Every March, about five million Danes can log in and find their tax already done. The annual statement, the årsopgørelse, has been assembled from what employers, banks, pension funds and mortgage lenders reported during the year. It shows the income, the deductions, the tax withheld, and the balance owed or refunded. If it’s right, the taxpayer does nothing. It stands. If it’s wrong, they change the line that is wrong, and a correction window runs until spring, with refunds starting to arrive in April.1
Denmark does your taxes for you.
Denmark began pre-filling returns in the late 1980s, the first country to do so.2 The Danish return mostly works. Where Danish administration fails is more instructive, because it fails on the same records, in the same state, and the technology can’t explain the difference between the two outcomes.
Reporting moved to the parties who hold the numbers
In most tax systems the filer is the one who has to fit a life into the institution’s categories. She collects the forms, finds the numbers, learns which box they go in, and certifies the result under penalty. The American version of that burden is large enough to support an industry that lobbies to keep it.3 In Denmark the work was moved onto the parties who already hold the numbers. Employers report wages monthly to the national income register, banks report interest and balances, and mortgage lenders report what was paid. By the time the statement is built, roughly ninety-five percent of all income has been reported by someone other than the person who earned it.4
That shift did more than save filers an evening. In a large audit experiment, Henrik Kleven and colleagues found that evasion on third-party-reported income was about 0.3 percent, against 37 percent on income people reported themselves.4 The state that asked less of people collected more accurately, because it had moved the reporting to the parties who couldn’t easily shade it.
Whether the record can be seen before it binds
The reform Denmark represents carries a risk. Once employers, banks and lenders do the reporting, the taxpayer is no longer the one who says what she earned. Her wages come from a payroll system and her interest from a bank’s ledger, each kept for its owner’s purposes, and an error in either could settle what she owes without ever passing through her hands. The missing half of the reform is a record shown to the person and correctable before it takes effect.
Denmark built the missing half. The statement is a proposal until the correction window closes, and every figure on it is visible and editable. After the window, a person can still ask to reopen the assessment until the first of May in the fourth year after the income year, on factual or legal grounds.5 The preliminary assessment, the forskudsopgørelse, sets withholding for the coming year and works the same way in advance. For 2025, more than a million people changed theirs within two months of it opening, a record, and the number of changes has risen by about a fifth since 2019.6
Those changes are work, and the work doesn’t fall evenly. The people changing their forecasts are disproportionately young, and the items they change are income, commuting deductions and interest: the parts of a life that move. A salaried employee with a mortgage and one bank account may never touch the system. A freelancer with three clients and a new flat touches it every quarter. The design moved the ordinary case upward and left the irregular case with the person. That allocation is right, since the irregular facts are the ones only the person holds. But it means the system’s smoothness is partly a property of the lives it was built around.
Refusal is where Denmark’s answer gets less comfortable. There’s no opting out of the income register. Employers report whether or not the employee wants them to; the personal identification number that joins the records is assigned at birth or on arrival. The arrangement cannot claim consent. Its legitimacy rests on what the state does with the join, and on the person’s ability to see and dispute what the join produces. As long as the record faces her, shown before it binds, the lack of exit is tolerable. The same join pointed somewhere she can’t see is a different matter.
The same records, pointed at fraud
It is pointed somewhere else.
Udbetaling Danmark is the agency that pays pensions, child benefits, housing support and other transfers, and it uses the same national records for fraud control. Its data unit has built up to sixty algorithmic models to flag people for investigation. Amnesty International spent more than two years investigating the system and published its findings in November 2024 under the title Coded Injustice.7 The models draw on residency and movements, citizenship, place of birth, family relationships, travel and ties to countries outside the European Economic Area. One, which Amnesty calls “Really Single,” estimates whether a person receiving a single person’s benefit is really living alone. Another flags people with connections abroad. Amnesty found that the models risk discriminating against migrants, people with disabilities, people on low incomes and racialized groups, and argued that parts of the system may amount to the social scoring the EU’s AI Act prohibits. The agency disputes that reading. In a focus group Amnesty held with disabled people at the Dansk Handicap Foundation, one participant said the constant monitoring of benefit claimants felt like “sitting at the end of the gun.”
Everything the tax statement gets right runs the other way here. The agency sees the pattern first, because it has every file. The person flagged bears the cost of the model’s errors: she must show she is single, or resident, or not what the score suggests. And the record is never shown before it binds. She doesn’t see the score that selected her, the features that produced it, or the rate at which the model is wrong about people like her. Detection shifted upward into ATP’s fraud-scoring algorithms, but the cost of the model’s errors stayed below. The same register that shows her tax figures every March, open to correction, selected her for investigation on criteria she never gets to check.
Denmark has failed this way before, on the tax side itself, when collection rather than assessment was automated. The tax agency started EFI, a system meant to automate collection of debts owed to the public sector, in May 2006. It was postponed six times and went live in stages in September 2013. By 2015 it was collecting debts that had already passed their limitation period, and failing to collect others. Automatic collection through EFI was suspended on 8 September 2015, and the tax minister said the system was so riddled with errors that there was no alternative to shutting it down.8 The state then spent years rebuilding by hand what the machine had been trusted to do. The property valuation system followed a similar arc: suspended valuations, a replacement promised within two years, and costs that multiplied many times over. By 2021 the state owed refunds estimated at about fourteen billion kroner, interest included, on some 715,000 homes taxed on valuations that were too high.9
In both failures, the state’s machinery acted on people, and the people found out afterward.
Which way the record faces
The annual statement and the fraud models run on the same income register and the same personal number, in the same high-trust state, under the same ministers. One is a model of how a state should use what it knows. The other is a textbook case of how it should not. What separates them is the direction the knowledge faces and who pays when it’s wrong. How much the state knows is the same in both.
Judged against the test in Who Corrects the Corrector?—that a new capacity is legitimate only when it pays for its own errors, makes them visible from outside, and triggers their correction without forcing the affected to stay vigilant—the Danish tax statement meets the first two and partly meets the third: it corrects the ordinary case without anyone watching, though the irregular case still depends on the person noticing. Its errors are visible from outside because it is shown to the person before it acts. The fraud models fail on all three counts.
Denmark’s tax base is broad and simple, its employers and banks report reliably, and its public administration starts from a level of trust that most countries don’t have. A country that copied the pre-filled statement without the correction window, or without reliable third-party reporting, would get the form’s disappearance and none of its replacement. And the Danish case says little about a hard question that stays open: who decides whether a burden recurs across a class, when that decision is itself a classification.
It does show that the choice is real. A state can know a great deal about people and put that knowledge in front of them, or know the same things and keep them to itself. Denmark does both, on the same records, and the difference is plain to anyone who has been on the receiving end of either. If a system built on records like these flags you, ask for what Denmark’s fraud models don’t show: the score that selected you, the features that produced it, and how often the model is wrong about people like you.
Notes
Skattestyrelsen, “Årsopgørelsen for 2025 er på vej” (March 2026), and skat.dk, “Relevant dates for the tax assessment notice.” For income year 2025 the statement opened on 23 March 2026 for about five million people, with refunds from 24 April. If nothing on it is wrong, no action is needed: Life in Denmark (borger.dk), “A general introduction to the Danish tax system.”
Tax Policy Center, “What are prepopulated tax returns?” Tax Policy Center Briefing Book (updated January 2024). See also OECD, Using Third Party Information Reports to Assist Taxpayers Meet Their Return Filing Obligations: Country Experiences with the Use of Pre-populated Personal Tax Returns (2006).
Justin Elliott and Lucas Waldron, “Here’s How TurboTax Just Tricked You Into Paying to File Your Taxes,” ProPublica, April 2019; Justin Elliott, “TurboTax Deliberately Hid Its Free File Page From Search Engines,” ProPublica, April 26, 2019; Justin Elliott and Paul Kiel, “Inside TurboTax’s 20-Year Fight to Stop Americans From Filing Their Taxes for Free,” ProPublica, October 17, 2019.
Henrik Jacobsen Kleven, Martin B. Knudsen, Claus Thustrup Kreiner, Søren Pedersen and Emmanuel Saez, “Unwilling or Unable to Cheat? Evidence from a Tax Audit Experiment in Denmark,” Econometrica 79, no. 3 (2011): 651–692.
Skatteforvaltningsloven § 26, stk. 2, as set out in Skattestyrelsen’s Juridisk Vejledning, section A.A.8.2.1.2.1, “Genoptagelsesfristen.”
Skattestyrelsen, press release, 17 January 2025, on 1.058 million citizens changing their 2025 forskudsopgørelse; and Skattestyrelsen, “Borgerne ændrer oftere i forskudsopgørelsen end tidligere,” on the rise from 4.1 to 4.9 million annual changes between 2019 and 2023 and the share made by people aged 20 to 39.
Amnesty International, Coded Injustice: Surveillance and Discrimination in Denmark’s Automated Welfare State (EUR 18/8709/2024, 12 November 2024). The fraud-control unit is operated for Udbetaling Danmark by the pension fund ATP.
Rigsrevisionen, Beretning til Statsrevisorerne om SKATs systemmodernisering (beretning nr. 5/2014, January 2015), paragraph 61, on the start in May 2006, the six postponements and the staged go-live in September 2013; Rigsrevisionen, fortsat notat til Statsrevisorerne om SKATs systemmodernisering (October 2017), on the suspension from 8 September 2015; DR, “Skandalen om Skats it-system: Efter 11 års fiasko koster det millioner at lukke EFI ned” (15 February 2016), on the time-barred debts and the minister’s statement.
DR, 24 November 2021, on the Statsrevisorer’s criticism of the Ministry of Taxation’s valuation system; see also Rigsrevisionen’s continuing reports on the property valuation programme. The refund estimate is Skatteministeriet’s factsheet “14 mia. kr. i for meget betalt boligskat tilbagebetales til boligejerne” (2021): about 14 billion kroner at 2021 levels, interest included, across about 715,000 owner-occupied homes taxed on valuations that were too high between 2011 and 2020.