FHFA Cut Its Own Watchdog by 80% — and the Capital Rules That Price Mortgage Credit Still Have No Date

In a report covered by Reuters on Oct 30, 2026, the FHFA's Office of Inspector General says the agency refused to provide records during at least three reviews of its supervision of Fannie Mae, Freddie Mac and other housing lenders, describing the obstruction as unprecedented. One of the blocked reviews concerned FHFA's supervision of a lender tied to multifamily loans that defaulted in June 2024. The inspector general's general counsel invoked the watchdog's statutory right of access to \"all records\" before FHFA cut the office's funding.

This is not a paperwork squabble. FHFA sets GSE capital floors, approves counterparties and administers Ginnie Mae, so the office that would audit the regulator's own conduct is the office being stripped of staff and money. Congress is already circling: a Senate Banking Committee letter from Sen. Elizabeth Warren dated Oct 6, 2026 presses the director over what she describes as FHFA's failure to comply with GAO information requests, and three senators had already asked the GAO to probe how mortgage-fraud referrals were handled.

The budget document is where the damage is quantified. In its Management and Performance Challenges for Fiscal Year 2027, published Oct 14, 2026, the inspector general says FHFA's funding reduction essentially eliminates its capacity to prosecute multifamily fraud and forces discontinuation of essentially all criminal investigations. Without that investigative capacity, the OIG says FHFA cannot develop the evidence needed to prosecute fraudsters at all — the regulator does not need to break the law to fail at its job, only to stop investigating it.

The gap is not theoretical. FHFA's 2025 Report to Congress said both Enterprises lacked adequate capital for their business models and did not meet the minimum regulatory capital requirements FHFA had set, and the OIG's own memo repeats it. Meanwhile, the counterparty books are growing: the Enterprises' combined multifamily portfolios reached almost $1.05 trillion as of Jun 30, 2026, on $150 billion of new business volumes in FY2025.

The capital proposals themselves are specific. The Fed, OCC and FDIC proposed on Mar 19, 2026 to set risk weights of 25% to 75% on residential mortgages under the standardized approach and 20% to 70% under the advanced approach, and to apply a 250% risk weight to every mortgage servicing asset held by banks outside the largest categories. Comments closed Jun 18, 2026, and the agencies set no overall implementation timeline; observers writing the same week expected final rules to arrive no earlier than late 2026 or 2027.

The agency that wrote the floors has already conceded that its charges fail them. FHFA's 2025 Report to Congress said both Enterprises lacked adequate capital and missed its minimums, and the inspector general's October memo repeats the finding in its FY2027 challenges. That is the gap an auditorless regulator is left to police alone: the entity setting mortgage capital floors is also the entity certifying that its own charges do not meet them.

One clarification matters for anyone trading this story. The scored-mortgage reforms are not the slow part. FHFA and HUD announced VantageScore 4.0 and FICO 10T implementation for FHA, Fannie Mae and Freddie Mac loans on Apr 22, 2026, the Enterprises opened VantageScore 4.0 to every approved lender on Sept 9, 2026, and FHA agreed to accept VantageScore 4.0 collateral from Jan 1, 2027. Discretion at this agency is fast when it is pushing a competitive narrative and slow when it is writing capital floors.

The mechanism runs down the chain. A score toll sits at the top, the guarantee and the capital floors sit in the middle, and private insurance, nonbank origination and housing sit at the bottom. When the regulator that writes the middle rules is also the regulator that cannot be audited, the uncertainty shows up as price, not as headlines: how much insurance a lender must buy, how much capital a nonbank must hold, and what a servicer is paid to keep a file clean.

One mix shift is easy to miss. Ginnie Mae's July 2026 market analysis shows VA loans packaged in Ginnie Mae MBS passing FHA production, at 48.6% of 2026 year-to-date volume as of June 2026 — and VA loans carry no private mortgage insurance. That moves Ginnie Mae's growth mix away from private-MI business, so a Ginnie Mae volume headline is a weaker read on mortgage insurers than it looks.

Two insurers, two different stories. At MGIC Investment, revenue was essentially flat across three fiscal years and trailing net income of $707.9M sits below FY2025's $738.3M, with the per-share line propped up by a falling share count. NMI Holdings is the growth name: revenue has climbed from $579.0M in FY2023 to $730.5M on a trailing basis, with trailing diluted EPS of $5.08 against $3.84 three years earlier.

Demand has a mechanical source: the Enterprises updated PMIERs effective Sept 30, 2026, and those are the requirements that set how much private mortgage insurance a lender must buy on a guaranteed loan. The counterweight is equally mechanical — FHFA says it lacks statutory authority to supervise mortgage insurers and reinsurers directly, and it suspended 76 counterparties in FY2026 against 6 in FY2025, so the layer below the guarantee is being policed harder by the agency while the layer above it is being audited less.

Nonbank economics have recovered hard: Rocket Companies posted net losses of $103.8M, $1.8M and $123.9M in the first three quarters of 2025, then $297M and $230M of net income in the first two quarters of 2026, with trailing twelve-month net income of $1.30B on $11.16B of revenue. But a nonbank's economics are a license, not a balance sheet, and the license comes from the same agency whose watchdog says it obstructed three reviews. GAO's February 2026 review of Ginnie Mae and FHFA monitoring of nonbanks closed with four recommendations, including procedures to assess the reliability of the nonbank data FHFA uses for monitoring.

The asymmetry is the risk. If a nonbank's data is wrong and the supervisor is slow, a single fraud finding can end a franchise in a way it cannot end a bank's. That is the mechanism by which an auditorless FHFA eventually reaches a mortgage insurer's balance sheet, and it is why the 76 suspended counterparties in FY2026 matter more than the headline.

The name collision is the governance story. Bill Pulte founded PulteGroup in 1950 and, as FHFA director, appoints himself chairman of both GSE boards. Inside the company, the third quarter of 2026 showed a builder with volume and margin pressure — revenue of $3.97B, down 9.8%, closings of 6,876, down 9%, net new orders of 6,208, down 6%, and gross margin of 26.9% against 27.8% a year earlier — and a mortgage arm that originated $1.92B of loans at an 84.9% capture rate on a 750 average FICO, generating $107.0M of revenue and $45.6M of pre-tax income.

It is worth being precise about where FHFA discretion is actually being used. The agency ships its deadlines while the capital rule has none: VantageScore 4.0 and FICO 10T were implemented for FHA, Fannie Mae and Freddie Mac loans on Apr 22, 2026, all approved lenders gained access to VantageScore 4.0 on Sept 9, 2026, FHA agreed to accept VantageScore 4.0 collateral from Jan 1, 2027, and FHFA released new pricing built on the alternative credit-scoring approach on Sept 30, 2026.

That contrast is the practical answer to \"does agency dysfunction matter?\" It matters most where discretion is exercised without verification, which is capital, counterparties and issuer eligibility — and least where a rule was published, dated and shipped. The score toll is priced by competitive grids; the capital floors are priced by an agency that has already admitted its own charges fail them.

The investable conclusion is a barbell. Own the fee pools whose economics keep moving — private mortgage insurance, where PMIERs effective Sept 30, 2026 set volume and the profit pools are already visible, and nonbank origination franchises that have just converted three quarters of losses into $527M of half-year profit. Be cautious on the balance sheets whose risk is set by a regulator nobody can audit: GSEs that FHFA itself says are below minimum capital requirements, and insurers sitting behind counterparties that FHFA admits it cannot supervise directly.

What this story is not is a rate call. The mortgage coupon is set by the Treasury curve and by a capital rule that has no date; what the obstruction report changes is the discount applied to the fee streams that sit behind the coupon — insurance, servicing, origination and score tolls. For PulteGroup specifically, the founder now runs the agency, which is a governance fact rather than an earnings fact: the third quarter's 9% decline in closings and 90 basis points of gross-margin compression came from the housing cycle, not from regulation, and the regulation is now being written by the company's own former chief executive.