Concession burn-off is showing up in year-two NOI projections as if it is permanent income, and I want to know how rooms are underwriting it.

The assumption doing the most work in most models I see is that concession burn-off is treated as a certainty timed to lease expiration, with gross rent flowing through clean from month one of the new lease. That works if the market cooperates. It does not work if you are in a Sun Belt submarket still absorbing supply, where the operator re-leases at face rent on paper but the next concession cycle starts eighteen months later.

The more defensible convention separates the two acquisition types exactly the way you are framing it. On a value-add deal, concession burn-off can be modeled as part of the value-add thesis, but it needs its own line and its own probability weight, not a smooth glide into stabilized NOI. On a stabilized acquisition, booking gross rent before concessions have actually burned is aggressive underwriting dressed as conservative underwriting, and that is the gap your going-in cap is hiding.

What I see sharper shops do: they run a concession persistence stress in the hold period, typically assuming some percentage of concessions re-emerge at lease two or three if the submarket vacancy is above a threshold they define at underwriting. If the IRR survives that scenario, the deal underwrites. If it only survives the burn-off-as-scheduled scenario, the entry price is doing too much work.

The exit-side risk you named is real and often underweighted. A buyer in year five underwriting those leases at gross rent less a market concession assumption will impute a lower NOI than your projection, and the bid reflects it. The spread narrows your exit cap cushion right when you need it most.

The unmentioned risk here is lease term stacking. If a third of the leases are already concession-laden and they cluster in expiration, re-leasing concentration amplifies both the burn-off timing risk and the re-concession exposure.

Does your model run a lease-expiration heatmap against your projected submarket vacancy curve, or are the expirations treated as uniformly distributed across the hold?

Once you call it stabilized, every concession on the rent roll gets treated like noise rather than structure, and the model never questions whether those six weeks come back at renewal.

What I have watched fail consistently is building the burn-off as a straight-line glide to gross rent with no corresponding stress on the renewal spread. You get to year three, renewal comes due, the market is flat or softer, and the operator re-offers two to four weeks just to hold occupancy, so the NOI never actually reaches the number the year-two projection was pointing at.

On the value-add side, do you at least have a parallel scenario where the concession re-offers at half the going-in level, or is the base case just assuming the market absorbs it clean?

The misconception worth correcting first: burn-off is not a timing problem, it is a comparables problem. The concession is not going away when the lease rolls, it is going away only if the market will bear gross rents without it, and that is a separate underwriting question that most models never ask.

On convention, the approach that holds up best at disposition separates effective rent from face rent at the lease level and runs the rent roll forward on effective rent only, with a line item called "concession premium" that decays according to a lease-by-lease roll schedule rather than a portfolio average. If the exit buyer can re-underwrite every unit on face rent, the premium collapses to zero at the comp sheet, not at some modeled midpoint. For value-add the standard is more forgiving because the story is margin expansion through renovation, so a rising effective rent trajectory has narrative support. For stabilized the defensible position is to cap the assumed effective rent at the trailing twelve average across the submarket, not the face rent in the lease, and to hold that number flat through year three before allowing any growth. The six-week concession on a third of the units in a 300-unit building is roughly 1.5 months of rent across 100 units, so at a $1,800 average monthly the gross-to-net gap is $270,000 annualized. That is real NOI that the exit model needs to either defend at the rent comp level or discount explicitly in the terminal cap. What is the submarket vacancy trend over the last four quarters, because the defensibility of holding effective rent flat depends entirely on whether absorption has been positive or the market has already started to give ground?