Are Property Taxes Regressive? Why the Denominator Changes the Answer

In September 2026, Personal Finance Club published a five-slide chart with a striking headline: the median US family pays 29 times more in property tax than a family in the top 0.1%, relative to net worth. The two bars read 2.2% and 0.08%. Personal Finance Club, September 2026.

The arithmetic in that chart is correct. We checked every step. The interesting question is what the ratio measures, because “relative to net worth” is a choice, and a different choice gives a different answer to the same question about the same tax.

Short version: property tax divided by net worth is mostly a measure of leverage and portfolio mix. An identical house with an identical bill scores 1% for an owner with no mortgage and 20% for an owner at 95% loan-to-value. Measured consistently, the gap between the middle of the US wealth distribution and the top 0.1% is real but closer to 6x. And the strongest evidence that property taxes fall harder on lower-value homes comes from how assessors value houses.

The claim, and its arithmetic

The chart builds two households and taxes both at 1% of home value:

Every one of those steps is right. $429,100 × 1% is $4,291. $4,291 ÷ $192,900 is 2.2245%. $141,000 ÷ $184,000,000 is 0.076630%. The ratio of those two is 29.03, so “29X” is accurate on the unrounded inputs.

One small note for anyone who tries to reproduce it: you cannot, from the digits printed on the slides. 2.22 ÷ 0.08 gives 27.75x, because 0.0766% was rounded to 0.08%. Printing the third digit would close that gap.

The tax rate cancels out

The chart applies 1% to both households, which makes it look like a statement about property tax. Write the ratio out and the rate disappears:

With $V$ the home value, $W$ net worth, and $\tau$ the tax rate. Set $\tau$ to 1%, 0.5%, or 4% and the 29x is unchanged. The number compares how much of each household’s balance sheet is a house: 222% for the constructed median family, 7.7% for the top 0.1%. That is a fact about portfolio composition, and the property tax rate is along for the ride.

Where the inputs come from

The top-0.1% side is sourced correctly, and it is worth saying so before criticizing anything. $184M and $14.1M are Federal Reserve Distributional Financial Accounts figures for 2026:Q1. The DFA reports $25.072 trillion of net worth held by 136,095 households in that group, which works out to $184,226,327 each, with $14,234,792 of owner-occupied real estate. ² Those numbers are real.

They are also means. The DFA publishes group aggregates and household counts; it publishes no medians anywhere. The top 0.1% runs from roughly $46M at the entry threshold up to centibillionaires, so the median household in that group holds far less than $184M. The chart compares a median on one side against a group average on the other.

The median side has larger problems.

Combining medians from different distributions does not produce a real household. The median of a ratio is not the ratio of medians, so a family at the median of the net worth distribution does not automatically own a median-priced house.

The internal inconsistency

Slide 4 adds that home equity makes up 42% of the median family’s wealth. That figure appears to come from a different sentence in the SCF report, which says about 42 percent of families held debt secured by a primary residence. That is the share of families carrying a mortgage, not a share of anyone’s wealth.

Take the chart at its word anyway and follow the arithmetic. Equity of 42% × $192,900 is $81,018. Against a $429,100 house, that implies a $348,082 mortgage, or an 81% loan-to-value ratio. Three independent measures put the real figure near 28%: the SCF’s median leverage ratio among families who have a mortgage was 29.2% in 2022, DFA aggregate home mortgages were 28.4% of owner-occupied real estate in 2026:Q1, and the Financial Accounts put owners’ equity at 71.9% of household real estate in 2026:Q2.

The constructed median family has to be about three times more levered than the American housing stock for 2.22% to appear. That is the mechanism behind the headline.

What a mortgage does to the ratio

Suppose a house is a household’s only asset. Net worth is then $V - M$, and with $L = M/V$ the loan-to-value ratio:

Same house, same street, same assessment, same rate, same bill. The only thing that changed is how the buyer financed it. A household closing on its first home with 5% down scores twenty times “worse” than the identical household that paid cash, and the difference vanishes over the next thirty years as the loan amortizes, again with no change in the tax.

Other assets move it just as hard. With $A$ of financial assets alongside the house, the ratio is $\tau V / (V - M + A)$. Adding $100,000 of retirement savings takes a 90%-LTV household from 10.0% down to 3.0%. As savings accumulate, everyone’s ratio falls toward zero. The top 0.1% score well on this measure because roughly 92% of their balance sheet is something other than a house.

Try it with your own numbers

The calculator below shows one property tax bill against three denominators at once. Move the mortgage slider and watch the bill stay fixed while the share of net worth swings. The effective rate and the share of income do not move at all.

Five things “burden” can mean

Each denominator answers a different question, and none of them is the single correct one.

One honest limitation: we cannot complete the income version of the chart’s own comparison. The DFA publishes no income figures by wealth group, and the SCF suppresses cells at the top of the wealth distribution. Property tax as a share of income for the top 0.1% is not computable from public data, so nobody should quote one.

Measuring both sides the same way

Use one source, one concept, and one quarter for every group, and the picture changes. These are DFA per-household means for 2026:Q1 at an assumed 1% rate.

On a consistent basis the multiple is about 5.7x for the middle of the distribution and 14.6x for the bottom half, against 29x in the chart. The gradient is real and it runs the direction the chart claims. It is not as steep as advertised.

The bottom-50% row carries the most information. That group’s mean real estate exceeds its mean net worth, which is another way of saying its housing is heavily mortgaged and it owns comparatively little else. That is a statement about who holds leveraged housing, and it is the real finding underneath the viral number.

Correcting the original comparison one step at a time gets to a similar place:

Three views of who actually pays

Economists separate statutory incidence, meaning who receives the bill, from economic incidence, meaning whose real resources fall once prices and behavior adjust. Property tax is one of the harder cases, and the profession has not settled it. George Zodrow put it plainly in 2006: “the debate over the incidence of the residential property tax has raged for at least the last thirty years, and is still far from resolved.” ⁶

Three frameworks compete, and they disagree about the answer.

• The traditional or excise view. Following Mieszkowski (1972), the tax on structures behaves partly like a tax on housing consumption. Since housing spending rises less than proportionately with income, tax-to-income ratios then look regressive.
• The benefit view. Tiebout (1956) modeled households sorting among jurisdictions by their preferred bundle of taxes and services. Hamilton (1975) showed that adding zoning can convert the property tax into something closer to a price for local services than a redistributive tax. Under this reading the distributional effect is roughly neutral, because the payment buys something.
• The capital tax or “new” view. Treating the national average property tax as a tax on capital makes it progressive, since capital ownership is concentrated at the top.

These are live positions. Oates and Fischel (2016) argued the benefit view applies to most of the US population; Zodrow (2023) answered with five specific objections and concluded that “many of these issues are still not fully resolved.” ⁸ A five-slide chart does not resolve a seventy-five-year argument.

Taxes and services show up in the price of the house

Expected future taxes and expected local services get capitalized into what buyers will pay. Cellini, Ferreira and Rothstein studied narrowly decided California school bond elections and found that marginal homebuyers were willing to pay “$1.50 or more for each $1 of capital spending,” with their most conservative estimate at $1.13. ⁹ That willingness to pay already includes the higher future property taxes the bond creates.

Two caveats travel with that result. The authors found test scores explain only about a sixth of the price effect, so this is not a clean measure of school quality. And Biasi, Lafortune and Schönholzer (2024) extended the design to 29 states and found that once state aid to districts is netted out, homeowner valuations roughly match local spending, which undoes the under-investment reading. ¹⁰

Renters pay some of it too

A homeowner-only chart leaves out the third of households who rent and still bear part of the tax through their rent. How much is contested. Sarah Baker used California’s Proposition 13 to generate quasi-random differences in landlords’ tax bills and estimated pass-through of $0.50 to $0.89 per $1 of tax shock to new tenants in Berkeley. ¹¹ That is one rent-controlled city, new tenants only, in a working paper, and the author frames it as a pricing anomaly explained by differences in landlord sophistication. It illustrates the gap between who writes the check and who bears the tax rather than settling the magnitude.

The stronger case for regressivity

Rejecting a weak argument for a conclusion is not the same as rejecting the conclusion. There is better evidence that US property taxation falls harder on lower-value homes, and it has nothing to do with net worth denominators. It comes from how assessors value houses.

McMillen and Singh (2020) documented that assessment ratios tend to run higher for lower-priced homes, enough to reverse the mild progressivity that homestead exemptions create, and enough to produce regressivity measured against income. ¹² Their 2026 follow-up calculated the homestead exemption that would eliminate regressivity in 9,091 municipalities and found 92.3% of them had at least somewhat regressive assessment practices. ¹³

Avenancio-León and Howard went further, with the clearest evidence of inequity in the whole literature. Their Quarterly Journal of Economics paper finds that, “holding taxing jurisdictions and property tax rates fixed, Black and Hispanic residents face a 10–13 percent higher tax burden for the same bundle of public services.” ¹⁴ Just over half of the gap arises between neighborhoods, because assessments respond less than market prices do to very local factors, and residential segregation then concentrates the error. The rest operates through appeals: minority homeowners are less likely to appeal, less likely to win, and receive smaller reductions when they do.

There is a genuine methodological fight here that a careful reader should know about. McMillen and Singh (2023) showed that regression-based measures of assessment regressivity carry a bias “that tends to imply regressivity even when it is not present,” because sale price appears in both the numerator of the assessment ratio and on the right-hand side of the regression. ¹⁵ Christopher Berry answered with tract-level correlations, Monte Carlo simulations, and repeat-sales evidence, arguing the measured regressivity is far too large to be explained by noise in sale prices. ¹⁶ Both camps now accept some bias exists; they disagree about its size relative to the effect. The finding survives in weakened form, which is a more useful thing to know than a verdict.

Worth noting that the OECD’s own review is more cautious than the headline suggests. It reports that several studies find recurrent property taxes regressive against current income, then argues those studies are limited, concluding that such taxes “may not be as regressive as generally thought, and may even have some progressive features.” ¹⁷ An earlier OECD working paper put the life-cycle version this way: “Over the life-cycle, the tax is likely to be neutral.” ¹⁸

Housing tax breaks and who receives them

Owner-occupied housing does get favorable federal treatment. The largest piece is invisible to most homeowners: the rental income you implicitly earn by living in a house you own is never taxed. Treasury scores that exclusion at $157.4 billion for fiscal 2026, its second-largest income tax expenditure by ten-year cost. ¹⁹ Sellers can also exclude up to $250,000 of gain on a principal residence, or $500,000 filing jointly, limits set in 1997 and never indexed to inflation. ²⁰

The Congressional Budget Office states the comparison plainly: “The tax code treats investments in owner-occupied housing more favorably than it does other types of investments.” A landlord deducts mortgage interest, property taxes, depreciation and maintenance but pays tax on net rental income and on gains at sale; a homeowner deducts mortgage interest and property taxes while never paying tax on the net rental value of the home. ²² CBO also finds the mortgage interest deduction “distorts the housing market by encouraging people to take out larger mortgages and buy more expensive homes, which pushes up housing prices.” ²³

Federal support extends past the tax code. CBO projects roughly $1.6 trillion of new housing and real-estate loans and loan guarantees issued in fiscal 2026, about 86% of all new federal credit assistance, most of it Fannie Mae and Freddie Mac mortgage guarantees. ²⁴ That figure is the face value of the credit extended. The budgetary cost is a separate and much smaller number: CBO scores the lifetime subsidy on that same portfolio as a $19.3 billion saving under standard federal credit accounting, and as a $15.3 billion cost on a fair-value basis.

Those benefits are not evenly spread, and they are not spread the same way as each other. Congressional Budget Office estimates for 2019 put 84% of the mortgage interest deduction’s value in the highest income quintile, including 25% in the top 1%. The home-sale gains exclusion in the same table went 44% to the top quintile and only 5.3% to the top 1%. ²¹ Two housing preferences, opposite distributions. Any claim that housing subsidies flow mainly to the middle class, or mainly to the rich, has to say which subsidy it means.

What we recommend

When you see a tax burden expressed as a percentage, find the denominator before you react to the number.

• For a question about a tax system, use tax over property value or tax over income. The first is the effective rate and is what assessors, budget offices, and most of the academic literature use. The second is the cash-flow burden and is the measure on which the regressivity case is usually made.
• For a question about your own balance sheet, tax over net worth is fine, as long as you read it as what it is. It tells you how large a recurring expense looks against what you own, which is a real thing to want to know while you are deciding whether a house is affordable. It is not an effective tax rate, and comparing yours to somebody else’s mostly compares leverage.
• Watch for mismatched vintages and populations. A 2022 median paired with a 2026 transaction price, or an all-families median paired with a homeowners-only asset, will generate a dramatic ratio out of nothing.
• Argue about assessments if you want to argue about fairness. Assessment ratios, appeal access, and assessment caps are where the measurable inequities in US property taxation live, and the research there is far stronger than any net worth ratio.

Frequently Asked Questions

Are property taxes regressive?

It depends on the denominator and on which incidence model you accept. Measured against current income, many property tax systems do look regressive, and regressive assessment practices make that worse. Measured against wealth, or under the capital tax view, the tax looks proportional or progressive. Over a life cycle the OECD describes it as roughly neutral. There is no single answer, which is why the question needs a specified measure attached.

Is the 29x figure wrong?

The arithmetic is right and the top-0.1% inputs are correctly sourced. The comparison is built from mismatched pieces: a 2022 net worth median, an August 2026 sale price, an all-families median against a group average, and a household whose implied 81% loan-to-value ratio is about three times the national aggregate. Measured consistently the multiple is roughly 5.7x for the middle of the distribution.

Why does the mortgage change the percentage so much?

Because the tax is assessed on the full value of the house while net worth counts only the equity. The ratio is $\tau/(1-L)$, which grows without bound as the loan-to-value ratio approaches 100%. A household at 95% LTV shows twenty times the ratio of an identical household that owns outright, on the same bill.

Do renters pay property tax?

Indirectly, to a degree that is genuinely uncertain. Landlords receive the bill and pass some share into rent. One quasi-experimental study of new-tenant rents in Berkeley estimated $0.50 to $0.89 per $1 of tax shock, which is specific to that market rather than a national figure.

How much revenue do property taxes raise?

About 30% of all state and local tax revenue, which was $630.2 billion in fiscal 2021 and $724.8 billion in fiscal 2024. Be careful with that share: as a percentage of state and local general revenue, which includes federal transfers and charges, it is closer to 15%. The tax is overwhelmingly local, supplying roughly 71% of local tax revenue and about 1.6% of state tax revenue.

Would a land value tax fix this?

It would change the incentives, since taxing land rather than structures removes the penalty on improving a property. It would not remove the measurement question this guide is about. A land value tax still has to be assessed, and its burden would still look different depending on whether you divided by land value, income, or net worth.

Key Takeaways

• Property tax divided by net worth is dominated by leverage and asset mix. At a 1% rate the same house and the same bill produce 1% for an outright owner and 20% at 95% loan-to-value.
• The tax rate cancels out of the 29x comparison. What remains is a ratio of housing shares of the balance sheet, so the statistic is not sensitive to property tax policy at all.
• The chart’s own inputs imply an 81% loan-to-value median household. Measured US aggregates put the figure near 28%, from three independent sources.
• Measured on one consistent basis, the gap is roughly 5.7x for the middle of the wealth distribution and 14.6x for the bottom half. Those are Federal Reserve group means for 2026:Q1 at an assumed 1% rate, and the gradient runs in the direction the chart claims.
• The measurable inequities in US property taxation sit in assessment practice. Black and Hispanic homeowners face a 10 to 13 percent higher tax burden for the same bundle of public services, holding jurisdiction and rate fixed.
• Economists have not agreed on who bears the property tax. The traditional, benefit, and capital tax views imply regressive, neutral, and progressive respectively, and the argument is still active in the journals.

Related Guides

• Should You Count Your Home and Cars in Net Worth?: the same denominator problem pointed at your own balance sheet rather than at a tax bill
• The True Cost of Owning a Home: where property tax sits among the carrying costs people forget to count
• How Much House Can High Earners Really Afford?: one question, three measurement frameworks, three different answers
• How State Taxes Affect Retirement: why states that lean on property and sales taxes have lower average burdens but more regressive ones
• Can 37% of Americans Really Not Afford a $400 Emergency?: another viral statistic that measures something narrower than its headline

Sources

1. Personal Finance Club. “The poor pay more” property tax carousel, September 2026. The claim under discussion.
2. Board of Governors of the Federal Reserve System. Distributional Financial Accounts. Net worth by wealth group, 2026:Q1. Group aggregates and household counts; no medians are published. Figures here are computed from the release file and reproduced by scripts/analyze_property_tax_denominators.py. DFA levels are allocated from Financial Accounts aggregates using SCF distributional shares, so post-2022 quarters extrapolate off the 2022 survey.
3. Aladangady, Bricker, Chang, Goodman, Krimmel, Moore, Reber, Volz and Windle. Changes in U.S. Family Finances from 2019 to 2022. Board of Governors of the Federal Reserve System, October 2023. Median family net worth $192,900; homeownership 66.1%; conditional median primary residence $323,200; median leverage ratio 29.2% among families with home-secured debt. The 2025 SCF was fielded through December 2025 and publishes in late 2026.
4. National Association of Realtors. Existing-Home Sales, August 2026, released September 10, 2026. Median existing-home sale price $429,100.
5. Board of Governors of the Federal Reserve System. Households; Owners’ Equity in Real Estate as a Percentage of Household Real Estate, Financial Accounts of the United States via FRED. 71.93% in 2026:Q2.
6. Zodrow, George R. “Who Pays the Property Tax?” Land Lines, Lincoln Institute of Land Policy, April 2006.
7. Tiebout, Charles M. “A Pure Theory of Local Expenditures,” Journal of Political Economy 64(5), 1956, 416–424. Hamilton, Bruce W. “Zoning and Property Taxation in a System of Local Governments,” Urban Studies 12(2), 1975, 205–211. Mieszkowski, Peter. “The property tax: An excise tax or a profits tax?” Journal of Public Economics 1(1), 1972, 73–96.
8. Zodrow, George R. “75 Years of Research on the Property Tax”, National Tax Journal 76(4), 2023, 909–940. Responds to Oates, Wallace E. and William A. Fischel, “Are Local Property Taxes Regressive, Progressive, or What?” National Tax Journal 69(2), 2016, 415–433.
9. Cellini, Stephanie Riegg, Fernando Ferreira and Jesse Rothstein. “The Value of School Facility Investments: Evidence from a Dynamic Regression Discontinuity Design”, Quarterly Journal of Economics 125(1), 2010, 215–261.
10. Biasi, Barbara, Julien M. Lafortune and David Schönholzer. “What Works and For Whom? Effectiveness and Efficiency of School Capital Investments Across the U.S.” NBER Working Paper 32040, January 2024. Working paper; peer-reviewed publication not confirmed.
11. Baker, Sarah S. “Property Tax Pass-Through to Renters: A Quasi-Experimental Approach”, Federal Reserve Bank of Philadelphia Working Paper 25-41, December 2025. Not peer reviewed and not an official Federal Reserve position.
12. McMillen, Daniel and Ruchi Singh. “Assessment Regressivity and Property Taxation”, Journal of Real Estate Finance and Economics 60(1–2), 2020, 155–169.
13. McMillen, Daniel and Ruchi Singh. “Assessment Regressivity and the Homestead Exemption”, National Tax Journal 79(2), June 2026, 393–407.
14. Avenancio-León, Carlos F. and Troup Howard. “The Assessment Gap: Racial Inequalities in Property Taxation”, Quarterly Journal of Economics 137(3), August 2022, 1383–1434.
15. McMillen, Daniel and Ruchi Singh. “Measures of vertical inequality in assessments”, Journal of Housing Economics 61, 2023, article 101950. The bias they identify applies to regression-based measures; they treat the price-related differential as the better measure.
16. Berry, Christopher R. “Reassessing the Property Tax”, University of Chicago Harris School working paper, March 2021. Cite as a working paper; journal publication not confirmed.
17. OECD. Housing Taxation in OECD Countries, OECD Tax Policy Studies No. 29, July 2022. Quoted passage at pp. 80–82.
18. Blöchliger, Hansjörg. “Reforming the Tax on Immovable Property: Taking Care of the Unloved”, OECD Economics Department Working Papers No. 1205, April 2015.
19. U.S. Department of the Treasury, Office of Tax Analysis. Tax Expenditures, Fiscal Year 2027, December 16, 2025. Exclusion of net imputed rental income, $157,410 million for FY2026. The prior edition scored the same year at $185,200 million before the 2025 tax law, so the edition matters. The Joint Committee on Taxation does not classify imputed rent as a tax expenditure at all.
20. Internal Revenue Service. Topic no. 701, Sale of your home. $250,000 and $500,000 exclusion limits, set by the Taxpayer Relief Act of 1997 and not inflation-indexed.
21. Congressional Budget Office. The Distribution of Major Tax Expenditures in 2019, October 2021, Table 2. CBO assigns incidence to the taxpayers directly liable and does not model behavioral responses.
22. Congressional Budget Office. “Convert the Mortgage Interest Deduction to a 15 Percent Tax Credit”, in Options for Reducing the Deficit: 2017 to 2026, December 2016, p. 136. Quoted from the option’s factual background. CBO presents options without recommending them, and this one predates the 2017 change to the mortgage debt limit.
23. Congressional Budget Office. “Eliminate or Limit Itemized Deductions”, in Options for Reducing the Deficit, 2023 to 2032, Volume I, December 2022, p. 80.
24. Congressional Budget Office. Estimates of the Cost of Federal Credit Programs in 2026, January 2026, p. 7 and Table 1. The $1.6 trillion is obligations and commitments on new fiscal-2026 lending, not outstanding balances and not budgetary cost.
25. U.S. Census Bureau, Annual Survey of State and Local Government Finances. Property taxes of $630.2 billion in FY2021 and $724.8 billion in FY2024, against total state and local tax revenue of $2,103.2 billion and $2,463.7 billion.

Editor’s note

Every figure attributed to our own calculation is reproduced by scripts/analyze_property_tax_denominators.py in the Summitward repository, which reads the committed Federal Reserve Distributional Financial Accounts extract. Published medians from the Survey of Consumer Finances and the National Association of Realtors are quoted from their sources rather than recomputed.