The Vermont tax that falls as income rises
Vermont’s own Tax Structure Commission matched education-tax bills to income-tax returns and concluded the education tax “is not a progressive tax” — its share of income climbs to about 3% near $140,000, then falls to roughly half a percent above $1 million. The finding comes first; the mechanics behind it follow: at the top of the ladder, income stops being wages, and a tax tied to a house cannot see pass-through profits or capital gains. Method & caveats · download the data.
The finding: Vermont matched taxes to incomes — the top pays a falling share
Under standing law, the Joint Fiscal Office and the Department of Taxes have matched homestead tax bills to income-tax returns at least three times. The Commission’s 2021 report, built on the 2017 match, recommended restructuring so the education tax “reduces regressivity that now occurs at high incomes” — Vermont’s own benchmark is a tax that rises with income, and its own data shows where the current one falls short. Our survey estimate follows; the matched data, the Commission’s modeled analysis, and our estimate all find the same fall at the top.
“Somewhat progressive up to incomes of about $140,000, and regressive at higher incomes… it is not a progressive tax.” Vermont Tax Structure Commission, 2021 · on the matched CY2017 data
MATCHED ADMIN DATA · CY2017 Median homestead education tax after the income adjustment, % of household income
From the Tax Structure Commission’s 2021 report (Figure 15; data: Joint Fiscal Office, all ~170,000 homestead filers). Solid points are values printed on the published chart; open points are read from it (±0.1 pt).
Behind the fall-off is a fact from the same match (Figure 13, printed exactly): the median housesite of a household making over $1,000,000 was $582,394 — about 4× the median for households under $47,000 ($144,896) — while incomes differ more than 20×. House value does not scale with income at the top.
SURVEY ESTIMATE · ACS 2019–2023 Property tax on the primary residence, % of household income — owner households
Our estimate from Census ACS microdata (11,323 Vermont owner-household records, replicate-weight standard errors as whiskers): all real-estate taxes on the primary residence ÷ household income, by owner-income decile plus the top 5% and 1%. Not administrative data; not the page’s tax-return percentiles.
First-decile ratios are high partly because the measure divides a house-linked tax by a single year’s income (retirees and loss years sit in low deciles with real houses). The estimate covers primary residences only — no second homes — and excludes renters. Income here is the ACS household concept, adjusted to 2023 dollars.
How “least regressive state” and “regressive at the top” are both true
Two claims about Vermont taxes sound like they cannot both hold. ITEP’s Who Pays? ranks Vermont 49th of 51 on its tax-inequality index — only Minnesota and the District of Columbia score more progressive — while Vermont’s own Commission calls the education tax regressive at high incomes. Both are true, and ITEP’s own Vermont table shows why. The full published rows:
MODELED · ITEP “WHO PAYS?” 7TH ED (2024) Vermont state & local taxes as a share of family income
ITEP’s microsimulation of 2024 tax law at 2023 income levels, non-elderly tax units (source). A model, not the matched administrative data above — its property row also counts renters’ passed-through tax, vehicle taxes, and business property. Values as published.
| Tax | Lowest 20%under $27.5k | Second 20%$27.5k–55.4k | Middle 20%$55.4k–83.3k | Fourth 20%$83.3k–135.9k | Next 15%$135.9k–240.2k | Next 4%$240.2k–588.5k | Top 1%over $588.5k |
|---|---|---|---|---|---|---|---|
| Sales & excise taxes | 5.9% | 3.8% | 3.2% | 2.7% | 2.2% | 1.3% | 0.8% |
| Property taxes | 2.2% | 3.6% | 3.8% | 4.4% | 4.3% | 4.2% | 3.0% |
| Income taxes | −2.0% | 0.6% | 2.4% | 3.3% | 4.0% | 4.7% | 6.2% |
| Other taxes | 0.2% | 0.2% | 0.1% | 0.1% | 0.1% | 0.1% | 0.1% |
| Total | 6.3% | 8.2% | 9.6% | 10.5% | 10.6% | 10.3% | 10.1% |
Three readings of this table reconcile the ranking with the finding on this page. First: the ranking is carried by the income tax, not the property tax. The income-tax row runs from −2.0% at the bottom — refundable credits pay out more than the tax collects — to 6.2% at the top, and Vermont’s sales & excise reliance is comparatively light (ITEP: eight of the ten most regressive states lean heavily on sales taxes). Second: the property tax’s bottom end is repaired; its top end is not. The lowest-income fifth pays the smallest property share of any group — 2.2% — the income-sensitized credit at work (the matched data above shows the same repair directly: roughly 13% of income before adjustment at $10–20k, 1.9% after). The same row then peaks at 4.4% in the fourth quintile and falls to 3.0% for the top 1% — ITEP’s model independently reproduces the fall the Commission found in the admin data. In ITEP’s words, Vermont is “progressive through the bottom part of the income distribution and regressive through the top part.” Third: the ranking compares whole systems across states; the fall at the top is a fact within Vermont’s. The total row rises to 10.6% of income at $135.9k–240.2k, then falls — 10.3%, then 10.1% for the top 1%. Vermont outranks 48 other systems because its income tax does the progressive lifting and its credits repair the bottom — not because its largest levy rises with income. Property taxes ($1.67bn education + $702M municipal billed) exceed the state income tax ($1.16bn), and their share of income falls as income rises at the top in every dataset on this page, ITEP’s model included.
Mechanics · 1 of 4Income at the top is not wages
The first mechanism behind the falling share: what counts as income changes up the ladder. In every group but one, wages are most of income — at the top 1%, business and investment income take over. A tax tied to a house can track a paycheck; it cannot track pass-through profits or capital gains, which is where top incomes live. First, who the top of the ladder is:
Income composition, bottom quartile to top 1% — tax year 2022
Each bar is one group of Vermont returns, ordered up the ladder. Segments are that group’s income by source.
TY2018 note: the IRS reported IRA distributions and pensions as a single merged line that year; “Retirement” here is that merged value.
“Other” is AGI minus the seven named sources: rents & royalties, taxable Social Security, unemployment, state refunds, alimony, and everything else not itemized.
Mechanics · 2 of 4The income the house cannot see
The same ladder read the other way: the income types that escape a house-linked tax are exactly the ones that concentrate at the top. If each kind of income were spread evenly, the top decile would collect 10% of each — the dotted line. Pass-through and capital-gains income run at 8–9× that line; wages at ~3×. The year control in part 1 applies here too.
Top decile’s share of the statewide total, by income type — tax year 2022
Each income type, split across the seven groups — tax year 2022
One bar per income type; segments are the share received by each group, darker = higher up the ladder.
Small net-negative components (business losses in the lower groups) are not drawn; the data table carries exact values, including negatives.
Mechanics · 3 of 4The two Vermont taxes have opposite shapes
Vermont’s income tax is progressive but small; its education property tax is larger and only tracks income below the sensitivity cap (≈$115,000) — above it, the bill follows the house, and house values stop rising with income. For every group the linked data covers, the property tax is the larger of the two.
ADMIN DATA · CREDIT CLAIMANTS ≤ ~$115K Vermont income tax and housesite education property tax, % of income — by income group
Vermont income tax: net tax ÷ AGI, all resident returns, TY2024 (VT Dept of Taxes percentile file). Education property tax: net housesite tax after the income-based credit, for the 88,304 credit-claimant households (FY2025–26 bills on 2024 household income), their $10,000 income bands mapped to the percentile groups by the 2024 AGI cutoffs; the rate is shown as a range computed from each band’s own bounds, never an income estimate. Claimant household income and AGI are related but not identical concepts.
Above the 90th percentile the property component is not published, and is not imputed here. The credit’s income cap (≈$115,000) sits below the 90th-percentile floor ($180,889), and property tax attaches to parcels, not returns. Vermont’s modeled analysis (the Tax Structure Commission’s 2021 report, JFO) put homestead education tax near 2.5% of household income on average, falling as income rises past the sensitivity cap. The full estimates — Vermont’s own 2017 matched data and our current survey estimate — are the two charts at the top of this page.
Mechanics · 4 of 4The bigger tax is the house-linked one
Vermont raises more from the education property tax than from its income tax — $1.67bn in FY2025–26 education property taxes (1.44× the $1.16bn of TY2024 income tax), plus $702M of municipal levies. Who actually bears it by income is only knowable where the state links the two: the income-sensitized credit households below.
ADMIN DATA · CREDIT CLAIMANTS ≤ ~$115K Housesite education tax and the income-based credit — by household income
Average FY2025–26 housesite education tax for the 88,304 credit-claimant households, split into what they pay after the credit and what the credit covers. Real administrative data (VT Dept of Taxes) — the only published link between property tax and owner income. The tan segment — the income link — covers about 90% of the average bill in the lowest band and about 4% above $100,000; beyond the ≈$115,000 cap it is gone entirely.
The top of the ladder is absent by construction. The credit’s income cap ends this data around $115,000 of household income — above it (all of the top decile), education tax follows house value, not income, and no published dataset links the two. The matched data at the top of this page shows what that does to the share of income paid.
ADMIN DATA · CREDIT CLAIMANTS ≤ ~$115K The municipal layer: housesite town taxes and their credit — by household income
Town government is funded from the same grand list — $702M of municipal levies in FY2026 — with an income adjustment that reaches far less of the ladder than the education credit does. In the Department’s own statistics the municipal credit averages $1,092 in the lowest income band and is zero above the $40–50k band; the average municipal bill itself varies only 1.6x across a more-than-tenfold income range.
The Commission called the unadjusted housesite tax “extremely regressive.” Above the $40–50k band, the municipal levy is unadjusted for everyone. Vermont’s property-tax burden overall is the second-highest in the nation as a share of personal income (Lincoln Institute, 2021 data).
Context: ten years of true-decile data · 2013 – 2022
The IRS has published percentile cuts by state since tax year 2013, plus a one-off 2006 release — shown detached; no state percentile data exists for 2007–2012.
Share of all Vermont AGI reported by the top 10%, 5%, and 1%
What the top decile’s income is made of, year by year
Capital gains swing with markets — the 2021 spike and the 2022 fall are the same asset cycle, not a structural change. Read any single year against the run of years.
Context: the longer view · 2004 – 2022, a fixed-dollar proxy
Income composition of $200,000+ returns, 2004 – 2022
Partnership & S-corp income is only itemized from 2009 (dashed line) — before that it sits inside “Other,” which is why “Other” falls sharply at the break.
Returns above $200,000 as a share of all Vermont returns
Read directly from the “$200,000 or more” class through 2009; summed from the 200–500k, 500k–1M, and $1M+ classes from 2010. This file’s universe includes dependent filers, so its totals sit a few percent above the percentile series’.
County by county
How each county’s education property levy compares with its residents’ Vermont income tax — larger in every county — plus what each county’s whole income is made of, its tax rates, and where its high-income returns sit. Vintages vary by metric; the chart title shows which.
Vermont counties — education property tax ÷ residents’ VT income tax · tax year FY2026 vs TY2024
| County | Value | $200k+ returns |
|---|
Statewide:
These are FEDERAL rates (income tax after credits ÷ AGI) — Vermont’s state income tax is separate and much smaller: net Vermont tax ran 3.7% of AGI overall, 5.5% for the top 10%, and 6.5% for the top 1% in 2022 (VT Dept of Taxes, resident returns). For scale: the federal percentile file’s own 2022 rates are 12.4% on all returns, 18.9% for the top 10%, and 24.8% for the top 1%. County-level top-10%/top-1% rates cannot be computed — no county percentile data exists; the $200k+ slice is the closest county view. County rates use a slightly different tax concept than the statewide figures (about 2% lower in total) — compare counties with counties, not against the statewide lines. The Vermont state-tax county metric is TY2024 resident returns (net Vermont tax ÷ federal AGI, VT Dept of Taxes); its county totals exclude towns with fewer than ten returns.
A comparison of county totals, not a rate anyone pays. Property taxes are levied on all property in the county — whoever owns it, resident or not — while the income figures cover county residents’ returns. Arithmetically, a high ratio can come from lower resident incomes, higher property-tax levies, ownership from outside the county, or any mix; this page does not attribute it. Property taxes are FY2025–26 bills (PVR); income figures are TY2024. Education + municipal taxes only; local-agreement and special-district levies excluded. “Non-homestead” is the statutory category (32 V.S.A. § 5401) for property that is not a declared primary residence — rentals, commercial and industrial property, camps, and second or undeclared homes; the split reports where the levy lands, not who bears it.
Vermont’s own count · through 2024
The Vermont Department of Taxes publishes percentile statistics for resident returns through tax year 2024 — the freshest numbers anywhere, on a different universe (Vermont residents, Vermont filing rules). Shown separately; never merged with the IRS series.
Top-decile share of AGI on Vermont resident returns, 2014 – 2024
The top-10% floor on this count rose from $133,283 (TY2018, first year published) to $180,889 (TY2024); the top-1% floor in 2024 was $586,634.
What one dataset would settle
In 2021, on 2017 data, Vermont’s own Commission concluded the education tax “is not a progressive tax.” Everything since — the survey estimate, the credit-band data, the composition ladder — points the same way, but all of it is estimate or partial view. And the municipal levy — $702M more, from the same grand list — carries no income adjustment at all above the $40–50k band. One updated match — education and municipal tax as a share of income, full ladder, current year — would replace every estimate on this page with administrative fact. The law to produce it is already on the books.
Method & caveats
Returns, not people
Every figure on this page counts tax returns. A married couple filing jointly is one return — two moderate salaries can clear the top-10% floor together. Nothing here counts high-earning individuals, and none of it counts households.
Three series, three universes
- IRS percentile data (TY2013–2022, plus a one-off TY2006): true percentile cuts of federal AGI over returns with positive AGI, excluding dependent filers. The only source of exact top-decile composition, and the spine of this page.
- IRS Historic Table 2 (TY2004–2022): fixed AGI classes; the $200,000+ class is a proxy, not a decile. Its universe includes dependent filers, so its totals run a few percent higher (336,880 vs 312,312 returns in 2022). Two documented breaks: partnership/S-corp and qualified dividends first itemized in TY2009; the class read directly through 2009 and summed from three classes from 2010. The county map draws on the same program’s TY2022 county file (the $200,000+ stub), so the same caveats apply county by county; disclosure-suppressed county cells, where present, are shown as “not disclosed,” never as zero.
- Vermont Department of Taxes (TY2014–2024): Vermont resident returns under Vermont filing rules. Percentile groups since TY2018; a direct top-decile row 2014–2017. A different count of a different population — charted alone, never merged with the IRS series.
Categories
Named sources are wages & salaries; partnership & S-corp (pass-through business entities); net capital gains; ordinary dividends; Schedule C business income; taxable interest; and retirement (taxable IRA distributions plus pensions — one merged IRS line in TY2018). “Other” is the residual: AGI minus those seven. It holds rents & royalties, taxable Social Security, unemployment, state refunds, and alimony, none of which the source files itemize consistently across the span. The same seven named categories are used in every chart, so “Other” means the same thing everywhere. The seven ladder groups are non-overlapping bands built by subtraction from the IRS’s cumulative percentile groups; they partition every positive-AGI return exactly (the build fails if they don’t). In the lower groups, net business losses can make a source slightly negative — charts draw the positive components and the data tables carry the exact values, negatives included.
Volatility
Capital gains are realized when assets are sold, so they swing with markets: the top decile’s gains share roughly doubled into 2021 and fell back in 2022. No single-year composition figure on this page should be quoted as structural without checking it against the ten-year series.
Sources & downloads
IRS Statistics of Income: AGI percentile data by state and Historic Table 2 · Vermont Department of Taxes income statistics. Every number on this page: tidy CSV · page JSON. The build re-derives the IRS’s own published totals from the raw files and fails if any check drifts.
More Vermont data tools: the School District Comparer · the Legislative Record.
Hunger Mountain Intelligence · IRS SOI (TY2004–2022) · Vermont Department of Taxes (TY2014–2024) · all data embedded in this page · no tracking.