Insights · Public finance · Property tax
When property-tax relief shows up, it usually sticks. That’s the expensive part.
We tracked 13 major state homeowner tax reforms since 2006 through 18 years of Census data. Where relief showed up, it mostly stuck. So did the bill for it, and that bill is what Florida votes on Nov. 3.
At a glance
Where a major state reform cut homeowners’ tax burden relative to other states, at least half of the cut was still there five years later in all five cases, and in four of five after ten years.
In North Dakota and Texas, keeping relief costs more over time. North Dakota’s relief budget grew 5.5-fold from 2009–11 to 2023–25, and $44.5 billion of Texas’s $51 billion relief budget only maintains earlier cuts.
Texas is the exception: it bought down school taxes while other local levies grew by more than half after 2019. Judge any relief plan by who pays, what it costs in year ten, and whether it covers the whole bill.
- Major reforms with measurable relief that still held at least half of it five years later
- 5 of 5
- Rise in North Dakota’s two-year relief budget, 2009–11 to 2023–25 ($325M to $1.8B)
- 5.5×
- Yearly local revenue loss from Florida’s Amendment 3 once fully in place (state estimate); the official analysis lists no replacement revenue
- $11.86B
Here is what I’d say at a party: property-tax cuts last longer than their critics claim, and they cost more than the first-year price tag suggests.
That is not the usual complaint. The usual complaint is that relief melts: rising home values eat the cut within a few years. I expected to find exactly that. Then I tracked 13 major statewide homeowner reforms since 2006 through 18 years of Census data on what owners actually pay. Where relief showed up, most of it was still there. What had changed was who was paying for it.
That matters now. On Nov. 3, Florida votes on Amendment 3. It would raise the homestead exemption from non-school taxes to $250,000 by 2028. The Legislature’s Revenue Estimating Conference puts the cost to cities, counties and special districts at $4.95 billion in the first year, rising to $11.86 billion a year. The amendment needs 60% of the vote. Neither the Legislature’s analysis nor the Senate sponsor’s announcement mentions any replacement revenue.
The scoreboard
Bills rise almost everywhere, because incomes and home values do. Between 2006 and 2024 the median homeowner’s bill rose in every state. So “did bills go up after the cut?” is the wrong test. The right one is whether a state’s homeowners ended up better off than homeowners elsewhere.
The measure here is simple. For each state and year, I take the median homeowner’s property tax as a share of household income, from the Census Bureau’s American Community Survey. Then I subtract the same figure for the median state. In 2006, for example, the typical Texas homeowner paid 3.3% of income in property tax, against 2.5% in the median state. Then, for each reform, two questions: did that gap fall, and was the fall still there later?
Five reforms produced a drop big enough to measure: Indiana’s 2008 overhaul, North Dakota’s state buydowns from 2009, Idaho’s 2006 school-tax swap, South Carolina’s Act 388 of 2006 and Florida’s 2008 Amendment 1. Three others (Pennsylvania’s Act 1 and Texas’s 2006 and 2019 packages) produced no measurable relief at all, and five pure caps on growth, such as New Jersey’s 2% levy cap, showed no change beyond what similar states did anyway. Of the five with measurable relief, all five had kept at least half of it five years later. Their five-year changes beat 83% to 95% of the 513 five-year stretches in states with no reform in our log. Ten years later, four of the five were still ahead of where they started. Only South Carolina’s, one of the smallest, had faded on this measure.
Exhibit 1
Where relief showed up, it was still there five years later. Texas’s never showed up until 2023.
Change in homeowner burden gap vs. the median state, from the pre-reform level · percentage points of income
Indiana
2008 overhaul: state takeover, sales tax 6%→7%, bill caps
−0.32 → −0.22
Five years after → 2024
North Dakota
2009 state buydown of school levies; 2013 buydown folded into school formula
−0.24 → −0.30
Five years after → 2024
Idaho
2006 swap: school levy cut, sales tax 5%→6%
−0.31 → −0.51
Five years after → 2024
South Carolina
2006 Act 388: school levy off homes, sales tax 5%→6%
−0.12 → −0.04
Five years after → 2024
Florida
2007 rollback + 2008 Amendment 1 (exemption doubled)
−0.42 → −0.18
Five years after → 2024
Texas
2006 school-rate compression; 2019 compression + 3.5% city/county cap; 2023 exemption $40k→$100k + compression
+0.63 → +0.04
2021 → 2024: no measurable relief until the 2023 package
How to read it: each line is a state’s median homeowner tax-to-income ratio minus the median state’s, re-zeroed at its pre-reform level. Below zero means homeowners are better off relative to other states than before the reform. Figures above each chart: change five years after the first reform → change in 2024, in points.
Source: Theoria analysis of ACS 1-year PUMS (owner-occupied units, housing weights), 2006–2024; 2020 not released. 2005 excluded (income validation failure). Limits: descriptive, not causal; reforms are not random; ACS taxes refer to “last year,” so timing can shift by a year.
Indiana is the cleanest case. In 2006 its median homeowner’s tax bill was 23% below the median state’s. By 2009 it was 37% below. In 2024 it was still 39% below.
Where relief showed up, it mostly stuck. So did the bill for it.
The subscription
In four of the five states where relief held, the law handed the bill to someone else: shoppers or the state budget. In North Dakota, the price rose in every budget.
North Dakota began buying down school levies with state money in 2009 and folded the buydown into its school-funding formula in 2013. Afterward, the state’s homeowners went from paying about the typical state’s share of income to noticeably less, and they stayed there. But the state was buying a slice of a tax base that keeps growing. Property-tax relief cost North Dakota $325 million in the 2009–11 budget and $1.8 billion in 2023–25. That is 5.5 times as much, over a stretch when consumer prices rose about 44%.
“It works,” state Sen. Dwight Cook told a Forum Communications reporter after voters rejected abolishing the tax in 2012. “Our only problem is the taxpayers have forgotten they’re getting it.”
Exhibit 2
North Dakota’s relief has held. Its two-year price rose 5.5× from 2009–11 to 2023–25.
State-funded property-tax relief by biennium · $ millions, nominal
Source: North Dakota Legislative Council, “Property Tax Relief Programs – 2007-09 Through 2023-25 Bienniums” (LC 25.9237.01000, April 2024).
Note: Totals include school mill-levy reduction grants and formula buydown, homestead and veterans’ credits and other relief. 2013–25 school amounts are Department of Public Instruction estimates. 2007–09 predates the school buydown and is mostly an income-tax credit. Not adjusted for inflation; consumer prices rose about 44% from 2010 to 2024 (BLS CPI-U annual averages).
Other states sent the bill to the checkout counter. South Carolina, Idaho and Indiana each raised the state sales tax by a point, to 6%, 6% and 7%, to replace school property taxes. Renters got no property-tax cut but paid the higher sales tax: Purdue economist Larry DeBoer estimated in 2008 that Indiana’s median renter would pay $36 to $69 a year more. Florida’s 2008 exemption added no tax at all. University of Florida economists predicted it would barely change total collections, because local governments would raise rates to make up much of the difference.
The case against: Texas
Texans will say relief never lasts. In Texas, the data mostly back them.
Texas cut school tax rates by a third in 2006–07, cut them again in 2019 and 2023, and raised the school homestead exemption from $15,000 to $140,000 in four steps since 2015. Yet its homeowners’ burden gap over the median state widened from 0.81 points in 2006 to 1.44 in 2021. Only the 2023 package pulled it back to 0.85, roughly where it started.
The levies suggest why. The state bought down one slice of the bill, school taxes, while cities, counties and special districts kept growing. Since 2019, school levies rose 15%; everything else rose by more than half. Schools’ share of Texas property-tax levies fell from 59% in 2006 to 47% in 2025. Meanwhile the state budgeted $51 billion for relief in 2026–27, and $44.5 billion of it only keeps earlier cuts in place. “We’re building a large obligation,” said state Sen. Charles Perry.
Exhibit 3
Texas bought down the school slice. Since 2019 school levies rose 15%; everything else rose by more than half.
Texas property-tax levies by type of taxing unit · $ billions, nominal, by tax year
Source: Texas Comptroller of Public Accounts levy data as compiled by Texas Policy Research (“Texas Property Tax Levies 1998–2025,” updated April 29, 2026). Relief budget: Legislative Budget Board, “Foundation School Program and Property Taxes” (Feb. 17, 2025).
Note: Only the five tax years shown were extracted, so lines between points are straight-line joins, not annual data. A Texas County Progress summary of the Comptroller’s 2018–19 report puts 2019 county levies at $11 billion against $10.4 billion here (likely a definitional difference); on either figure, non-school levies grew by more than half from 2019 to 2025. Levies include business and other non-homestead property, so this shows where the bill grew, not who paid it.
Texas suggests that relief is likely to last only for the part of the bill someone has agreed to keep paying.
What to ask on Nov. 3
Amendment 3 answers a real squeeze. On published Census figures, Florida homeowners’ median bill went from 11% below the median state’s in 2014 to 2% above it in 2024. “Affordability is on the forefront for every Floridian,” said Florida Realtors president Chuck Bonfiglio Jr. Gov. Ron DeSantis argues that local governments have benefited from fast-growing property-tax collections and should control spending before asking homeowners for more. To the extent they can, some of the cost will come out of local budgets rather than other taxpayers.
But on this record, the cut is likely to last, and so is its cost. If it passes, $11.86 billion a year has to come from somewhere: higher rates on property without the exemption, fees, state help or lower spending. Jacksonville’s council auditor puts the city’s loss at $200 million in 2027, about a tenth of its operating budget. Of a cut that size, Chris Hand, a former chief of staff to Mayor Alvin Brown, said, “the word scalpel doesn’t really come into it.”
Florida is not alone. Indiana, Ohio, Wyoming, North Dakota and Texas all passed relief in 2025. This fall Oklahoma and North Carolina vote on new limits, and Wyoming on a new homestead exemption. Legislators writing these plans should do what budget offices do for any permanent program: name the payer in the law, publish the cost in year ten, and say whether the plan covers the whole bill or one slice of it. Voters, and companies that own property in these states, should ask the same three questions. In Florida, a Tax Foundation analysis estimates taxes on rental property could rise about 15% by 2031 if local governments raise rates to make up the loss.
A property-tax cut is a subscription, not a purchase. Before cheering one, find the name of whoever has agreed to keep paying.
What would change our mind
- The 2025 and 2026 Census releases show Indiana’s, North Dakota’s and Florida’s gaps drifting back to their pre-reform levels. That would mean the durability we measured was a phase, not a rule.
- A county-level rebuild (assessor rolls rather than survey medians) finds that the five “durable” reforms were offset within five years by rate increases on the same homesteads.
- Florida’s cities and counties absorb Amendment 3 in their FY2027–28 budgets mostly through efficiency, without raising rates on non-homestead property, adding fees or cutting services.
Sources, method and every number checked
Where each number comes from
Provenance audit completed October 22, 2026
Main sources: U.S. Census Bureau American Community Survey 1-year microdata and published tables B25103 and B25119 (2006–2024; no standard 2020 release), processed by Theoria; official documents from the Florida Legislature, the Texas Legislative Budget Board and House Research Organization, the North Dakota Legislative Council, the South Carolina Attorney General and state revenue notices; university research from Purdue and the University of Florida. Secondary sources, used where no primary source was reachable: Texas levy totals compiled by Texas Policy Research from Texas Comptroller data (a second summary differs on 2019 county levies, disclosed under Exhibit 3); CPI-U values from an Idaho Department of Labor reprint of BLS tables; reported quotes and local estimates from Forum Communications, the Texas Tribune, Jax Today and CBS12; ballot listings from the Tax Foundation and Ballotpedia. Key judgments, with their confidence: relief is likely to last only for the part of the bill someone has agreed to keep paying; Amendment 3’s cut is likely to last. All results are descriptive; no causal effect is estimated. Audited Oct. 22, 2026: 73 claims checked; 71 verified against their sources, recomputed by script or corrected, and 2 removed.
| Claim | Kind | Source |
|---|---|---|
| When property-tax relief shows up, it usually sticks. | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| We tracked 13 major state homeowner tax reforms since 2006 | Official / primary | Theoria (from U.S. Census Bureau microdata); Theoria |
| 18 years of Census data | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Where relief showed up, it mostly stuck. So did the bill for it. | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria; ND Legislative Council; Texas LBB |
| At least half of the cut was still there five years later in all five cases | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| ...and in four of five after ten years | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| In North Dakota and Texas, keeping relief costs more over time. | Official / primary | ND Legislative Council; Texas LBB |
| North Dakota's relief budget grew 5.5-fold from 2009-11 to 2023-25 ($325M to $1.8B) | Theoria analysis | ND Legislative Council; Theoria |
| $44.5 billion of Texas's $51 billion relief budget only maintains earlier cuts | Official / primary | Texas LBB |
| Texas bought down school taxes while other local levies grew by more than half after 2019 | Theoria analysis | Texas Policy Research; Theoria |
| $11.86B yearly local revenue loss from Amendment 3 once fully in place (state estimate) | Official / primary | Florida Legislature |
| Neither the Legislature's analysis nor the Senate sponsor's announcement mentions any replacement revenue (key fig: 'The official analysis lists no replacement revenue.') | Official / primary | Florida Legislature; Florida Senate |
| Property-tax cuts last longer than their critics claim, and they cost more than the first-year price tag suggests. | Official / primary | Theoria (from U.S. Census Bureau microdata); Theoria; Florida Legislature; ND Legislative Council; Forum Communications; Texas Tribune |
| The usual complaint is that relief melts: rising home values eat the cut within a few years. | Reported / secondary | Texas Tribune; Forum Communications |
| On Nov. 3, Florida votes on Amendment 3. It would raise the homestead exemption from non-school taxes to $250,000 by 2028. | Official / primary | Florida Legislature; Tax Foundation |
| REC: $4.95 billion in the first year, rising to $11.86 billion a year; cost to cities, counties and special districts | Official / primary | Florida Legislature |
| The amendment needs 60% of the vote. | Official / primary | Florida Legislature; Tax Foundation |
| Between 2006 and 2024 the median homeowner's bill rose in every state. | Theoria analysis | U.S. Census Bureau; Theoria |
| Measure: median homeowner property tax as share of household income (ACS), minus the median state | Official / primary | Theoria (from U.S. Census Bureau microdata); Theoria |
| In 2006 the typical Texas homeowner paid 3.3% of income in property tax, against 2.5% in the median state | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Five reforms produced a drop big enough to measure (IN 2008, ND 2009+, ID 2006, SC 2006, FL 2008) | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Their five-year changes beat 83% to 95% of the 513 five-year stretches in states with no reform in our log | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Three others (Pennsylvania's Act 1 and Texas's 2006 and 2019 packages) produced no measurable relief at all | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Five pure caps on growth, such as New Jersey's 2% levy cap, showed no change beyond what similar states did anyway | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Only South Carolina's, one of the smallest, had faded on this measure. | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Exhibit 1 title: Where relief showed up, it was still there five years later. Texas's never showed up until 2023. | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria; Texas LBB |
| Exhibit 1 chips: IN -0.32/-0.22; ND -0.24/-0.30; ID -0.31/-0.51; SC -0.12/-0.04; FL -0.42/-0.18 (pp, yr 5 / 2024) | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Exhibit 1 event labels (IN 2008 overhaul, sales tax 6%->7%, bill caps; ND 2009 buydown, 2013 formula; ID 2006 swap 5%->6%; SC Act 388 5%->6%; FL 2007 rollback + 2008 Amendment 1; TX 2006, 2019 + 3.5% city/county cap, 2023 $40k->$100k) | Official / primary | ND Legislative Council; Texas LBB; Texas House; Sales Tax Institute; S.C. Attorney General; Deseret News; Sales Tax Institute; Purdue University; UF BEBR; Bryant Miller Olive (law firm); TRERC |
| Exhibit 1 note: 2020 not released; 2005 excluded (income validation failure); ACS taxes refer to 'last year' | Official / primary | Theoria; Theoria (from U.S. Census Bureau microdata) |
| Indiana's median bill: 23% below the median state's (2006), 37% below (2009), 39% below (2024) | Theoria analysis | U.S. Census Bureau; Theoria |
| In four of the five states where relief held, the law handed the bill to someone else: shoppers or the state budget. | Official / primary | ND Legislative Council; Sales Tax Institute; S.C. Attorney General; Deseret News; Sales Tax Institute; Purdue University |
| In North Dakota, the price rose in every budget. | Official / primary | ND Legislative Council |
| ND began buying down school levies with state money in 2009 and folded the buydown into its school-funding formula in 2013. | Official / primary | ND Legislative Council |
| Afterward, ND homeowners went from about the typical state's share of income to noticeably less, and stayed there. | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| The state was buying a slice of a tax base that keeps growing. | Official / primary | ND Legislative Council |
| $325 million in 2009-11 and $1.8 billion in 2023-25; 5.5 times | Theoria analysis | ND Legislative Council; Theoria |
| Consumer prices rose about 44% (2010 to 2024) | Theoria analysis | BLS via Idaho DOL; Theoria |
| Dwight Cook quote, told a Forum Communications reporter after voters rejected abolishing the tax in 2012 | Reported / secondary | Forum Communications; ND Secretary of State (via handoff workbook) |
| Exhibit 2 bars and labels: $134M (2007-09), $325M (2009-11), $1,802M (2023-25); '5.5x the 2009-11 level' | Theoria analysis | ND Legislative Council; Theoria |
| SC, ID and IN each raised the state sales tax by a point, to 6%, 6% and 7%, to replace school property taxes | Official / primary | Sales Tax Institute; S.C. Attorney General; Deseret News; BSPR / Idaho Ed News; Sales Tax Institute; Purdue University |
| DeBoer estimated in 2008 Indiana's median renter would pay $36 to $69 a year more | Official / primary | Purdue University |
| Florida's 2008 exemption added no tax at all; UF economists predicted it would barely change total collections, because local governments would raise rates to make up much of the difference | Official / primary | UF BEBR |
| Texas cut school tax rates by a third in 2006-07 | Official / primary | Texas House |
| ...cut them again in 2019 and 2023 | Official / primary | Texas LBB |
| ...and raised the school homestead exemption from $15,000 to $140,000 in four steps since 2015 | Official / primary | Texas Secretary of State; Texas Tribune; Fox 7 Austin; Texas LBB; Texas Legislature / Ballotpedia |
| Texas burden gap widened from 0.81 points (2006) to 1.44 (2021); 2023 package pulled it back to 0.85 (2024) | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Since 2019, school levies rose 15%; everything else rose by more than half | Theoria analysis | Texas Policy Research; Texas Association of Counties; Theoria |
| Schools' share of Texas property-tax levies fell from 59% in 2006 to 47% in 2025 | Theoria analysis | Texas Policy Research; Theoria |
| The state budgeted $51 billion for relief in 2026-27, and $44.5 billion of it only keeps earlier cuts in place | Official / primary | Texas LBB |
| Charles Perry: 'We're building a large obligation' | Reported / secondary | Texas Tribune |
| Exhibit 3 points and labels: school $20.8B/$36.1B/$41.7B; other $14.2B/$30.5B/$47.8B (2006/2019/2025); end labels 'from $30.5B / $36.1B in 2019' | Theoria analysis | Texas Policy Research; Theoria |
| Texas suggests that relief is likely to last only for the part of the bill someone has agreed to keep paying. | Official / primary | Theoria (from U.S. Census Bureau microdata); Texas LBB; Texas Policy Research |
| Florida homeowners' median bill: 11% below the median state's in 2014 to 2% above in 2024 | Theoria analysis | U.S. Census Bureau; Theoria |
| Bonfiglio: 'Affordability is on the forefront for every Floridian' | Reported / secondary | Jax Today |
| DeSantis argues local governments benefited from fast-growing property-tax collections and should control spending before asking homeowners for more | Reported / secondary | CBS12 |
| On this record, the cut is likely to last, and so is its cost. | Official / primary | Theoria (from U.S. Census Bureau microdata); Theoria; Florida Legislature |
| $11.86 billion a year has to come from somewhere: higher rates on non-exempt property, fees, state help or lower spending | Official / primary | Florida Legislature |
| Jacksonville council auditor: city's loss $200 million in 2027, about a tenth of its operating budget | Reported / secondary | Jax Today |
| Chris Hand, former chief of staff to Mayor Alvin Brown: 'the word scalpel doesn't really come into it' | Reported / secondary | Jax Today |
| Indiana, Ohio, Wyoming, North Dakota and Texas all passed relief in 2025 | Reported / secondary | Pew |
| This fall Oklahoma and North Carolina vote on new limits, and Wyoming on a new homestead exemption | Reported / secondary | Tax Foundation / Ballotpedia |
| Tax Foundation: rental-property taxes could rise about 15% by 2031 if local governments raise rates | Reported / secondary | Tax Foundation |
| Median incomes match B25119 within the 25% QA rule (2006+) | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| PUMS tax medians run below B25103, by 8% or less in two-thirds of state-years; largest gaps in LA, MS, AL, AK | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| 513 windows; median +0.08; p10-p90 -0.23 to +0.30; slope -0.07 | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| SC median bill 52% below median state's (2006) and 54% below (2024) | Theoria analysis | U.S. Census Bureau; Theoria |
| Robustness: IN, ND, ID hold on published-median ratio and tax/value; SC timing-sensitive; growth limits no change beyond convergence; PA Act 1, TX 2006/2019 no measurable relief | Theoria analysis | Theoria (from U.S. Census Bureau microdata); U.S. Census Bureau; Theoria |
| Reform log retention values (IN 85%/77%, ND 58%/64%, ID 135%/63%, SC 73%/-15%, FL 346%/269%; initial relief -0.37,-0.41,-0.23,-0.17,-0.12; TX -0.09, +0.04; PA -0.06) | Theoria analysis | Theoria (from U.S. Census Bureau microdata); Theoria |
| Pennsylvania Act 1 (2006): gaming-funded homestead exclusion; who pays: expanded-gaming revenue | Official / primary | Commonwealth of Pennsylvania |
| Florida 2007 rollback (HB 1B) | Official / primary | Bryant Miller Olive (law firm); Florida Legislature |
Method, comparisons and the reform log
Method
Burden. For each state and year, the weighted median across owner-occupied households (ACS 1-year PUMS, TEN = 1 or 2, weight WGTP) of annual real-estate tax divided by household income adjusted with ADJINC. 2006–2019 and 2021–2024. The Census Bureau did not release standard 2020 1-year data, and 2005 income failed validation. Taxes are survey bins at midpoints through 2017 and TAXAMT from 2018. Because that coding change hits every state at once, all results use the gap: the state value minus the same-year median of 50 states + DC. The panel was built in an earlier phase of this project and audited here. Median incomes match published table B25119 within the 25% QA rule, and PUMS tax medians run below published B25103, by 8% or less in two-thirds of state-years; the largest gaps are in Louisiana, Mississippi, Alabama and Alaska, none of them reform states here.
Reforms. Logged from statutes and official summaries, independent of the outcome data: broad, enacted, statewide homeowner relief or limits. Event year = first ACS year whose “last year” tax period contains bills paid under the new law. Relief = lowest gap in the event year and the next two years, minus the mean gap of the two prior years. Measurable = at least 0.10 point. Retention = change at year 5 ÷ initial relief. These rules were written down before the event results were computed, but after the state-year table had been viewed, so treat them as fixed in advance, not blind.
Comparison. 513 five-year windows in states without a reform in our log within five years (Texas excluded); unlogged reforms elsewhere could sit in this group. Their 5-year gap change: median +0.08 point, 10th–90th percentile −0.23 to +0.30. States below the median drift up slightly (slope −0.07 per point of starting gap), so reform states starting below the median faced a headwind, not a tailwind.
Dollar comparisons. Statements that a state’s median bill was X% above or below “the median state’s” compare its published B25103 median real-estate tax with the median of the 51 state medians that year.
Robustness. On the ratio of published medians (B25103 ÷ B25119) and on the tax-to-home-value ratio, Indiana, North Dakota and Idaho hold on every measure. South Carolina holds on the dollar measure (its median bill sat 52% below the median state’s in 2006 and 54% below in 2024) but is timing-sensitive on the main measure. Florida’s 2008 relief overlaps the housing crash and is not causally separable from it. Pure growth limits (New Jersey’s and New York’s 2% caps, Rhode Island’s 2006 levy cap, Oklahoma’s SQ 758, Arizona’s Prop 117) showed no change beyond what similar states did anyway. Pennsylvania’s Act 1 and Texas’s 2006 and 2019 packages produced no measurable relative relief.
Not claimed. No causal effect is estimated. Medians are not representative bills. The ACS measures homeowners only, so renters’ and businesses’ burdens are not measured here.
Reform log
| State | Reform | Event yr | Initial relief (pt) | Retained, yr 5 | Retained, yr 10 | Who pays / design |
|---|---|---|---|---|---|---|
| Indiana | HEA 1001 (2008): state takes over school general fund and county welfare levies; 1% homestead cap | 2009 | −0.37 | 85% | 77% | Sales tax 6% → 7% (Apr. 1, 2008) + state budget |
| North Dakota | Mill-levy reduction grants (2009); formula buydown (2013) | 2011 | −0.41 | 58% | 64% | State general fund, growing ($325M → $1.8B per biennium) |
| Idaho | HB 1 (2006 special session): school M&O levy removed (~$260M) | 2007 | −0.23 | 135% | 63% | Sales tax 5% → 6% (Oct. 1, 2006) |
| South Carolina | Act 388 (2006): owner-occupied homes exempt from school operating millage | 2008 | −0.17 | 73% | −15% | Sales tax 5% → 6% (June 1, 2007) |
| Florida | 2007 rollback + Amendment 1 (Jan. 2008): homestead exemption $25k → $50k, portability | 2008 | −0.12 | 346% | 269% | No new revenue; local rates (housing crash overlaps) |
| Texas | HB 1 and companion tax bills (2006): school M&O rates cut one-third by 2007 | 2007 | −0.09 | Not measurable | Margins and cigarette taxes + state budget | |
| Texas | HB 3 + SB 2 (2019): compression to $0.93; 3.5% voter-approval rate | 2021 | +0.04 | Not measurable | State budget | |
| Pennsylvania | Act 1 (2006): gaming-funded homestead exclusion | 2009 | −0.06 | Not measurable | Expanded-gaming revenue | |
Closest earlier work
Pew’s Fiscal 50 project (Feb. 24, 2026) surveys the 2025–26 wave of state relief and local fiscal strain but does not measure whether past relief lasted. The Tax Foundation’s Amendment 3 analysis (Oct. 20, 2026) argues the cost would shift to non-homestead owners and renters; it does not track earlier reforms over time. Neither offers a cross-state, household-level test of how long relief lasts, which is what this piece adds.