The contrarian case for office writes itself in 2026: buildings trading below replacement cost, sellers who are tired, caps wide enough to absorb bad news, and a buyer pool thinned out by everyone who swore off the asset class in unison. Markets pay people for standing where others will not. But the payment is conditional, and the condition in office is specific: the deals reward honest rollover pricing and punish everything else, because rollover is where the entire bear case lives.
Understand what actually broke. Office income did not collapse because rents fell. It eroded lease by lease, at each expiration, where downsizing tenants took less space, departing tenants took none, and retaining any tenant required a package of improvements, free rent, and commissions that consumed years of that lease's income before contributing a dollar. The distress is not in the rent roll's face numbers. It is in what each expiration costs to survive, which means it is invisible to any model that does not price expirations one at a time.
So the 2026 discipline is a per-suite interrogation. For each tenant: what are the honest odds they renew, in this building, given their actual utilization, not their lease. If they leave, how long does this suite sit, in this submarket, at this floor plate, given what is genuinely competing for the same tenant. What does the replacement lease really cost all-in, and what does the effective rent look like after the concessions that market requires. Sum those honest answers across the rent roll and you have the building's real income path, which is frequently a different building than the one in the offering memorandum, and occasionally, this is the point, a better one than the price implies.
The bifurcation is the other half of the underwriting. Office in 2026 is not one market. Buildings that compete on commute, parking, light, and finished product are leasing, and commodity space is not, and the line between them runs through submarkets, through buildings, sometimes through floors of the same building. The question is never what office is doing. It is which side of the line this building is on and what it would cost to move it, and a rollover model is where that answer becomes arithmetic instead of narrative.
This asset class is the strongest argument that exists for lease-level modeling, because averaging office assumptions does not soften the picture, it deletes the deal. PropCalc™ underwrites office at the suite level, per-tenant renewal odds, downtime, TI, and free rent, with rent-weighted WALT and effective rents carried live, and there is a fictional office deal on this site built to 2026 norms, linked below, worth opening just to watch what honest rollover assumptions do to a wide going-in cap. Sometimes the deal survives them. Those are the ones the market is actually paying contrarians to find.