The average-income test: designations, flexibility and the compliance edge

The average-income test is the third minimum set-aside election under IRC § 42(g)(1)(C), added in 2018 alongside the classic 20-50 and 40-60 tests: at least 40% of units must be income- and rent-restricted, each unit is designated at a level in 10-point bands from 20% to 80% of AMI, and the average of the designations may not exceed 60% AMI.

For four years the election was legally live but practically radioactive — sponsors feared one over-income unit could break the average and take down the whole project's credits. Treasury's 2022 final regulations at 26 C.F.R. § 1.42-19 defused that, and state agencies layered their own election rules on top — in California through the CTCAC regulations.

The minimum set-aside election is made on Form 8609 and is irrevocable — average income is a placed-in-service decision, not something to revisit later. Each restricted unit carries a designated level (20, 30, 40, 50, 60, 70 or 80% AMI), and that designation drives everything unit-specific: the income limit for qualifying the household and the rent ceiling, both computed off the AMI schedules built on HUD's Part 5 income definitions. A 30% unit and an 80% unit sit in the same project under one election, with rents roughly proportional to their bands.

The regulations require designations to be recorded and communicated in the manner they prescribe — this is not an internal spreadsheet convention but a compliance fact about each unit, and the state agency's procedures govern how and when designations are reported. Sloppy designation records are the single most preventable audit exposure in an average-income project.

The 2018-era fear was a cliff: if one designated unit fell out of compliance and dragged the average above 60%, the reading then in circulation suggested the project could fail its minimum set-aside entirely — every unit's credits at risk from one bad file. The 2022 final rules rejected that. A noncompliant unit has unit-level consequences, but the project keeps its minimum set-aside so long as a qualified group of units still satisfies the 40%-and-average test. The rules also allow designation changes within limits — including changes that restore the average — where the pre-2022 framework treated designations as effectively frozen.

• Underwriting watch-outs:
• Designation changes are permitted within the regulation's limits and procedures — document every change contemporaneously and per the agency's rules, or the flexibility evaporates in an audit.
• CTCAC layers its own election and reporting rules on federal minimums; California deals comply with both or neither.
• The available-unit and next-available-unit mechanics run off each unit's designation, which is materially more complex than a flat 40-60 project — train property management before lease-up, not after the first over-income recertification.
• Some investors and lenders still price average-income deals conservatively out of 2018-era memory; expect diligence questions the regulations have already answered.

The election exists for cross-subsidy: 70% and 80% AMI units carry rents well above the 60% ceiling, and that extra income funds 30–40% AMI units the project could not otherwise support. That makes average income the natural fit for mixed-income submarkets, and the deeper targeting it enables — real 30% units instead of a uniform 60% project — is exactly what state scoring systems reward.

The trap is the high band. An 80% AMI rent ceiling only binds where market rents sit above it — in soft submarkets the "restricted" 80% unit rents at market like any other apartment, and the restriction adds compliance cost without revenue. Underwrite every band to the lower of the achievable market rent or the ceiling, and demand a market study that breaks out achievable rents band by band rather than blessing the ceilings wholesale.

• Underwriting watch-outs:
• If market rents undercut the 70–80% ceilings, the cross-subsidy engine that justified the election does not exist — rerun the deal as a 40-60 project and compare.
• Administration is a permanent operating cost: designation tracking, band-specific income limits, and utility-allowance math across seven possible levels.
• An average pinned exactly at 60.0% leaves no cushion; many sponsors designate to a slightly lower average so one unit swap does not threaten the test.