Spreads Are Tightening. Our Underwriting Isn't.

We quoted three hotel bridge deals last week. All three sponsors led with rate. Two of them had already collected a term sheet 15 basis points inside ours, and both of those term sheets came in at 58% of cost on a deal that needs 65% to work.

That is the market in one sentence. Money is cheaper than it was a year ago and harder to get enough of.

What the second quarter actually showed

Spreads on fixed-rate, five-to-ten-year permanent loans came in to 204 basis points in the second quarter, 21 basis points tighter year over year, with multifamily at 162 basis points (CBRE). Loan counts rose 11% and average loan size rose 5%. On its face, that is a loosening market.

Look at the credit metrics from the same report and the picture inverts:

Source: CBRE Lending Momentum Index, Q2 2026

Lower leverage, higher required coverage, higher required debt yield, tighter spread. Lenders are buying market share with price and refusing to buy it with proceeds. CBRE says it plainly: competition is on price, not leverage.

Meanwhile the benchmark is working against everyone, and it has gotten worse in the last three weeks. The ten-year Treasury closed at 4.72% on August 17 and the thirty-year at 5.31% (Federal Reserve H.15), the long bond's highest close since 2007 (CNBC). As of late July the thirty-year had already closed above 5% on 27 days this year, including twelve consecutive sessions, and futures were pricing roughly 80% odds of at least one Fed increase by the September 16 meeting (CREFC). Any underwriting model still carrying a rate-relief assumption into 2027 is a model, not a plan.

What it means when we underwrite

Three consequences, in the order they hit a file.

Debt yield sets your proceeds, not LTV. At a 10.2% market debt yield, a hotel throwing $4.2M of stabilized net operating income supports roughly $41M of senior debt regardless of what the appraisal says the building is worth. Sponsors still arrive anchored to a value and a percentage. We start from the income and work backward, and the gap between those two methods is where most of our repricing conversations happen.

Coverage at 1.43x means the ramp has to be real. A tighter coverage test punishes back-loaded revenue assumptions. If the business plan needs 2028 to hit the number, the loan has to be sized to 2026 cash flow with a reserve that carries the asset until 2028. That is not conservatism for its own sake. It is the only version of the deal that survives a takeout underwrite.

The exit is the hard part now. CMBS delinquency hit 7.86% in July, the highest reading since November 2020, and 66% of newly delinquent balances were non-performing matured balloons — loans that simply could not refinance at maturity (CREFC). The effective rate including performing matured balloons is 9.62%. We now stress every bridge exit at a takeout coupon 75 basis points above today's market and a debt yield 50 basis points above today's requirement. If the deal still clears, the structure is sound. If it only clears at today's terms, the sponsor is underwriting our patience rather than the asset.

Where we're saying yes

Hospitality fundamentals are cooperating, which is why we are still leaning in. STR and Tourism Economics raised the 2026 U.S. RevPAR growth projection to 4.4%, with ADR up 3.1% and demand up 1.7% (CoStar). CoStar reported in June that first-quarter RevPAR was the highest on record (CoStar). Rate-led growth is exactly the kind we can underwrite, because it flows to the bottom line faster than occupancy-led growth.

We are actively quoting select-service and upper-midscale hotels in secondary markets with a completed or fully funded PIP, resort and mixed-use assets in Florida and the coastal Southeast where we understand the seasonality, and 2021-to-2022 vintage bridge loans coming due where the sponsor is moving early rather than waiting for the maturity date. On that last category, early is doing real work. We are seeing fewer lenders willing to kick the can with another extension, which is visible in the July CMBS data: 66% of newly delinquent balances were non-performing matured balloons.

Where we're saying no

Ground-up hospitality with a hard-cost budget assembled before 2024. Nonresidential materials costs are still more than 55% above early-2020 levels and some electrical equipment backlogs exceed two years (CREFC). A stale budget is not a pricing problem. It is a completion-risk problem, and no interest reserve fixes it.

Also passing on: deals where the interest reserve stops before stabilization, sponsors whose liquidity after closing is thinner than the reserve itself, and any file where the ADR ramp assumes market-leading rate within twelve months of a brand conversion. We have declined all three of those in the last month, and in each case another lender issued a term sheet. We will be interested to see how many of those close.

The box this week

• Loan size: $5M floor, $20M–$100M sweet spot, flexible above
• Asset focus: hotels, resorts, mixed-use hospitality; bridge and construction
• Leverage: to 65% of cost on transitional hospitality, sized off debt yield
• Structure: interest-only, interest reserve funded through stabilization, extension options priced not promised
• Timing: term sheet in 72 hours on a complete package, closing in 45–60 days

If you have a hotel file that needs proceeds a bank will not reach and a decision faster than a committee calendar allows, send it. We will tell you the same week whether we can do it, and if we cannot, we will tell you why.