Why Agency Take-Out Failed on 23% of 2025 Bridges

If you closed a bridge loan in late 2023 or 2024 with a 24-to-36-month term and a Fannie, Freddie, or HUD take-out penciled into your exit model, you are likely aware by now that penciling something in and executing it are different activities. Based on tracking from Trepp’s Q1 2026 bridge maturity analysis and corroborated by MBA originations survey data, roughly 23% of multifamily bridge loans with anticipated agency take-outs that matured in 2025 did not close into permanent agency financing on schedule. Some extended. Some sold. Some are still sitting in forbearance conversations with lenders that would rather not discuss their warehouse line.

This piece is not a postmortem for its own sake. It is a checklist for operators now under LOI on a 2026 vintage bridge — or trying to refi out of a 2025 bridge that already missed its window.

By the Numbers

Sources: Trepp Bridge Maturity Wall Report Q1 2026; MBA Commercial/Multifamily Originations Survey Q4 2025. Figures are estimates from aggregate reporting, not audited loan-level data.

The DSCR Gap Nobody Priced In

The most common failure mode, and the one most operators seem surprised by, is the mismatch between the DSCR standard used to underwrite the bridge and the DSCR standard the agency actually applies at take-out.

Bridge lenders — especially the mid-market debt funds operating in the $5M–$25M balance space — underwrote 2024 vintage deals to a stabilized DSCR in the 1.20x–1.25x range on pro forma NOI. That pro forma assumed rent growth of 4–6% annually in markets like Phoenix, Charlotte, and Tampa, which were the darlings of the 2022–2023 acquisition cycle. Fannie Mae’s actual DSCR test for a conventional multifamily take-out on a 30-year amortization at Q4 2025 rates — call it a 6.40% all-in coupon on a 10-year fixed with 30-year am — required stabilized NOI sufficient to produce 1.25x on the permanent loan amount, stress-tested to the note rate, not a blended forward rate.

For a deal that went in at 70% LTC in 2024 and was supposed to refi at 65% LTV on stabilized appraised value, that math worked when cap rates were 5.0% and rent growth was running. By late 2025, Phoenix vacancy was running closer to 9–10% in Class B product, rent growth in the Sun Belt metros that dominated 2023–2024 bridge volume had gone flat to mildly negative in real terms, and Fannie’s DUS lenders were applying the updated Q4 2025 multifamily guidelines that tightened IO periods and stressed DSCR tests more conservatively in markets with elevated new supply. The pro forma NOI from the bridge underwriting, applied to real market conditions 24 months later, often came in 8–14% below the number needed to size the permanent loan at the LTV the operator needed.

That delta — 8–14% on NOI — is the difference between a clean exit and a 12-month extension at 200bps over your current spread.

T-12 Trailing Income: The Timing Problem

Even when operators hit their NOI targets, a material subset of 2025 take-out failures came down to documentation timing rather than fundamental underperformance. Fannie Mae and Freddie Mac SBL both require a T-12 trailing income history at the property — not pro forma, not forward-looking, not a 3-month annualized run rate. Twelve months of documented, signed leases and collected rent.

Here is where the 18-month bridge creates a structural problem. An operator who closed a bridge in October 2023, began rehab in November 2023, and stabilized by August 2024 might have 10–11 months of solid T-12 data at the intended take-out date of April 2025. Freddie SBL guidelines require a minimum occupancy of 90% for 90 days, but the DUS lenders underwriting Fannie deals in practice want to see the full year. Submitting in April 2025 with an October 2023 rehab start means your cleanest T-12 window doesn’t open until October 2025. The operator’s options at that point: extend the bridge (hope the lender agrees and the extension fee doesn’t consume the NOI upside), sell (into a thin Q1 2025 transaction market), or submit early and accept a lower loan proceeds number sized off a partial-year income history.

Several composite cases in the MBA’s Q4 survey data reflect operators who submitted to Fannie at month 22 of a 24-month bridge and received proceeds roughly 6–8% below the amount needed to fully retire the bridge balance. The gap had to be covered with cash-in refi proceeds or seller carryback structures that the original underwriting never contemplated.

The fix is not complicated, but it requires admitting it at loan origination: if your stabilization timeline runs past month 12, you almost certainly need a 36-month bridge or a structured extension option with pre-negotiated terms — not a 24-month bridge with a handshake understanding that the lender will extend.

Rate Cap Requirements and the Rate Environment Shift

The third major failure mode is less about operator execution and more about how the interest rate environment moved relative to what lenders required operators to purchase.

In 2023 and into 2024, bridge lenders required SOFR-based rate caps as a condition of closing — typically capped at SOFR + 200–250bps strike, with a 2-year term matching the initial loan period. When those caps matured in 2025, operators seeking extensions discovered that replacement cap premiums had moved substantially. A 2-year SOFR cap at a 5.0% strike for a $10M floating-rate bridge loan that cost $180,000 in early 2023 was pricing at $310,000–$370,000 by mid-2025, reflecting both the forward curve and tighter cap market liquidity.

More importantly, lenders issuing 12-month extensions in 2025 started requiring new rate cap purchases — not cap extensions — at current market pricing, and in several cases required the strike to move to match the current SOFR forward curve, which effectively made the extension cost significantly higher than the original cap cost. Operators who had not reserved for replacement cap costs in their cash flow projections found themselves either absorbing a six-figure unbudgeted capital call or negotiating aggressively with lenders who had no particular incentive to accommodate them.

The 10-year Treasury as of late 2025 was trading in the 4.60–4.90% range, which kept agency all-in rates elevated enough that even deals with adequate NOI were finding that the permanent loan proceeds they could support at a 1.25x DSCR test were materially lower than the bridge balance they needed to retire. HUD 223(f), which processes more slowly and does not share Fannie/Freddie’s rate sensitivity in the same way, saw FY2025 endorsement volume come in roughly 18% below FY2024 levels, partly reflecting borrower reluctance to lock long-term at prevailing rates and partly reflecting processing delays at the regional offices.

What to Design For in a 2026 Bridge

The takeaway from the 2025 cohort is not that bridge-to-agency is broken. It is that the structure only works if you design the bridge backward from the agency underwriting box rather than forward from the acquisition price.

Specifically:

Build the DSCR test first. Take your market cap rate (not your pro forma cap rate — the actual cap rate for your asset class and vintage in your MSA as of today), apply it to your realistic stabilized NOI, and compute what loan balance that NOI will support at 1.25x DSCR on a 30-year am at a rate 50bps above current Fannie pricing. If that number doesn’t retire your bridge balance, you either need more equity at acquisition, a lower purchase price, or a longer bridge term with a realistic stabilization timeline.

Price the T-12 clock into your bridge term. If you’re buying a 1978-vintage, 24-unit in a secondary market and your rehab takes 14 months, you need a 30-to-36-month bridge minimum to have a full T-12 available at take-out. “We’ll extend if needed” is not a term — it’s a verbal agreement that will be re-priced against you at the moment of maximum leverage.

Reserve for replacement rate cap costs. The number is not symbolic. On a $7.5M bridge at 75 LTC, budget $250,000–$400,000 for the replacement cap if you extend. If you are not holding that in reserves or structuring it into the initial loan proceeds, you are self-insuring against a liability you have not quantified.

Stress your take-out against the rate environment, not the in-place rate. Fannie and Freddie size permanent loans off actual market rates at the time of commitment, not the rate from your bridge term sheet. Model the take-out at note rate + 75bps as a base case. If the deal still works, proceed. If it doesn’t, the problem is not the bridge lender — the problem is the acquisition price.

The MBA’s Q4 2025 commercial originations survey showed agency multifamily volume contracting in the second half of 2025 even as overall multifamily fundamentals were stabilizing in coastal and Midwest markets. The contraction was not a demand story — it was a proceeds story. Operators couldn’t get the loan balance they needed at rates that penciled. In many cases, that mismatch was visible at the time the bridge was originated in 2023. It just wasn’t priced in.

Design against it now, or design for an extension conversation in 2028.