The average conduit loan still looks familiar on paper. The leverage number has hardly moved. Yet beneath the headline 57.6% loan‑to‑value ratio, the economics of new CMBS debt have shifted decisively against marginal borrowers, as rising debt service coverage and debt yield requirements quietly re-frame what that leverage really buys.
Stronger metrics at the same LTV
On a recent episode of the Trepp podcast, Trepp's research team walked through first‑quarter conduit issuance and highlighted a pattern that will be immediately recognizable to borrowers who have been back in the market this year. For the 10 conduit deals that priced in the latest quarter, the average LTV was 57.6%, up only slightly from 57.1% a year earlier. On its face, that is a rounding error.
The change is in the coverage. According to the episode, hosted by Trepp Chief Product Officer Lonnie Hendry, head of research at Stephen Buschbom, the average DSCR on those same loans rose from 1.76x to 1.99x, while the average debt yield climbed from 12.5% to 13.3%.
"The more telling metrics are the debt service coverage ratio and debt yields," Buschbom said on the show. "DSCR increased to 1.99 times from 1.76, and debt yield climbed to 13.3% from 12.5% — those are some pretty noticeable increases."
For seasoned CMBS users, the implication is straightforward: lenders are still willing to write loans around the same nominal leverage band, but they are doing so against higher levels of stabilized net operating income and tighter definitions of "in‑place" cash flow. That combination reduces the wiggle room for business plans built on aggressive rent growth, rapid lease‑up or optimistic expense assumptions.
Underwriting the income line harder
The dynamic reflects two overlapping forces. One is a simple property‑level improvement. Some assets — particularly those that continued to grow rents through the rate shock — are generating enough incremental NOI to satisfy higher DSCR and debt‑yield hurdles without cutting proceeds. Hendry noted that the stronger metrics "highlight some pretty strong performance at the property level," even as leverage has remained broadly stable.
The other force is a shift in how originators translate NOI into underwritten cash flow. In an environment where inflation has proven stickier than many expected and refinancing risk remains top of mind, lenders are leaning further into haircuts on non‑credit tenants, rolling shorter leases at market and embedding higher vacancy and expense loads. That effectively adds more stress to the numerator of both DSCR and debt yield, while leaving the maximum LTV guidelines largely intact.
The result is that two loans with the same 57.6% LTV can now behave very differently from a risk perspective. A sponsor bringing a stabilized, well‑leased asset with clean rollover and documented operating history may clear the higher coverage metrics with little friction. A borrower with a story asset — more vacancy, more near‑term roll, heavier capex — will find that the same nominal leverage band supports a smaller absolute dollar amount of debt than it did a year ago.
Proceeds pressure on marginal deals
For investors trying to size proceeds on the margin, the new coverage thresholds matter more than the LTV headline. Higher DSCR and debt yield requirements at effectively unchanged leverage imply that coupon income and structure must carry a greater risk-management burden than they did when rates first moved higher.
That is most evident in segments where performance has flattened or is only starting to recover. In those cases, stronger coverage metrics tend to be achieved through smaller loan balances rather than better cash flow. Because DSCR is a function of both NOI and debt service, lenders can generate the 1.99x coverage Trepp cites by sizing to lower proceeds at today's all‑in coupons. The same is true for debt yield: if the numerator is constrained, the denominator has to fall.
Buschbom framed the shift as a sign that the market has, at least for now, chosen to tighten via underwriting rather than via visible leverage compression.
"Without seeing significant changes in the leverage points, this actually feels pretty strong," he said, referring to the combination of coverage ratios and stable LTVs.
For conduit buyers, that is reassuring. For borrowers, it is a reminder that the proceeds conversation now starts with income resilience rather than an LTV target.
What sponsors should expect in 2026
The TreppWire data points align with what many sponsors are encountering in term sheets this year: structures that look familiar, paired with more scrutiny of line‑item NOI and less tolerance for pro forma-heavy stories. In practice, that means several things for experienced borrowers heading into the balance of 2026.
First, LTV caps are a poor guide to achievable proceeds on weaker assets. A sponsor that underwrites to a 60% leverage assumption may find the final loan amount effectively sized by a 13%+ debt‑yield constraint or a 1.90x–2.00x DSCR test, not the nominal LTV.
Second, there is more value in pre‑emptively presenting conservative cash‑flow cases — including stressed rent rolls and realistic expense growth — than in pushing underwriters to accept aggressive assumptions that will likely be pared back in credit.
Finally, the Trepp numbers underscore that capital is available and willing to take risks, but on terms that reflect the lessons of the last two years. Issuance remains robust and, as Hendry put it, the quarter's statistics "highlight some pretty strong performance here at the property level," yet the market is signaling that it wants more cushion in the income line at the same leverage point.
For sponsors with genuinely durable cash flows, that is a manageable adjustment. For those trying to refinance or acquire marginal assets, the arithmetic embedded in a 1.99x DSCR and 13.3% debt yield may be the constraint that matters most.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.