Five years of office financials retire the income-recovery refinance

Five years is long enough to settle an argument, and the one about office income growing its way back to refinanceability has now lost on the numbers. Trepp's review of property-level financials for office properties backing CMBS loans found operating expenses outgrew revenues in every year from 2021 through 2025, compounding across the period to 2.7% a year for expenses against 1.3% for revenues and 0.2% for net operating income. At two-tenths of a percent a year, a building's income statement is holding still while the loan around it ages.

Revenue growth of 1.3% a year, compounded, means the cohort's median office collected more rent in 2025 than it did in 2021, but the cost of owning the building rose faster and the gap compounded the wrong way in five consecutive years.

Lenders size a refinance off net cash flow, which feeds the debt yield, and a borrower arriving at maturity carries an income stream not meaningfully larger than the one the original loan was sized against: the chained medians for net cash flow show cumulative growth of just 1.1% over five years. Trepp's illustration puts that increase against a starting debt yield of 8.00% and moves it to roughly 8.09%. Nine basis points is what half a decade of income growth bought.

Underneath the aggregate sits the line item doing most of the damage: property insurance grew at an implied 6.1% a year, more than double the overall expense pace, and the report identifies it as the single largest driver of expense pressure over the period. Insurance reprices on the carrier's calendar rather than the owner's, which means the most corrosive cost in the office stack is also the least responsive to better management or a better rent roll, and the one with the longest renewal lead time.

These are office properties backing CMBS loans, a population shaped by what the conduit market financed rather than the whole asset class, and the figures are annual medians, which compress the distance between a well-leased tower and a half-empty one. The illustrated walk is also the report's own hypothetical rather than a loan-level result; what is not hypothetical is the direction of the chained series, which ran the same way in each of the five years.

The maturity calendar turns a research finding into a negotiating position. Trepp's loan-level data from August flagged a hard-maturity cohort twice July's size with more than half its balance below an 8% debt yield, and the September cohort that followed carried more than a quarter of its maturing balance under a 6% debt yield. A borrower whose net cash flow has grown 1.1% across five years cannot close a gap of that size by operating the building better; Trepp's own summary says as much, more politely, that the annual medians suggest cash-flow growth provided limited additional support for refinancing.

That leaves the stack: structured extensions, preferred equity and stack compression are resolving the wall rather than distress sales, and Trepp's medians are the arithmetic underneath that pattern. With no income-side exit to underwrite, the exit gets paid for in new equity or bought with time. An extension converts a maturity into a duration trade: the lender holds a coupon, the sponsor writes a check, and both sides are betting that insurance costs, taxes and the rate environment look different when the loan comes back around. That is a defensible trade and, for the 2021 and 2022 vintages, close to the only one available.

None of this overturns the split visible all year. Trophy assets with leases rolling to market can refinance above their prior loans and will keep doing so, because the median describes the middle of the cohort and the middle is where the math fails. Clearing trades happen at the top, and separately at the bottom, where a sponsor balance sheet sets the first bid. A refinance market that clears at both ends and stalls in the middle is exactly what a flat median looks like from the inside.

Office is also not alone in this squeeze. Trepp's multifamily medians, covered last month, showed expense growth finally cooling in 2025, with revenue slowing by the same amount, so NOI growth weakened anyway and the cost-revenue gap persisted. When the identical pattern turns up in apartments, a different demand profile with a different capital stack and a different set of lenders, the diagnosis shifts. This looks less like one sector's demand problem than a cost-of-ownership problem that underwriting models have been slow to price, and the multifamily medians suggest the office figures are not merely an office story.

The takeaway for a lender is subtractive. Income growth drops out of the refinance model for the office CMBS vintages that need it most, and what remains is an exercise in weighing how much new equity a sponsor will write against how much maturity a lender will extend. Nine basis points over five years is the number to hand a borrower who wants to know why the new loan is smaller than the old one: the property has not earned the debt it carries, and the difference has to come off somebody's balance sheet. Watch the insurance line. If it turns, this arithmetic turns with it.