Underwriting
What A Rent Comparable Actually Proves
A comparable rent records what one unit achieved under specific conditions, and the adjustments made between that unit and the subject property carry most of the analytical weight.
Rent comparables are the foundation of almost every underwriting model, and they are also the input most often treated as fact rather than estimate. A comp is a single observation, not a market rate.
A comp is a transaction, not a price level
Every asking rent is a hope and every signed rent is a transaction. The two diverge, and only the signed lease tells you what a tenant was willing to pay under real conditions.
A signed rent also embeds terms that never appear in a listing. Free months, a waived deposit or an unusually long lease all shift the effective rent away from the headline figure.
This is why concessions matter more than they look. A unit advertised at a strong rent with two free months collects considerably less over the year than the number implies.
The adjustments are where the analysis lives
No two units are identical, so a comp is always adjusted before it applies. Square footage, floor level, outdoor space, parking and renovation quality all move the number.
The problem is that adjustments are estimates layered on estimates. If four adjustments each carry a small error, the adjusted comp can drift far from anything a tenant would sign.
Underwriters who adjust in one direction repeatedly should treat that as a warning. Consistent upward adjustment usually means the comp set is stronger than the subject property, not that the subject is undervalued.
Recency has a short half-life
Leasing conditions change faster than sale conditions because leases turn over constantly. A comp from eighteen months ago describes a different set of conditions than one signed last month.
In markets where new supply has arrived, older comps are especially misleading. They record a period before the competition existed, and the units they describe may now be leasing at different terms.
The practical response is to weight recent signings heavily and treat older ones as context. A thin recent sample is uncomfortable, but it is more honest than a large stale one.
Who collected the comp changes what it says
Comps assembled by a seller's broker are selected to support a price. Comps assembled by a leasing agent working the same submarket are selected to close tenants. Neither set is neutral.
None of this implies dishonesty. Selection happens through ordinary professional incentive, and the person assembling the set genuinely believes the units they chose are the relevant ones.
The defence is to rebuild the set independently, including the comps that undercut the thesis. A comp set with no weak observations in it has almost certainly been filtered.
What a comp cannot tell you
A comp describes what was achieved, not what will be. It contains no information about future supply, employment conditions or how a specific tenant pool will behave.
It also says nothing about durability. Two units can achieve the same rent while one holds tenants for years and the other turns over annually, and turnover is a real cost.
Treating comps as evidence about the recent past, rather than as forecasts, keeps the model honest. The forecasting happens elsewhere, explicitly, where it can be tested.
