v8 · Colorado TABOR reform, state through special district

What TABOR Reform Would Do to
Budgets and Services, Top to Bottom

An interactive model of Colorado's revenue cap: pick the index the cap grows on, choose a reform option, and trace the result from the state general fund down through counties, cities, school districts and special districts — including the H.R.1 federal cost shift landing between October 2026 and January 2028.

Adjust anything in the model panel; the bar below follows you down the page. Pin a scenario to compare two configurations side by side.

Why H.R.1 belongs in a TABOR modelbackground

Federal cost shift, constitutional cap. Federal funds are exempt from TABOR's revenue limit. General Fund dollars are not. Every provision of H.R.1 that shifts cost from Washington to Denver converts TABOR-exempt revenue into TABOR-subject demand — the obligation rises, the cap does not move. Worse, H.R.1 phases the hospital provider fee from 6.0% down to 3.5%, shrinking CHASE — the very enterprise Colorado created in SB17-267 by permanently cutting its own cap by $200M. The cap reduction stays; the enterprise it bought does not. And because Colorado is one of ten county-administered states, the new eligibility workload lands on county human services departments, including counties like Routt that cannot raise revenue in response. H.R.1 is modeled here as added cost of maintaining services, which flows through all four layers below.

Correction carried forward. An earlier version treated school district de-Brucing as a gain for the district. Under Colorado's school finance formula it usually isn't. A district's local share is calculated first and the state backfills the difference up to total program funding — so when local property tax revenue rises, the state's obligation falls by roughly the same amount and the district ends up close to flat. School district de-Brucing mostly benefits the state, not the district. Only voter-approved mill levy overrides, which sit outside total program, are additive for the district. This is exactly why HB21-1164's mill levy correction was described as relieving the state's growing share of K-12 costs.

How the model is wired — four layersbackground

Layer 1

State fiscal capacity

Reform sets cap growth. Against the cost of maintaining services, that yields a squeeze or relief.

Layer 2

Transmission down

Five channels: HUTF, senior homestead backfill, human services county admin, K-12 state share, severance and mineral lease.

Layer 3

Local budgets and services

Each jurisdiction's revenue mix, state dependence, and its own TABOR status.

Layer 4

Spillovers

What local de-Brucing does back to the state, to taxpayers, and to inter-county equity.

01The model

Pick an index and a reform, set assumptions, toggle a recession. Everything below recalculates.

H.R.1 cost shift active

SNAP error rate under 10%

Model a recession in year 4

All 13 capped counties de-Bruce

H.R.1 federal loss backfilled 70%

How much of the provider fee and directed payment loss the state replaces rather than cuts. The single largest policy choice in the model.

County share of H.R.1 admin burden 35%

Colorado is county-administered. Arapahoe alone projects +10,000 work hours.

Index the cap grows on

Inflation 2.9%

Set by the index above; drag to override manually.

Population growth 0.7%

FY2027-28 state forecast is 0.2%.

Service cost growth 5.2%

Medicaid ~7%+, K-12 ~4.5%, other ~4%.

Assessed value growth 7.0%

Routt rose ~74% in the 2025 reassessment alone.

New construction growth 1.2%

Raises allowed local revenue; reassessment does not.

Projection horizon 10 yrs

Through 2036.

Returns sliders to defaults, clears recession and the 13-county toggle, and selects Current law.

02Index comparison

Every candidate, on level and on stability. Sorted by average growth — but read the volatility column before drawing conclusions from the first one.

Two results that invert the obvious ranking. First, national CPI-U would tighten the cap, not loosen it — it grew 2.55%/yr against Denver's 2.92% over 1992–2024. Denver ran hotter than the country, so the problem is the consumer basket, not the geography. Second, NHCCI's long-run average is below current law — highway construction costs rose 67% from 2003 to 2024, about 2.47%/yr, slightly under CPI over the same span. But it moved 6.5% in a single quarter in 2024 while CPI moved 0.3%, and rose 68% in the four years after 2020 Q4. Indexing a constitutional revenue cap to NHCCI would not have raised it much over two decades; it would have made it impossible to budget against. That is the case for treating volatility as a first-order criterion rather than a footnote.

Volatility — the same indexes, showing plausible year-to-year range

03The efficient frontier

Every point is a possible index or blend. Up is faster cap growth; left is more predictable. The green line is the frontier — blends below and right of it are dominated, meaning some other combination gives more growth and less volatility. There is no reason to ever choose a dominated option.

Single index Possible blend Efficient frontier Your current blend

Why blending never raises the level — and what it does buy. BEA's IPD is already a weighted composite of what state and local governments actually purchase, with weights that update automatically from real spending. Blending it with medical CPI or ECI double-counts components already inside it; blending it with consumer CPI pulls the result back toward the very basket the reform is trying to escape. Every blend therefore lands below IPD alone on growth. What blending genuinely buys is stability — an equal-thirds CPI/medical/IPD blend cuts volatility roughly 28% at a cost of about 0.31 points of growth — and political durability, since a formula that still contains consumer inflation is harder to attack than a pure government index. Blend for defensibility, not for level.

Bars show mean annual growth; whiskers show approximately ±2 standard deviations. Standard deviations are estimates calibrated to observed behavior, not published figures — treat the ordering as reliable and the exact widths as indicative.

Scenario comparison

Layer 1 — State position

Layer 2 — What flows down

The senior homestead backfill is the channel to watch. The state reimburses counties for property tax exempted under the senior and disabled-veteran exemptions — $203.3M statewide in FY2026-27. When there is a TABOR surplus it's paid as a refund mechanism; when there isn't — which the June 2026 forecast projects for FY2025-26 — it comes from the General Fund and competes with everything else. The General Assembly may reduce it in any year the budget doesn't allow, and counties have no recourse.

Layer 3 — Jurisdiction impacts

Green border = de-Bruced. Red = still under its own TABOR cap. School districts show the state-backfill offset described above.

04Historical backtest

Everything above projects forward. This runs candidate formulas backward against five years of published TABOR results — the only test that uses actuals rather than assumptions.

Fixed baseline — published TABOR results, FY2020-21 through FY2024-25

Certified growth rates and caps per Colorado state certifications. FY2024-25 cap ($19,153.4M) and growth rate (5.9%) match LCS Table 10B exactly; official revenue differs by $13.8M from the LCS June 2026 figure ($19,463.2M), likely a certification-versus-forecast vintage difference. Population is derived as certified growth minus the lagged Denver CPI, and reproduces the LCS-published 0.6% for FY2024-25 within rounding.

05Layer 4 — Spillovers from local de-Brucing

The question v1 didn't ask: if Routt de-Bruces, who else is affected?

→ Effect on the state

—

County de-Brucing produces no state offset — county revenue doesn't enter the school finance formula. The state's structural gap is completely unchanged.

→ Effect on other municipalities

No direct effect

Colorado has no recapture or equalization mechanism between counties. Routt keeping more of its own property tax does not reduce anyone else's revenue. The effect on other jurisdictions is political and competitive, not fiscal — see the equity problem below.

→ Effect on Routt taxpayers

—

De-Brucing is not free money. It means property owners stop receiving the credit and pay the full levy.

→ Effect on school districts

≈ Neutral for districts

Because of the backfill offset, a district that de-Bruces its base mills largely transfers money from the state to itself and back again. The state is the net beneficiary. Only mill levy overrides are genuinely additive at the district level.

The incidence point that makes Routt unusual. Bell Policy Center analysis finds that as much as half the housing stock in Colorado mountain resort communities is second homes or investment property. If that holds in Routt, a large share of the cost of de-Brucing falls on owners who do not vote in Routt County. That is politically favorable and worth naming plainly rather than leaving implicit — it is a meaningful part of why de-Brucing may be more achievable here than in a county whose tax base is mostly resident-owned. Routt-specific ownership data isn't published; county assessors generally don't track owner-occupancy, so treat the resort-wide figure as indicative only.

The equity problem with "just de-Bruce locally"

De-Brucing lets a jurisdiction keep what its own tax base generates. That means its value scales directly with how rich that base is — and Colorado's bases are wildly uneven, including within Routt County. Effective property tax rates run 0.28% in Steamboat Springs and 0.76% in Hayden, because Steamboat's values are so much higher that far fewer mills raise the same money.

So a strategy of "every jurisdiction should just de-Bruce" is not neutral. It delivers a great deal to resort counties with enormous assessed value and very little to property-poor rural counties with the same legal right. The model's representative property-poor county shows the gap. If local de-Brucing becomes the de facto answer to TABOR, inter-county disparity widens — and the state, which is the only entity that could equalize, is precisely the entity whose capacity is unchanged by it.

06Routt County: the ratchet

Routt County — published figures

2025 budget TABOR property tax and revenue credit	31.65%
2026 estimated credit (at 4% → 3% TABOR limit growth)	38.65% → 39.65%
2025 reassessment — Routt County valuation increase	~74%
2025 reassessment — statewide median	~40%
Assessed valuation increase used for the 2026 budget	11%
Effective property tax rate (state average 0.50%)	0.28%–0.32%

Routt County Board of Commissioners, Property Tax Limitations; Routt County Assessor. TABOR binds Routt more tightly than the separate 5.5% statutory limit.

Why it only moves one way. Allowed revenue grows at inflation plus new construction. Potential revenue grows with total assessed value, which in a resort market grows far faster. The credit is the gap, and because each year's limit builds on the last, the gap never closes on its own. Every reassessment ratchets it up; nothing ratchets it down.

What this model still doesn't capturelimitations · 7 items

Impact channels the model handles poorly or not at all — listed so the gaps are explicit rather than invisible.

Deferred capital, not operating cuts. Real jurisdictions under revenue pressure defer bridges, facilities, and fleet long before they cut staff, because the consequences are invisible for years. Denver's pre-2012 experience is the canonical case. This model converts shortfalls into operating FTEs, which almost certainly overstates near-term service loss and understates the long-term liability being accumulated.
The 3% emergency reserve. Applies to every jurisdiction regardless of de-Brucing status, and cannot be spent on an economic downturn — only a declared emergency. It is dead capital exactly when it would be most useful.
Debt and voter approval timing. Multi-year debt needs a vote regardless of de-Brucing, which ties capital planning to November election cycles and drives use of certificates of participation and lease-purchase structures that avoid the "debt" label. Real cost, hard to model.
Fee substitution and its regressivity. Locals shift from taxes to fees, which don't require a vote. Fees are generally less progressive than property or income tax, so the same total revenue lands differently across households. None of that shows up in a dollar-level model.
State assessment-rate policy as a direct local lever. SB24-233 and the 2023 special session changed local tax bases directly. A state that can't fix its own cap can still move local revenue substantially — a channel that operates independently of anything modeled here.
Population decline. If a jurisdiction's population falls, its local TABOR growth factor can approach or hit zero while costs keep rising. Several rural Colorado counties are in that position now, and it is the one scenario where the local cap tightens without any boom at all.
Interaction with the 5.5% statutory limit. A separate pre-TABOR property tax limit that still binds many counties independently. Routt's own materials note TABOR is the more restrictive of the two for them — but that isn't universal, and for some counties relief would require addressing both.

07Findings

Grouped by question rather than by when they were discovered.

On choosing the index

Index choice is worth about as much as the entire reform debate. Holding everything else constant, switching from Denver CPI to a floored BEA S&L IPD cuts the squeeze on state programs from −16.4% to −9.5% and improves Routt County from −$14.4M to −$10.4M a year — a larger swing than most structural reform options, achieved by changing one defined term.
Two candidates are worse than current law, and both look appealing at first. National CPI-U (−19.4%) and NHCCI (−20.0%) both tighten the cap relative to Denver CPI's −16.4%. National CPI-U because Denver ran hotter than the country; NHCCI because its long-run average sits below CPI despite recent spikes. "Index it to what roads actually cost" would, on a twenty-year record, have made Colorado's problem worse and far less predictable.
The two highest-growth indexes are the two you cannot use. CPI education (−5.5%) and ECI state and local compensation (−8.2%) top the table on level. Education is a narrow sub-index; ECI is circular, since government pay raises would raise the index that sets the cap that funds government pay raises. Ranking by growth rate alone selects for the options opponents dismantle fastest.
Every blend containing IPD underperforms IPD alone. Equal weighting gives 3.32%, budget-share 3.42%, IPD-dominant 3.51% — all below IPD's 3.65%, because blending a government basket with consumer indexes pulls toward consumers and double-counts categories already inside the deflator. The optimizer can beat IPD on both axes, but only by loading up on ECI and the education sub-index. The mathematically optimal answer is the politically indefensible one.
What blending buys is stability, and it is not free. Inverse-volatility weighting cuts volatility 39% against IPD alone at a cost of 0.21 points of growth; budget-share cuts it 27% for 0.23 points. A worthwhile trade if predictability matters more than level — decide which problem you are solving before picking weights.

On state reform versus local de-Brucing

Routt's best option is local de-Brucing, and it isn't close. Every state reform moves Routt's ten-year position by roughly $14M. Local de-Brucing moves it by more than $300M.
But it does nothing for the state, and that matters strategically. Local de-Brucing leaves the state's structural gap exactly where it was, and functions as a release valve: the jurisdictions with the strongest tax base and the most campaign capacity opt out first, steadily removing the most capable advocates from any state-level coalition. The 80–99% local de-Brucing rate is both evidence the system works and part of why state reform keeps failing.
School district de-Brucing is not what it appears. The state, not the district, is the main beneficiary of de-Brucing base mills. Anyone building a school-funding argument on local de-Brucing should check whether they are actually arguing for state budget relief.

On H.R.1

H.R.1 makes the state problem worse and the local problem barely different. At default assumptions it adds about 2.3 points to the squeeze on state-funded programs — roughly $1.05B/yr by 2036 and $7.7B cumulatively. But Routt County moves only from −$15.8M to −$16.2M. It is overwhelmingly a state-budget event that reaches counties as a thin administrative mandate, about $157K/yr for Routt, plus degradation of every transfer channel. The reform rankings survive it; the urgency changes, not the order.
The backfill decision dwarfs every reform option here. Slide backfill from 70% to 0% and H.R.1's annual cost falls from $1.05B to $398M — a larger swing than the gap between current law and full state de-Brucing. Whether the General Assembly replaces lost federal money or lets the cuts land is the most consequential fiscal choice in this window, and it is being made through budget mechanics rather than as visible policy.

On equity and coalitions

Local de-Brucing widens inter-county inequality. It delivers in proportion to assessed value, which is precisely backwards from need. Routt gains enormously; a property-poor rural county exercising the identical right gains a fraction as much.
The two failure modes are opposite, which is why coalitions fracture. Sales-tax jurisdictions fail in recessions; property-tax jurisdictions under the cap fail in booms. Turn on the recession toggle and watch Steamboat Springs and Colorado Springs invert relative to Routt County. No single reform serves both, and any statewide campaign has to hold both groups together anyway.

Scale figures remain estimates. De-Brucing status, Routt credit percentages, HUTF shares, the school finance backfill mechanic, and all state figures are sourced. Jurisdiction budget totals are working estimates used to size the model — replace them with adopted budget figures before using dollar outputs externally. The property-poor county is a representative archetype, not a named jurisdiction.