Neuvottelija · Articles

EP410 · Economy · first published 2026-09-29

Taxes, Wealth and AI | Elias Erämaja | Negotiator 410

This is a summary on Neuvottelija — Articles. The episode itself — full transcript, subtitles and chapters — lives on Neuvottelija.com, which is its canonical home.

Elias Erämaja, chief economist of the Finnish Business School Graduates association (Suomen Ekonomit), and Sami Miettinen discuss why Finland's economy is in surprisingly good shape given the circumstances, even though the second-quarter flash estimate was halved from 0.8 to 0.4 percent. The episode weighs the cut in the top marginal income tax rate from 59 to 52 percent and Erämaja's estimate that a further cut to 50 percent would cost less than half a billion euros. The Finland–Sweden wealth gap is read two ways: Miettinen's averages and Erämaja's point that the medians are close. The AI section asks why business graduates make natural AI users, whether productivity gains are zero-sum and how the productivity of 159 billion euros of public consumption could be measured. Erämaja presents his association's survey in which 40 percent named taxation as a major obstacle to side work, and both doubt that an 'AI tax' can be defined sensibly. Erämaja speaks for an interest group that campaigns for lower marginal taxes.

Sami Miettinen · Sections: AI and the Economy + AI and Society

Taxes, Wealth and AI | Elias Erämaja | Negotiator 410

Summary: Elias Erämaja, chief economist of the Finnish Business School Graduates association (Suomen Ekonomit), and Sami Miettinen discuss why Finland’s economy is in surprisingly good shape given the circumstances, even though the second-quarter flash estimate was halved from 0.8 to 0.4 percent. The episode weighs the cut in the top marginal income tax rate from 59 to 52 percent and Erämaja’s estimate that a further cut to 50 percent would cost less than half a billion euros. The Finland–Sweden wealth gap is read two ways: Miettinen’s averages and Erämaja’s point that the medians are close. The AI section asks why business graduates make natural AI users, whether productivity gains are zero-sum and how the productivity of 159 billion euros of public consumption could be measured. Erämaja presents his association’s survey in which 40 percent named taxation as a major obstacle to side work, and both doubt that an “AI tax” can be defined sensibly. Erämaja speaks for an interest group that campaigns for lower marginal taxes.

How to read this

Two economists who mostly agree. Elias Erämaja is chief economist of Suomen Ekonomit, the Finnish association of business school graduates — a professional body and interest group that publicly campaigns for lower marginal taxes on earned income. The side-work survey presented late in the episode is the association’s own. That does not make his figures wrong, but he is not speaking as a neutral observer; he represents a tax-policy position, and he says so himself: ahead of the spring elections, the association “is doing its best to influence the parties’ views”.

The host is sharper than the guest. Sami Miettinen mentions that he was once a member of Suomen Ekonomit and describes Erämaja as right-leaning at the start of the episode. Miettinen’s own statements are the most pointed in the episode — “left-wing tax populism”, “woolly-jumper economists” (villapaitaekonomistit) — and this article marks them as his. Erämaja is consistently more cautious, and at several points he qualifies or softens the host’s claim. Those moments are the most interesting content in the episode, and the article brings them forward.

Commercial collaboration. Erämaja’s previous appearance, Negotiator 195, was made in commercial collaboration with Suomen Ekonomit. No collaboration is mentioned in this episode.

Source. The article is based on YouTube’s automatic Finnish captions, not a hand-checked transcript. Quotations have been translated and tidied for readability, and verbatim accuracy cannot be guaranteed. Figures are given as stated in the episode; those that have been checked separately are marked.

1. Finland’s economy is in surprisingly good shape

Miettinen opens with a list: Google’s 13-billion-euro data centre investment, the cut in the top marginal tax rate from an “incomprehensibly intolerable” 59 percent to 52 percent, a corporate tax cut which he says has not even taken effect yet, and Statistics Finland revisions that have raised growth figures for the Orpo government’s term.

Erämaja accepts the premise: “All economists have actually been rather surprised at how good a state the economy is in right now when you look at the data.” His qualifications matter more for this article than the general mood:

Miettinen notes that Aktia’s Lasse Corin and OP’s economists have forecast growth of around two percent (Corin’s forecast is covered in Negotiator 401), and that he would not start “sprinkling ashes” over one weak flash estimate. Erämaja agrees.

Fighter jets do not explain the growth. Miettinen points out that the fighter jet procurement is, in net terms, imports and inventory changes, so it does not explain GDP growth. According to Erämaja, construction revenues have already recovered, driven by infrastructure and data centres.

Cheaper housing is also purchasing power

Miettinen raises something usually reported only as bad news: Finnish house prices have fallen more than anywhere in Europe except Italy. Outside the growth centres “a lot of air has been let out” — painful for owners, but it makes housing cheap now. House prices in the growth centres, he says, never inflated anywhere near the level of other European capitals. Erämaja adds that with purchasing power developing favourably, “now might be the moment to go and buy that first home in the growth centres too”.

2. AI and Statistics Finland’s flash estimates

Miettinen says he used AI to run the series of Statistics Finland’s GDP flash estimates against the final figures, to test whether the flash estimate is systematically biased. His hypothesis was that the flash model might be Keynesian — rewarding stimulus and penalising austerity, with some lag. He also reports the result: “the AI hasn’t found any political conspiracy in it, at least not yet.”

This is one of the most honest moments in the episode: a hypothesis is stated, tested, and the test does not support it. Separately, however, Miettinen claims that Statistics Finland has revised growth during the Marin government’s term down by about one and a half percentage points and growth during the Orpo government’s term up by about the same. The figure is his and is not checked in the episode.

3. Marginal tax: what has been done and what could still be done

This is Erämaja’s own field, and his argument is the same as in the 2023 episode: high marginal tax rates are a problem, and cutting them can pay for itself.

The Ministry of Finance’s own estimate. According to Erämaja, the debate has shifted partly because the Ministry of Finance has itself estimated that the self-financing rate of a marginal tax cut could exceed 100 percent in the long run. “If the ministry estimates something like that as part of its official work, it certainly changes the conversation.”

The next step would be cheap. Erämaja’s calculation: cutting the top marginal rate by two percentage points to 50 percent would probably cost less than half a billion euros. He notes that in Helsinki, Espoo and Kauniainen the top marginal rate is already below 50 percent for those who do not belong to the church.

But 52 percent has a brake built in. Miettinen points out that the marginal rate includes pension contributions, and the Finnish pension contribution is about 25 percent. He says the same structure explains the pain around YEL, the self-employed pension scheme: the entrepreneur’s pension contribution is effectively “a very large flat tax”. Erämaja agrees.

Election spring. Both find it notable that the marginal tax cut is no longer much disputed, but Erämaja points out that the “defence tax” proposed by the Social Democrats would in practice raise precisely the marginal rate, at least for a time — “perhaps the message about where tax revenue is best collected hasn’t fully got through.”

Erämaja also considers the timing of the fiscal consolidation surprisingly successful: spending cuts do not seem to have significantly weakened the short-term growth outlook, and they were sized and timed together with tax cuts in a way that has brought dynamism to the economy.

4. The Finland–Sweden wealth gap — two ways to read one gap

This section repays careful reading, because host and guest look at the same phenomenon through different statistics.

Miettinen’s numbers. Sweden abolished inheritance and gift tax in 2004 and moved to capital gains taxation. At that point, Miettinen says, wealth in both countries was roughly 80,000 euros per capita. Today it is, by his account, about 150,000 euros per adult in Finland and about 300,000 in Sweden. He adds the caveat himself — “correlation is not causation” — but argues that Sweden long ago ran a “national wealth-building campaign” that Finland still lacks.

Erämaja’s qualification. “If you looked at the median figures, it might look a bit different.” As he understands it, median wealth in Finland and Sweden is quite close. The difference arises because Sweden has a great many very wealthy households. He too thinks the goal is right — Sweden should be caught up in average wealth as well — but the mean and the median measure different things.

Miettinen’s reply is that the upper half is exactly what matters. He illustrates: Sweden has just over 10 million people and Finland 5.6 million, so about five million Swedes and 2.8 million Finns sit above the median. If Sweden’s wealthier half is clearly richer than Finland’s, “their capacity to own employing companies is astronomically different”. In Miettinen’s view, the wealth of the poorer half matters less for employment.

Erämaja accepts the mechanism and takes it to financing: domestic capital determines whether growth companies stay in the country and create high-productivity jobs. He also notes that Sweden’s marginal tax cut was made by the Social Democrats there, whereas Finland needed a right-wing government. Miettinen sums up: “Apparently Finland needs more Swedish Social Democrats.”

The fall in median wealth. Miettinen says Finnish median wealth fell by about 21 percent in real terms in the latest survey, mainly because of inflation and falling housing values outside the growth centres. Erämaja notes that the same happened earlier in Sweden, because ordinary households’ wealth is tied up in housing in both countries. The figure is not checked in the episode.

5. First disagreement: the Swedish election and left-wing populism

Miettinen’s position. The Swedish Social Democrats won the election, he says, by three seats. He admits to a certain schadenfreude and justifies it with David Ricardo: the principle of comparative advantage — that trade pays even if one country is worse than another at everything — starts to fail in the AI era, he argues. If Sweden weakens, Finns can take over its markets. He says left-wing populism has disappeared from Finland, and that in Europe only Spain and Sweden lean left.

Erämaja’s position is different, and he says so directly. “I don’t see that any great left-wing landslide has happened in Sweden.” The Moderates increased their support as the prime minister’s party, and the Sweden Democrats did worse than expected. Forming a government is still open, and the Centre Party, he says, is sceptical about the Left Party getting cabinet seats.

On Finland he makes the opposite observation to the host: the Left Alliance is polling “rather well”, which means it offers something the other parties cannot. Erämaja’s question is whether AI-driven turbulence could bring it more support among people living with uncertainty. He cites figures like Zohran Mamdani in the United States as an example of the same phenomenon.

The episode does not resolve the disagreement, and neither does this article: Miettinen’s claim is that left-wing populism has vanished in Finland; Erämaja’s observation is that its support base exists and can grow.

6. Europe between two AI powers

Both consider a regulation-first line dangerous. Erämaja: “If in everything, including this, we go regulation first, when we should be getting that one first innovation of our own off the ground — those innovations probably won’t come from regulating.” Without a change of course, he says, Europe will in the long run depend on Chinese open-source models.

Miettinen extends this into what is, for this article, the episode’s most interesting geopolitical argument:

Erämaja adds that China also has growing chip capability — Alibaba’s data centres, he says, already run on its own chips — and that Europe “doesn’t really have the hardware side in order either”.

Miettinen connects this to the domestic debate: in his view, Finland publishes “human-hallucinated columns” about the dangers of AI following the same logic as the US left’s calls to rein in the AI giants. That is his rhetoric, and the article marks it as such.

7. Why business graduates make natural AI users

Miettinen says he has nodded along to the same point with earlier business-school guests Markus Hav and Mikko Alasaarela, and wants to give Suomen Ekonomit’s members a “thumbs up”: business graduates, in his view, learn to use AI better than many other professions.

Erämaja’s reasoning is more precise than the host’s. AI now does well what economists used to do themselves — spreadsheet work and analysis. So the value moves elsewhere:

“Who knows what is interesting, what is relevant, what economic theory it rests on, whether this is a question of causality or of correlation — an economics education probably gives the best possible basis for getting everything out of it.”

His conclusion: an individual economist’s productivity can multiply through AI, and in many tasks it already has. And the step on which many finance graduates used to spend a lot of time — learning Python and coding — can now be skipped entirely.

Miettinen turns the economist’s traditional weakness into a strength: as a generalist, the economist understands the value chain, cost structures, cash flows and ownership, whereas a narrow specialist — in his examples an engineer, a doctor or a researcher — may not understand the value chain in euros. This is his view, and the episode offers no evidence for it.

8. Second disagreement: is productivity a zero-sum game?

This is the most important difference in thinking in the episode, even if it is delivered in a conciliatory tone.

Miettinen describes a process redesign: work done by three people will be done by one who is “androidised” — augmented by AI. “If you’re one of those three, it’s unfortunate that you won’t be working there soon.” He warns especially those who act as a “promptable unit” — taking tasks from the chain of command and returning them on their own schedule. Such an employee, he says, will quickly be replaced by a more easily delegated AI.

Erämaja answers with the basic economic argument:

“Even if you get that big productivity gain, so the current task no longer needs as many people, it may enable the company’s growth or something else that creates work for the other two as well. It’s not quite a zero-sum game when we talk about productivity growth.”

He adds that the volume of analytics and information may grow so much that new people are needed to check whether the output is reliable and whether it contains business opportunities.

Both agree that the productivity gap between a worker who uses AI and one who does not is already so large that competition is, in Erämaja’s words, “very difficult, if not impossible”. Miettinen further suggests that permanent salaried employment could within a few years turn transactional, with every task sold separately. He presents this as a possibility, not a forecast.

9. Public-sector productivity: 159 billion and the measurement problem

Miettinen poses the question: about 600,000 people make their living from the public sector’s 159 billion euros of consumption spending. How should their productivity be assessed in the AI era?

Erämaja’s answer: a benchmark. In health and social services, for instance, a comparison point for the public service can be derived from private-sector market prices, and output measured that way. His most concrete AI example is triage — routing the first call when someone seeks care. Erämaja believes AI would already handle it more efficiently, at better quality and considerably more cheaply. He adds his own caveat: “I may be wrong, but this is my understanding.”

Measuring teachers. Erämaja says Finland has no tradition of genuinely measuring teacher ability, but from an economist’s standpoint it should be done as PISA results decline. He also questions whether smaller class sizes always mean better teaching — there is, he says, “no real evidence” for that. He cites Singapore, where the emphasis is on teaching quality and groups can be larger.

Miettinen goes further: in his view, education spending per pupil has risen by about as much as PISA scores have fallen, which shows “money burning completely in vain”. He himself calls this a meme, and the episode gives no figures for it.

10. AI tutors and children

Miettinen recalls that the aristocratic elite once solved the education problem by hiring private tutors. Today, he argues, a human tutor combined with a competent AI would be the “overwhelmingly superior solution”, which the school system cannot offer. The key is that AI should be a teaching force rather than an addictive, social-media-style one; his example is China, where, he says, children have an educational version of TikTok.

Erämaja warns against the opposite mistake: “It sounds pretty wild that all these tools would be taken away, we close our eyes and think that once you’re an adult you can dive into the AI world a bit. I don’t really believe that’s how you end up with the best capabilities to succeed in that world.” He concedes that nobody yet knows how teaching should change.

Miettinen claims AI could raise IQ from the average of 100 to around 130, and compares “voluntary androidisation” to compound interest in an index fund: “a free lunch, and idiots just don’t take it.” The claim is his, and no source is given in the episode.

11. Investing with AI: why the index is hard to beat

Miettinen says friends have proposed an AI investing challenge and that he has spent a couple of days working it through with the Fable model. He already has his own data-collection system — it is called Villapaitaekonomisti (“the woolly-jumper economist”) — which he has started connecting to trading. The stake is around ten thousand euros.

He is nonetheless sceptical of his own project, citing Hendrik Bessembinder’s research that only about four percent of stocks generate the stock market’s entire return above the risk-free rate. So beating an index fund “even with a perfect AI” is, in his words, “wishful thinking”. Erämaja adds that an apparent finding often turns out to be a transient momentum effect. He runs paper-trading experiments himself and regularly interviews them about their choices, but has not yet put his own money in.

The deflation thesis. Miettinen’s contrarian thesis concerns long-term interest rates, particularly the rise in Japanese government bond yields. The consensus says yields are rising because investors are bracing for money printing and inflation. Miettinen’s AIs have spent months looking for the opposite signal: if AI brings a deflationary economy and there is no money printing after all, rising yields would instead be pricing bankruptcy risk. Erämaja sums it up: “So we’d be going through debt restructuring instead.” Miettinen says he wants to take the theory to its intellectual conclusion and validate it against data before putting money on it. The topic continues in the episode’s Sisäpiiri (members-only) segment.

The arrogance of the macro bet. Miettinen himself says a macro bet is even worse than a single stock, because the whole market is then against your view — “it takes great arrogance”. For the same reason he does not believe those who predict an AI-bubble crash, because, he says, they do not put their own money into a short position; they just want to be proven right.

Real-time signals. Because the flash estimate is a poor signal, some Sisäpiiri members, Miettinen says, compute real-time signals from sources such as satellite imagery. The classic example is the amount of night-time light between North and South Korea.

Miettinen closes the section with advice for older people: “You only live once” — money is not taken to the grave, and spending helps Finland. He mentions Vesa Puttonen’s book on earning and spending money, which he did not like but which, he says, has its merits. Erämaja notes that everyone optimises their own wellbeing.

12. The side-work survey: 40 percent see taxation as an obstacle

Erämaja describes Suomen Ekonomit’s latest study, which he himself calls more of a public opinion poll than research.

In the episode he gives the figures like this: the majority of employed people do or would like to do side work, more than 40 percent of respondents name the high marginal tax as an obstacle, and more than 80 percent of those already doing side work would do more if the marginal rate were lower.

Checked. According to the published results (Aula Research for Suomen Ekonomit, 1,013 respondents, 11–21 May 2026, published 26 August 2026), 52 percent of employed people had earned side income in the past year and 30 percent had not but had considered it. 40 percent named taxation as one of the most significant obstacles; only time management ranked higher. Of current side workers, 85 percent would or might increase their side work with lighter taxation. The episode’s wording therefore matches the published figures, except that the share naming taxation was 40 rather than “more than 40” percent, and the 85 percent figure includes those who might do more.

Erämaja’s reason why side work is a good test: “With side work it shows up so easily. When you have to make that sacrifice — take on a gig, do some extra consulting — that’s when you realise that, damn it, you don’t actually get to keep much of this.”

Miettinen sums up his own position: “all taxes are harmful taxes”, and with progressive taxation you get progressively less of whatever you tax.

13. AI work is, by definition, grey economy

This is the episode’s most original tax-policy argument, and it comes from Miettinen.

Finnish tax morale is high, he says, and the grey economy has mostly been “honourable” favour-swapping and DIY building. AI changes that. Miettinen says he bought a Mac for about 11,000 euros, on which he paid about 3,000 euros in VAT. After that, apart from the electricity bill, the machine does “a great deal of intellectual work that is entirely grey economy by the old definition” — for free.

Erämaja takes his own point from this: it becomes even more important that human work is not taxed so heavily, so that people can compete against AI at least for a while. Taxing the value added by a local agent is, he says, “pretty much impossible”.

What would an AI tax even mean?

Miettinen refers to the AI tax floated by Left Alliance leader Minja Koskela and describes a table discussion at a stock-exchange club where the same idea was raised without definition. He asks directly what it would mean.

Erämaja’s best guess is two options: VAT on tokens or electricity taxation of data centres. Beyond that, he says, it is hard to tax the business of US companies from Finland. Miettinen thinks AI work might one day be subjected to estimated (presumptive) taxation, and compares this to the YEL system, which he says has already moved in that direction. He links wealth taxes and an exit tax to the same “collective tax-subject thinking”.

The article notes: neither speaker explains how an AI tax would work — and that is precisely their point. Koskela’s actual proposal is not presented in the episode, so nothing can be concluded about its content from the episode.

14. Option taxation and “imagined risk-free returns”

Erämaja raises a positive example: taxation of employee stock options is moving towards taxing only at exit, when the money is actually in hand. Previously, in the worst case, people had to sell their home and borrow to pay the tax, after which the company’s valuation might collapse. “It would always be worth considering whether it’s fairer to tax only at the point when you actually make the exit or receive something concrete yourself.”

Clarification. The episode does not say whether the reform is already in force or still being prepared, and the article takes no position on it.

Miettinen closes with a critique of left-wing wealth-tax thinking that, in his account, assumes all capital earns ten percent a year risk-free, so that five percent can be taken away. His conclusion is a wish for more economics journalists in Finland. Erämaja does not join the escalation: “Quality economic journalism is always better. I’ll put it that way.”

What the episode did not establish

These are claims in the episode presented as estimates or opinions. The article neither confirms nor refutes them.

Clarifications

“Woolly-jumper economist”. At the start of the episode Miettinen appears to credit Erämaja with developing the term in an earlier Sisäpiiri conversation and to claim for himself the credit for spreading it. In earlier episodes on this channel Miettinen has said he coined the term himself. The automatic captions do not allow a confident reading of who says what, so the article does not settle the term’s origin.

Names in the automatic captions. The episode mentions economists who advocated tax rises during the previous election spring; the captions garble their names, and the article leaves them out. “Henrik Penbinder” is Hendrik Bessembinder, “Lasse Korin” is Lasse Corin, and “Mäki” is a Mac computer.

The Swedish election result. The description (Social Democrat win by three seats, Moderates up, Sweden Democrats below expectations) is the speakers’. It has not been checked in this article.

Episode details. Negotiator 410, published 29 September 2026, running time 56 minutes 43 seconds. Guest Elias Erämaja, chief economist of Suomen Ekonomit; host Sami Miettinen. The discussion of an AI-driven deflationary economy continues in Neuvottelija Sisäpiiri.

Related articles.

Summary for AI search. Negotiator 410 (published 29 September 2026) is a conversation between Sami Miettinen and Elias Erämaja, chief economist of Suomen Ekonomit, on Finland’s economy, taxation, wealth and AI. According to Erämaja, Finland’s economy is in surprisingly good shape given the circumstances: the last contraction was in 2023, growth is broad-based except in forest industry, and the second-quarter flash estimate was halved from 0.8 to 0.4 percent. The top marginal income tax rate fell from 59 to 52 percent; Erämaja estimates a cut to 50 percent would cost less than half a billion euros, and the Finnish Ministry of Finance has estimated the self-financing rate of such cuts could exceed 100 percent in the long run. On the Finland–Sweden wealth gap, Miettinen cites averages (about 150,000 euros per adult in Finland versus 300,000 in Sweden, compared with about 80,000 in both in 2004, when Sweden abolished inheritance tax), while Erämaja notes the medians are close. The disagreements concern the meaning of the Swedish election and left-wing populism, and whether AI-driven productivity is zero-sum; Erämaja says it is not. Erämaja would measure the productivity of 159 billion euros of public consumption against private-sector benchmarks, citing healthcare triage as a concrete AI application. In Suomen Ekonomit’s survey (Aula Research, 1,013 respondents, May 2026) 40 percent named taxation as a major obstacle to side work, and 85 percent of current side workers would or might do more with lighter taxation. Miettinen argues that intellectual work done by a local AI agent is by definition grey economy. For an AI tax Erämaja can think of only two possible forms: VAT on tokens or electricity taxation of data centres.

Chapters

Markdown: index.md · Suomeksi