How Scandinavia Actually Pays for Itself, and What Others Can Learn From It

In 2025 Denmark's top tax rate started at 1.24 times the average wage. In the United States, 8.73. Both countries have a top bracket; the word does entirely different work in each. Denmark raises 25.0% of GDP from personal income tax, the highest in the OECD, and 0.06% from social contributions. Read every labour tax together and Denmark, Sweden, Finland, Germany and France land within 1.56 points of each other.

In 2025 the OECD put Denmark’s top statutory rate on labour income at 55.9% of gross wage earnings. That is the number everyone quotes. The one next to it is the number nobody quotes: that rate began at 1.24 times the average wage. In the United States, the top federal bracket began at 8.73 times average pay.

Both countries have a top bracket. The word does entirely different work in each, and before you argue about whether a country taxes too much you have to know who it points at. I went looking for how the Nordic model is financed, and the revenue tables kept correcting me.

Where the top bracket begins

Dot plot on a log scale showing the earnings at which each OECD country's top statutory income tax rate begins, as a multiple of the average wage, for 36 of the 38 members in 2025. Estonia is lowest at 0.31, Belgium at 1.03, Sweden 1.15, Ireland 1.16, the Netherlands 1.17 and Denmark 1.24 cluster just above one times the average wage, the United States sits at 8.73 and Colombia is highest above 50. — OECD Tax Database, Table I.7, 2025. The series is OECD’s own published factor of the average annual wage, not a ratio computed here. The Nordic countries do not behave as a bloc: Sweden 1.15, Denmark 1.24, Iceland 1.33, Norway 1.76, Finland 2.97, a range of 2.58 times end to end. Hungary is excluded because its top rate applies from the first unit of income, so a multiple of the average wage is not defined for it. Latvia is excluded for a different reason: it reports both a zero threshold and a zero top rate in 2025, which is a missing value rather than a flat tax. In 2024 it reported 3.91 times the average wage at 31.2%. A low threshold and a high rate are different things, and this chart shows only the threshold.

The Nordics are not a bloc. Sweden starts at 1.15 times the average wage, Denmark 1.24, Iceland 1.33, Norway 1.76, Finland 2.97, a range of 2.58 times end to end. Finland sits closer to Canada than to Sweden, so any sentence beginning “the Nordics” has to survive it.

And a low threshold is not a Nordic invention. Belgium begins at 1.03, Ireland 1.16, the Netherlands 1.17, all at or below Denmark. Nor is it only Northern European: Hungary taxes from the first unit of income. The lowest threshold actually on the chart is Estonia’s, a single 22% rate from 0.31. Hungary is off it for that reason, Latvia because its 2025 threshold and rate are both reported as zero, so the chart holds 36 of the 38. A low threshold and a high rate are different things; this chart shows the threshold.

One change matters more than a footnote should. In January 2026 Denmark split its single top bracket, a 15% topskat, into three tiers, all stated after the 8% labour market contribution: a middle tax of 7.5% above DKK 641,200, on personal income plus positive net capital income over DKK 55,000 for a single taxpayer and DKK 110,000 for a couple; a top tax of 7.5% above DKK 777,900 on personal income, stacking back to 15; and a top-top tax of 5% above DKK 2,592,700. Its highest rate now starts far higher up and is higher, about 60.5% of gross wage earnings against 55.9 in 2025. For income between DKK 611,800 and 777,900 the reform is a cut, though DKK 641,200 is only DKK 611,800 indexed forward. The 2025 figure above is DKK 611,800 of personal income, DKK 665,000 gross, against a DKK 537,071 average wage.

OECD has published no 2026 figure, so I will not draw a line across the break. If it reports the new top-top tier the multiple jumps from 1.24 to about five, my own calculation from the statutory threshold and a projected 2026 average wage. The lesson is not about Denmark: an indicator built on “the threshold of the highest rate” can quadruple because a country added a small surtax at the very top, while the person on average pay notices nothing.

How the money is actually raised

Stacked horizontal bar chart of tax revenue by type as a percent of GDP in 2021 for ten countries plus the OECD average. Denmark's personal income tax band is by far the longest at 25.0 percent and its social contributions band is invisible at 0.06. Germany and France show the mirror image, with modest income tax bands and social contribution bands near 14.8. Denmark's total is 47.4, France 45.2, Finland 43.2, Sweden 42.7, Norway 42.4 and the United States 26.5. — OECD Revenue Statistics, 2021, general government. Bands sum to the published total by construction; the residual is income tax that a country does not split between individuals and corporations, plus other taxes. The OECD average is an unweighted mean computed here, because the source file publishes an OECD total with no component breakdown. Across Denmark, Sweden, Finland, Germany and France the personal income tax line ranges from 9.47 to 25.05% of GDP, a factor of 2.64. Count social contributions and payroll taxes as well and the same five countries land between 24.89 and 26.45, a spread of 1.56 points. Norway’s corporate band is mostly petroleum, taxed there under a separate regime at a 78% marginal rate rather than the ordinary 22.

Denmark raised 25.0% of GDP from personal income tax in 2021, the highest in the OECD and not narrowly: second-placed Iceland is at 14.5, the average 8.3. It also raised 0.06% of GDP in social contributions. Not a rounding artefact: essentially zero.

Germany raised 10.5% in income tax and 14.8 in contributions. France, 9.5 and 14.8. Read one line and Denmark looks like a wild outlier. Now read every labour tax together:

	Personal income tax	Social contributions	Payroll	Together
Denmark	25.05	0.06	0.26	25.36
Sweden	12.40	8.93	5.12	26.45
Finland	12.86	12.03	0.00	24.89
Germany	10.47	14.77	0.00	25.25
France	9.47	14.82	1.84	26.13

Percent of GDP, 2021. Figures are rounded individually, so two rows do not appear to add; totals are computed unrounded.

Their income tax lines differ by a factor of 2.64, their total taxation of labour by 1.56 points. Against Germany and France, Denmark’s income tax is not extra burden but the same money routed through one channel instead of two.

Two qualifications, because that table is a comparison I chose. The convergence belongs to those five: swapping France for Norway widens the spread to 6.06 points, the five actual Nordics span 8.63, and across all 38 the ratio between largest and smallest income tax line is 18.96. And routing is not the whole of Denmark: its income tax runs 16.7 points above the OECD average while the contributions it skips are worth 9.0, so routing accounts for about 54% of the excess. Denmark ranks fourth of 38 on labour taxes and first on total tax, a heavy-taxing country, just not in the way the income tax line alone suggests.

What the revenue buys, and what it only appears to buy

Three scatter panels plotting life expectancy, the poverty rate after taxes and transfers, and poverty removed by the state against total tax revenue as a percent of GDP. Each panel carries two fitted lines, one across all OECD members and one excluding the five with the lowest GDP per capita, which are highlighted in amber. In the life expectancy panel the fit falls from an R squared of 0.166 to 0.023 once those five are removed. The poverty panels show a stronger downward and upward relationship respectively. — Tax revenue is the mean of 2017 to 2021, the most recent five years in which every OECD member reports. Single years are unreliable here: Denmark’s total falls from 47.4% of GDP in 2021 to 41.9 in 2022, outside this window. Poverty is the share below 50% of median disposable income, total population, current income definition, latest year per country, 2019 to 2025, from the OECD Income Distribution Database. The five excluded countries are chosen by rule, not by hand: they are the OECD members with the lowest GDP per capita. Greece is sixth and stays in. Colombia reports no income-distribution data, so the two poverty panels cover 37 members and drop four of those five rather than all five. The third panel is the weakest of the three, because poverty removed is defined as market poverty, which is not plotted here, minus the disposable-income rate in the middle panel, so a larger fiscal state having a larger fiscal effect is close to arithmetic.

Plot tax revenue against life expectancy across all 38 OECD members and you get R-squared 0.166. Taxing more, it appears, buys longer life.

It does not. Remove the five members with the lowest GDP per capita, Colombia, Mexico, Costa Rica, Chile and Türkiye, and it falls to 0.023. The relationship was never about tax; it was that the low-tax end of the OECD is also the low-income end. Control for log GDP per capita, so no cutoff has to be defended, and tax adds 0.034, noise on 38 countries.

The same test belongs on the other side: life expectancy against income alone is 0.291 across all 38 and 0.143 once those five leave. The income story survives the cut better than the tax story, but it halves too.

Denmark is the concrete case. It collects more than any other OECD member on the five-year average this chart uses, and lives 81.9 years, twentieth of 38, behind Japan 84.7, Korea 84.3 and Switzerland 84.0. It does take market poverty from 22.2% to 6.3.

The finding I have to argue against myself

That last number is weaker than it looks. “Poverty removed by the state” is market poverty minus disposable poverty, and disposable income is market income plus transfers minus taxes. So the thing explained is the output of the tax and transfer system, and the thing explaining it is that system’s size. Market poverty alone, measured before the state acts, predicts it slightly better (0.583) than the tax level does (0.548), and high-tax countries have more of it to remove in the first place, at a correlation of 0.45.

It is not purely circular. The fit is 0.55, not 1.0, and the counterexample is in the data: France collects 45.5% of GDP and removes 27.4 points of poverty, Denmark 46.2 and 15.9. Size correlates with redistributive output; it does not determine it.

So the honest headline is not the symmetry I first wrote. How rich a country is predicts how long its people live. How much it taxes tracks how much levelling its transfer system does, a smaller and more mechanical claim, and the next section is where it stops being circular.

Why there is no schooling panel

PISA lets countries exclude up to 5% of their target population. In 2022 Denmark excluded 11.6%, and the OECD says dyslexic students using assistive devices the test does not accommodate appear to be a major cause of the rise from 5.7% in 2018 (OECD, 2023).

That is an upward bias on a country 12.7 points of GDP above the average on this chart’s x axis. Lifting the y value of a high-x country tilts the fitted line upward, pushing the panel toward showing that tax buys learning. A panel I would have to caveat into uselessness is worse than a paragraph explaining its absence.

Where the equalising actually happens

Horizontal stacked bar chart of the reduction in the Gini coefficient from market income to disposable income for 35 OECD countries, split into the part done by cash transfers and the part done by direct taxes and employee contributions. The transfer share is labelled at the end of each bar and runs from 49 percent for Israel to 93 percent for Greece and Chile. Finland has the largest total reduction, Switzerland exceeds 100 percent and is drawn in red. — OECD Income Distribution Database, latest year available per country, 2019 to 2025, total population, current income definition. Market income to gross income adds cash transfers; gross to disposable removes direct taxes and employee social contributions. The percentage at the end of each bar is the transfer share of that country’s total reduction. The median country does 77.0% of its equalising through transfers, and 34 of 35 countries do more than half of it that way. Switzerland exceeds 100% because its disposable-income Gini is measured slightly above its gross-income Gini, so the tax step subtracts rather than adds. That is what the OECD publishes, and this post does not speculate about the cause. Israel, at 49%, is the only country where transfers do less than half.

Split the journey from market to disposable income into the part done by cash transfers and the part done by direct taxes and employee contributions. Across 35 OECD countries the median does 77% of its inequality reduction through transfers, 34 of 35 more than half. Denmark 76%, Finland 79, the United States 62.

This locates the previous section’s weak result without rescuing it. The tax level tracks the transfer step (0.497) and barely the tax step (0.066); decomposed properly, 86% of the covariance runs through transfers. The same near-definitional pairing, at a lower fit.

The 77% is a different claim, and it is not circular. It is a composition inside each country, with no tax level on either side: of the equalising a country does, this much comes through transfers. Nothing in the accounting requires 77 to 23 rather than the reverse.

What travels

A universal benefit needs a universal bill. If a state covers everyone the money has to come from the middle, because that is where the taxpayers are. That is my prior, not a finding: nothing here measures what share of the bill each income group carries, only how much is collected and what it does.

The testable version does not survive. If collecting a lot forced a low top bracket, the two would move in opposite directions. Across the 31 richer OECD members the correlation is minus 0.06, the right sign and no size: it accounts for 0.3% of the variation. France collects 45.5% of GDP with a top bracket at 12.94 times the average wage, Austria 42.4% at 17.44, both top-ten collectors. Two caveats: across all 36 members the correlation is minus 0.47, carried by the same five low-income countries excluded above; and this threshold is the income tax schedule’s alone, the least comparable line in the table. Denmark chose a broad base.

What does look transferable is the other half, equalising through what the state pays out rather than how steeply it taxes. What does not is the capacity to collect and the consent underneath it.

Indonesia’s entire tax take averaged 11.2% of GDP across 2017 to 2021, of which personal income tax was 1.1. Denmark’s income tax alone, over the same five years, was 24.7% of Danish GDP: 2.2 times Indonesia’s whole tax take, 22.7 times its own income tax.

The argument about tax I most often hear at home is an argument about the rich, and the Danish tables are uncomfortable about that: the rich are not where a universal state gets its money, in Copenhagen or Jakarta. The harder question is not the rate anyone will set. It is whether a state can reach ordinary earners at all, and whether what it hands back is worth what it takes.

Method notes

Five time bases. Thresholds 2025, revenue 2021, life expectancy 2023, a 2017 to 2021 mean tax level, income-distribution data at each country’s latest year (2019 to 2025). Revenue is 2021 because 2022 drops Japan, Australia and Greece. The tax level is a five-year mean because Denmark’s single year falls from 47.4% of GDP to 41.9 between 2021 and 2022, outside this window, about 38% of that its pension yield tax following markets and most of the rest nominal GDP growth; Denmark ranks first on either window, leading France by 0.77 points on 2017 to 2021 and 0.05 on 2018 to 2022.

The threshold is OECD’s published series, not a ratio I computed: Danish law states it net of the 8% contribution while the average wage is gross, so dividing one by the other returns 1.139, not 1.238. The DKK 665,000 gross equivalent is derived, not legislated.

Definitions sit on the figures: the poverty line and income definition on the third, Norway’s petroleum and the Switzerland exception on the second and fourth. Personal income tax throughout is the wider OECD category, including capital gains, reported by all 38.

These are associations, not causal estimates, including in “What travels”. The strongest claim I will defend: tax revenue does not predict longevity once you know how rich a country is, explaining 0.023 among the 33 richer members. What it predicts is how much levelling the transfer system does.

The analysis, the four charts and every number above are reproduced by the code in the data-stories repository. The source files are not redistributed; a fetch script downloads them from the OECD and Our World in Data.

References

Eurostat, Organisation for Economic Co-operation and Development, International Monetary Fund, & World Bank. (2026). GDP per capita, PPP (constant 2021 international $) [Data set]. With minor processing by Our World in Data. https://ourworldindata.org/grapher/gdp-per-capita-worldbank

Norwegian Offshore Directorate, & Ministry of Energy. (2026). The petroleum tax system. Norwegianpetroleum.no. https://www.norskpetroleum.no/en/economy/petroleum-tax/

Organisation for Economic Co-operation and Development. (2023). PISA 2022 results (Volume I): The state of learning and equity in education. OECD Publishing. https://doi.org/10.1787/53f23881-en

Organisation for Economic Co-operation and Development. (2026a). OECD income distribution database [Data set]. https://sdmx.oecd.org/public/rest/data/OECD.WISE.INE,DSD_WISE_IDD@DF_IDD,1.0/all

Organisation for Economic Co-operation and Development. (2026b). OECD tax database, Table I.7: Top statutory personal income tax rates [Data set]. https://sdmx.oecd.org/public/rest/data/OECD.CTP.TPS,DSD_TAX_PIT@DF_PIT_TOP_EARN_THRESH,1.0/all

Organisation for Economic Co-operation and Development. (2026c). Revenue statistics: Comparative tables [Data set]. https://sdmx.oecd.org/public/rest/data/OECD.CTP.TPS,DSD_REV_COMP_OECD@DF_RSOECD,1.0/all

Organisation for Economic Co-operation and Development. (2026d). Revenue statistics in Asia and the Pacific: Comparative tables [Data set]. https://sdmx.oecd.org/public/rest/data/OECD.CTP.TPS,DSD_REV_COMP_ASAP@DF_RSASAP,1.0/all

Riley, J. C., Zijdeman, R., Human Mortality Database, & United Nations Department of Economic and Social Affairs, Population Division. (2025). Life expectancy at birth [Data set]. With major processing by Our World in Data. https://ourworldindata.org/grapher/life-expectancy

Skatteministeriet. (2024). Lov nr. 482 af 22. maj 2024 om ændring af personskatteloven, ligningsloven og forskellige andre love. Retsinformation. https://www.retsinformation.dk/eli/lta/2024/482

Skatteministeriet. (2025). Satser og beløbsgrænser i lovgivningen: Personskatteloven. https://skm.dk/tal-og-metode/satser/satser-og-beloebsgraenser-i-lovgivningen/personskatteloven