Ask for the Roof, Get the Window

1. Is the “far right” rising or falling?

On 6 September 2026, Saxony-Anhalt held its state election. The Alternative for Germany (AfD) took 43.8% of the vote and 39 of 83 seats, three short of a majority on its own. Three days later, Chancellor Merz stood up in the Bundestag and twice described the AfD’s “remigration” as ethnic cleansing by origin and skin colour. The AfD filed a criminal complaint against him.

Four days after that, Sweden voted. The Sweden Democrats (SD) got 17.48%, three points down on four years earlier, and lost 11 seats. It was the first time the party had gone backwards since entering parliament in 2010. The four parties backing the government ended up with 173 seats against the opposition’s 176. The government lost its majority.

Look only at those numbers and you get two opposite stories: Germany is lurching right, Sweden is swinging back. Look at policy and both countries are heading the same way. Germany’s first-time asylum applications halved in 2025, and the current government has suspended family reunification for one group of refugees. Sweden has raised the residence requirement for citizenship from five years to eight, and asylum seekers are at a forty-year low. Magdalena Andersson, leader of the Social Democrats (S) and likely the next prime minister, said plainly during the campaign that she would keep migration policy strict.

In both elections, immigration was the issue that split the parties most. In Germany the mainstream still refuses to work with the AfD, a line Germans call the Brandmauer, and the main reason the AfD sits on the wrong side of it is its migration agenda. You can roughly line the parties up on migration: the AfD is the hardest, the Christian Democrats (CDU/CSU) lean strict, the Social Democrats (SPD) sit in the middle, and the Greens and the Left Party (Die Linke) lean open. Sweden is the same. What divides its two blocs, in the end, is whether you are willing to lean on the SD, and the deal the SD signed with the government is about migration and crime at its core. Who can govern with whom is largely drawn along the migration line.

Still, migration isn’t necessarily what voters care about most. In an ARD poll this June, Germans ranked the economy (27%) above migration (21%) as the most important problem. A post-election analysis in Saxony-Anhalt found that 89% of AfD voters worried about paying their bills, against 70% of all voters. In Sweden’s exit poll, the top four issues were healthcare, schools, the economy and law and order. Migration wasn’t among them. So calling these two “migration elections” is wrong. Migration is the issue that splits the parties hardest and decides who can form a government, but it is not the only reason people vote the way they do.

Which raises a question: if migration decides who gets into power, how much does migration policy actually change when power changes hands? My answer: between the mainstream parties, far less than the slogans suggest. An S government won’t take Sweden back to 2015. The real variable is the coalition partners. To govern, S would need the Left Party (V), the Swedish Greens (MP) and the Centre Party (C), and all three want looser rules in one way or another. Migration is exactly the kind of thing that gets traded away in coalition talks. The AfD is shut out at federal level and has no majority in Saxony-Anhalt. Even if it got into a state government, what it could actually touch is mostly state-level business like enforcing deportations.

Both parties get called “far right”. One rose in Germany, the other fell in Sweden, and yet migration policy in both countries is a lot stricter than it was ten years ago. I can only think of one explanation. The biggest use of parties like these is often not getting into power themselves, but driving the price up so everyone else has to move toward them. Overshoot a little, and things land about right. Yes, that’s easy to hear as “the more extreme the better”. That’s not what I mean.

2. Aim for the top

“Aim for the top and you’ll end up in the middle” is often credited to the I Ching or the Analects. It comes from neither. The more reliable source is the Di Fan, a book of advice the Tang emperor Taizong wrote for his heir: “Take the highest as your model and you will reach only the middle; take the middle as your model and you will end up at the bottom.” The Song dynasty critic Yan Yu said something similar about learning to write poetry. The meaning is simple: set the bar high and you’ll probably get something decent; aim for decent from the start and you’ll do worse. Lu Xun, in a 1927 lecture in Hong Kong called “Silent China”, put it more vividly: “The Chinese temperament loves compromise and the middle way. Say this room is too dark and we need a window here, and everyone will refuse. But propose tearing off the roof, and they will come round to a compromise and agree to the window.” Lu Xun was talking about the Chinese, but the mechanism isn’t Chinese. In any system where decisions come out of bargaining, the opening offer moves the final price. Political science has two names for this. One is the Overton window: the range of policies a society considers “discussable” at a given moment. Ideas outside it are treated as crazy; only ideas inside it get to the negotiating table. The other is the radical flank effect: when a more radical wing appears in a movement, the moderates suddenly look acceptable. But it cuts both ways. A radical wing can just as easily drag the whole movement’s reputation down.

Borrowing the phrase doesn’t mean radical positions are smarter. “The top” here just means a more distant opening bid, not a better plan.

But there are two kinds of overshooting. One is overshooting the ask: demanding more than you actually want, putting the opening bid far out so the deal lands in the middle. The other is overshooting in power: getting into office and treating the opening bid as the final plan, pushing it through even when it crosses the line of the rule of law and equal citizenship. The first is ordinary political bargaining. The second is the rule of law breaking down. I’m only talking about the first.

Overshooting the ask only works because something in the middle discounts it. Coalition partners haggle, courts review, constitutions put some things off-limits, the state machine has a capacity limit, and voters get another say a few years later. If none of that existed, the opening bid would be the deal, and overshooting would be genuinely dangerous. So the phrase carries a hidden assumption: the discounting in the middle exists and mostly works. I call those discounting mechanisms filters.

3. Why the old consensus moves slowly

If the mainstream parties could all see the migration problem, why didn’t they fix it themselves instead of waiting for a party they call extreme to push them? At least three reasons.

First, stock and flow are different things. Migration debates are mostly about inflow, an argument over how far to open the tap. What voters feel is how much water is already in the bath. Germany’s first-time asylum applications in 2025 came to 113,236, down 50.7% on the year before. The tap really was turned down. In the same year, the number of foreigners legally required to leave Germany rose by more than 11,000, reaching 232,067 by year end, over 190,000 of them on a Duldung, a temporary suspension of deportation. Policy got tighter, and the stock kept growing.

Sweden is similar. Only 6,735 people sought asylum in 2025, the lowest in forty years. But first-time residence permits ran at about 103,000 in 2023 and still about 90,000 in 2025. Asylum is just one of many taps; family reunification, work and study are all running. Turn off one and the bath drains slowly.

Second, there are a lot of veto players. Germany is usually governed by a two- or three-party coalition, much legislation also has to pass the Bundesrat, which represents the states, above that sits the Constitutional Court, and above that EU law and the European Court of Human Rights. Sweden’s minority governments need parties outside government to pass a budget, and changing the constitution takes two votes in two separate parliaments. Every extra person who has to say yes adds inertia. These are all filters. They block radical proposals, and they also block the mainstream’s own corrections.

Third, the old consensus had its own gatekeepers. The open door of 2015 wasn’t a left-wing solo act. Germany at the time was run by Merkel’s CDU/CSU and SPD coalition, Sweden by Löfven’s coalition of S and MP, and Sweden itself started tightening that November. Business was broadly optimistic too. The consensus crossed party lines, so loosening it took pressure from across party lines.

Today the defenders of the old consensus are mainly on the left. On 27 June 2025 the Bundestag voted 444 to 135 to suspend family reunification for people with subsidiary protection for two years. The Greens and Die Linke voted against; the Greens’ home affairs spokesperson called the law cruel, and Die Linke tabled a motion to relax family reunification rather than suspend it. In Sweden, the three potential partners of S, namely V, MP and C, all want more immigration in their own ways. They each have their reasons. But it shows the old consensus won’t step aside on its own.

4. What “remigration” actually means

“Remigration” is the most overstated word in Germany’s whole argument, in both directions. Critics say the AfD wants to ship two million people to North Africa. Supporters say it just means deporting people who are here illegally. Neither is accurate. The word actually has three layers.

The first layer is the AfD’s own definition. In January 2024 its federal executive published a position paper defining “remigration” as the lawful deportation and encouraged return of foreigners who are obliged to leave. It states explicitly that the AfD makes no distinction between German citizens with and without a migration background, and rejects expelling citizens because of one. On that definition, we’re talking about the 230,000-odd people mentioned above, which is basically a question of enforcing existing law.

The second layer is the AfD’s concrete demands. The same paper came with seven measures, and together with its 2025 federal manifesto and its 2026 Saxony-Anhalt manifesto, the scope is much wider than the definition: people with subsidiary protection who should go home once the war is over, Syrians and Afghans, people who can’t support themselves, Ukrainians in Saxony-Anhalt, and dual nationals convicted of serious crimes. Many of these people are here legally. In its own case before the Cologne administrative court, the AfD estimated that this layer covers more than two million people.

The third layer is the plan of the Austrian activist Martin Sellner. He laid out his ideas at a private meeting in Potsdam in November 2023, which the investigative outlet Correctiv later exposed. His targets included asylum seekers, foreigners with residence rights, and what he called “non-assimilated citizens”, meaning people who are naturalised but, in his view, not assimilated enough. He estimated that five to six million German citizens with a migration background “might qualify”. The idea of a “model city” in North Africa for around two million people is also his, not AfD policy. But the link can’t simply be waved away: Ulrich Siegmund, the AfD’s lead candidate in Saxony-Anhalt, attended that meeting. Correctiv’s reporting led to several court cases with mixed results, but on the core quotes of Sellner’s own remarks, the Hamburg regional court found them accurate.

Only the third layer clearly includes citizens. In June 2025 Germany’s Federal Administrative Court examined Sellner’s plan in a ruling and found that the part aimed at German citizens with a migration background violates human dignity, because it sorts citizens into different classes. That ruling was about Sellner, not about the AfD’s programme.

So what about Merz’s “ethnic cleansing”? That’s a politician characterising an opponent, not a court’s finding. Hold the AfD’s programme up against the definition of ethnic cleansing used by the UN Commission of Experts on the former Yugoslavia, and the wording doesn’t fit. Sellner’s plan, which targets the makeup of the population, comes much closer. Some conservative German scholars publicly rejected Merz’s comparison. Interestingly, in the same speech Merz also said Germany has to deport more people with no long-term right to stay. His real disagreement with the AfD is over the scope of the second layer, not over whether to deport at all.

The AfD also has a part that is hard to pin down: its people regularly go beyond the official definition. Some talk about deporting millions, some about changing the makeup of the population. There are voices the other way too. Maximilian Krah has publicly warned against touching citizens, and in January 2026 the party leadership called off an event that had invited Sellner. As for whether the party as a whole is extremist: Germany’s domestic intelligence agency classified it as a confirmed right-wing extremist organisation in May 2025, the Cologne administrative court suspended that classification in urgent proceedings in February 2026, and the main case is still pending. Five state branches remain classified as confirmed extremist.

So the real negotiating table for “remigration” is the second layer. The first layer is existing law, with not much to argue about. The third runs straight into the constitution’s firewall. Saying the AfD has decided to ship two million people to North Africa isn’t supported by the evidence. Saying “remigration” just means deporting illegal residents isn’t accurate either, because the AfD’s own demands already reach large numbers of legal residents. Once a party with these intellectual roots gets near power, the question worth asking is: where can the system hold it in check, and how far are voters willing to let it go?

5. The filters

The filters in the middle come in roughly three kinds: firewalls, soft spots and bargaining.

The firewall is in the constitution. Article 1 of Germany’s Basic Law says human dignity is inviolable, and Article 79(3) says that article can’t be amended at all, the so-called eternity clause. Article 16 forbids stripping anyone of citizenship; loss of citizenship can only happen under law, and not against a person’s will if it would leave them stateless. Article 3(3) forbids discrimination by descent, origin or faith. Changing either of those last two takes a two-thirds majority in both the Bundestag and the Bundesrat. Article 1 can never be changed. The part of Sellner’s plan aimed at citizens runs straight into the one that can’t be changed.

Soft spots are in administrative discretion. Outside the firewall, existing law leaves a lot of room. Section 35 of the Nationality Act, for instance, lets naturalisation be revoked within ten years if it was obtained by false statements, and a state government that wanted to could push clauses like that hard. Legal scholars went through the Saxony-Anhalt AfD’s programme line by line: of the 31 migration demands a state could act on, 18 breach higher law. The remaining 13 are doable within the current framework, and that is the real room an AfD state government would have.

The most common filtering happens in bargaining between parties. The AfD took 43.8% in Saxony-Anhalt, but 39 seats falls short of the 42 needed for a majority. It invited the other parties to talks, and the only one willing to sit down was the Sahra Wagenknecht Alliance (BSW). The two together have 44 seats, exactly the same as all the rest combined. Between coming first and governing, there’s a maths problem.

Sweden gives a more detailed sample, because for the past four years the SD wasn’t in opposition. It was a partner backing the government through a written agreement, named after the castle where it was signed: the Tidö Agreement. Here’s what happened to some of the SD’s demands on the way to becoming law.

1. Mosques. In 2023 SD leader Jimmie Åkesson proposed confiscating and demolishing mosques that spread anti-democratic propaganda. Prime Minister Ulf Kristersson immediately said Sweden doesn’t tear down places of worship. Åkesson backed off to closing such places and expelling hate preachers.
2. Citizenship. The SD had planned to write into its statement of principles that immigrants who betray their civic duties could be stripped of citizenship. That amounts to two classes of citizen by birthplace. The constitutional amendment parliament actually passed applies only to dual nationals, and only for citizenship obtained by fraud or crimes that seriously threaten the country’s fundamental interests. Under the Swedish constitution, it still has to be approved again by the newly elected parliament.
3. Age of criminal responsibility. A government inquiry proposed lowering it from 15 to 14 for serious crimes, for a five-year period. The government’s proposal went for 13, and the Council on Legislation said that breached the Convention on the Rights of the Child. In June 2026 the government withdrew the 13 because it couldn’t find a majority; on 13 August parliament passed 14 by 281 votes to 66, with S voting in favour.
4. The salary threshold for work permits. The inquiry proposed 100% of the national median wage. Parliament passed 90%, with around twenty occupations set at 75%.

All four cases have the same shape: ask far out, settle in the middle. And the discount doesn’t only come from opponents. Kristersson is the SD’s partner, and the age of 14 came from the government’s own inquiry.

Filters also fix collateral damage. A group of young people who grew up in Sweden lost the residence they had held through their parents when they turned 18 and faced deportation, including children of work-permit holders. In March 2026 the Tidö parties paused those deportations, and in June they agreed to raise the age limit to 21 from 1 October, with those already deported able to reapply.

Filters can be rewritten too. In 2023 Giorgia Meloni’s government in Italy signed a deal with Albania to build centres there for asylum seekers rescued at sea. Italian judges repeatedly ordered people brought back to Italy, the sea transfers stopped, and from April 2025 the Albanian centres became pre-deportation detention facilities. That August the European Court of Justice backed the Italian judges. That wasn’t the end of it. On 12 June 2026 the EU’s new Pact on Migration and Asylum took effect, and the return rules agreed after it let member states set up return hubs in third countries. Austria, Denmark, the Netherlands, Greece and Germany are all planning them. The courts blocked one country’s version, and the EU then rewrote the rules. Rewriting a filter just takes a bigger majority and a lot longer.

Filters have no politics. That matters just as much to people who want tighter policy. Germany actually deported 22,787 people in 2025, and in the same year roughly 35,000 deportation attempts failed. More failed than succeeded. The machinery that stops Sellner also stops the federal government’s own deportations. Expecting courts and bureaucracies to filter only the policies you dislike is wishful thinking.

6. Stress test: Trump

Germany and Sweden are parliamentary systems, where power is spread out by design. To see how much filtering is left where power is most concentrated, the US makes a ready-made stress test: one president holds executive power, and the Republicans control both houses of Congress.

Here’s the reality. The fact-checking site PolitiFact tracked 75 promises from Trump’s second term. By February 2026 about 19% had been kept, 3 were compromises, about 45% were in progress, about 31% had stalled and 1 was broken. More interesting than the numbers is which ones got done and which didn’t.

What sits purely within presidential power happened fastest. On day one he pardoned the January 6 defendants. In February 2025 he signed an executive order ending paper straws, but it only covers federal purchasing, not the states.

What needs Congress got discounted. “No tax on tips” became a deduction of up to $25,000 a year, with income limits, and only for 2025 to 2028. “No tax on Social Security” became an extra deduction of up to $6,000 a year for people over 65, again time-limited and income-tested.

What ran into the limits of the law got struck down. The sweeping tariffs he imposed under the International Emergency Economic Powers Act were ruled unlawful by the Supreme Court 6 to 3 on 20 February 2026, with more than $170 billion facing refunds. The replacement tariffs he then levied under another law were struck down by the Court of International Trade.

What ran into the Constitution got blocked. His executive order ending birthright citizenship was ruled a violation of the Fourteenth Amendment by the Supreme Court, 6 to 3, on 30 June 2026, with his own appointee Justice Barrett in the majority.

What ran into federalism got blocked too. In August 2025 he announced he would end mail-in voting, and in March 2026 he signed an order putting the postal service in charge of mail ballots. In September the Supreme Court kept that order frozen, and mail voting in the midterms is going ahead as usual. Incidentally, ending mail-in voting wasn’t among the 20 promises in the 2024 Republican platform, which asked for voter ID, paper ballots and proof of citizenship.

What depends on state capacity hit physical bottlenecks. In fiscal 2026, up to 11 July, ICE deported 356,389 people, close to the 2013 record under Obama but nowhere near “millions”. Detention beds, deportation flights, court dockets, whether receiving countries cooperate: every one of them has a ceiling.

What needs the other side’s cooperation is hardest of all. During the campaign he said he could end the war in Ukraine in 24 hours. By May 2026 there had been a single three-day ceasefire.

The test shows two things. First, even in the most concentrated system there are several layers between campaign slogans and actual policy: the Constitution, the courts, Congress, the states, state capacity, and opponents. Second, the reverse also holds: where presidential power is big enough, radicals can change things fast and deep. Those tariffs had been in force for nearly a year before they were struck down. A filter isn’t a fuse that blows by itself when the current gets too high. It’s an ongoing fight, and sometimes it loses.

7. Ratchet and thermostat

If the filters discount everything, where does a radical party’s value come from? From the mainstream parties.

Start with Sweden. S is still the largest party after this election and likely to govern again, but it has no plans to go back to 2015. Andersson said during the campaign that she would keep migration policy strict. In May 2026, when parliament voted on the constitutional amendment to strip dual nationals of citizenship, all 106 S MPs voted yes, V voted no and C abstained. To be fair, that amendment was voted on as a package with putting abortion rights in the constitution and restricting the right of criminal organisations to associate. In August, S backed 14 as the age of criminal responsibility. Earlier, in 2023, it said it could drop its “100% renewable” electricity target and accept the governing coalition’s “100% fossil-free”, which leaves room for nuclear. In the new parliament, the opposition can only reverse Tidö-era laws with S votes, and judging by what Andersson says, S isn’t planning to.

Germany is following, more slowly and with clearer limits. The Merz government suspended family reunification, and its coalition partner the SPD voted yes through gritted teeth. In March 2026 the Bundestag replaced the citizen’s income with a stricter new basic income support, and the government wants Ukrainians who arrived after April 2025 to move onto asylum seekers’ benefits, about a fifth lower. On energy, though, the German mainstream didn’t follow: the coalition agreement has no nuclear restart, and the CDU/CSU parliamentary group called restarting nuclear a “shop-window debate” in committee.

Denmark moved first. Around 2019, the Danish Social Democrats (Socialdemokratiet) under Mette Frederiksen took over a strict migration line. The Danish People’s Party (DF), which had 21.1% in 2015, fell to 8.7% in 2019 and 2.6% in 2022. When the mainstream takes the radical party’s issue, the radical party shrinks. That works like a ratchet: once the gear turns one way, it’s hard to turn it back.

But the ratchet is only half right. DF bounced back to 9.1% in the March 2026 election and is now polling at 13 to 14%, while Socialdemokratiet got just 21.8%, its worst result in more than a hundred years. Political science has a better image for this: a thermostat. When the public thinks migration policy is too loose, it demands tightening. Once policy tightens and the problem eases, or people feel it’s tight enough, demand for more drops and the radical party loses ground. If people feel the problem still isn’t solved, the demand comes back. DF’s comeback has more than one cause, but the thermostat explains at least part of it.

Sweden’s result reads the same way. The SD’s lost votes mostly went to the Moderates (M), and among voters aged 19 to 21 it fell by about 7 points. SD MP Tobias Andersson’s explanation was that the SD is a great opposition party but struggles in the role of propping up a government. Put another way: once M took over the SD’s issue, the SD lost its monopoly on it.

The ups and downs elsewhere fit the same frame. The Dutch Party for Freedom (PVV) became the largest party with 37 seats in 2023 and joined government, couldn’t push its asylum agenda through, walked out of the coalition in June 2025 and brought the government down. In the October election it fell to 26 seats and was left out of the new government, while JA21, also hardline on migration, went from 1 seat to 9. Austria’s Freedom Party (FPÖ) came first in 2024 with 28.8%, but coalition talks collapsed and a three-party coalition of the others took office. Hungary’s Orbán fell from power in April 2026 after 16 years. The US too: according to Gallup, the share of Americans wanting less immigration was 55% in 2024, fell to 30% in 2025 after Trump took office, and stood at 29% in 2026.

Radical parties get filtered too. The SD dropped its call to leave the EU in 2019 and its opposition to joining NATO in 2022. The AfD’s 2025 draft programme called for leaving the EU; the version adopted turned that into “transitioning from the EU to a European union of nations”. That’s aiming high inside the party: shout about tearing the roof off, then settle for the window yourself. And not everything gets absorbed. The AfD’s calls to lift sanctions on Russia and repair Nord Stream are still rejected by the German mainstream.

It also isn’t fair to put all of this down to radical parties. The 2015 refugee wave, gang violence in Sweden and how integration actually turned out all changed voters’ minds on their own. Radical parties are more of a catalyst than the only cause. And the AfD’s high vote in the east isn’t only about migration: the east-west income gap, pensions, energy prices and distrust of the federal government are all in the mix.

So “Europe’s turn to the right” needs saying more precisely. It isn’t right-wing parties marching from victory to victory. They rise and fall; some lose ground once in power, some win elections and still can’t form a government. What has moved is the policy consensus. On migration, citizenship, welfare eligibility and law and order, Europe’s policy centre has shifted a long way toward restriction, and what has shifted rarely shifts back. Once a radical party has done the pushing, it often loses its monopoly on the issue. Bad news for the party. Not necessarily for the policies it pushed.

8. The economics: two ledgers, not one

Bottom line first: the biggest confusion in the economic argument is lumping refugees and labour migrants together as “immigrants”. One side says immigrants fuel the economy, the other says they’re draining the welfare state, and both are arguing from a single average. Split it, and the two ledgers point in opposite directions. Refugees and asylum migrants are, on average, a long-term net cost. Labour migrants selected for their skills are a net contribution, and they are exactly what Europe is short of.

Start with the optimism of 2015. That September, Daimler’s CEO Dieter Zetsche said that at best, refugees could lay the foundation for Germany’s next economic miracle. Less than a year later, the Frankfurter Allgemeine Zeitung surveyed the 30 DAX companies: between them they employed 54 refugees, 50 of them at Deutsche Post. The idea of refugees as a ready-made workforce hit a wall within a year.

The fiscal ledger splits cleanly by channel. In 2025 Sweden’s National Institute of Economic Research (KI), commissioned by the government, calculated immigrants’ net contribution to public finances from 1983 to 2022. “Other foreign-born” here means people who came through non-refugee channels: work, family, study.

The table says it plainly: the refugee group is negative, the other group is positive, and together they leave a small plus of 6 billion kronor in 2022. Look only at the total and you’d think immigration roughly breaks even. Split it and you see a loss and a gain cancelling each other out. Refugees’ net contribution was negative in every single year from 1983 to 2022. Over a lifetime, a refugee turns net positive about 15 years after arrival, peaks at around 30 years, and turns negative again after 40 as they retire. The overall figure turned positive in 2022 for the first time in more than thirty years, mainly thanks to high-skilled labour migrants. The study has limits: it’s a static calculation, it leaves out indirect effects on wages, jobs and growth, and it counts the second generation as native-born. But it’s the most authoritative number Sweden has.

Dutch and Danish studies point the same way. One Dutch study puts labour migrants’ lifetime net contribution in the positive and asylum migrants’ at around minus €400,000 on average. Denmark’s finance ministry calculated that immigrants and their descendants from the Middle East, North Africa, Pakistan and Turkey had a net cost of about 24 billion Danish kroner in 2018. Three countries, one conclusion: what decides whether the ledger is positive or negative isn’t whether someone is an immigrant. It’s which door they came through.

Germany’s ledger is harder to work out, because the federal government and the states count differently. At federal level, refugee-related spending was €29.7 billion in 2023, 6.4% of the federal budget, falling to €24.8 billion in 2025. The often-quoted “nearly €200 billion on refugees” is the federal total for 2015 to 2023, and it includes foreign aid. The states separately pay benefits under the Asylum Seekers’ Benefits Act, about €6.7 billion in 2024, but that can’t simply be added to the federal figure, because federal transfers to the states already cover part of it. An official total covering states and municipalities doesn’t exist.

Crime splits by channel too. A 2025 research review by Sweden’s National Council for Crime Prevention found that in Nordic studies, people from high-income countries who came for work or study offend at low rates, even lower than the native-born, while people from low-income countries who fled war offend at higher rates, with access to the labour market a key factor. The big asylum wave around 2015 is linked in several European studies to a short-term rise in crime. German research shows the increase in violent crime was confined to cases with foreign victims, and German victims actually declined. But the review also says plainly that current research can’t establish immigration’s net effect on crime in Sweden. Admitting a problem exists and pinning it on a single cause are two different things.

So when does “immigrants take our jobs” hold true? The evidence says: when the immigrants are actually working, have skills close to the locals’, and arrive in a concentrated wave. In the 1990s Germany distributed large numbers of ethnic German returnees around the country by quota, and research found roughly three locals lost their jobs for every ten returnees in work. When Czech workers were allowed to commute across the border into German border regions, local wages dipped slightly and local employment fell more noticeably, concentrated among people in routine jobs: older locals lost jobs, younger ones took pay cuts. In other words, the immigrants who really compete with locals for jobs are the ones who are working and have similar skills. Not refugees.

Refugees are a different story. Studies of the 2015 cohort found no evidence that they pushed Germans out of jobs, for a simple reason: they struggled to find jobs themselves. A Danish study even found that after refugees were dispersed around the country, low-skilled locals moved into more complex jobs and their wages went up. A German study found that when some districts lifted employment restrictions on asylum seekers, the people affected were not locals but other foreign residents. KI’s review of the Swedish research reaches a similar conclusion: immigration’s direct effect on native employment is small, with a negative but limited short-term effect on natives without a degree and lower wages for the bottom fifth of earners. Earlier immigrants also take a hit from new ones. Put simply, refugees go through the fiscal channel, working low-skilled migrants go through the competition channel, and the competition lands first on earlier immigrants.

That also explains an apparent paradox: Germany has a labour shortage and mass unemployment at the same time. In the first quarter of 2026 there were 1.15 million vacancies; unemployment in August stood at 3.061 million, a rate of 6.5%. Only 24% of vacancies require no vocational qualification, yet 48% of job seekers are looking for exactly that kind of work. 37.2% of foreign employees do helper-level jobs, against 12.5% of Germans. In Sweden, unemployment is 16.4% among the foreign-born and 6.2% among the native-born. What’s missing is engineers, nurses and skilled tradespeople. What’s in surplus is unqualified labour. So a country can perfectly well have “too many immigrants” and “not enough immigrants” at the same time: too many people who can’t get into the labour market, too few who can step straight into the vacancies.

The strongest counterargument is demographic. Germany’s Institute for Employment Research (IAB) estimates that without net immigration, the working-age population would shrink by about 16 million by 2060, and that around 400,000 net immigrants a year are needed to roughly hold the workforce steady. Since 2010, about 70% of employment growth in Germany has come from foreign workers. Those numbers are real. But what they support is selecting skilled workers from outside the EU, not expanding asylum. In KI’s data, it was high-skilled labour migrants who pushed the foreign-born as a whole into the black. The demographic gap is real, but it has to be filled with skilled migrants picked for the jobs that are short. Expecting the asylum system to fill it on the side just doesn’t add up.

Put it all together and the conclusion is blunt: refugees and labour migrants are two different things and shouldn’t be counted as one “immigration”. In high-welfare countries, asylum migrants’ average lifetime net fiscal contribution is negative. Employment comes slowly and at low rates, and it takes well over a decade to turn positive. Skilled migrants selected to fill shortages are, on average, net contributors, and exactly what Europe’s labour markets lack. So asylum is humanitarian policy, not labour-market policy, and how many to take and how much to spend should be argued on humanitarian terms. Defending more asylum with “we need workers” uses the labour-migration ledger to cover for the refugee ledger. And using the refugee ledger to reject all immigration is the same confusion the other way round.

9. Why someone putting down roots can back aiming high

How should a foreigner who came to Europe to work and plans to stay for good look at this wave of tightening? Instinct says the parties that want less immigration are every foreigner’s opponent. But that instinct treats immigration as one thing, and the data keeps showing that different channels have completely different ledgers.

First, the fiscal ledger is on his side. The tightening mainly targets asylum and welfare-dependent channels, and his channel is a net contributor to public finances. Sweden’s new welfare rules require newcomers to have lived in the country for a total of five years within fifteen before they can claim residence-based benefits like child benefit, housing allowance and parental benefit, but working speeds up eligibility. The rule punishes not working. It rewards exactly what he does.

Second, the competition ledger is on his side too, uncomfortable as that is to say. New, unselected low-skilled workers hit earlier immigrants doing similar work first, not locals. A foreigner who has already put down roots has every reason to want newcomers screened the way he was. Critics will call that pulling up the ladder behind you. The charge only sticks in one case: if the ladder was only ever lowered for him. If newcomers face the same rules he did, judged by the same behaviour, that’s not pulling up the ladder. That’s asking everyone to climb the same one.

Third, the thermostat. If the old consensus refuses to adjust, voters’ demand doesn’t go away. It comes back at the next election in a cruder form, and a crude backlash doesn’t distinguish between channels. It lands on every foreigner. In that sense, backing a controlled tightening is buying insurance against an uncontrolled one.

But the line has to be drawn clearly. How do you judge “willingness to integrate”? If you take it to mean embracing every aspect of the new place and becoming culturally local enough, you fall into Sellner’s trap: that standard has no objective answer, so in the end whoever holds power decides, and that is exactly why the Federal Administrative Court ruled it violates human dignity when applied to citizens. The only workable standard is observable behaviour: supporting yourself, speaking the language, obeying the law, living here a set number of years. That’s precisely what Sweden’s new citizenship rules test. And the test can only happen before naturalisation. Once you’re a citizen, you’re an equal citizen, with no second exam. That line is where my “aim high” and Sellner’s “remigration” part ways.

This position has costs, and there’s no pretending otherwise.

The first cost is that the rules land on you too. Sweden’s residence requirement for citizenship went from five years to eight with no transition rule, so someone who has lived here five years and is about to apply has to wait three more. The salary threshold for work permits went up to 90% of the median, and the good-conduct requirements got stricter.

The second cost is tail risk. Not every party you’d be backing draws its lines by behaviour. The SD once planned to split citizens into two classes by birthplace, people in the AfD talk about changing the makeup of the population, and Sellner sat in the same room as AfD members. Backing their opening bid is a bet that the filters will strip those parts out. So far they have: the constitution stopped Sellner, a coalition partner stopped the mosque proposal, and parliament narrowed the citizenship clause to dual nationals. But filters can be rewritten too. It just takes bigger majorities and more time. This bet isn’t risk-free.

“Aim for the top, end up in the middle.” The real goal was always the middle. Anyone backing the people shouting to tear off the roof needs to be clear that all they actually want is a window, and needs to make sure the roof doesn’t really come off. Because the house that ends up with the window is one we all live in, including the people who moved in a bit later.