to permit something the note does not allow.

The CFPB's 2016 servicing amendments require a servicer to treat a confirmed successor in interest as a borrower for communication and loss-mitigation purposes. The Bureau's own commentary makes the limit explicit: a servicer “may not require a confirmed successor in interest to assume the mortgage loan obligation under State law to be considered a borrower.” That is the converse of a right to assume. These are disclosure and communication rules. They do not move the debt.

The sentence above is about the note. It is not the end of the analysis, and treating it as the end is the most common error in this subject after conflating assumption with release.

A conventional note may contain no assumption provision. The party who would object to an assumption anyway is not the note; it is the noteholder. On a loan sold to Fannie Mae or Freddie Mac, that is the GSE, and both of them publish what their servicers must do when a divorce transfer arrives.

The authority for this is not the note. It is the servicing contract. Garn-St. Germain removes the noteholder's right to accelerate on a divorce transfer; the investor then authorizes its servicer to substitute the obligor by written agreement. It is a novation permitted by the noteholder, which is why no assumption clause is required. The party who would otherwise object is the party granting permission.

Two limits on that, both easy to miss. Fannie's separate creditworthy-purchaser route at D1-4.2-02 reaches adjustable-rate loans and a small set of fixed-rate portfolio loans purchased under commitments dated before November 1980. It does nothing for a modern conventional fixed-rate divorce case, which runs entirely through the exempt-transfer channel above. And where the loan has already been modified, or is delinquent, the servicer must go to the GSE rather than decide alone.

That sentence is the strongest language either guide contains on this subject, and it is conditional in four places. Anyone quoting it to an attorney should be able to name all four.

Two loan conditions take the file off this track entirely. A delinquent loan, or one already modified, requires Freddie Mac's own approval under 8406.1(a)(ii) and the simultaneous assumption and modification provisions at 9207.1(b).

None of this applies until the owner of the loan is known, and the servicer on the statement is rarely the owner. Fannie Mae and Freddie Mac each publish a loan lookup tool, and the Mortgage Electronic Registration Systems database covers loans registered there. Where those come back empty, Regulation X provides the faster route: a request for information asking the identity and contact details of the owner or assignee must be answered within ten business days, at 12 C.F.R. §1024.36(d)(2)(i)(A). That one is a consumer right, with a deadline, and it is enforceable in a way the guides are not.

The published part is the credit side. Both guides set out documentation, qualification standards and fees. The fee schedules are identical: $100 where no credit review is required, otherwise the greater of $400 or one percent of the unpaid balance, capped at $900, plus actual third-party costs. Freddie expressly prohibits yield-enhancement and lost-opportunity fees as a breach of its guide. Freddie's five percent contribution requirement does not apply to a divorce transfer, which is carved out at 8406.2(b)(iii). Neither GSE permits a rate change on assumption outside a simultaneous modification.

The unpublished part is everything about delivery. Freddie requires the servicer to notify the transferee of the decision within twenty-five days of a completed application, and that is the only turnaround standard either GSE publishes. Fannie publishes none at all. Neither publishes an intake route, a staffing expectation, an escalation path for the transferee, or any approval-rate reporting. And neither imposes a duty to tell a divorcing borrower that this pathway exists. The guides govern what a servicer must do once asked. They say nothing about what it must volunteer.

It is widely repeated among practitioners that conventional servicers have become more open to divorce-incident assumptions over the past two or three years. We could not verify it. There is no GSE statement, no regulator finding, no trade association survey and no servicer disclosure supporting it.

The one adjacent development is real but narrower than it sounds. Freddie Mac added assumption underwriting to Loan Product Advisor in August 2024, which lets a servicer underwrite an assumption through the automated system instead of manually. That lowers the cost of processing one, which is the right mechanism. Freddie itself declined to characterize it as a loosening, stating that it is “not an expansion or change to Freddie Mac's mortgage assumption requirements.” Fannie has issued no counterpart.

The evidence pointing the other way is more direct. The CFPB's issue spotlight of December 2024, published eight months after that capability went live, documents servicers pressing homeowners to refinance at current rates rather than assume, waits of months or years, and repeated requests for documents already supplied.

And the claim cannot be settled either way, because the GSEs publish no conventional assumption volume at all. FHA and VA release counts. Fannie and Freddie do not. So a practitioner who believes servicer behavior has improved, and one who believes it has not, are both arguing from anecdote about a population nobody measures. That is the same hole this report opened with, in a different wall.

A quitclaim deed conveys the departing spouse's interest in the property. It has no effect whatsoever on the note. The client who quitclaims without obtaining a release has given away the asset and kept the entire liability, which is the worst of the available positions and the one consumers most often report landing in.

The Consumer Financial Protection Bureau publishes every consumer complaint whose narrative the consumer consented to release. Querying that database on August 27, 2026 produces the only public time series that tracks this problem at all.

Three Things This Shows

Successor-in-interest complaints doubled, then doubled again. Seventy in 2024, 150 in 2025, and 150 already by August of 2026 with four months still to run. As a share of narrative mortgage complaints, they went from 1.07 percent to 2.07 percent, so this is not simply an artifact of rising complaint volume overall.

Assumption complaints rose roughly fourfold from the trough, from 78 in 2020 to 328 in 2025, tracking precisely the period in which an existing low-rate loan became something worth fighting for.

Complaints that merely mention divorce are flat to falling, from 338 in 2016 to 228 in 2025. The growth is specific to the assumption and successorship mechanics. It is not a story about more divorce. It is a story about the same divorces colliding with a rate environment that changed what the house is worth keeping.

No fourth state has enacted a divorce assumption mandate, and no bill on the subject is pending in the 119th Congress. In the forty-seven remaining states the position is the one set out in section four: Garn-St. Germain prevents acceleration on the transfer, the note governs whether an assumption exists at all, and for most conventional fixed-rate loans it does not.

How statutes were verified. Each of the three was read in its enacted form rather than in secondary coverage. California Civil Code §2951 was read as codified. Virginia §6.2-419 was read on the Virginia Law site, where the amendment history records “2026, c. 962”; note that at least one major secondary code service still displays only subsections (A) through (D) and does not yet show the 2026 amendment. Maryland was read from the enacted chapter text of Ch. 202 (2025).

How the complaint series was built. The CFPB Consumer Complaint Database was queried on August 27, 2026, filtered to the Mortgage product. The search term matches the complaint narrative field only, so the correct denominator is mortgage complaints with published narratives, not all mortgage complaints. The term “assumable” was excluded because search stemming collapses it with ordinary uses of “assume” and inflates the count.

What we could not establish. No agency, enterprise or commercial source records the reason for an assumption. No state among the three collects data on assumption requests, approvals, denials or timelines, and none of the three statutes creates a reporting mechanism. California will produce no data for years by construction, since it governs loans originated from 2027 forward.