"Which competition are we entering — and once we file, can we change our mind?"
What you are actually choosing
9% credits are rationed by a fixed annual ceiling. 4% credits are as-of-right, but the private activity bond volume cap that comes with them is competitively awarded — the bond award and the 4% credit award are decided together, on one application.
| 9% (competitive) | 4% (QRRP, bond-financed) | |
|---|---|---|
| 2026 statewide ceiling | $136,381,450 federal ($3.416 per capita × 39,355,309 population, + $1,943,714 returned/surplus) + $67,256,563 state credit | $4,829,170,308 bond volume cap — 90% of a $5,312,966,715 state ceiling, + $56,203,593 unallocated carryforward |
| Per-project cap | $2,800,000 max annual federal credit, any round (4 CCR § 10325(f)(9)(C)) | No per-project bond cap |
| Rationing mechanism | Fixed annual ceiling, allocated by round | Bond volume cap awarded competitively, 3 rounds a year |
| Applicant of record | Developer applies directly | The bond issuer (§ 5102(a)) — the developer is the "Project Sponsor" |
| 2026 issuer market share | — | California Municipal Finance Authority ~70% of applications (86 of 122 Round 1, 73 of 122 Round 2); CalHFA second (15, then 24) |
One vintage warning before citing anything: CDLAC comprehensively reorganized and renumbered its regulations on December 10, 2025 — Section 5231 is now Section 5106(f), and Section 5230 is now Section 5105. Most industry material still uses the old numbering, and the adopted text itself carries a stale internal pointer at Section 5106(f)(1)(D)(iii)(bb) back to "Section 5230(m)." CTCAC adopted the same day, but a line-by-line diff of Section 10325 against the December 11, 2024 version found every point value and threshold unchanged.
The 2026 break that reset the 4% math
The One Big Beautiful Bill Act (P.L. 119-21) § 70422(b) amended 26 U.S.C. § 42(h)(4)(B) to add a second path that waives the competitive-allocation requirement for 4% credits, based on how much of a building's basis is bond-financed.
| Path | Threshold | Condition |
|---|---|---|
| 50% path | ≥ 50% of aggregate basis (building + land) | No additional condition |
| 25% path | ≥ 25% of aggregate basis | At least one bond issue dated after 12/31/2025 must itself finance ≥ 5% of aggregate basis — a token 2026 issue bolted onto pre-2026 bonds does not qualify |
Measured per building, not per project — a scattered-site or multi-building deal can clear 25% in aggregate and still fail on one building. The denominator is depreciable basis plus land, not eligible basis (which excludes land) and not total development cost.
| Award timing | Cap |
|---|---|
| Round 2 of 2025 or earlier | 55% of aggregate depreciable + land basis (§ 5108(a)) |
| After Round 2 of 2025 | 30% of aggregate depreciable + land basis (§ 5108(b)) |
| Executive Director exception | Up to 40%, on a showing that permanent financing supports the larger award and recycled bonds aren't obtainable |
| Pre-2026 award extending expiration into 2026+ | Must restructure down to the new cap (§ 5108(c)(2)) |
Across 122 applications in 2026 Round 1, bond request ÷ total project cost had a median of 26.2%, a minimum of 22.5%, and a maximum of 37.2%.
One test is unresolved in the source material: whether the 25% path also carries a "placed in service in taxable years beginning after 12/31/2025" condition. One document asserts it does; another read the statutory text and found no such condition; two more state the test with no placed-in-service condition at all. Read P.L. 119-21 § 70422's effective-date section directly before building a stack on either reading.
The 9% side grew in the same act: § 70422(a) multiplies the state 9% housing credit ceiling by 1.12 for calendar years after 2025, on top of inflation. And on the bond side, the minimum set-aside under 26 U.S.C. § 142(d)(1) is 20-50 or 40-60, elected by the issuer at issuance — there is no average-income option on the bond side; that exists only at § 42(g)(1)(C) for the credit.
Points are a gate. The tiebreaker is the competition.
Points barely separate applicants in either program — the tiebreaker, not the score, decides who gets funded.
Every one of the 19 CTCAC-recommended 2026 Round 1 projects scored exactly 109.00; CDLAC's own maximum dropped to 112 points after it eliminated the leveraged-soft-resources scoring category. Time spent chasing the last two points is time not spent on what actually ranks you.
| Gate | Requirement | Citation |
|---|---|---|
| Lowest Income Targeting floor | Must score at least 45 of this category's 52 points, or the project is ineligible for 9% credits entirely | § 10325(c)(6) |
| Energy efficiency threshold | Not a scoring category — a pass/fail requirement. Rehab projects need a 10% post-rehab energy improvement (5% per site for scattered sites), electrification of 2 of 3 residential end uses, or electric-ready status. Older QAP summaries still describe it as worth points. | § 10325(f)(7)(A) |
The two tiebreakers reward opposite things
The two competitions don't just have different formulas — they reward opposite behavior. Both results are reported as a percentage.
| Component | What it rewards |
|---|---|
| (A) Leveraged soft resources | Soft funds defraying residential cost, × a size factor (0.75 + new-construction/adaptive-reuse tax credit units ÷ 200, only above 50 such units, capped at 150 units for a 1.50 max factor), ÷ total residential development cost |
| (B) Low eligible-basis request | (1 − requested unadjusted eligible basis ÷ total residential development cost) ÷ 2 |
| (C) Opportunity Area bonus | +20 non-rural Highest Resource, +10 non-rural High, +10 rural Highest, +5 rural High (percentage points, not a ratio) |
| (D) Rural set-aside bonus | +5 for rural set-aside projects in counties with no tax-credit applications in the prior 5 years (percentage points) |
(A) and (B) are ratios; (C) and (D) are percentage points. CTCAC publishes the sum as a percentage — 112.212%, for example — so an implementation that doesn't normalize units is silently wrong by a factor of 100.
| Numerator component | Value |
|---|---|
| Bedroom-adjusted unit multiplier | 0.9 studio · 1.0 one-bed · 1.25 two-bed · 1.5 three-bed (capped at 30% of units) · 1.75 four-bed+ (capped at 10% of units) |
| Production | $50,000 per bedroom-adjusted unit |
| Rent savings | (HUD FMR − AMI gross rent limit) × 180 months |
| Extremely-low-income unit | $20,000 per unit, up to half the tax credit units |
| Resource area bonus | $30,000 / $20,000 / $10,000 per bedroom-adjusted unit for Highest / High / Moderate Resource |
| Mass transit proximity | $25,000 |
Denominator: (bond request + state credit request) × (1 − w − c − d − s), where the bond request is the greater of the actual tax-exempt loan request or 27.5% of depreciable + land basis. Project-wide prevailing wage cuts the denominator 15%; adding a PCC § 2500(b)(1) project labor agreement or a GC apprenticeship-plus-healthcare commitment cuts it 18% (§ 5106(f)(2)(B)) — on the 9% side, labor is a cost line, not a scoring input.
That asymmetry is the whole decision: the 9% tiebreaker rewards soft money brought to the deal and a low eligible-basis ask; the 4% tiebreaker rewards deep targeting, family bedroom mix, location, and asking for less allocation. A deal already carrying prevailing wage for other reasons is structurally better suited to 4%. One cliff to know: if a 4% deal's average non-assisted-unit affordability drops below 40% AMI, every non-assisted unit is computed as though targeted at 40% AMI for rent-savings purposes — below that average you keep buying § 5105(d) points and stop buying tiebreaker benefit.
| County | 2BR limit | Delta | Effect on denominator |
|---|---|---|---|
| Riverside / San Bernardino / Imperial | $531,200 | −5.14% | d = −1.29% (denominator increases) |
| San Francisco | — | +30.00% (published cap) | d = +7.5% (denominator falls) |
Two things about this term are contested and unsettled in the source material. The published table caps the delta at 30%, but the underlying worksheet (Attachment 40, cell G155) computes the raw ratio × 0.25 with no visible cap — for San Francisco that's 18.1% rather than 7.5%. And while the arithmetic implies negative deltas apply as shown (subtracting a negative increases the denominator), no regulation text or memo confirms it. Reconcile against a published award before relying on a self-computed 4% tiebreaker in either direction.
Bucket election, and the doors that only open one way
Bucket election matters more than the score. Both agencies fund a mix of set-asides/pools and geographic regions, and moving between them after a loss isn't always allowed.
| Set-aside | Share of ceiling |
|---|---|
| Nonprofit | 10% |
| Rural | 20% (14% RHS/HOME/CDBG-DR sub-apportionment, 10% Native American sub-apportionment of the rural set-aside) |
| At-Risk | 5% |
| Special Needs | 4% |
| First supplemental | 3% |
| Second supplemental | Executive Director discretion |
Rural is mandatory and exclusive — an eligible rural-tract project must compete there and cannot enter a geographic region unless its region drew zero other eligible projects all year. If not awarded in a set-aside, a project competes in one of eleven geographic apportionments (unnumbered in the regulation): City of Los Angeles 17.6%, Balance of LA County 17.2%, Central Valley 8.6%, San Diego 8.6%, Inland Empire 8.3%, East Bay 7.4%, Orange 7.3%, South and West Bay 6.0%, Capital 5.7%, Central Coast 5.2%, Northern 4.4%, San Francisco 3.7% (shares of the ceiling as of February 1).
| Bucket | Lowest funded tiebreaker |
|---|---|
| At-Risk set-aside | 16.330% |
| Rural set-aside | 71.888% |
| Special Needs set-aside | 87.111% |
| Nonprofit set-aside | 107.092% |
| Central Valley region | 57.825% |
| Orange region | 50.680% |
| City of Los Angeles region | 64.122% |
| San Diego region | 65.521% |
| Balance of LA County region | 86.491% |
| East Bay region | 99.369% |
A 6.6x spread between the lowest (At-Risk) and highest (Nonprofit) cleared tiebreaker. Inland Empire had $1,030,264 available and zero recommended projects — its only applicant, San Jacinto Vista II in Perris, won through the At-Risk set-aside instead, at a verified 18.319%. San Francisco also had no recommended projects. This is one round (n=60), and Round 1 is historically smaller than Round 2 — these are observed histories of specific buckets, not a ranking, and shouldn't be used to run a counterfactual on your own self-computed tiebreaker.
| Pool | 2026 amount |
|---|---|
| Acquisition/Rehabilitation | $1,000,000,000 |
| New Construction ELI/VLI set-aside | $780,000,000 |
| New Construction Homeless set-aside | $650,000,000 |
| New Construction State-Funded Mixed Income | $310,000,000 |
| New Construction geographic apportionments | $1,749,424,098 |
| BIPOC | $130,000,000 |
| Rural New Construction | $130,000,000 (capped at 10% of QRRP, § 5101(a)(2)) |
| Supplemental | $79,746,210 |
The pool structure changes annually — 2025's separate Preservation and Other Rehabilitation pools merged into Acquisition/Rehabilitation for 2026, and Mixed Income moved from $479.5M to $310M. Geographic apportionment of the NC pool (§ 5101(d)) is Coastal 21%, Bay Area 21%, City of Los Angeles 17%, Balance of LA County 16%, Inland 16%, Northern 9%. Inland is Fresno, Imperial, Kern, Kings, Madera, Merced, Riverside, San Bernardino, Stanislaus and Tulare; Orange and San Diego are Coastal. Los Angeles County splits into two competing buckets on both sides.
| Rule | Consequence |
|---|---|
| CDLAC Rural / Acquisition-Rehab loss (§ 5106(a),(b)) | Cannot be reconsidered in the New Construction pool the same round |
| CDLAC BIPOC loss (§ 5106(c)(2)) | Can fall through to other pools — the one pool where losing isn't fatal |
| CDLAC set-aside order (§ 5106(d)(1)) | Homeless, then ELI/VLI, then Mixed Income — the smaller, apparently less-competitive pool can be strictly worse |
| CDLAC skip rule (§ 5106(e)) | Last project in a pool/region is skipped unless ≥80% of its request remains (100% in the year's final round); a skipped project can only jump the line if within 1 point of, and ≥75% of the tiebreaker of, the first project skipped |
| CTCAC housing-type skip (§ 10315(h)) | If a round's housing-type goal is already met (Large Family 65%, Large Family New Construction in High/Highest Resource 30%, Special Needs 40%, SRO 15%, At-Risk 15%, Seniors 20%, Rural Acq/Rehab 30% of rural credits), a project is skipped for a same-scoring application whose type goal is unmet — unless it's the highest-ranked in its own set-aside, apportionment, or region |
Neither skip rule is knowable at election time. Both are disclosed risks, not forecasts.
What actually binds in 2026 is state credit, not bond cap
The binding constraint on the 4% side in 2026 is not bond cap — it's state credit. Section 5106's opening clause voids the entire bond allocation if state credits were requested and are not scheduled to be awarded.
| Available | Awarded / requested | |
|---|---|---|
| State credit requested (122 applications) | — | $586,426,503 |
| New construction state credit | $195,265,536 | $191,935,441 awarded (~$3.3M balance, effectively exhausted) |
| State Farmworker Credit | $25,000,000 | $25,000,000 awarded (100% exhausted) |
| Bond volume cap (all of CDLAC) | $1,553,798,433 | $1,295,593,706 across 74 awards |
State credit ran roughly 3x oversubscribed while over $258M of bond cap sat unused.
| Region | Available | Allocated | Awards |
|---|---|---|---|
| Coastal | $128,582,671 | $129,020,531 | 10 (fully subscribed) |
| City of Los Angeles | $104,090,734 | $103,180,934 | 6 (fully subscribed) |
| Bay Area | $128,582,671 | $75,227,685 | 2 |
| Balance of LA County | $97,967,749 | $49,620,987 | 5 |
| Inland | $97,967,749 | $26,500,000 | 1 |
| Northern | $55,106,859 | $12,763,156 | 1 |
Don't conclude bonds are easy in general: 2025 Round 3 allocated $2,060,351,376 of $2,069,300,891 available (essentially fully subscribed), and 2026 Round 2 drew $2,276,692,657 of requests against $1,863,798,433 available (1.22x oversubscribed on bonds) with $695,011,237 of state credit requested. One and a half rounds of post-OBBBA data is not a trend — check the actual round-by-round subscription rate rather than a general belief about the market.
One structural mercy: the performance deposit is fully refunded if the only reason an allocation goes unused is failure to secure the state tax credit (§ 5006(b)(2)).
Calendar, cost of entry, and what follows the sponsor
| Round | Deadline | Scheduled meeting | Actual |
|---|---|---|---|
| Round 1 | February 3 | May 12 | Awards actually taken up June 22, 2026 |
| Round 2 | May 19 | September 1 | — |
| Round 3 | September 8 | December 9 | — |
Published meeting dates are soft — CTCAC's and CDLAC's own 2026 schedules even disagree on whether 4% Round 2 awards land August 18 or September 1. The authoritative artifact is the posted meeting agenda (published 10 days before each meeting), not the annual schedule PDF.
| Milestone | 9% (CTCAC) | 4% (CDLAC) |
|---|---|---|
| 2026 Round 1 filing → decision | Applications due Apr 7, 2026 → recommendations published Jun 22, 2026 (~11 weeks) | Deadline to award runs ~14–19 weeks |
| Assembly time before filing | 4–6 months | 3–5 months from "we want to apply" to a filed application |
| Award → financial close | A competitive project realistically underwrites to close within 6–9 months of award | Bond issuance no later than Dec 31 of the year following the award year (§ 5010(a)(2)); readiness compresses this — construction must start within 180, 201, or 222 days of allocation (§ 5105(h)) |
| Requirement | Window | Citation |
|---|---|---|
| Title report | Dated within 90 days of the deadline | § 5102(b)(1) |
| Market study | Inspected within 180 days | § 5107(b)(3) |
| Enforceable financing commitments | ≥ 50% of construction or permanent financing | § 5102(b)(5) |
| BIPOC pool prequalification | ≥ 15 business days before the deadline | § 5106(c)(1)(A) |
CDLAC allows a one-time reuse of a title report and CNA from the immediately preceding unsuccessful round, but the reuse is round-specific and expires.
| Fee | Amount | Citation |
|---|---|---|
| CDLAC filing fee | $1,500, non-refundable | § 5003(e)(1) |
| CDLAC second installment | 0.00035 × allocation actually used, due within 30 days of issuance | § 5003(e)(2) |
| CTCAC filing fee | $1,500 ($1,700 for resyndication and scattered-site) | — |
| Performance deposit | 0.5% of allocation requested, capped at $100,000, due within 20 calendar days of award; forfeited pro rata if under 80% of the allocation is used | § 5006(a), (c) |
| Appeals | $500 | § 5005 |
| Supplemental filing fee | $600 | § 5109(d) |
Program election isn't a per-deal decision at the sponsor level. CDLAC negative points (10 per occurrence per year, for two to three years) follow the Project Sponsor and Related Parties across deals, including mirrored CTCAC determinations (§ 5105(m)(6)-(7); CTCAC mirrors back at § 10325(c)(2)(T)). In a field where 76% of CDLAC applicants sit within one point of the ceiling, a single 10-point deduction is a two-year exclusion from competitiveness on both sides. Taking a Supplemental Allocation on an award from Round 2 2022 or later also triggers a tiebreaker reduction for the sponsor's next round (§ 5109(c)).
Revisions run both directions — CA-26-037 Kashia Windsor fell from 88.206% to 71.888%, while La Joya Commons II gained 2.236 points — so this isn't simply applicants inflating; it reflects genuine ambiguity about what counts as a leveraged soft resource. Treat your own tiebreaker as an estimate.
Hybrid, and what the sources do not say
The hybrid 9%/4% election belongs in this phase. The inputs the sources name are the simultaneous-phase structure and the combined developer fee limits, and the identified benefit is a size-factor boost to the 9% tiebreaker — because the size factor scales with new construction and adaptive reuse tax credit units up to the 150-unit cap, splitting a large project can move a phase from below the 50-unit floor or lift its factor toward 1.50. CTCAC's per-round applicant list carries a hybrid flag, so the population of hybrid filers is observable in public data.
What these sources do not contain is a worked hybrid structure, the combined developer fee arithmetic, or any published funding-line data broken out for hybrid deals. Treat the mechanics as a structuring question for bond counsel and the syndicator rather than something to model from the regulations alone.
What is knowable is the exposure. A hybrid puts you in both competitions at once, which means both sets of one-way pool doors, both readiness clocks with their separate rescission consequences, CDLAC's state-credit precondition attaching to the 4% half, and the cross-mirroring of negative points between the two agencies. The upside is real but it is not a way to hedge the election — it is a way to be subject to all of it.
Where this goes wrong
- Electing the Rural or Acquisition/Rehabilitation pool at CDLAC and losing. Section 5106(a), (b) bars those applicants from consideration in the New Construction pool in the same round. BIPOC is the only pool whose losers fall through (Section 5106(c)(2)). The smaller pool is not automatically the safer one.
- Running the aggregate-basis test at the project level. It is measured per building — a scattered-site or multi-building deal can clear 25% in aggregate and fail on one building.
- Using total development cost, or eligible basis, as the denominator of the bond test. It is depreciable basis plus land. Eligible basis excludes land; TDC includes items that are not in basis at all.
- Bolting a small 2026 issue onto pre-2026 bonds and assuming the 25% path is satisfied. The post-2025 issue must itself finance at least 5% of aggregate basis.
- Requesting state credit on a 4% deal without modeling its scarcity. Section 5106's opening clause voids the bond allocation entirely if state credits were requested and are not scheduled to be awarded — and in 2026 Round 1 state credit was roughly 3x oversubscribed while CDLAC allocated only $1,295,593,706 of $1,553,798,433 in bond cap.
- Deepening a 4% deal's average affordability below 40% AMI expecting more tiebreaker benefit. Below that average, every non-assisted tax credit unit is computed as though targeted at 40% AMI for rent savings; you keep buying Section 5105(d) points and stop buying tiebreaker.
- Treating a self-scored tiebreaker as the number you will be ranked on. In 2026 Round 1, 13 of 22 recommended and alternate 9% projects had theirs revised by CTCAC staff, by as much as −17.014 percentage points, in both directions.
- Choosing 9% for a project whose credit need exceeds $2,800,000 of annual federal credit — that per-project cap applies in any round and is not waivable through bucket selection.
- Assuming a geographic apportionment will fund because it has money in it. The Inland Empire 9% region had $1,030,264 available and zero recommended projects in 2026 Round 1; its only regional applicant carried a 33.37% self-scored tiebreaker.
- Treating published award-meeting dates as fixed. CDLAC's 2026 Round 1 meeting was scheduled for May 12 and the awards were taken up June 22; CTCAC's and CDLAC's own 2026 schedules disagree on the Round 2 4% award date.
- Hardcoding the CDLAC pool list. 2025's separate Preservation and Other Rehabilitation pools merged into Acquisition/Rehabilitation for 2026, and Mixed Income moved from $479.5M to $310M.
- Running a 9% score-optimization exercise. 96.7% of the 2026 Round 1 field scored 109 of 109 and every recommended project scored exactly 109.00 — the points are a gate, not a differentiator.
- Citing CDLAC Section 5231 for the tiebreaker. It was renumbered to Section 5106(f) on December 10, 2025, and the adopted text still contains a stale cross-reference back to the old numbering.
- HUD
- LIHTC
- State QAPs
- IRS § 42
- Housing Finance Agencies