Which constraint should bind when sizing mezzanine debt into a 2026 refinance gap

A sizing exercise worth running from four directions, since it tends to produce four different answers. Take a suburban office-to-flex conversion, stabilized, in-place NOI $2.35m. A maturing loan at $24m was written when the coupon was under 4%. New senior quotes come back at $17.8m, so the sponsor needs about $6.2m plus costs and wants to fill most of it with mezz rather than write equity. Run it four ways. Debt yield on the total stack. Senior alone is a 13.2% debt yield. Cap total stack debt yield at 10% and the whole stack is $23.5m, which supports $5.7m of mezz. That is the most generous answer and it ignores whether current cash flow can pay the coupon. Combined DSCR at a stressed refi constant. Price the senior at today's coupon and the mezz at 12% current pay plus accrual, hold combined DSCR at 1.10x on in-place NOI, and the mezz shrinks to somewhere near $3.4m. That answer keeps the borrower current through a soft year, and it leaves the sponsor writing a real equity check. Stressed exit LTV. Take in-place NOI at a 7.75% exit cap instead of the appraised value, get about $30.3m, cap total stack at 75% of that, and mezz comes in near $4.9m. Blended cost against unlevered yield. If in-place yield on cost is 8.6% and the blended stack cost crosses that, the sponsor is borrowing at a rate the asset does not earn, which caps mezz well under $4m regardless of what the other three tests allow. Every one of these is defensible, and they land about $2m apart. The DSCR test is generally the one worth holding to in credit, since it is the only one anchored to what the property can actually pay through a soft year rather than what a stack can theoretically support.