mortgagetape^™

Written by Mortgage Tape Team — a group of industry analysts leveraging our proprietary mortgage-domain language models to synthesize and decode housing data.

📌 Takeaways by role

• CRE and whole-loan investors: Between year-end 2023 and year-end 2025, MetLife’s jumbo commercial mortgage book (loans with recorded investment ≥$50M) dropped from $25.9B to $19.7B across 240 → 185 loans. TIAA’s dropped from $31.0B to $23.2B across 253 → 193 loans. The two drawdowns are statistically parallel in shape, size, and timing: $14.0B of jumbo commercial exposure out, in a 24-month window, from two of the oldest institutional mortgage lenders. Within those two filers, 149 of 169 departed jumbo loans are gone from the balance sheet entirely (not amortized down) — the loans were sold, paid off, or otherwise disposed, not just paid down below the jumbo threshold.
• Insurance-dedicated investors: The industry total isn’t shrinking — it’s rebalancing. The top 13 life-insurer filers hold ~$294B of mortgage loans at Dec-2025. Northwestern Mutual leads at $60.8B, MetLife $44.5B, Corebridge (American General) $37.0B, TIAA $30.6B. MetLife’s own book has run off $10.5B (-19%) over five years — $55.0B at Dec-2020 to $44.5B at Dec-2025 — while Pacific Life grew +82% over the same window, Metropolitan Tower Life (MetLife’s commercial subsidiary) grew loan count 2.4×, and Corebridge-AGL ran its book from $23B (2022) to $37B (2025). Capital reallocation is happening within the insurer channel, not out of it.
• CRE market-structure analysts: The geography of MetLife’s and TIAA’s departures tells the same story as the aggregate: concentration in major-CBD office and retail loans — California, Texas, New York, Massachusetts, DC — plus cross-border exposure (UK, Mexico, Netherlands at MetLife; UK at TIAA). The vintages shed are predominantly 2015-2019 originations: the pre-COVID commercial-real-estate boom tier that now faces simultaneous valuation pressure and refinance-rate shock. When two of the most sophisticated institutional whole-loan lenders independently de-risk the same vintage and geography on the same timeline, the pattern is a mark-to-market signal, not a coincidence.

The data nobody reads

Every March, every US life insurer files a statutory annual statement with its domiciliary state’s Department of Insurance. The filing — the NAIC “Blue Book” — runs 1,500 to 3,000 pages. Deep inside, Schedule B Part 1 lists every mortgage loan the insurer owned at the prior year-end, loan by loan: location, state, origination date, note rate, book value, and property valuation.

The information is public. The format is a 2,500-page PDF per insurer. Nobody reads it because nobody can.

We parsed it. For the top 13 life-insurer filers we have usable 2025 data, that produces ~418,000 loan rows across all available vintages and $293.7B of mortgage AUM at December 31, 2025. For nine of those filers we have multi-year panels going back to 2020, 2022, or 2023 — enough to see what each book did during the post-COVID rate cycle.

Here’s the Dec-2025 landscape:

$293.7B is a bigger number than most observers expect from the insurer channel. For comparison, that’s about 13 percent of what US banks collectively hold in 1-4 family mortgage servicing — but entirely on balance sheet, as whole loans, not MSR. These are owned assets paying spread over Treasury, not servicing contracts earning fees on behalf of others. The insurer channel is a quiet, mortgage-industry-sized liability-matched investment program that most mortgage analysis simply skips.

The MetLife and TIAA drawdowns

MetLife’s and TIAA’s commercial mortgage programs are 70+ years old. They are not transient venture plays; they are core strategic allocations. So when both books contracted at the exact same time, in the same tier, by nearly identical dollar amounts, that is worth paying attention to.

The pattern is clean. In both insurers:

• Residential and small-commercial loans (under $1M, and $1-10M respectively) sit essentially flat across the three years in count and in UPB.
• Mid-commercial loans ($10-50M) also hold near-steady.
• The jumbo bucket — anything with book value ≥$50M — contracts sharply, both in loan count and in dollar exposure.

Said simply: MetLife and TIAA aren’t shrinking their mortgage platforms. They are each specifically unloading large commercial mortgages, nothing else.

The departure mechanism matters. Of the 80 MetLife jumbo loans that left the >$50M tier between 2023 and 2025, 66 are gone from the balance sheet entirely — the loan number does not appear anywhere in the 2025 Schedule B. Only 14 were paid down from above $50M to below $50M while remaining on the book. For TIAA the ratio is even starker: 83 of 89 departed loans are gone entirely; 6 were paid down.

This is not amortization. The loans were sold or fully paid off. 2023-2025 is a 24-month window in which two of the biggest institutional mortgage lenders each actively removed roughly $7B of jumbo exposure from their books. $14.0B of combined dollar exposure, 169 loans, two insurers, same window.

Where the exposure went

The states and jurisdictions of the departed jumbo loans trace a map of the loans most under valuation pressure right now:

• MetLife top 10 departed states/jurisdictions by UPB: US-portfolio ($1.34B), Mexico ($1.21B), NY ($0.85B), UK ($0.75B), FL ($0.57B), CA ($0.40B), TX ($0.38B), NLD ($0.27B), GA ($0.26B), MN ($0.22B).
• TIAA top 10 departed states/jurisdictions by UPB: TX ($1.94B), CA ($1.77B), UK ($1.09B), MA ($0.80B), NY ($0.62B), DC ($0.59B), GA ($0.55B), FL ($0.51B), NJ ($0.46B), MI ($0.38B).

Nearly every departed-loan dollar lands in either a major US CBD or a cross-border market — exactly the segments where commercial real estate values have faced the longest and widest-reported post-COVID correction. The UK exposure parallel between the two insurers is striking: a combined $1.84B of UK commercial mortgages left both books in a 24-month window.

The vintages are consistent, too. For MetLife, departed-loan UPB clusters in 2015-2022 originations ($5.9B total). For TIAA, 2014-2019 originations dominate ($6.9B). Both insurers were net sellers of the vintages that locked in low-rate commercial debt during the 2015-2019 boom — the vintages now struggling to refinance into a 7-handle rate environment.

The industry isn’t shrinking — it’s reshaping

The aggregate insurer-held mortgage total is holding roughly steady, which is the first thing a casual look at the Dec-2025 landscape chart will suggest. But under the surface, dollars are moving between filers.

Pacific Life sits in the “growing-growing” quadrant: +73.8% in loan count, +81.9% in UPB from 2020 to 2025 — $11.3B to $20.5B on a consistently diversified book. Mean loan size has barely moved ($23M), medians are stable ($1.6M). This is a deliberate book-build, not opportunistic jumbo pickup.

Metropolitan Tower Life — the MetLife commercial subsidiary — grew loan count from 1,924 to 4,545 (+136%) and UPB from $3.7B to $6.7B (+81%) between 2020 and 2025. The sub-filer is adding small- and mid-sized commercial loans while the parent sheds the mega-commercial tier. Within the same holding-company group, two different books moved in opposite directions over five years.

Corebridge-AGL shows the sharpest platform build-up in the dataset: loan count 5,417 → 36,255 across 2022-2025 (+569%, driven mostly by residential origination), and UPB $23.3B → $37.0B (+58%). Corebridge spun off from AIG in late 2022; the mortgage book has grown steadily since.

Nationwide (NLIC + NLAIC combined) added 3.4 billion dollars (+21%) on a stable 1,192-loan book — classic steady-state institutional growth.

MetLife and TIAA sit in the “shrinking-shrinking” quadrant. MetLife’s five-year arc runs $55.0B → $44.5B (-19.1% UPB, -20.5% loan count). TIAA runs $38.1B → $30.6B over 2023-2025 (-19.6% UPB, -4.4% loan count). VALIC sits in an unusual spot: +51% in loan count, -19% in UPB — adding many smaller loans while the dollar book shrinks, which the sub-filer identity (a Corebridge annuity subsidiary) partly explains.

The capital isn’t leaving the insurer channel. It’s moving within it — from the oldest, largest commercial-mortgage platforms toward the mid-sized growth programs.

What this signals

Three things are visible in these filings that an earnings-call summary would miss:

1. Sophisticated capital is actively marking down 2015-2019 commercial real estate — by exit, not by writedown. Fair-value marks for life-insurer commercial mortgages are notoriously conservative. The clean way to express a bearish mark-to-market view is to sell the loan at or below book value and remove the position. $14B of jumbo exposure gone from two insurer books, with the vintage and geography concentration of the departures, is a mark-to-market signal that no quarterly fair-value adjustment would convey as clearly.

2. The international CRE angle is unusually present in these numbers. US CRE repricing coverage is dominated by domestic office and retail. These filings make a quieter point: insurer-held UK and Mexican commercial exposure is being pulled back in parallel, by both insurers independently. The international pattern suggests the mark-down thesis is cross-border, not just US-centric.

3. The counterweight — Pacific Life, Metropolitan Tower, Nationwide — is building books in exactly the segments the giants are stepping back from. This is not just divestiture; it is a transfer of balance-sheet exposure between institutions with different investment horizons, portfolio sizes, and risk tolerances. Smaller or faster-growing books take advantage of the opportunities the big books are offloading.

The data to see all of this is already public. It just lives in 2,500-page PDFs filed with state insurance regulators, one filer at a time, once a year. Pulling the top-13 panel into a single analytical dataset — 418,000 loan rows across 50 insurer-year vintages — makes the pattern obvious. Without the panel, the drawdown looks like earnings-call color on one filer at a time. With it, the mechanism is a chorus.

---

On the data: Figures in this article come from Mortgage Tape’s parse of NAIC Statutory Annual Statement Schedule B Part 1 — the loan-level mortgage ownership disclosure life insurers file each March for the prior calendar year. Twelve filers’ Dec-2025 vintages are included; multi-year panels are available for a subset. All UPB figures are book value / recorded investment at the balance-sheet date, not fair-value marks or market-settlement values. New York Life, Prudential, and Northwestern Mutual Dec-2025 figures appear in the landscape chart but not in the multi-year panel because their 2021-2024 statements for Schedule B Part 1 were not readily available on state-DOI public portals.