Rent comps for multifamily underwriting: which numbers survive a closer look?

Start with the decision your comp set must support

Before collecting listings, write the decision in one sentence. For example: can an unrenovated, roughly 850-square-foot two-bedroom apartment support a $2,000 asking rent without an incentive? That question determines which units belong in the comparison, what terms must be recorded and what would count as conflicting evidence.

Define the inclusion rules before viewing the prices: unit type, a defensible size range, property condition, location and relevant amenities. Keep excluded observations in the research file with a reason. Otherwise it is too easy to remove the cheapest evidence after seeing the answer. If an important difference cannot be priced from evidence, keep it visible instead of inventing an adjustment.

Separate how often concessions appear from how large they are

RealPage reported that 16.9% of stabilized U.S. apartment units offered concessions in May 2026, with an average discount of 10.9%. These describe different features of the market: the prevalence of incentives and their depth. They are dated national context, not a discount to subtract from every local comp.

For your comparison, record the actual offer for each unit: advertised monthly rent, lease length, free-rent amount or credit, eligibility and observation date. Keep mandatory fees separate so a base-rent comparison does not quietly become a total-cost comparison. Asking offers also remain distinct from executed leases and collections.

Source: RealPage, 9 June 2026: May 2026 concession use and discount depth

A worked comparison that changes the rent conclusion

The following seven observations are fictional. Assume a target of unrenovated two-bedroom units between 800 and 900 square feet, similar locations and 12-month leases. Each row represents a different property. All free months waive base rent only; required fees and utilities are excluded.

Average monthly base rent after incentives = (monthly asking rent × lease months − total base-rent incentive) ÷ lease months. This is a lease-period comparison, not a forecast of annual property cash flow.

Read the difference between the headline and the evidence

Across all seven offers, the asking-rent average is $2,060.71. Restricting the set to the five pre-defined core comparisons reduces that average to $1,975. After incentives, the core mean is $1,875 and the median is $1,879.17. The core observations range from $1,787.50 to $1,975 after incentives.

That does not prove the subject can achieve $1,875. It shows why an unqualified $2,060.71 market average is a weak basis for a $2,000 assumption. A higher subject rent now needs a specific explanation: superior features, evidence of stronger demand, or recent executed leases on comparable terms. Keep those reasons separate from the arithmetic.

If one property supplies twenty available listings while another supplies one, a simple listing average lets the first dominate. Summarize comparable unit types within each property before comparing properties, or show both calculations and explain the weighting. Repeated advertisements for the same unit should count once.

Test whether the conclusion depends on one observation

In the fictional core set, leaving out each property in turn produces after-incentive means between $1,850 and $1,896.88. None reaches $2,000. This is a useful stability check for this small example, not a statistical confidence interval or proof of achieved rents. New lease evidence could still change the conclusion.

Then test the incentive assumption itself. At a $2,000 asking rent on a 12-month lease, zero, one and two free months produce average base rents of $2,000, $1,833.33 and $1,666.67. Model what is evidenced today separately from what you expect after incentives end. Do not turn a temporary promotion into a permanent discount without saying so.

Keep the accounting consistent when transferring the result to the property model. If you enter an already discounted rent and also deduct the same concession in a separate line, you count the cost twice. Conversely, entering asking rent with no allowance ignores a documented incentive. Keep vacancy, bad debt, renewals and operating expenses as distinct assumptions.

Build the comparison in five steps

The minimum useful output is one dated table, a supported comparison range and a short list of evidence that could change it. Add columns for source, observed date, inclusion reason, asking or achieved status and unresolved terms.

1. Write the subject unit definition and inclusion rules at the top of the sheet. Select an observation window and keep older evidence in a separate group.
2. Capture each offer once, preserving the source and all lease terms. Resolve unclear incentive eligibility before using it as a fact.
3. Calculate after-incentive base rent and compare similar units within each property. Show the number of properties and observations, not only the average.
4. Compare asking and after-incentive results, then remove one property at a time. Explain differences that survive these checks and gaps that still matter.
5. Transfer the chosen assumption and its treatment of concessions into the existing model. Assign a refresh date and retain the evidence behind any later change.

What this analysis can support

This process helps challenge a rent assumption and prioritize follow-up. It does not replace property-level diligence, executed lease evidence or the team’s investment judgement. Its commercial value is repeatability: the next property should not require rebuilding the same collection and checking process from scratch.

Sources and further reading

Sources checked 12 October 2026. Examples and workflow suggestions are Dart editorial guidance.

• RealPage, 9 June 2026: May 2026 concession use and discount depth