Restricted rents and the operating pro forma — California

What happens, and in what order

The structural risk is that nothing in this sequence has a natural human checkpoint. A utility allowance error propagates silently into net rent, NOI, supportable debt, the equity gap and the credit request.

Step one: which table applies

Industry consensus is that electing PIS is usually a mistake, because the default already preserves the higher of the two.

A separate absolute 10 percent cap, said to apply since FY2024 and to bind at 10.0 percent for FY2026, is reported but was not confirmed against HUD's own FY2026 income limits methodology document, which returns a bot challenge to automated retrieval. Treat it as a hypothesis until someone pulls that PDF through a browser session.

The formula, and the two ways spreadsheets get it wrong

A unit is rent-restricted if gross rent does not exceed 30 percent of the imputed income limitation applicable to the unit (IRC Section 42(g)(2)(A)). max_gross_rent_monthly = FLOOR(imputed_income_limit × 0.30 / 12).

FLOOR, not ROUND — an off-by-one error over the limit is a Section 42 noncompliance finding.

The 100% AMI row does not follow the formula: HUD derives it from the already-truncated 50% rent rather than recomputing it. It is not a LIHTC tier — Section 42 caps at 80 percent — and it appears only for manager and market-rate reference. It needs a separate code path.

"60% AMI" is not 60 percent of AMI

This is the single most consequential misunderstanding in the domain, and it bites hardest in exactly the high-cost markets California developers work in.

Every MTSP tier derives from the Very Low-Income Limit — the 50% column — which itself carries high-housing-cost adjustments, a state non-metro floor, national maximums and a cap on annual change.

HCD says the same thing in its own 2026 briefing Q&A, explaining why the very low-income limit does not equal 50 percent of AMI. In Los Angeles County for 2026 the low-income (80%) limit exceeds the AMI outright, which forced HCD to adjust its own moderate-income limits in Los Angeles, Santa Barbara and Santa Cruz counties to avoid an inversion.

Never compute a tier by multiplying a median. Read the published tier.

Utility allowances — the deduction that decides the deal

Under 26 CFR Section 1.42-10, gross rent includes a utility allowance for any utility other than telephone, cable or internet paid directly by the tenant. Actual-consumption submetering counts as tenant-paid (Section 1.42-10(e)).

California narrows (E). CTCAC Section 10322(h)(21) requires that any energy consumption model be the California Utility Allowance Calculator, most recent version, developed by the California Energy Commission and signed by a CABEC Certified Energy Analyst. CUAC use is limited to new construction, rehab meeting Section 10325(f)(7)(A), and existing credit projects adding photovoltaics with HERS-rater site verification. It is approved only upon completion of field verification; until then the project must use another Section 1.42-10 source.

The 90-day rule, Section 1.42-10(c)(1): when the applicable allowance changes, the new allowance applies to gross rents due 90 days after the change, and owners using methods (B) through (E) must submit estimates to the agency and make them available to tenants at the beginning of that window. Annual review is mandatory under Section 1.42-10(c)(2).

The allowance is a pure deduction from revenue, which is why it carries disproportionate weight.

That last figure is illustrative arithmetic, not an agency-published number, but the order of magnitude is the point.

Which revenue counts

Section 42(g)(2)(B)(iv): gross rent does not include any payment under Section 8 or a comparable rental assistance program. A project-based unit can therefore collect contract rent far above the LIHTC limit, because only the tenant portion is tested. That is why PBRA transforms LIHTC economics.

CTCAC draws a different line for its own underwriting. Section 10327(f) counts committed federal, state and local rental subsidies in cash flow after debt service, but expressly excludes income generated by tenant-based rental subsidies. Underwriting tenant-based vouchers as revenue is a common and expensive modeling error.

Project-based HAP contract rents adjust annually by HUD's Operating Cost Adjustment Factor.

The pro forma is a parameter set, not a model

All of the following is 4 CCR Section 10327, and all of it is versioned data that changes by rulemaking — none of it belongs hardcoded.

Analysts optimize for the floor and forget the ceiling. Over-performing on the pro forma reduces the credit award.

Section 10327(f) requires positive cash flow after debt service for a fifteen-year minimum from stabilized occupancy, and excludes expenses that do not persist all fifteen years from the feasibility, DSCR and cash-flow tests. Section 10327(g)(7) bars residential income from supporting negative commercial cash flow, and the reverse.

Where HCD money is in the stack, model the codified waterfall rather than a generic real estate cash flow.

One more thing that reaches back into this phase from year fifteen: the qualified-contract exit is legally unavailable in California, per H&SC Section 50199.14(f) and 4 CCR Section 10338(h). Any residual value assumption premised on year-15 conversion to market rate is not aggressive, it is impossible.

The operating expense line has no defensible source

Section 10327(g)(1) requires expenses at least equal to minimums "published by the Committee staff annually," by region and project type. Those minimums exclude property taxes, replacement reserves, depreciation and amortization, compliance monitoring and lender fees, and site and service amenity costs. The Executive Director may allow up to 15 percent less where the equity investor and permanent lender have agreed in writing. Partially-special-needs projects prorate.

They are not being published annually. The CTCAC 2026 application page links a file named 2022-operating-expense-minimums.pdf, re-hosted under a 2026 path. Downloaded and diffed against the original 2022 file, the content is byte-identical. The 2022 table is the operative underwriting floor for 2026 applications.

Treat that table as a regulatory floor to clear, never as an estimate of real expenses.

Insurance is now the most volatile line and there is no free published benchmark for it at all — it is quote-only. A pro forma set at the CTCAC minimum clears the regulation and distresses in year three.

There is also an unresolved detail with real dollars attached. The regulation says the minimums "shall include expenses of all manager units and market rate units," which reads as multiplying the per-unit minimum by total units rather than credit units. That reading is probably right but no confirming guidance was located; on a deal with a meaningful market-rate or manager component the difference is not trivial. Ask CTCAC staff rather than guessing.

Two more rent regimes, and what is genuinely unsettled

Notice the mismatch. The 2.5 percent income trending in Section 10327(g)(1) is an underwriting assumption. Section 10336(a) is a binding operational cap on what you may actually charge a sitting household. On acquisition/rehab and resyndication deals these interact directly, and a model that trends in-place rents at 2.5 percent per year can overstate year-two income — or describe something you are not permitted to do.

Unsettled, and worth saying plainly: nothing in the available sources resolves whether the AB 846 cap resets when a household turns over. The cap is expressed per household and the enumerated exceptions include transfers, which points toward a new household resetting to the AMI-derived limit. If that is right, the compounding drag is bounded by tenure and the modeling burden is modest. If it is not, the cap compounds across a 55-year term and the economics are materially different. Resolve it against Section 10336(a) and the CTCAC Rent Cap Memo before building anything that depends on the answer.

That June 23 memo expressly replaces a prior 2026 version dated May 29, 2026. HCD revised its own limits mid-year. Any cached copy is wrong, and the publication date belongs stored next to the values, not just the year.

Also contested: whether AB 846's amendment to Section 50053 makes CTCAC rent limits control for qualifying projects, superseding the Section 50053 methodology even where prior deed restrictions specify otherwise. That proposition comes from counsel summaries rather than from the amended statutory text, and it decides whether a California deal needs one rent engine or two. Fetch the current text of H&SC Section 50053 before relying on it.

Finally, a claim that circulates and does not survive arithmetic: that HCD's 1.20 first-year DSCR ceiling and CTCAC's 1.15 floor leave no feasible band. The interval 1.15 to 1.20 is not empty; the two tests do not bind in the same year, since CTCAC requires 1.15 in at least one of the first three years while HCD tests first-year 1.10 to 1.20; and DSCR rises under the prescribed trending, so a year-one 1.12 can satisfy HCD and still reach 1.15 by year two or three. The useful question is which constraint binds and in which year, which requires working the interactions through rather than asserting a conflict.