So white collar work, where you’re sitting down at a computer, either being a lawyer or an accountant or a project manager or a marketing person, most of those tasks will be fully automated by an AI within the next 12 to 18 months.

Assume for a moment they are right about the technology. Models keep improving, and tasks that used to need a junior analyst or programmer can now be done by a prompt. Does it follow that the jobs disappear?

There are two answers on offer. Alex Imas emphasises the demand side, in an excellent new essay I recommend. I want to offer here the supply side one, which my co-authors Jin Li and Yanhui Wu and I develop in our forthcoming book, Messy Jobs.

Imas asks: if AI makes a wide range of cognitive work cheap, where does spending go next? His answer: spending flows toward goods and services where the human origin is part of what customers buy.

When a sector becomes more productive, spending shifts away from it toward sectors with higher income elasticity. For instance, agriculture employed forty percent of the American workforce in 1900 and under two percent today. As societies get richer, they spend relatively less on food and more on other goods and services.

Human origin itself is, in some markets, part of the value. René Girard argued that human desire is mimetic: we want what others want, and we want it more when they cannot have it. Imas and Kristóf Madarász show experimentally that the winner pays a premium for possessing what others cannot. Imas and Graelin Mandel found that when the scarce good is produced by a machine, the exclusivity premium falls by half.

So as AI cheapens commodity production, real incomes rise and spending shifts toward what Imas calls the relational sector: teaching, care, hospitality, craft and live performance, where the human element is part of what is being bought. Imas concludes that the durable jobs of the future will be nurses, therapists, teachers, boutique fitness instructors, personal chefs, bespoke tailors, craft brewers, live performers, spiritual guides, childcare workers, and hospitality workers.

Many will find Imas’ list a bit frightening. If the relational sector is where human work survives, what happens to the hundreds of millions of people who are neither artisans nor caregivers? What will office workers, consultants, engineers, and middle managers do?

I think the relational shift is part of the story. But not the full answer. Many jobs survive because firms do not buy isolated tasks. They buy bundles. And because organizations do more than process information. They allocate authority and settle conflicts. AI can help with all of this. It does not follow that it can replace the people who do it.

Bundles

Most of the discussion of AI and labour markets starts from task exposure: if AI can perform more of the tasks in an occupation, that occupation should lose employment or earnings. But labour markets price jobs, not tasks.

A job is a bundle of tasks. The real question is not whether AI can perform one component of the bundle. It is whether that component can be separated from the rest at low cost, as we discuss in a recent working paper. When that separation is cheap, the bundle is weak: AI takes a piece, the human role narrows, and labor loses share. When separation is expensive, the bundle is strong: AI helps with part of the work, but the human still sells the full service and keeps the larger share of the revenue.

Many thought travel agents would be eliminated by online booking. As Ernie Tedeschi of Stripe Economics showed this month, travel agent employment is now more than 60% below its dot-com peak. For most of what agents used to do —searching flights, comparing hotel rates, issuing tickets— the bundle was weak. Separating the booking task from the human was cheap, and once it was cheap, the task was gone. But something else happened to the agents who stayed. They moved upmarket, charged planning fees, and joined luxury consortia that offer upgrades and personalized itineraries. In 2000, average weekly earnings at travel agencies were 87% of the private-sector average. By 2025, they had reached 99%. The surviving agents earn more per hour than they used to, precisely because the machine took the weak part and left them the strong one.

Most of the individual tasks that make up accounting involve the diligent completion of spreadsheets. These tasks individually seem easy to replace. In 2013, a study by Carl Frey and Michael Osborne put the probability that accountants and auditors would be automated at 94 percent. A decade later, the US Bureau of Labor Statistics counts 1.6 million accountants and auditors employed, median pay of $81,680, and projects the occupation to grow another 5 percent through 2034, faster than the average for all jobs. By contrast, the BLS category “bookkeeping, accounting, and auditing clerks” is falling, a projected 6 percent over the same decade. The clerical task, writing down transactions and reconciling ledgers, is a weak bundle. The accountant’s job is a strong bundle. She interprets tax law as it applies to a specific client, signs the audit that the bank and the SEC will rely on, and carries the legal exposure.

Three traits make a bundle stronger.

The first is unpredictable demand. If you could tell in advance which tasks a customer will need, you could assign each to the right worker, human or machine. In practice, you often cannot. How often have you placed a seemingly routine customer service call about a bug only for it to turn up a much thornier problem with your account? AI handles the routine well and the delicate poorly, and constant handoffs are expensive. My coauthor Yanhui Wu spent time in Hangzhou with a leading Chinese customer service firm that has deployed a domain-specific model for two years. Human agents are still indispensable on the two dimensions that matter most: recognising what the customer is not explicitly saying and avoiding the kind of interaction that ends up on social media. As one of their senior managers put it, unbundling those tasks from the routine ones would require “frequent switches between the AI and human modes. The coordination cost would be too high.”

The second is production spillovers. Some tasks belong together because doing one makes you better at the other. A radiologist who has already spoken with the patient and reviewed the clinical record reads the scan better than one who sees only the image.

The third is the measurement problem. Armen Alchian and Harold Demsetz pointed out that when several inputs jointly produce an output, their separate contributions are hard to disentangle. If AI drafts and a human signs, and the final product goes wrong, who is legally responsible? Banks, regulators, boards, and clients need someone to blame.

Where these conditions hold, AI may raise the productivity inside the bundle without replacing it. A nurse practitioner with AI diagnostic support may handle cases that used to require a doctor, or an entrepreneur with AI tools may run a company that used to require a team, without either being displaced.

Authority

Bundling understates the case. Organisations are not production functions. They are coalitions of people with legitimate but conflicting goals.

Managers exist to solve conflicts within the firm and to allocate scarce resources between competing ends: marketing would like to buy advertisements; engineers would like more tokens; lawyers would like to add processes; and the financial side would prefer to do more of everything with less money. They cannot all be satisfied at once. Someone has to decide who loses. The question is whether AI can play that role.

AI optimists say AI will do this too. Billions of AI agents will negotiate, draft contracts, and transact at machine speed. Hayek’s insight was that the price system aggregates enormous amounts of dispersed knowledge without anyone needing to understand the whole. If AI makes that system faster and smarter, why do we need managers at all?

Kenneth Arrow, in The Limits of Organization, argued that information is not a commodity like coffee: you cannot inspect it before you buy it, because once you have inspected it, you already have it. The people who hold information have reason to distort it, and what checks that distortion is not a better contract but trust—and trust cannot be traded:

Trust is an important lubricant of a social system: It is extremely efficient; it saves a lot of trouble to have a fair degree of reliance on other people’s word. Unfortunately this is not a commodity which can be bought very easily. If you have to buy it, you already have some doubts about what you’ve bought.

Trust in this sense is not a belief about competence. It is the expectation that the other party will bear a reputational cost if it defects, and will be around tomorrow to pay it.

Oliver Williamson added a second half. Once one party has made a relationship-specific investment, the other can hold it up. Before xAI built Colossus in Memphis, it could walk away to any site. Once it had sunk billions into the facility, the city could demand concessions, and did. A human manager carries a reputation with counterparties who will deal with her again. The firm can fire her if she fails. An AI agent has neither. It cannot be sued, cannot be replaced with fanfare to signal a reset.

There is also the problem of the unforeseen. Every project involves thousands of situations nobody specified in advance. A construction contract says the project must be delivered in March, but it does not say what happens when the electrician and the plumber both need access to the same wall on the same day. The manager resolves this by exercising what Oliver Hart and John Moore called residual decision rights: the authority to decide matters that contracts and processes have not specified in advance.

These residual decision rights are not a cognitive task, but an institutional one. The decider must hold tacit and relational knowledge the people around them will not share, because communicating it would give away their bargaining power. They must bear consequences, meaning they can be sued, fired, or publicly blamed when things go wrong. When a Sonos product launch fails, the CEO is fired, and the organisation moves on. They must confer legitimacy on the decision. This is why organisations hold so many meetings. Meetings are not information exchanges. They are rituals of procedural fairness. People accept decisions they dislike when the process looks legitimate and the decider is accountable for the result.

Could AI acquire this standing? In principle, yes. The machinery that lets human managers do this work, such as corporate registries, professional licensing, courts that can compel testimony, took centuries to build. No equivalents exist for AI. Until they do, some human will have to hold the residual decision rights.

Why Amodei is wrong

AI will not produce mass unemployment in rich economies that can fund the transition. Spending will shift toward work where humans add value. Labour share may fall in aggregate even as human-intensive sectors grow as a share of the economy.

Imas thinks most surviving employment will be in the relational sector where customers demand human origin. I think much of the surviving employment will sit in strong-bundle, AI-augmented work and in the political-organizational core of firms. The future includes more therapists, tailors, personal trainers, and craft brewers, but also more managers whose value lies in handling ambiguity, integrating context, reconciling conflicting interests, and bearing the consequences of decisions.

The disruption of junior career ladders is real, and we have written about it. But the argument that “half of entry-level white-collar jobs be gone in five years” confuses task automation with the extinction of jobs. The real world is messy. The mess is what happens when human beings with competing interests try to get things done together. The economy Imas describes is the economy of what customers want. The economy I am describing is the economy of what organizations need to do. The second is larger.

These ideas draw on ongoing work with Jin Li and Yanhui Wu for our forthcoming book, Messy Jobs.

In a worker survey that the authors undertook in the US and six European countries (Germany, the UK, France, Italy, Sweden, and the Netherlands) they found that, in early 2026, 43 percent of US workers were using generative AI for their jobs, compared with 32 percent on average in the six European countries. The UK leads the surveyed European countries at 36 percent. France and Italy trail at 28 percent and 26 percent.

The gap in how intensely workers use the tools is even larger: 5.2 percent of total US work hours involve AI, roughly double the rate in Northern Europe and triple that of Germany, France, and Italy. These are self-reported figures for conscious use of generative AI and the authors argue they should be read as lower bounds on actual adoption, as AI could be used in the background, without workers’ awareness, as well. They also report firm level comparisons, but they caution that it is less clean, since the US and EU surveys differ in question framing, reference period, and coverage. On the most comparable measure, AI use in the production of goods or services, 7 percent of US firms report adoption versus 4 percent for the EU average.

But what jumps out at me from this picture, more than the US gap is the within-Europe gap, which repeats the gaps we saw in the adoption of ICT. Britain, Sweden and the Netherlands beat Germany, which beats France and Italy. What drives these gaps? Part of it reflects compositional differences in education, occupations, industries, and firm size. A statistical decomposition of the effect across the six surveyed countries finds that demographics, occupation, industry, and firm size explains 55 percent of the average gap. But roughly half remains unexplained by those factors.

Management matters

The paper’s most novel claim is that management accounts for much of that residual. World Management Survey scores, which measure management practices in an homogenous way across countries, are highly correlated with AI adoption rates not just between countries (correlation: 0.81), but also within countries. Within the UK, only 2 percent of firms in the bottom management decile adopt AI, versus 20 percent in the top decile. Using their own worker surveys, the authors construct a personnel management index, based on whether firms reward performance, promote on merit, and address underperformance. A one standard deviation increase in this index is associated with 9.6 percentage points more AI adoption, controlling for country, demographics, industry, and firm size.

This is not the first time that the same technology, available at similar prices, has been adopted at very different rates. The ICT revolution followed an identical pattern. Nick Bloom, Raffaella Sadun, and John Van Reenen showed in 2012 that US firms invested more in ICT and got higher returns from it, and that differences in management practices explained a majority of this advantage.

To understand how and why management matters, it helps to drill down on one case study. The new paper by Bick et al (2026) cites a study I did with Paul Heaton, in which we analyzed how management affects the adoption and use of information technologies by police departments, by using data from nearly all US police departments between 1987 and 2003 to measure adoption and its effects on crime.

American police departments spent heavily on computers during the 1990s and 2000s.

We found that the productivity gains appeared only when technology was combined with deep organizational change. The model that worked was CompStat, the management system pioneered by Bill Bratton in the New York Police Department. If you have seen The Wire, you know how Compstat works: crimes get geocoded, the precinct commanders are trained on the system and made accountable for what their cops are doing and where they are deployed. We found departments that adopted IT with the full CompStat bundle saw crime clearance rates increase by 3.2 percentage points, while those that did not adopt the bundle did not experience an improvement, even if they purchased the computers and software.

NYPD Compstat meeting. NBC News

Peter Moskos, a sociologist who also served as a patrolman in Baltimore, recounts in a recent book how the credit for CompStat belongs to a cop named Jack Maple. Maple was a working-class transit lieutenant from Queens, who wrote an unsolicited memo to Bill Bratton, the NYC police commissioner, advocating the use of geocoded data to solve crime. Bratton read it and later pulled him up through the ranks to be the manager in charge of design, deployment and implementation of the CompStat system. He later called Maple the smartest cop he ever met.

Richard Corkery/NY Daily News Archive

Maple was obsessed with geolocating crime. Before the digital tools existed, he had covered 55 feet of wall with maps of the transit system and tracked crimes and clearances with crayons, by hand, in what he called the “Charts of the Future.” With Bratton backing him, he used the data to hold commanders accountable. The CompStat meeting became what Bratton called “the grand theater where precinct commanders are called to account,” sometimes “browbeaten” in front of their peers.

Maple’s lesson spread unevenly through the US: departments that bought the same computers but did not change their management saw no improvement. The technology was necessary, but it was reorganization that made it productive.

Why Europe has fewer Maples

Apart from providing an overall score, the World Management Survey run by Bloom, Sadun, and Van Reenen has been analyzing the frequency of specific management practices across countries for two decades i Their findings provide a possible explanation for why the Bick et al. results look the way they do.

American firms score higher on average on management practices than European firms. The gap is significant, and it persists even when you compare firms of the same size, in the same industry. Three forces drive it. First, competition: US product markets are more contestable, and badly managed firms die faster, which raises the average. Second, ownership: a larger share of European firms, especially in Southern Europe, are family-managed (as opposed to family-owned but professionally managed), and family management is strongly associated with lower management scores. Third, labor regulation: rigid employment protection makes it harder to reward high performers and address low performers, which is precisely the personnel management index that predicts AI adoption in the Bick et al. data.

Italy is the extreme case. Bruno Pellegrino and Luigi Zingales (2017) linked bad management practices to slow IT adoption and argued this failure is the main reason Italy’s productivity has stagnated:

We find that Italy’s productivity disease was most likely caused by the inability of Italian firms to take full advantage of the ICT revolution. While many institutional features can account for this failure, a prominent one is the lack of meritocracy in the selection and rewarding of managers. Unfortunately, we also find that the prevalence of loyalty-based management in Italy is not simply the result of a failure to adjust, but an optimal response to the Italian institutional environment. Italy’s case suggests that familism and cronyism can be serious impediments to economic development even for a highly industrialized nation.

In Pellegrino and Zingales‘ view, management practices respond to an institutional environment where weak contract enforcement and dense networks of personal relationships made loyalty-based management rational for individual firms, even as it proved disastrous for aggregate productivity. Familism and cronyism, they concluded, can be serious impediments to economic development even for a highly industrialized nation.

The Bick et al. paper today points to the same past pattern repeating with AI. Italy has the lowest worker AI adoption rate among the six countries surveyed (26 percent), the lowest share of firms encouraging AI use, and the lowest share providing AI tools. The authors argue that management practices that blocked ICT diffusion are blocking AI diffusion. The technology changed, but the institutional environment that made IT adoption fail is placing the same obstacles to the adoption of artificial intelligence.

Share

The organizational gap

Technology adoption is often not a technological problem, but an organizational problem. The gains from a new technology go to firms and institutions that reorganize around it. But doing so requires decentralizing authority, holding people accountable for results, and actively pushing workers to experiment. Firms that buy the technology but do not change how they work see little benefit.

While the Bick et al. paper cannot prove this story causally, the pattern is consistent across methods, geographies, and decades. Much of the heterogeneity in artificial intelligence adoption is not just within the EU, but within European countries, and much of it traces back to how firms are managed. This will be much harder to solve than repealing the AI Act or GDPR: it points to deep institutional failures in many countries, especially in the European periphery.

References:

Alexander Bick, Adam Blandin, David J. Deming, Nicola Fuchs-Schündeln, and Jonas Jessen, “Mind the Gap: AI Adoption in Europe and the U.S.,” NBER Working Paper 34995, March 2026. Forthcoming in the Spring 2026 Brookings Papers on Economic Activity.

Bloom, Nicholas, Raffaella Sadun, and John Van Reenen. "Americans do IT better: US multinationals and the productivity miracle." American Economic Review 102, no. 1 (2012): 167-201.

Garicano, Luis, and Paul Heaton. “Information technology, organization, and productivity in the public sector: Evidence from police departments.” Journal of Labor Economics 28, no. 1 (2010): 167-201.

Moskos, Peter. Back from the Brink: Inside the NYPD and New York City’s Extraordinary 1990s Crime Drop. Oxford University Press, 2025.

Pellegrino, Bruno, and Luigi Zingales.”Diagnosing the Italian disease” NBER working paper No. 23964. National Bureau of Economic Research, 2017.

The hundreds of billions that European governments spend on defense, health, education, and public administration barely touch AI. Directed well, this spending, which accounts for up to half of European GDP, would improve the efficiency of Europe's sclerotic public sector. But it could also create the conditions for the appearance of a layer of firms doing the messy work of turning general-purpose AI into sector-specific products, which Europe currently lacks and badly needs.

In Part I of this essay, published last July, we argued that the EU should stop focusing its strategic efforts on chasing the frontier in foundational AI and instead focus on becoming a fast and smart adopter. Our argument was based on the conjecture that intelligence is on its way to becoming a commodity.2 The plummeting costs per token, the constant change of the leading model between the three top AI labs (OpenAI, Anthropic and DeepMind), and the existence of an open-weights ecosystem, running a few months behind the frontier, still suggest that this may be the path. In a world where intelligence becomes cheap and widely available, value accrues not to model builders but to firms that integrate AI most effectively. Our conclusion was that the European Union’s job was to deregulate AI adoption, force open standards, build data commons, and let its world-class industries do the rest.

We believe that our diagnosis was correct. However, our prescription was incomplete. We identified implementation and diffusion as the correct layer, but assumed that companies would emerge to exploit it if the European Union cleared the regulatory excesses and kept foundation models competitive. The European open standards and interoperability rules we proposed remain necessary. Without them, European users become captive to a few American gatekeepers. But they are not sufficient. Europe needs a mechanism to get the missing implementation layer started.

We propose here using Europe's public procurement power and a redesigned digital tax to seed this layer, relying on joint ventures pairing foreign AI companies with European partners. Does this kind of forced partnership work? China's auto industry is widely seen as a useful precedent.

Turning failure into success

Under the 1994 Automobile Industry Policy any foreign automaker manufacturing in China had to partner with a Chinese firm, with foreign equity capped at 50%. Import duties that had reached up to 260% in the mid-1980s, and remained prohibitively high through the 1990s, made local production the only way into the market. VW, GM, Toyota, and Peugeot complied.

The policy’s objective was to transfer internal combustion engine technology to Chinese state-owned enterprises. In this sense, it failed. The foreign partners transferred the outcome of their technological capability (the vehicle blueprints) but not the capability itself (the engineering capacity to design the cars). The immense profitability of just producing and selling VW and GM vehicles turned SOEs like SAIC, FAW, and Dongfeng into rent-collectors. The joint venture strategy was widely considered a failure given the inability of Chinese carmakers to stand on their feet. He Guangyuan, industry minister when the policy was approved in 1994, was sharply critical of its results in September 2012:

It's like opium. Once you've had it you will get addicted forever…From central authorities to local governments, everyone has been trying hard to bring in foreign investment. But so many years have passed and we don’t even have a one brand that can be competitive in the auto world…I feel red-faced.

Yet the joint ventures worked as incubators for supply chains and talent. J.D. Power quality data shows that in 2001, domestic models had 65% more malfunctions than joint venture models; by 2014, only 33% more. An excellent paper by Bai et al. (2025) identifies two channels. Workers who changed jobs moved from joint ventures to affiliated domestic firms, carrying production knowledge. Suppliers who built to global joint venture standards then sold the same high-quality parts to independent Chinese companies, producing firms like CATL in batteries and Yanfeng in interiors.

When the paradigm shifted to electric vehicles, the existing joint ventures, as well as their Chinese SOE parents, lost huge market share in 2024/2025. The companies that succeeded were all independent domestic brands rather than joint ventures offspring: BYD, Geely, NIO. But they could not have existed without the supply chain the joint venture policy generated.

The lesson for AI joint ventures is that the way the transfer takes place matters. In China's auto joint ventures, the foreign partner could hand over blueprints without handing over the ability to design cars. In AI the organizational capability is how to clean and structure data, how to redesign the way work is done. This kind of knowledge is hard to withhold. An American AI company that creates a joint venture to build, say, an AI-powered clinical decision support system for European hospitals with European engineers cannot keep them in the dark. They learn by doing.

The SOE failure, teaches the usual lesson: incentives matter. The Chinese partners were the result of state patronage, content to collect fees. The European partners of the AI diffusion companies must have skin in the game, and must face the discipline of a competitive market. The goal is not to create national champions sheltered from competition, but to seed an initial firms that can eventually stand on their own.

The Ukrainian market for drones

Public procurement is the second important lever that European governments have. Here, we think they should look to Ukraine for a model.

Before Russia’s full-scale invasion in 2022, Ukraine had seven drone manufacturers. By 2025 it had over 500, producing four million units a year. The industry now attracts over $100 million a year in private investment.

The architecture that made this possible has three features worth studying. First, the government acted as a committed buyer. The Ministry of Defense allocated over $2.5 billion to Ukrainian drone manufacturers in 2024-2025, and 96% of state FPV drone purchases went to domestic producers. Second, procurement was decentralized. Military unit commanders could order directly from certified manufacturers through the Brave1 marketplace or even directly from manufacturers, choosing which systems to buy based on what worked. Third, the entry barriers were kept low. Brave1 provided over 540 grants totalling $50 million, letting small teams prototype and test systems without the overhead of a traditional defense contractor.

As drone warfare is shifting from remote-controlled flight to autonomous navigation and machine vision, the firms that mastered hardware integration are moving naturally into software. AI-guided drones with autonomous target recognition are currently deployed at the front, and drone swarm software attracted the largest single investment in Ukrainian defense technology.

Of course, there is a huge obstacle to generalizing these lessons: war. An existential threat eliminates bureaucratic resistance; the front provides the best real life testing environment, and good incentives for the buyers, since you will not use your dollars to buy the dodgy drone if your life and security depends on the quality of your hardware.

But some lessons do transfer in our view to the AI implementation problem: the state set the demand floor, kept standards open, decentralized purchasing decisions to the users who knew what they needed, and refused to shelter incumbents. Europe’s proposed AI procurement programs in health, education, defense and public administration could adopt the same architecture: unified standards at the EU level, purchasing authority devolved to army units, hospitals, schools, and tax offices that will actually use the tools, and competitive entry open to any firm that meets the specification. The feedback loop will be slower than a battlefield, but it will be faster than a centralized procurement agency in Brussels or Athens deciding what software a clinic in Thessaloniki needs.

Sticks and carrots

For this approach to work, Europe needs both a stick and a carrot.

The stick is a European Digital Tax on large digital services companies. But what we propose is a special kind of tax. Companies subject to the tax could discharge part of their obligation not in cash, but by capitalizing joint ventures with European partners to create European AI diffusion companies. This aligns incentives. A digital company that pays its tax by investing in a JV has a financial interest in making that JV succeed. Its equity stake is worth more if the European partner becomes a competitive firm. Since many of the large digital services companies are also behind the frontier foundational AI models, they will be investing in their own customers. When every major AI company is scrambling to find revenue a guaranteed European channel should be attractive.

Some firms will prefer to pay the tax in cash. That is fine; the revenue still funds public AI procurement.

The carrot is public demand. The EU creates standardized specifications and interoperability requirements for AI-powered tools in four areas where public sector control is strong: defense, health, education; and tax compliance and social security administration.

Three conditions would apply. First, the tools must be built on open interfaces: standardized data formats, portable configurations, and documented APIs, so that any institution can switch providers without rebuilding its systems. Where open-weights foundation models are available and competitive, they should be preferred. But the non-negotiable requirement is substitutability, not open source. A hospital that adopts an AI diagnostic tool must be able to replace it with a competitor’s tool without losing its data or retraining its staff. In practice, this means the software is built so that the underlying AI model can be swapped out the way you change a light bulb: the rest of the system stays the same, only the intelligence behind it changes.3 If a better or cheaper model appears, the hospital switches in a day.

Second, the joint ventures receiving public contracts must demonstrate European capability transfer, by specifically hiring, training and retaining Europe-based engineers. Third, procurement decisions must be decentralized.

Is this industrial policy? Of course. The question, as always, is whether the conditions are right for success. Here, unlike in the car tariff case Luis analysed in an earlier post, the conditions we discussed then are better. These are public contracts, so these decisions belong in the public sector in any case. Moreover, there are no entrenched incumbents to protect, as this AI diffusion layer barely exists in Europe. Finally, the objective is clear: build implementation capability. The risk is the usual one: that the joint ventures specialize in collecting rents rather than in building capabilities. But doing nothing means Europe will import expensive AI services while its data and domain knowledge flow to the firms providing them. The competitive advantage of our industry will follow the data out of Europe.

References:
Bai, Jie, Panle Jia Barwick, Shengmao Cao, and Shanjun Li. “Quid pro quo, knowledge spillovers, and industrial quality upgrading: Evidence from the Chinese auto industry.” American Economic Review 115, no. 11 (2025): 3825-3852.

Share

On March 18, the European Commission will unveil its 28th regime for European companies. The regime is supposed to be the most significant step yet in the post-Draghi attempt to strengthen the single market, give European firms better access to 450 million consumers, and increase economic growth.

So far, it seems unlikely that this will happen. A draft of the 28th regime proposal was leaked a few days ago. We hope that the final version undergoes further revision, as, in its current form, it leaves a lot to be desired.

The main feature in the leaked proposal is an EU registry that will allow firms to incorporate digitally, within 48 hours, for a maximum fee of €100. Commissioner Michael McGrath, whose portfolio includes the proposal, has said he wants to apply this new regime to the widest possible scope of companies, not just innovative growth firms. The impact assessment reportedly estimates the reduction in the regulatory burden on companies of around €400 million over ten years. This is a rounding error for a continent whose internal trade barriers cost it hundreds of billions annually.

The message we hear everywhere in Brussels is a call for patience: bear with us, we are just getting started, and we will be more ambitious later. This is starting to become grating. Why should we expect there to be better chances in the future? It won’t be easier to create a supranational regime than in the immediate aftermath of the Draghi report, after the Parliament voted 492 to 144 in favour, and after France and Germany made it a joint priority.

As one of us has noted before on this blog, Europe does not actually lack startups. It lacks large, dynamic companies — startups that have scaled. It is the frictions around scaling that the 28th regime is supposed remove. A company expanding from Lisbon to Berlin currently may need a new legal entity, a new employment structure, kafkaesque rules on cross-border labour contracts, a new compliance apparatus, and a new VAT registration before it even closes its next financing round. Winding down a business means navigating different creditor hierarchies in each country. And, just as importantly, Europe is bad at allowing founders to exit successfully from their start up by selling it or going public. Between 2008 and 2021, 147 European unicorns were founded. Forty relocated their headquarters abroad, the vast majority to the United States.

The idea behind the 28th regime, which was proposed in both the Letta and Draghi reports, was to create an outside option: a single set of European rules that firms could opt into if their national system held them back. The current proposal doesn’t do that. Even though it takes the form of a regulation directly applicable to all member states, it does not create a supranational company. It is incorporated in member states, governed by the law of the member state of registration, with national law filling any gap not covered by regulation. Notarial preventative control is still required for every formation and amendment of the articles of association, preserving the gatekeepers the whole regime was meant to bypass. By having incorporation take place through national registers, subjecting it to national preventive controls, and deferring to national law on every matter possible outside the scope of the regulation, it will de facto result in 27 different 28th regimes.

This has happened before. The European Company (Societas Europaea), introduced in 2004, was supposed to achieve the same goal. But, out of 23 million EU corporations, only a few thousand are SEs. Why? The law contains so many referrals to national law that it is more complex than the national systems it was meant to avoid.

To be clear, this is not really the fault of Brussels. Many member states do not want a 28th regime that competes with their own systems. The whole point of a successful outside option is that it reveals which national rules promote efficiency and which ones protect incumbents. That is what governments fear. So, every national hobby horse gets carved out from the new regime. McGrath has already promised unions that national rules will be unchallenged on all collective bargaining, codetermination, and labour questions.

The current proposal isn’t any good and the structural forces described above mean that any similar attempt from Brussels, however well-intentioned, will also fail. But it would be a mistake to conclude that no deepening of the single market for start-ups is possible. The key, we suspect, is to trade ambition about scope for ambition about content.

A 28th regime open to all companies ensures that every member state will imagine the regime cannibalizing its own corporate and labour order. The features that matter are then stripped out in the name of safeguards.

The difficulty with a narrower but more ambitious statute has been the fear of not ‘catching’ innovative companies. As Fiona Scott Morton and Reinhilde Veugelers have argued, even predictable criteria such as minimum R&D ratios or patent requirements miss many of the innovative firms we want to help. Facebook launched without patents. Amazon took three years to file its first one (1-click) after foundation. Netflix six.

The regime should apply to young, independent, non-listed companies that are cross-border scalable. That is the class of company for which fragmentation is uniquely destructive. Eligibility should be based on evidence such as risk-capital backing or cross-border activity.1 To ease concerns about labour protection, the regime could apply only to founders and key employees above certain income and ownership thresholds. The regime could also start as a regulatory sandbox and then be reviewed after five years to see whether it has worked in practice.

Once the scope is defined that way, the new regime can afford real substance. We can give this small subset of firms a genuine EU registry, standard venture documents, standard share classes, a simple path from formation to dissolution, one VAT registration, filing and refund channel for firms fully inside the regime.

The Commission’s own proposal may contain a proof of concept. The reported plan includes an EU-wide stock option regime, with common standards for exiting from equity compensation that all countries must respect, based on internal market law. This would be a significant improvement. It shows that Brussels can be ambitious when it narrows the target. Startup associations are putting huge emphasis on this component. This logic simply needs to be made more general: if you can harmonize the way you treat shares for founders and key employees, why not do the same with some labour provisions, tax filing and compliance?

When the United States allowed companies to bypass state securities laws through federal regulation, late-stage firms became four times more likely to attract out-of-state investors. Europe could do the same, and create a regime that opens the door to the same growth opportunities. Countries that wish to maintain their legal traditions can keep them. Young, new firms seeking scale can bypass them.

If it doesn’t decrease its ambition for who it includes, EU Inc. will be another Brussels concept with a logo and a website. Precisely by excluding more companies, the 28th regime could become something that matters, and give Europe’s best firms a reason to be born, to hire, to grow, and to stay in Europe. That is the outside option the continent needs.

The simplest explanation for Europe’s weak technology sector points to lack of capital, and in particular, venture capital. A recent open letter from a coalition of startups asked for ‘a European Venture Capital Initiative’. A recent IMF blog by Arnold, Claveres, and Frie concludes;

Europe’s shallow pools of venture capital are starving innovative startups of investment and making it harder to boost economic growth and living standards.

Europe, this story goes, lacks startups because it spends much less money on them than America. Give European entrepreneurs more money, whether public or private, and the gap will close.

But if we look at what explains the success of the more innovative European countries, the venture capital explanation doesn’t make much sense.

Sweden has ten million people and 48 unicorns (tech companies worth over one billion dollars). This makes it the country with the fourth most unicorns per capita in the world, behind only Israel, Iceland, and the United States. Stockholm itself produces more unicorns per capita than superstar cities like New York or London (but fewer than San Francisco). Sweden has the fourth most unicorns in Europe in absolute terms, only behind the UK, Germany, and France, countries with populations five to eight times larger. Italy’s GDP is more than four times Sweden’s, yet it has one-third of the unicorns, the same ratio as Spain, which has five times Sweden’s population and twice the GDP.

But most of Europe’s venture capital is cross border, and Sweden does not have access to a much larger domestic market for capital than Germany or France. The lack of venture capital activity in most of Europe is not a supply problem, but a demand problem: there are simply not as many promising companies for venture capitalists to invest in.

The reason is not that Sweden has a much greater supply of venture capital than the rest of Europe. It has a much greater supply of investable opportunities.

Subscribe now

Venture capital is not the problem

European venture capital has grown dramatically over the last two decades. Europe’s share of global venture capital rose from roughly 5 percent to 20 percent by 2022, and the performance of European venture capital investments has not been noticeably worse than that of the US. The number of unicorns has surged, and entrepreneurial activity has spread beyond the old hubs: 65 cities across 25 European countries were home to at least one unicorn as of July 2023.

Venture funding for European startups has a substantial cross-border component, and US participation has increased over time. In Dealroom data summarized by Bradley and coauthors, 44 percent of venture capital invested in European companies comes from outside Europe. Cross-border reliance was especially pronounced in some national markets: in 2016 Austria, Ireland, Sweden, and Spain each received over 70 percent of venture capital investment from foreign investors.

Foreign participation is also visible in Europe’s largest rounds: by 2017 Europe had 17 rounds which raised more than €100M, and the paper notes that most included foreign investors, including cases such as Improbable (€502M, Japanese investors) and Farfetch (€397M, Chinese investors). Many of Europe’s most successful unicorns emerged in countries without large domestic venture capital markets, funded primarily by foreign investors. This means that if Italy or Spain had a pipeline of promising startups, the capital would find them.

But investable companies at the right stage are in short supply. To understand why, consider what venture capital funds actually look for.

What venture capital funds need

Venture capital funds have a limited life, typically ten years, during which they must invest, develop portfolio companies, and exit through a sale or IPO. Fund managers earn a fixed management fee (often 2 percent of capital) and a share of profits (typically 20 percent), amounting to total fees of 5–7 percent of invested capital. On top of this, since venture capital investments are illiquid, investors expect to earn a return roughly 2–3 percent above liquid investments.

To generate the returns their investors require, venture capital investments need to produce very high expected gross returns, on the order of 10 percent above listed equity when adjusted for risk. These returns also need to materialize quickly.

In addition, fund management exhibits increasing returns to scale, because many of the costs of running a fund, things like research, systems, compliance, reporting, are fixed or only rise slowly with size, and once a manager has built credible distribution in the form of relationships with platforms, advisers, consultants, and institutional allocators, the effort and expense of raising additional assets typically increase much less than proportionally. This makes it hard for new venture capital teams to raise a first fund, and pushes established funds toward later-stage investments.

The result is that entrepreneurs must rely on other sources of funding such as personal savings, informal capital and angel investors before they reach a stage where venture capital becomes viable. Without a pre-venture capital financing ecosystem, venture capital has nothing to invest in.

Sweden’s secret

Sweden’s success is that it does provide those investable companies. It does so thanks to a virtuous cycle built on angel investors.

Angels are individuals investing in startups on their own account. They have become the most important source of equity finance at the seed and early stages. In fact, there is now more early-stage angel financing than venture capital financing in the US and elsewhere.

Angels range from informal capital providers with limited wealth to ultra-wealthy individuals with entrepreneurial backgrounds. But only a few of these are actually helpful. An effective angel needs access to dealflow, the ability to screen investments, and the capability to add value through expertise and networks.

Sweden had a significant, but not astonishing, presence in the first wave of tech companies. But those alumni have gone on to found a huge number of new companies. Former employees of Spotify, Klarna, Skype, King and iZettle have built (and backed) a wave of unicorns, from finance and payments (Anyfin, Brite, Wise) and business software (Gilion, Stravito) to energy‑grid platforms (Fever) and new game studios (Resolution Games, Snowprint).

Part of this is due to a 2003 tax reform, where corporate tax on capital gains from selling shares in unlisted companies can be postponed as long as the proceeds are reinvested in other unlisted companies. Since Swedish founders and employee shareholders hold their equity through a corporation, this created a strong incentive to recycle proceeds into new ventures after a successful exit. In contrast, a broader tax incentive introduced in 2013 for individuals investing in unlisted companies has been largely ineffective.

Part of the problem with replicating Sweden’s success is that Sweden did not manufacture angel investors directly through subsidies. It needed the conditions (exit opportunities, tax incentives for reinvestment, a culture that treated serial entrepreneurship as normal) for successful founders to become the next generation’s backers.

Sweden abolished its wealth tax in 2007 and inheritance and gift taxes in 2005. It introduced the ISK (Investment Savings Account) in 2012, a simplified flat-tax vehicle that further increased household participation in equity markets. Swedish market capitalisation reached 169 percent of GDP in 2023, more than double the EU27 average of 69 percent (CEPS, 2025).

Investors in startups need to sell successful portfolio companies, and Sweden has deep, multi-tiered equity markets. Between 2016 and 2023, Sweden recorded 823 IPOs, the highest among EU member states, with nearly 90 percent on SME growth markets.

The Swedish government has occasionally invested in startups, but over time its participation has decreased. The participation of the Swedish government in total venture capital investments declined from 25 percent on average between 2011 and 2013 to just 6 percent in 2021. Saminvest, established as a fund-of-funds in 2016, has committed to venture capital funds and angel syndicates rather than investing directly.

Sweden benefits from a virtuous cycle where the founders and early employees of successful companies become angel investors, mentors, and serial entrepreneurs themselves. Guiso, Pistaferri, and Schivardi use Italian survey and census data to show that higher exposure to entrepreneurial activity in late adolescence predicts both entry and performance later on: a one‑standard‑deviation increase in firm density where someone lived at age 18 raises the probability of becoming an entrepreneur by about 1.5 percentage points and it is associated with about 11 percent higher entrepreneurial income.

What government can do (and has done)

Angel investor tax credits are widely used across Europe, but they rarely target experienced investors the way Sweden’s 2003 tax law did, and thus have little effect. Indeed evidence from tax credit adoption across 31 U.S. states suggests they mostly induced participation from investors who did not fit in the profile of engaged, expert angels. In fact, fewer than one percent of tax-credit recipients identify as professional investors. As a result, while the credits led to a one-fifth increase in the total volume of angel investing, they didn’t increase any meaningful economic outcome, including young-firm employment and job creation, startup entry, patenting, and the incidence of successful exits.

The lesson from Sweden is that Europe does not need more money for developed startups. If there is a case for government intervention, it is to provide tax relief to early, experienced investors in regions that lack them along the lines of the 2003 Swedish reform. Despite the cross-border nature of European venture capital, the first rounds do still typically come from locals.

If one must use the government to incentivize venture capital, there are a design principles that would make this less wasteful. Indirect government investment such as fund-of-funds that co-invest in new venture capital firms is more successful than government entities investing directly. Broad geographic and sectoral scope outperforms narrow targeting. And, requiring private capital to co-invest with public funds rather than the government being the majority investor is associated with better performance.

But there are no real shortcuts. Much of Sweden’s performance can be explained by experienced angels recycling proceeds from past exit. None of this can be conjured by a single policy or a new public fund. Like the Swedes in 2003, the parts of Europe that do not have start-up activity need to start putting these conditions in place today for a thriving entrepreneurial culture one or more decades down the road.

References

Axelson, Ulf, and Milan Martinovic, 2016, “European Venture Capital: Myths and facts.” Discussion Paper no 753, London School of Economics.

Bach, Laurent, Ramin Baghai, Per Strömberg, and Katarina Warg, 2023, “Who becomes a business angel?,” working paper, Stockholm School of Economics.

Bai, Jessica, Shai Bernstein, Abhishek Dev, and Josh Lerner, 2022, “The Dance Between Government and Private Investors: Public Entrepreneurial Finance around the Globe,” working paper, Harvard Business School.

Bradley, Wendy, Gilles Duruflé, Thomas Hellmann, and Karen Wilson, 2019, “Cross-Border Venture Capital Investments: What Is the Role of Public Policy?,” Journal of Risk and Financial Management 12(3), 1-22.

Dealroom, 2023, “European tech ascendancy: Unlocking a continentʼs innovation potential,” Research report, Dealbook.co and Creandum. Available at https://dealroom.co/reports/european-tech-ascendancy (accessed 2/28/2024).

Denes, Matthew, Sabrina Howell, Filippo Mezzanotti, Xinxin Wang, and Tin Xu, 2023, “Investor tax credits and entrepreneurship: Evidence from U.S. states,” Journal of Finance 78(5), 2621-2670.

EIB, 2024, “European Investment Bank Investment Report 2023/2024: Transforming for competitiveness,” European Investment Bank.

EIF, 2023, “EIF VC Survey 2023: Market sentiment, scale-up financing and human capital,” EIF Research and Market Analysis Working Paper 2023/93, European Investment Fund.

Guiso, Luigi, Luigi Pistaferri, and Fabiano Schivardi, 2021, “Learning entrepreneurship from other entrepreneurs?,” Journal of Labor Economics 39(1), 135-191.

Kaplan, Steven, and Josh Lerner, 2010, “It ain’t broke: The past, present, and future of Venture Capital,” Journal of Applied Corporate Finance 22(2), 36-47.

Kaplan, Steven, Berk Sensoy, and Per Strömberg, 2009, “Should investors bet on the jockey or the horse? Evidence from the evolution of firms from early business plans to public companies,” Journal of Finance 64(1), 75-115.

Lerner, Josh, 2009. Boulevard of Broken Dreams: Why Public Efforts to Boost Entrepreneurship and Venture Capital Have Failed – and What to Do About It. Princeton, New Jersey: Princeton University Press.

Lerner, Josh, Antoinette Schoar, Stanislav Sokolinski, and Karen Wilson, 2018, “The globalization of angel investments: Evidence across countries,” Journal of Financial Economics 127, 1-20.

Lerner, Josh, and Ramana Nanda, 2020, “Venture Capital’s role in financing innovation: What we know and how much we still need to learn,” Journal of Economic Perspectives 34(3), 237-261.

Maurin, Vincent, David Robinson, and Per Strömberg, 2023, “A Theory of Liquidity in Private Equity,” Management Science 69(10), 5740-5771.

OECD, 2016, “The role of business angel investments in SME finance.” In: Financing SMEs and entrepreneurs 2016: An OECD scoreboard, OECD Publishing, Paris.

Reshid, Abdulaziz, Ulrika Stavlöt, Peter Svensson, and Barbro Widerstedt, 2023, “Evaluation of the tax incentive for private investors in Sweden,” Working Paper 2023:01, Swedish Agency For Growth Policy Analysis (Tillväxtanalys).

Strömberg, Per, and Christian Thomann, 2024, “Real impact of private equity,” manuscript, Stockholm School of Economics.

This makes it disappointing that the coalition’s big energy plan is to build 40 gigawatts of wind capacity financed by contracts for difference.

The North Sea is as good a place for offshore wind as exists anywhere. Most of it is only 15–30 meters deep, and wind speeds average over 10 meters per second. The surrounding countries are some of the most densely populated in the world (England, with 450 people per square kilometer, is not quite as dense as the Netherlands, with 541), and have some of the cloudiest weather, which makes them less suitable for solar.

But the economics of wind are quite poor, and the nature of contracts for difference means that the high costs of wind power now will lock in high electricity prices for years to come.

Optimistic predictions had the levelized cost per megawatt-hour of offshore wind down to $40 by 2030, decreasing much like solar has. Unfortunately, this hasn’t happened. The National Renewable Energy Lab in the US does good annual analyses of the cost of offshore wind. In the most recent one, it costs over $110 per megawatt-hour, and the price has increased in recent years.

The cost of solar has decreased so much because a lot of the cost is manufacturing the cell. This isn’t the case for offshore wind. The turbine is just 24 percent of the lifetime cost, and a significant chunk of that goes to raw materials. 35 percent of the cost is support infrastructure, including the foundations, undersea cables, substations, and installation, and 27 percent goes to operations and maintenance after installation. Even if the turbine gets more efficient it won’t help much, as more efficient turbines are often larger and so more expensive to install.

Rising costs have caused trouble for wind manufacturers. Ørsted, the world’s largest developer of offshore wind, has lost 88 percent of its value over the last five years, including 76 percent before Trump took office. Vestas, which makes turbines, has lost 36 percent of its value over the last five years.

Despite the political excitement, few offshore developers are that enthusiastic about North Sea wind anymore. In October last year, the Netherlands ran a one gigawatt tender for offshore wind in the North Sea and received no applications at all.

Part of the reason for that failure, I imagine, was that 42 percent of the points in the application were granted for contributions to sustainability and the ecosystem: you could get more points for minimizing porpoise disturbance days (26) than you did for proving that you, the foundation manufacturer, the foundation installer, the wind turbine manufacturer, the wind turbine installer, the cabling manufacturer, and the cabling installer each had done this more than fifty times before (22). Also, only one kind of foundation (a monopile) was permitted by the Environmental Impact Assessment.

The other reason was that the tender was unsubsidized, and developers are unwilling to back North Sea wind turbines without significant public monies attached.

Contracts for difference

To subsidize wind developers, North Sea countries – first Britain and now the Netherlands – have turned to the contract for difference (CfD). The CfD first became common in the financial world in the 1990s, as a way for hedge funds to get exposure to price movements without owning the underlying asset.

In the CfD, two parties agree to a strike price for a good along with a settlement period. If the market price in the settlement period is below the agreed price, the purchaser of the CfD pays the difference between the market price and strike price to the seller. In a two-sided contract, the same happens, but the seller also pays the difference to the purchaser if the market price of the goods exceeds the strike price.

The CfD in the energy world is usually two-sided. The government promises to pay the producer of the electricity the strike price minus the market price over the next 12 to 20 years, depending on the country. This guarantees that the seller will always receive the same amount for the power, no matter what happens elsewhere in the market.

In the best case scenario, CfDs create certainty and lock in favorable prices for consumers. In 2022, Britain sold seven gigawatts for under 50 pounds per megawatt-hour – a good deal for the government.

But if the price of the underlying good is high at the time of auction – in this case power generation with wind turbines – the CfD just locks in high prices.

In 2023, developers refused to bid on Britain’s CfD auction, even after the maximum price was raised. To keep building wind capacity, the government raised prices in 2024, and even further in 2025.

In the most recent round, the British government offered £91 per megawatt-hour for 8.25 gigawatts, locked in until the late 2040s. That price is above the price of generating electricity with gas today, despite gas being extremely expensive compared to historic levels. The recent auction guarantees that Britain will have some of the most expensive energy in the world for the next two decades.

This will be the Netherlands’ problem if the coalition’s plans proceed. As a trial, the government just announced a one-sided CfD for one gigawatt for up to €104 per megawatt-hour. The price of generating electricity in the Netherlands is only roughly a third of the price faced by consumers, so at the maximum price this trial will buy electricity a few percent more expensive than what we have today, for the next 15 years.

There are other issues with wind beyond its costs. The number of prime sites is limited; as those get taken, the costs of installation increase and the returns decrease. And there is the problem of intermittency: when the wind doesn’t blow, power doesn’t get generated.

But these problems will probably get solved. The learning curve for wind is not as steep as for solar but still negative; installation technology and more efficient turbines will make worse sites viable; batteries, nuclear, and transmission will help alleviate the problems with intermittency.

But CfDs mean that the prices for the next decades are the ones we face today, not the ones we may face in the future. The Netherlands has an installed electricity generating capacity of 63 gigawatts. The proposed 40 gigawatts will add two thirds to that.

The Netherlands’ plans with wind are extreme, but representative of a larger group of North Sea countries. The first wind turbines that generated more than a megawatt were built in Denmark in 1978. Along with the Netherlands, Britain, and Denmark, Germany and Ireland all want to hit their climate goals primarily with wind power.

I feel bad spending as much time as I do on this blog criticizing Europe’s climate efforts. But the scale of the resources being allocated, the degree to which Europe is alone in this effort, and the role of expensive energy in Europe’s productivity difficulties make getting this right hugely important.

The exciting thing about the new Dutch government is that it understands zero-sum transfers are taking place between generations, and it has decided to rebalance them towards the young. Making the choice to lock in high electricity prices for decades to appease climate-conscious voters today would do the opposite of that.

1. From a game-theory point of view, Trump always tries to defect from the ‘cooperative equilibrium’. He does not just play ‘defect’ under the old rules (the WTO, the UN, etc.), but also under the new rules he proposed. The Turnberry agreement concluded last July between the US and EU was presented by the Administration as a model for a new trading system and did not even last six months.

2. Even if ‘cooperating’ is less painful in the short run, against an opponent inclined to defect, the optimal response is retaliation. It is better to do this sooner rather than later, as Trump will keep exploiting Europe’s tendency to cooperate for increasingly megalomaniac purposes.

3. Europe is the main source of finance for America’s huge foreign debt. Europe owns twice as many US bonds and equities as the rest of the world combined.

4. Robin Wigglesworth and Toby Nangle from FT Alphaville have argued that dumping treasuries would cause as much pain to Europe as to the US, that it would be hard to find buyers for ten trillion dollars in assets, and that ‘most of these assets are not actually owned by European governments’.

5. The objections are well taken, but not definitive. Europe does not have to dump all the treasuries at once: the market is wobbly, as we saw after yields spiked in April. And, while many of the assets are not held by the public, a large share are also held by semi-public investors (such as Dutch and Swedish pension funds, and the Norwegian sovereign wealth fund).

6. Certain American tech companies have driven much of its stock market overperformance, are closely associated with the Trump administration, and make large shares of their revenues in Europe. Some of them — like Meta, X, and Uber — exist primarily due to network economies, where everyone uses them because everyone else is on them. If they did not exist, substitutes would rapidly develop. There is no reason why Europeans should pay a toll for communicating with each other to people like Zuckerberg, who have used their proximity to Trump to punish the European Commission.

7. There are four ways to target these companies. The DSA allows for fines and the suspension of Very Large Online Platforms. The DMA applies to gatekeepers such as Meta, and allows for even larger fines — up to 20 percent of global turnover — and also the forced divestiture of repeat offenders. Third, GDPR can be more strictly enforced. Fourth, Europe’s anti-coercion instrument allows for market access to be withdrawn.

8. Europe’s bazooka – the anti-coercion instrument – is unlikely to be triggered. It requires 15 of the 27 member states to vote in favor and many smaller and more exposed member states are probably unwilling to risk angering Trump further.

9. The difficulty of triggering the anti-coercion instrument reveals an important weakness of the ‘one country, one vote’ model of the European Council. In extremis, the thirteen smallest countries, which combined have 10 percent of the EU’s population, can block a major foreign policy avenue for the remaining 90 percent of Europeans.

10. In the long run, the best response to Trump, apart from pursuing general pro-growth reforms, is to make a concerted effort to draw top talent out of the US. Chinese universities are offering top researchers millions of dollars. So far, no European institution has been willing to make exceptions to its civil servant pay scales for top American-based talent.

11. China is not Europe’s friend and its values are still much further from Europe’s than America’s values are. However, building leverage means Europe must follow Canada in a pivot towards China. It is hard to negotiate without an outside option.

12. The second mover AI strategy – which we have repeatedly argued for on this blog – seems bankrupt in a world where the world’s sole provider of frontier models is willing to extort other countries and cannot be counted on to abide by the agreements it makes. A true ‘sovereign AI’ push seems worthwhile.

13. Given the quality of Europe’s leaders and the political constraints they face, by far the likeliest outcome is failure. But the payoff is such that it still seems worth spending a trillion euros on it (six percent of GDP) — not more than Europe has been spending fighting climate change every two years.

14. The reason for the surrender at Turnberry was Ukraine. But Ukraine is currently on a grim trajectory. American contributions have also declined, with aerial defenses providing less and less protection against Russian missiles. A just and sustainable peace seems more likely with muscular European guarantees than with weak American ones. The former is less inconceivable now than it was two weeks ago.

15. The danger of Trump is that it leads to negative polarization, where Europeans defend policies that do really need to reform. Europe emits only six percent of global emissions, and is still planning to drive its industry off a cliff in order to reduce those. They should consider instead the examples of the leaders who have stood up successfully to Trump: Mark Carney in his speech to Davos highlighted his ‘investments in energy’ and Canada’s place as ‘an energy superpower’ and Claudia Sheinbaum is trying to increase Mexico’s oil output by 31 percent by 2030.

16. Despite everything, Trump’s recent actions could work out well for Europe. Its biggest problems are still in the long-term (the original post planned for this slot was on pensioners and fiscal dominance). Few things focus the mind like the threat of external invasion. If Trump succeeds in uniting Europe around a hard-headed and pragmatic program of reform, Europe may well be in a much better position twenty years from now.

    Thanks for reading Silicon Continent! Subscribe for free to receive new posts and support our work.

But the proposed amendment isn’t much better. The current ban requires the average emissions of a carmaker’s new vehicles in 2035 to be 100 percent lower than the 2021 level. The proposed rule lowers this to 90 percent. (So far, carmakers have reduced emissions by just 7 percent.) The additional emissions allowed by the new rule must be fully compensated with green steel and renewable fuel credits.

In a sense, this change is in line with what we have advocated on the blog. Some uses of a given technology are very valuable and therefore costly to abate. In order to discourage other uses but still allow those, policymakers should use taxes rather than a ban.

You can debate whether a level of tax is too high or too low – in this case, a tax for 100 percent of the remaining emissions after a 90 percent reduction still seems quite steep – but a tax is in any case much better than a ban.

But in another sense, this isn’t actually that kind of tax. Instead of buying compensation credits on the marketplace, automakers are forced to obtain them by actually using ‘Made in Europe’ green steel and by using renewable fuels.

Even if you support this level of tax on the remaining combustion cars, this is a bad version of it. It spends money on reducing emissions only through the channel of steel manufacturing. It makes carmakers buy a single product from a single set of producers to meet their environmental obligations. And it forces consumers to subsidize an inefficient part of European industry.

Because the credits are paid for by the automakers, these costs will presumably be shared across car purchases, regardless of whether they are combustion or electric.

The changes to the combustion engine ban are part of a much larger package. Besides some smaller ideas, like a new category of small cars (which would be subsidized) and changed requirements for mandatory speed limiters, the other big proposal actually tightens emissions rules.

Sixty percent of all cars and 90 percent of all vans in Europe are leased or otherwise owned by companies. The deregulatory package creates new, binding emissions targets for these vehicles.

These targets vary by country. By 2030, 90 percent of all leased vehicles in the Netherlands must be either low or zero emission, while 48 percent must be in Greece. By 2035, 95 percent of all leased vehicles in the Netherlands or Germany must be zero emission (Portugal or Poland get away with 56 percent).

I’m not sure what method was used to figure out the ‘fair’ amount of combustion cars in each country. But even if one would like to electrify the leasing industry, a tax would be much preferable to the quantity targets proposed.

The proposal also introduces a rule that would prohibit public subsidies for leased cars not made in Europe.

This protectionism is almost certain to get much worse. The French government wants the Commission to oblige automakers to source at least 75 percent of electric cars parts in Europe. In parallel, the Commission is also considering ‘Made in the EU’ requirements for the components of electric vehicles as well as other technologies like solar panels.

This proposal would create chaos in their supply chains. It would also make Europe’s cars more expensive and less competitive abroad. The sourcing requirements in particular replicate the worst parts of Trump’s tariffs, by raising the costs of the inputs into industrial processes they want to protect.

The auto omnibus is a nice example of what we have warned in a previous post: the Commission that created all the green rules would find it hard to stop creating them.

It is presumably unpleasant for leaders to unwind a project that they spent years working to create. The bureaucracy has a momentum of its own, having spent the best part of a decade thinking of all policy through this lens. Trying to impose new priorities requires a strong signal from above, a requirement that hasn’t been met by the half-hearted walk-backs of the Green New Deal we’ve seen so far.

Commission officials and supporters of the transition like to talk about learning from China. It’d be good to learn from what China is actually doing.

The country is building lots of renewables and nuclear power. But it is also building enormous amounts of coal, enough that its emissions per capita are now higher than Europe’s.

As Matt Yglesias pointed out, the Chinese government is not behaving as if its goal is emissions reductions. It is behaving as if it wants the ability ability to survive a blockade of oil and gas imports.

Europe continues to do the opposite of that. There are times when the energy transition and growth coincide, such as by allowing solar and building transmissions lines. But in cases where they don’t, like the car industry, growth still loses out.

One way of thinking about this is that all knowledge work varies along one important spectrum: messiness. On one end, there is one defined task to execute, say helping clients fill their taxes. You get the expenses and payslips on email, you use some rules to put them on a form, you obtain a response. Over time, you become better at this task, and get a higher salary. On the other end of the spectrum, there is a wide bundle of complex tasks. Running a factory, or a family, involves many different tasks that are very hard to specify in advance.

Subscribe now

The risk of the single-task job is that artificial intelligence excels at single tasks. Humans are still often in the loop, since the rate of errors in many fields is still too high to allow for unsupervised artificial intelligence. But the rate of errors is rapidly decreasing.

You may bet that fields vary in their tolerance for errors. Certain simple tasks, like content moderation, involve a high tolerance for risk: tech companies are comfortable with ‘shooting first, asking questions later’. But many others, from diagnostics to corporate communications, involve extreme risk-aversion on the part of the customer.

As long as a human is still needed to check the outputs, some value accrues to them. If AI drafts a contract that a lawyer reviews, signs, and takes responsibility for, the lawyer remains the supplier of legal services.

But the models are continuously getting much, much better at single tasks. If there aren’t any legal requirements to keep humans in the loop, even risk averse fields will eventually switch to unsupervised AI. When AI can produce finished code for straightforward tasks without human intervention, junior developers who once supplied that work compete with systems that produce it nearly free. The supply of programming services is no longer limited by human time, and the price of the service collapses to 0.

The result is that workers with simple tasks will become continuously more productive (and richer), until their work is worth nothing. A junior customer support agent gets more and more effective while the AI provides her the accumulated knowledge of senior customer support agents, as in the recent Brynjolfsson, Li; Ramond (2025) paper, until the AI is good enough that she can be replaced.

Notice that the autonomy threshold is not just given by technology. Firms and governments have a huge say in this, and they may choose to block adoption. The pressure to adopt better technology is strong, but do not overestimate it. Remember that many European countries still de facto ban Uber (don’t get my friend Nicolas Petit started about getting a cab in Florence!), despite the enormous quality of life improvement over traditional taxi monopoly, and that notaries are not a requirement of current technology, but a legal constraint that their strong lobby will always protect in continental legal systems. Even the most single-tasked civil servants will still have a job for a long time.

The end of work? Not so fast

The other option is to go for a messy job, where the output is the product of many different tasks, many of which affect each other.

The head of engineering at a manufacturing plant I know well must decide who to hire, which machines to buy, how to lay them down in the plant, negotiate with the workers and the higher ups the solutions proposed, and mobilise the resources to implement them. That task is extraordinarily hard to automate. Artificial intelligence commoditizes codified knowledge: textbooks, proofs, syntax. But it does not interface in a meaningful way with local knowledge, where a much larger share of the value of messy jobs is created. Even if artificial intelligence excelled at most of the single tasks that make up her job, it could not walk the factory floor to cajole a manager to redesign a production process.

A management consultant whose job consists entirely of producing slide decks is exposed. A consultant who spends half of her time reading the room, building client relationships, and navigating organizational politics has a bundle AI cannot replicate.

In 2016, star AI researcher Geoffrey Hinton leaped from automation of reading scans to the automation of the full radiologist job, and gave the advice to stop training radiologists.1 But even fields that can look simple from the outside, like radiology, can be quite messy. A small study from 2013 (cited in this Works in Progress article) found that radiologists only spend 36 percent of their time looking at scans. The rest is spent talking to patients, training others, and talking with the nurses and doctors treating the patient.

A radiologist’s job is a bundle. You can automate reading scans and still need a radiologist. The question is not whether AI can do one part of your job. It is whether the remaining parts cohere in a manner that justifies a role.

To me, a key characteristic of these “messy jobs” is execution. Execution is hard because it faces the friction of the real world. Consider a general contractor on a building site. Artificial intelligence can sketch a blueprint and calculate load-bearing requirements in seconds. That is codified knowledge. But the contractor must handle the delivery of lumber that arrived late, the ground that is too muddy to pour concrete, or the bickering between the electrician and the plumber.

Or consider the manager in charge of post-merger integration at a corporation. Again, the algorithm will map financial synergies and redraw org charts, but it will not have the “tribal” knowledge required to merge two distinct cultures and have the tact to prevent an exodus.

Corporate law is increasingly vulnerable to automation because contracts are essentially code, but I would expect trial attorneys to subsist.

AI implementation itself could be the ultimate messy job. Improvements will require drastically changing existing workflows, a process that will be resisted by internal politics, fear, and legacy business models. For instance, law firms have always relied on “billable hours” to charge clients, a concept that will be useless in an AI world. But this organizational inertia is a gift: the transformation will be messier and more delayed than the charts suggest and it will require a lot of consultants, managers and workers, well versed in what AI can do, but with sufficient domain knowledge to know how to use it and how to redefine the process.

In the extreme instances, the feared AI transformation may not take place. Jobs defined by empathy, care, and real-time judgment will become the economy’s ‘luxury goods.’ In these fields, artificial intelligence is not your competitor; it generates the wealth (and lowers the costs of goods and services) that will fund your higher wages.2

More leverage

Usually, better companies are not created by transforming old ones, but by starting new ones. Starting a firm is the messiest job in the world, and artificial intelligence gives you a lot more opportunity to compete head on with larger companies..

For a century, large organizations dominated because only they could afford the fixed costs of specialized functions. If you needed both engineering and marketing, you needed scale to justify the overhead. AI reduces both the fixed and variable costs of that specialization. A single professional can now relinquish supporting tasks to AI and operate as a generalist. In a world where single tasks in areas like HR and finance are automated, you can ride a good idea all the way to a large company.

In our new book, we write about Base44, an AI-powered app builder. Founded by Maor Shlomo, a 31-year-old Israeli programmer, the company was built as a side project. He invested about $15,000 of his own money. He hired no employees, and he used Claude to write 90 percent of his frontend code and push updates daily, without any of the overhead that bogs down traditional engineering.

Within six months, Base44 had attracted over 250,000 users, generated $189,000 in monthly profit, and signed partnerships with major companies. Wix acquired it for $80 million.

Shlomo had no sales team, no marketing department, no HR. The tasks that once demanded human specialists were absorbed by the technology.

Conclusion

Given all the above, several investments appear to matter if you want to engage in careers in knowledge-intensive work (as opposed to the many crafts and trades, such as hair-dressing, plumbing, playing piano or being a chef, that are likely to remain untouched for a long time):

First, build deep, substantive knowledge in your field. As models get better, fewer humans will be good enough to add any value at all. Artificial intelligence can predict that a scan shows a tumor with 94 percent probability. But should the patient undergo surgery, radiation, or wait and monitor it? That decision depends on tradeoffs, some of which the algorithm cannot weigh. To exercise this type of judgment in a helpful way, you must know a domain deeply.

Deep domain knowledge will also make you good at AI implementation. Being able to change a given company’s processes to take advantage of the new technology is likely to employ a large number of people for a very long time.

Second, openness to new experiences, and the ability to learn new things quickly, will become even more important than it already is. Choose the job where you will learn most, but also the one where you will learn to learn. You will have to reinvent yourself throughout your life. New jobs will appear that we have not even imagined. If knowledge is the largest constraint that we face, then cheap knowledge will change everything, from medicine or the law to every field of research. The specific knowledge you learn today will depreciate faster than ever. What matters is the slope of your learning curve and your ability to adapt. For instance, in a job such as an early-stage founder or first employee, you do not have a fixed role; you have a set of problems. You must learn enough law or sales by the afternoon to survive, distinguishing essential signals from noise. Or, as a management consultant, you enter a room where everyone knows more than you, so you must absorb a new industry’s logic in days to structure vague problems. In tech, the role of product manager sits between the engineers and the sellers as a translator.

Third, seek leverage. In the past, your output was limited by your time—a chef can only cook for so many people. AI breaks this constraint. It allows a single writer, coder, or entrepreneur to serve a global market without a massive support staff. The constraint is no longer your production capacity; it is your ability to direct the machine. There is never been a better time to undertake entrepreneurial projects.

Fourth, if you do want to do a task that can plausibly be done by a model, location is more important than ever. A small number of cities, starting with San Francisco, Paris, London, and New York, are where almost everyone working and thinking about artificial intelligence is based. Go to these cities, or the closest approximation of them available to you, not just to work on these problems but to understand what possibilities may come your way.

Fifth, install Twitter. Twitter is a huge time sink, but it is also where all progress is happening, out in the open. You will have a huge advantage over virtually everyone not on the platform in understanding what is happening.

Sixth, learn to supervise the machine. The essential new skill is meta-cognition: recognizing when the AI is hallucinating, directing it toward the right problems, and verifying its outputs.

Finally, if all of these changes in the nature of work do happen, we are going to have much more leisure. In a recent paper, Betsy Stevenson points to the Japanese concept of ikigai, “that which makes life worth living”. The ability to derive meaning from sources other than your work is itself a form of human capital. Try to cultivate the habit of reading fiction. Try to spend less time watching videos and other forms of TV. Pick up hobbies: the hobby I would recommend most is starting a blog!

Happy new year and thanks for reading Silicon Continent!

We also discussed Europe’s predicament. The EU AI Act and GDPR have created a regulatory environment that makes AI implementation costly and slow, precisely when Europe needs growth most. The continent faces an“R without G” risk, paying higher global interest rates driven by the AI investment boom while failing to capture the productivity gains that would make those rates sustainable. Our tricky demographics make the problem worse. We discussed where value will accrue in the AI stack, why a “smart second mover” strategy might be Europe’s best option, and whether the new geopolitical pressures from the US might finally wake European policymakers from their regulatory slumber. I’m not optimistic on that, but the conversation was fun.

We discussed a lot of recent research, which is linked in the podcast notes below the transcript, at least nine papers published in 2025.

Watch the episode here on Youtube, Apple, or Spotify, or else you can read the transcript below. I hope you enjoy it.

Also listen on:

Apple podcasts

Spotify

Transcript:

Will AI trigger explosive growth? [00:00:00]

Anson

Hi, I’m Anson, I’m a researcher at Epoch AI. Today I’m joined by my co-host Andrei, who is an assistant professor at the University of Edinburgh. And I’m also joined by Luis Garicano, who is a professor at LSE studying economics. Luis, thanks for coming on the podcast.

Luis

That’s my pleasure. It’s great to be here.

Anson

I’d like to start with explosive growth very briefly. One thing that we briefly discussed on Twitter was whether or not we’re likely to see a massive acceleration in Gross World Product growth rates. And one point that I think is somewhat underrated by economists is that if we look at the last 200 years, growth has been exponential. But if we look much longer throughout history, it seems like there has been an acceleration. So shouldn’t we think that accelerations in growth aren’t that implausible after all?

Luis

The probability that we get a very large acceleration of growth exists —I’m not going to dismiss that. You were arguing that was potentially the case with your task-based model. My view was that there were several things that are likely to make that take a long time or slow it down. So the first obstacle I was pointing out — and in R&D for example it’s very clear — you can develop as many new ideas for proteins as you want so for biotech and for solutions to biological problems. But if you don’t manage to get them approved by the FDA, you don’t have a medicine. And if you don’t get doctors to use it and you don’t get people to learn it — so there are a lot of bottlenecks that slow things.

So that was my first objection, that people in Silicon Valley who are only observing the very best application of technology, which is coding. They’re extrapolating from which tasks we have, how many tasks we’re performing, and run the risk of overestimating how easy it is for organizations and institutions to accommodate these tasks.

Andrei

Just a question. I think we’re kind of on the same page that sustained explosive growth is perhaps not that plausible. What about kind of an explosive growth spurt, a shorter run thing where you have five, ten years of much, much faster growth than we’ve recently expected? Just because, you know, we start from an initial condition where AI seems to be good at exactly lots of things that humans are bad at. So you start with this high productivity sector being initially relatively large. Could we have that?

Luis

I think so. I’m an optimist in AI in spite of our disagreement on that. I do believe that, unlike people like Daron Acemoglu or others who in not just ten years, but even in the longer run, their models don’t predict large growth spurts. I think the good way to see it is: does it get autonomous? I think the key distinction is between autonomous and non-autonomous AI. As long as the AI needs your supervision because it makes lots of mistakes, then the bottleneck is the human. And the human is not improving much. I mean yeah, the AI is helping the human do it a little bit faster, but the human is bottlenecked by their own time. I’m a better lawyer, I’m doing better at my tasks, but okay, that’s just a qualitative difference.

The moment you get the AI lawyer, the moment the AI becomes autonomous, I think there you get a jump, a discrete jump. So we could easily have a situation where we see very small steps where the AI is helping us. We’re doing a little bit better. Think of Brynjolfsson’s customer support chatbots. There the chatbot is helping the juniors be better customer support agents. They suggest answers, the junior uses them, but it’s still a junior doing it.

We know because the paper is published in 2025 (but the experiment is from a little bit before). We know now that the chatbot is precisely one of the areas where it’s likely — and in fact, we are already seeing in some of the data — that the humans can be earlier removed from the production function. Because at the end of the day, there is a set of questions that are relatively repeated and common, and then you can do a lot of the customer service fast, reliably, etc. And you could always have a layer like in my knowledge hierarchies type work where basically you have the routine tasks done by some agents and the exceptions done by experts. That’s kind of how stuff is produced. The high value tasks are done by the high level consultant, and the entry level analyst does the routine jobs. You could still have that layer of people who get big leverage. If all of these tasks that are more junior get replaced and you get that expert that you’re expecting.

So I would think that it could easily be that we are all thinking that nothing’s happening… and then boom! Something happens in one particular profession, something major, like the type of spurt that you’re mentioning.

Short-run macroeconomic effects [00:06:26]

Andrei

We’re all working in this kind of long-run macro way of thinking about the effects of AI, but what about the short-run macro of it? So what would we expect to happen to things like unemployment, inflation?

Luis

I think the short-run macro is the problematic one. Let’s suppose that we have two sectors, let’s say sectors A and B, and sector A basically gets produced for free. So the price of sector A is zero. So the short run effects are you need to reallocate the labor and the capital to sector B. Now the first thing that is clear, I think we will all agree, is that welfare is going to improve. If, for example, let’s say sector A is medical services and legal services. This is autonomous AI. We get medical and legal services have zero price.

So first, we get a huge increase in consumer surplus. Fantastic, right? All my illnesses I can diagnose myself, I can get all my legal problems. You know, I need to buy a house, the AI does it, you sign it, it all goes in the crypto chain. It’s all automatic and perfect. So consumer surplus goes up.

But what happens to GDP and what happens to employment? Let’s talk about the short-run. Let’s say that you need a neurosurgeon to become somebody in sector B who is maybe a plumber, just to make the extreme example clear to our listeners. Then you have somebody who has very specific human capital, who has been completely depreciated, who was used to earning several hundred thousand dollars, now has to start working in a new sector that doesn’t have any — I don’t think any of this human capital is going to be very valuable. The machines, all the things that were complementary with the lawyer or the doctor are useless. We need to do depreciate it, and we need to redeploy them.

So we have an increase in supply in sector two. We have an increase in demand. In the short run, only the increase in demand. So the supply is reassigning itself. It’s really hard to get these machines to be useful. So in the short run, I would imagine that prices in sector two are going to go up. But in the long run, I don’t know.

I wouldn’t talk about this as inflation. This is a change in relative prices in sector two. I mean, we could have deflation if all of these people are unemployed, etc. But when it’s a price shock, I am reluctant to talk about inflation. It’s really just the price shock, which is that all of those skills and all of that capital is worth nothing. And people in this new sector have to accommodate this extra demand and this extra labor and capital. That would be how I would see this situation.

Obviously, the problem in the short run of my scenario is that the very short run completely contradicts it, which is the lawyers will get the bar association to say it’s illegal to sell your house without a lawyer signing, and the doctors will get the medical association —

Andrei

But I guess one of the intuitions I had and I struggled to reconcile in my head is like, you know, you have this situation where in sector A productivity has gone nuts and the price is almost zero, but wouldn’t we actually be worried that in the short run we’d have a recession, right? I mean, all these people would be worried about their jobs and would stop spending. So there’s this demand side thing happening in the short run. How do we reconcile those?

Luis

That’s why I said deflation, if you want to call that price shock as — because in this first sector there is a lot of consumer surplus. But in terms of actual P — so we have the price of sector one times X in sector one plus the price in sector two plus the quantity in sector two. The price in sector one is zero by assumption. So that part of GDP has fallen off a cliff and that capital and labor is unemployed. So yes, I think the short run effect until you get this reallocation is, I think, a big increase in welfare. Probably still, a lot of people are very happy. You are in Ghana and you don’t have access to good medical services in some rural village, and you suddenly can just get a doctor, an AI doctor. That’s great. But that increasing welfare doesn’t necessarily translate into a GDP increase. Indeed. And definitely those people who have to be reassigned could be in long term unemployment, a lot of them, because many of them might, depending on what their old skills were, might find it very hard to readjust to the new world.

The decline of junior jobs [00:11:29]

Andrei

So one thing that I also wonder about is the distributional consequences of these potential shocks. And here I sense a little bit of tension both when I read the news about the entry level job market and what’s happening to the entry level job market, and also when I read papers worrying about deskilling. So on some level, we expect AI to be bad for entry level workers and less skilled workers, at least within skilled professions. On the other level, we’re worrying about deskilling. So will AI be good for less skilled workers than skilled professors or bad for them? How do we think about that question?

Luis

It’s a great question and one that is really being played out right now. I joked in the NBER conference on AI in Stanford a few weeks back about Notion versus Brynjolfsson. There is a Stanford AI—

Andrei

That sounds particularly problematic.

Luis

I think we can reconcile it. There is a Stanford economist who had two really important papers. One is the one I was referring to before, which is in the Quarterly Journal of Economics earlier in the year, giving software chatbot assistants to the customer service support agents. And indeed, he finds big increases in the productivity of the most junior ones, because basically, you get into the job and you get already a tool that allows you to solve most of the problems. They actually also get trained faster, they seem to learn faster. So when you eliminate these, you turn it off. They seem to have picked up stuff. So in all dimensions they provide more quality, the clients are happier, etc. You get the more junior of them are helped. And there is also a field experiment. So this is one field experiment.

There’s another interesting field experiment with software developers (by Cui et al.) that goes also into that direction. Finds some gigantic increase in productivity, maybe 20 something percent, from August this year. So it says, look, we gave in three companies these tools. And we saw the software developers increase productivity a lot, particularly junior ones. So that’s your side. That’s like okay, it’s not deskilling.

Then when we look at the aggregate data, two very recent papers, one by Erik Brynjolfsson and co-authors find something very different already. So this is not in the big macro data that the Fed finds and the Fed economists haven’t found it. These are not big shocks that we would have expected in 2022. Let me tell you the two findings. So one is from early September. This paper by Lichtinger and a co-author (Seyed Mahdi Hosseini Maasoum) is called Seniority-Biased Technological Change. And basically what it finds using something like 62 million workers — so it’s very significant — in the AI exposed occupations, you don’t see anything happening to senior employment (you see it growing). But you see junior employment really dropping. And the way it’s dropping is through hiring. It seems like a lot of people are not hiring junior employees.

The logic behind it seems to me clear if you talk to a McKinsey partner, which I have done on exactly this question in person, recruiting for them. He was telling me things like the deep research does the job that the junior researcher could do. The PowerPoint slides, you can do them automatically quite well. A lot of the junior tasks can be done by the software. And so you get this replacement of juniors that you don’t hire anymore. And we’ll talk later probably about some work I’ve done on this training, the missing training ladder.

So these junior jobs are gone. And so you’re hiring less. You’re not hiring people. So that’s why I say this is subtle. This is the seniority based technological change. The Erik Brynjolfsson paper from August this year is the Canaries in the Coal Mine. Basically, it finds something similar. It finds for workers between 22 and 25 years old—so again, let’s look narrowly. Let’s be careful and let’s look at AI exposed versus not AI exposed provisions. We again see pretty clear drops and pretty robust with aggregate data.

Now how do we reconcile this? I would reconcile it with the following two ideas.

One is this idea that I was arguing before that you get like, “oh, I’m a better customer support agent” for a while… then “oops, I don’t have a job!”. Because the AI has been helping me become better until the moment the AI is sufficiently better that I am not needed anymore. That is one idea that autonomy kind of — we start with non-autonomous AI that enhances and complements our skills. So Ide and Talamas have a recent Journal of Political Economy paper, where they contrast autonomous and non-autonomous AI at different levels of the skill distribution. And basically part of the argument is the autonomous AI is going to basically pin down the wage distribution. It replaces people at that point and produces an enormous supply shock on that point. Everybody below that is going to have to compete with AI. It’s going to have to earn less than the AI charges or the AI is worth. And so the moment it becomes autonomous, things change. And that’s, I think, one way to reconcile it, autonomous versus non-autonomous.

And the other way to reconcile it is, of course, the level of the AI, which is very related to autonomy. As the AI advances, I think we’re going to see the complementarity in some of these lower end jobs become substitutability. Now, this does not necessarily yet affect the higher end jobs.

I think if you’re on the higher end, your leverage increases, the knowledge hierarchy becomes more productive. You have this superstar effect where if you are an AI — we see these salaries for the AI engineers that have been offered $100 million and things like that, like football players. When Messi is watched in the World Cup final or in the Champions League final, he’s watched by 500 million, a billion people. So being a little bit better of a player gives you a huge market size because many people are going to want to pay a little bit more. Multiply that by 500 million people, orwhatever it is. Now that gives you superstar effects. Sherwin Rosen, who was a very important labor economist, makes this point. When there is limited substitution between quality and quantity (I cannot substitute 20 players by Messi, I cannot substitute 100 players by Messi. There is 11 and there’s only one field that is like that. Any number of players is not going to replace Messi.), and when you have markets that have joint consumption, that one person can reach a lot of people (we can all consume the same football game), hen you get the superstar effects, and these superstar effects are affecting the top of the wage distribution. A very good AI programmer with lots of AI, developer with lots of actual AIs, LLMs that are being deployed by him or by her, can have enormous leverage and can reach very large market size. So the extra skill they can add is really very valuable.

So I think on the top distribution we could see this bifurcation between on the bottom getting this substitutability, on the top getting this complementarity. And I think of course as the supervisory threshold, the threshold of things that the AI can do on its own goes up, those that are actually getting the superstar gains will become smaller.

The missing training ladder [00:20:21]

Anson

So one thing I’m curious about is that if I’m an entry level worker and I want to try to figure out how I can get into this job and learn the skills I need to be valuable in this job, there’s sort of like a strange situation, right? It’s like if I get to the point where I can be valuable, you know, to become an expert, then that’s great. But there’s like a period in between where I would normally do these routine tasks, but then right now I’m not able to do them as often because the AIs are doing them for me. So how do I know when it’s worth it for a company to hire me if I’m an entry level worker?

Luis

Yes, that’s a question I’ve been thinking about with Luis Rayo, my coauthor from Kellogg. I like to think of this as an AI Becker problem. So let me tell you, Gary Becker was a famous economist who developed the theory of human capital. And he made this distinction between general and specific training by companies. And he said, look, a company can always give you specific training because they’re going to appropriate it, but are they going to give you general training? Well, general training can only be given if the company can recover it afterwards. But once you’re trained, you can just walk and get all the benefits from the training.

So he would say, how is this going to work? Well, either there’s a market failure because we don’t get enough training in the economy or basically somehow the workers pay for the training. And with Luis Rayo , we basically wrote an analysis that appeared in the American Economic Review. We basically say, look, the way that these contracts are going to work is that there’s a master and apprentice, and the master is going to basically slow down the training so as to extract all the value of the surplus from the apprentice, while the master is giving little nuggets of training. So I’m giving you just enough that you want to stay because you want to be an expert, but not so much that I train you very fast and you walk out. So that’s kind of the solution that we proposed.

Now in that solution, the AI, as you are hinting, is going to create a problem, which is that it basically devalues the currency with which the apprentice is paying. The apprentice is basically paying not in dollars — it’s paying in menial tasks.

Suppose you’re a lawyer and you’re working for Cravath, and it really is not worth your time to spend all your time reviewing all these contracts. I mean, it’s boring as hell, but okay, you’re learning something and you’re receiving, but it’s basically menial work. Or suppose you’re in McKinsey and you’re the smartest person in your class or an investment bank, and you’re the smartest person in your generation. And there you are, doing silly spreadsheets that many other people could do. But that menial task is the way you pay for getting this training.

Now, if the AI can do the basic research at McKinsey, can do the contract review at Cravath or whatever law firm this is, and can do the basic accounting at an accounting firm or basic programming, then how do you pay for your training? So our argument is that the AI devalues the currency with which you pay. And as a result, makes the firm reluctant or the expert reluctant to get the worker in the first place. [In the past they could say,] “okay, I get this worker is going to be a pain and so on, but you know, I’m going to get paid for training them through their work.” But now it’s so cheap to do with an AI that the value of the worker is devalued.

So basically we show in the paper we build a very simple model in which these exchanges are happening. And we show that there are two basic things that are happening, and the ratio between those two is what is crucial. One is the AI, the substitution aspect of the AI that is basically devaluing the currency with which the worker is paying. So basically the AI as it gets better, the worker basically has less to add to this production function of the partner or the more expert person. But at the same time, the fully trained worker is worth more. So that means that the traineeship is still worth it.

So the basic result that we have is that there is a ratio, a key ratio, which is how much the AI complements the expert. An expert, fully trained expert with AI. How much has that gone up relative to how much the AI replaces the untrained person? If the expert with AI value is going up a lot, then even though the untrained person is not worth a lot, you can extract so much from that value that the contract still exists. So basically that ratio determines whether you are going to want to employ that worker and to train.

In the absence of that, then the training ladder disappears, and we have a big societal market failure. Imagine all of this tacit knowledge, a lot of this training that happens on the job is not in any manual, right? If it was in a manual, it would be taught in law school. It’s about how you deal with the client. It’s about how you are really precise with the contract. It’s hundreds of things that are hard to describe. Tacit knowledge is the idea that there is a lot that we know but we can’t describe. And if the worker is not acquiring this tacit knowledge, because all this training is not taking place from the master, this transfer of knowledge from the master directly — he’s the one, she’s the one who has this knowledge. Then the economy has a problem in the longer run.

To the extent that the AI is not perfect, we don’t have those experts that can supervise the AI in ten years or in fifteen years. Then we have a hole in our growth model. Growth depends on human capital, and suddenly this pipeline of intermediate people acquiring skills has disappeared. And that’s actually a potentially big consequence of AI, a problem that AI could eliminate those lower rungs on the ladder.

And as I was arguing before with the Canaries in the Coal Mine and the Seniority-Biased Technological Change papers, I think there is a lot of anecdotal evidence from these companies that these very junior employees are not really being hired. But there is in these two papers—there’s already from August and from September — starts to be systematic evidence that this could be happening.

Anson

What do we know about the value of this ratio? Like, do we have any empirical evidence?

Luis

No, it’s a theory paper and we are suggesting that people should look into this empirically. We are inviting people to analyze it empirically. I think we’re seeing both. I think we are seeing senior people really complimented and more productive. Look at the $100 million checks that we were referring to. Senior AI experts, AI engineers are getting big paychecks, which would be unimaginable without AI. So they’re being complemented. I think that in our jobs, we already can see that the productivity is increasing with AI. We are also seeing substitution. So the question is how big is that ratio in different professions. And the larger the ratio, the more the training ladders will remain.

Anson

One thing I’m a little worried about with trying to estimate this is that, you know, if we had tried to do this exercise of estimating the ratio three years ago, the models were so different and so much worse that the ratio might have been pretty different. And I’m worried that if we try to do it today, three years in the future, it’s going to be also similarly irrelevant.

Luis

I think you’re right. But this is true for all AI, right? It’s also true for all the micro models that are trying to estimate how much is compute transforming to advance. We have some general patterns in some general scaling laws, but these things are—we don’t really know how much we can extrapolate. We are in a period of massive technological change. And the good news is that it’s massive. And the bad news is that we have to peek into the future with really just in the dark with a little bit of light. You guys are trying to help people see further into the future and we are all trying to use the best tools that we have. But the truth of the matter is, if this is as revolutionary as we expect, the future could give us big surprises. Yes, I do agree with that.

Anson

How much does this model depend on the tasks that are hard for humans also being the tasks that are hard for the AIs as opposed to some kind of different skill distribution for the AIs, which seems to be the case. Like it’s kind of like Moravec’s paradox in AI. The things that are easy for humans are hard for the AIs.

Luis

So I think the paradox is a huge discovery for all of us. We discover it every day, right? Things that we find impossible to do, the computer is doing perfectly. And then we end up spending time fixing some stupid mistake the computer, the AI is unable to fix. And so it goes the opposite way in some sense, as you’re suggesting. We are indeed studying a situation where the AI is little by little replacing things that the lower skilled worker can do. I think the reason why it makes sense in this context is because the AI makes mistakes. And I like to refer to this cutoff as this supervision threshold. So you need to be smarter than AI in order to be able to correct the AI.

Think of a kid who is now going to school and they can do the ChatGPT. They can make ChatGPT make the essay much better than them. So ChatGPT — they just do the essay and they hand it in. They can’t see where the mistakes are or the things that are actually not perfect, so they are never going to arrive at the supervision threshold. They’re never going to arrive to the point where they are able to read the essay and see the mistakes, because they basically spent all their years — and you have a young kid. My kids are already out of this, but you have a young kid, and this is going to be an issue, right?

I have a friend who’s a high school teacher of English, and he tells me, you know, how do I make these kids want to write and read? They read Hamlet quickly in the morning with the LLM. They take the key questions that have to be answered in the class, and they kind of BS their way through their answer and they don’t read anything. So I think that the reason that we are thinking of this is we are in a context where in a law firm, in a consulting firm, etc., as you are acquiring this seniority, you are acquiring the ability to add value and be above what the AI can do.

To the extent that it’s the opposite, to the extent that AI is doing all the difficult tasks and anybody can do the correction of the points, then this would be a different world. Indeed, I think companies will have to think of training in different ways. Maybe they have to think of, okay, we’re going to train the workers by — maybe we hire less of them, but the ones we have, we train them by going over the AI output and reviewing it. So that there is actually a way that you’re still improving, but you’re not going through all these routine tasks that, at the end of the day, don’t have any value at all anymore.

Andrei

So in response to this AI Becker problem, could there be more of an equity type arrangements involving human capital, where firms have some sort of exposure to the human capital they help create?

Luis

I mean, the human capital is inherently with the person and the person — you have a big moral hazard problem, right? So once somebody has invested in you, you are going to decide how much you work. You could decide not to work because you are not getting the upside. The company is getting the upside. So it has been historically very hard to find market solutions to this.

It’s similar with loans. I mean, there are loans for MBAs for certain high end things. But loans again, it’s hard to see how you secure the loan against the human capital. You cannot secure with the human beings because of slavery being forbidden! And you cannot pledge yourself as collateral. So human capital transactions — I don’t say they are impossible because they exist. Often these loans are government programs. In the US, there’s a lot of government guarantees. In the UK, government guarantees. I think that equity has proven really hard.

Maybe with football players, right? Maybe with football players you get the upside. You train a football player, you sell it to another team, etc. It has an equity-like arrangement, but it’s the only context where the firm trained the football player, which is the — I don’t know if it happens in the US, professional sports is able to get a fee, a transfer fee for having trained that person. But it’s a very unusual context, I would say. Equity is hard. Debt is more promising. But even debt is tricky because of moral hazard and repossession and all that.

Andrei

Going back to the bigger picture a bit on AI and training. Do we have a sense at the moment if AI is making training of humans—I should note—easier or harder? Because on some level you were mentioning that there’s all these learning tools, AI powered learning tools that you could tailor to the student, assuming regulation allows you, and that could be helpful. But on the other hand, you know, I’m an instructor myself and I can’t get my students to read anything. I can get them to read the AI summary of the AI summary of something. That seems bad. Is there any evidence you know of?

Luis

I haven’t seen evidence. I think we are all observing exactly what you’re observing. We’re all observing that students have been using AI for cheating. Let me tell you what I do with AI. So my view on education is summarized by what I do with AI in my two classes. I teach the microeconomics class in the first year of a master’s program, and my view is that if you want to be thinking in the future, you need some basic models, some basic facts, and some basic tools, and that is not going to change. Otherwise you cannot think, right? We are trying to triangulate is $400 billion big or small. Is that big valuation or — you need to have something in your brain to use things.

So at a basic level, I want them to use the old blue books, write the problem sets, and the exam is going to be written. And I tell the students, these are the basics you need in order to operate in life. So there I think AI is an enemy of us because with AI, I can do the problem set automatically. Why would I go through the problem set? And then you get to June and you have the exam and you’re like, “oh, what is this exam about?”. So there is an enemy. But indeed there are tools. And I try to tell the students, you can ask for help. You can ask Claude to explain. If you don’t understand what it is, you do it two ways, you do it three ways until you learn it. Okay.

On the other side, let me tell you what I do to my second year class. My second year class is — it could be called “what I learned in politics that I didn’t know before as an economist”. So I start from the policy. What is the policy that you’re looking at? So a group is looking at Tegucigalpa. They have a huge water problem. The water runs out all the time and there’s only a few hours of water a week. Okay, that’s the economics, the economic policy. But now you want to look at the politics. What is the political economy? Who are the interest groups? Who are in favor? Who are against? You want to then talk about the narratives. How do you discuss this in public? How do you give a speech? What is the message that you give? What do you want people to hear? People don’t hear what you say. They hear something else. What are the preconceptions?

And then you want to talk about implementation. How are you going to implement your solution? Well, in this class I tell students AI use is obligatory. They need to for all of these things: the analysis, the politics, the narrative, all of these things. They need to build models. They need to understand the data. They need to actually figure out stuff that three years ago would have been unthinkable. They couldn’t have been doing. So I think that my view on how AI is working in education is that we need to make sure that they are learning the basics and that is going to be a struggle. And I agree with you. But at the same time, we need to be able to get our students to do enormously more than they could have done.

So if you’re teaching a macro international class like you do, and the students can actually do a trade model of the Ukraine sanctions, they could actually change the substitution. I mean, they could do these things that before, the amount of computing and programming that you would need would be the task of a PhD! So I think that the way that training is going to work has to radically change in using the AI tools to learn and using the AI tools to get much further. But at some basic level, we need to be able to persuade the students. That’s the difficulty — that they need to learn the basics. Maybe your papers will be written by an AI, but if you don’t learn to write, you’re not going to learn to think. I know that argument is difficult to make, but if I have a child, a seven year old like you have, I would try to hammer home that argument somehow.

Maybe for this part, what we are going to have to do is homework in the classroom, right?

Andrei

Just notebooks and adults in rooms.

Luis

Maybe we have two hours and, you know, in the library of the school from 2 to 4, which is homework time, no phones, no computers. And you guys have to do homework for these basics part. And then we need to also use the AI. I mean I believe in both. I don’t think it’s either or.

Europe’s AI regulation problem [00:39:31]

Andrei

So this is fascinating. Should this make us a little bit pessimistic in the sense that… my sense is that there was this more optimistic line of thinking that I would associate with Daron Acemoglu, which is that we have options. There is directed technical change — we can choose to develop technologies to keep them complements with human labor, and then we won’t have so many problems. Whereas here it sounds like almost something inherent is happening, where as the AI gets more advanced, it becomes a substitute — so we don’t have a choice. We either accept advanced AI, but we accept substitution, or we don’t accept advanced AI. So there’s this advanced AI, and “no substitution” might not be on the menu.

Luis

I think that’s my view. Indeed. I think Daron Acemoglu has excessive optimism about two aspects of this. One is how much can we control this runaway train? We are in a game theoretical situation between China and the US. And I mean, there is a strategic interaction between them. If the US decides not to develop, then China is going to develop anyway. So I don’t think the possibility of slowing things down exists.

Second, I always think when he says—so actually it was two parts, but I want to make three. So one is the interaction [between the US and China]. The second is the “we can direct technology” — because who is “we”? Is “we” China? Is it the US? Is it firms? Is it workers? Is it lawyers? Is it truck drivers? Who is “we”?

All of those people have very different interests. Is it the people in the AI industry, which is now generating a big part of the growth in the US? Is the US not going to want to have this growth? So “we” is always kind of hidden away a little bit. Though as somebody who is as super sophisticated about political economy, hee knows better than me. He’s written a whole book and lots of papers about the institutions and how they mediate this “we”.

The other thing is that I think that the risk of trying to interfere is many unintended consequences. So I want to tell you about Europe because that’s what I know well. Apart from being an economist, I spent a few years as a politician. I was in the European Parliament and Europe has made a very Acemoglu effort. In fact, let me tell you. Let me tell you that this letter that Acemoglu and Elon Musk and many others signed—the Future of…

Anson

Future of Life Institute.

Luis

The Future of Life Institute was this February or March 2023, something along these lines? This letter actually came in the middle of the elaboration of the EU AI Act. So the draft was finished in November of 2022, the two drafts. But then the drafts have to be reconciled and the act was passed, I think, in the spring of 2023. In between, when they were just finishing, there was ChatGPT. The introduction, if you remember, of ChatGPT was in November of 2022, and that moment was the moment where there was this existential risk pandemic. Everybody was like, oh, we’re going to get turned into paperclips and humans won’t exist anymore. And so they wrote this letter and there was a moment of panic in Europe.

And the person who actually wrote the law from the Commission has given an interview to a Swiss newspaper. I wrote about it in my blog. If somebody wants to see it in my Silicon Content blog, it’s called “Why is the EU AI Act So Difficult to Kill?”. And basically he argues that it was a bad moment for that letter because really, Europe decided, “this is too risky. Let’s put all these guardrails all over the place.” And the consequence for Europe is that, as you were hinting, a lot of the productivity gains that we could be getting from AI are not possible to get.

So let me give you an example. The AI Act is built on four risk categories. So there are forbidden uses, which includes detecting emotions — that’s not allowed — or government controlled surveillance and point system social scoring systems. That’s forbidden. But emotions detection is also forbidden. Second, social scoring by the government’s public sector.

Second, high risk uses which involve energy infrastructure decisions that the legislature says AI shouldn’t be taken without a lot of steps. Now those decisions include education and health. So in education you would very much want, for example, your students in Edinburgh, you would want them to take an AI quiz and to help you see how they’re doing. Probably eventually, it’s going to be possible for them to do the problem sets in a customized way so they can jump a step, etc. But the AI Act says these things are high risk, and so the fact that they are high risk means that when you train a system, you have to make sure that all the data are correct, that all the data is free of errors to the extent possible, that it is unbiased, and that you have the relevant data.

Now error-free training data doesn’t exist. The training corpus right now is the internet. Errors must be all over the place. Somehow, for some bizarre reason that I don’t know if anybody understands, after all of this is aggregated, all the errors get washed out like in a law of large numbers kind of effect. So it kind of works. But the training data has to be unbiased and free of errors. Now you need to keep detailed logs on these high risk applications. You need to keep your records. You need to keep documentation of everything of the system for ten years. You need to prove accuracy and security. You need a conformity assessment and you need to register with the EU authorities.

Now there are 55 EU authorities that will do that. And these authorities are supposed to have personnel that are highly qualified in AI, highly qualified data protection, etc. Now if you’re an entrepreneur, you’re starting your company, your little startup with education — you have to do all this, plus GDPR! The General Data Protection Regulation. Businesses know it because of the cookies. It’s a pain, right? I mean, you know, in economics people often disagree about things. I can tell you there’s been 15 papers on the GDPR, and all of them find less venture capital investment, less startups, higher compliance costs. Every single thing tells you the GDPR has been bad for EU business, and now we’re adding the EU AI Act for startups.

So part of the risk is you try to control the technology and you end up without technology, which is kind of the world where Europe has a risk of finding itself. We don’t have foundation models. We have great researchers. We have a huge savings pool. For many reasons that we could go, if you guys care to go into that. We have the researchers, we have the ideas. Businesses don’t scale in Europe. We don’t have foundation models. I think there are something like two foundation models in Europe compared to 15 in the US. The numbers are really very disproportionate. And we have very little AI implementation. So that’s a problem.

Anson

I’m curious what you would say to a person who is like, “oh no, but I actually think that these risks are really serious.” And even if we don’t think all the way to immediately turning all humans to paperclips, they think that, “oh, actually, if you have a ton of AI systems that are deployed throughout the economy, if they’re not optimizing for the things that humans care most about, then you could slowly, gradually just shift things off the rails.” And so maybe they would say, “well, the Act’s most serious risks, the systemic risk for AI systems or general purpose AI systems, these are the models that require over 10 to the 25 training FLOP and maybe a bunch of other requirements. And so surely this is just applying to the most capital intensive or most capital rich companies. And so maybe for most other people, this particular thing doesn’t matter so much, but it’s just this particular group of actors that need to be subject to additional scrutiny.” What would you say to that?

Luis

The systemic risk category is maybe a different story. I was talking about systems more broadly. There is a systemic risk category indeed. As you’re pointing out, I think it’s 10 to the 24 FLOP, but it’s… well it’s 24 or 25.

Anson

I think it’s 25. It’s based on GPT-4.

Luis

So indeed GPT-4 is above. Llama is above. So we have already—

Andrei

The previous generation systems are already above.

Luis

The previous generation are already above. And yes, they have to be subject to adversarial tests. You have to prove and make sure that they are safe. To me, this kind of existential risk issues in these very large systems — they do deserve additional scrutiny.

Andrei

So in that context, are you more or less optimistic about the EU when you think about AI? Do you become more or less optimistic about the European Union?

Luis

I am desperately worried. I think that we are in a situation where if these kinds of effects that we were discussing before of big productivity growth, big welfare gains for many citizens that can get AI to do (for example) their driving. They’re not going to get killed in a car crash. They’re going to be able to do their contracts. They’re going to get legal advice to do smarter things that they would have done, negotiating with their landlord. Many, many things that are not going to happen potentially because they will not be allowed. I think that productivity growth will suffer. Growth will suffer. Welfare gains will not happen. And I think that Europe has a demographic problem and a high debt problem. So Europe needs growth more than many other places to pay its bills.

Europe is in a very tricky situation. Look at what’s happening in France. On the one hand it’s not growing and on the other hand they have this big debt and explicit debt and implicit debt for the pension liability. So it desperately needs growth. And I fear that the European Union has overregulated itself and is not going to get that growth.

Anson

I’m curious how much of what you say also applies to the UK, because, for example, the UK does have a frontier lab, which is Google DeepMind. So how much of what you say also applies to the UK?

Luis

Let me just start by saying that I don’t like Brexit and I don’t think Brexit is a good idea. But Brexit was bad for the UK, but it was bad for Europe because the UK was the force that was pushing Europe in a more free market and open minded way. It was. The UK was the motor for the single market project that was about making sure that Europe had an integrated market. And so once the UK left, we had this divergence.

The UK has a very pro-AI posture. It hasn’t diverged in other areas, in environmental areas, etc. It’s still applying the EU rules, but in other areas it’s actually taking a more positive posture. I’m a professor here. I’m at the London School of Economics because I actually believe in the UK. I do actually think that the UK has a very bright future. It’s just the governments are not making the decisions at the speed necessary to profit from them. But if you think of AI, you have capital, you have nuclei of talent, which are Oxford and Cambridge, which are at the cutting edge. You have DeepMind, you have all of these other labs around it. I think the UK could be Silicon Valley. I don’t see why that could be impossible. Maybe the risk taking mentality is the one that is missing. It’s not quite there.

Who captures AI value? [00:52:46]

Andrei

So thinking a bit about the AI value chain, you were saying that there was this infrastructure layer, this lab layer and this implementation layer. How do you think about where the value will go? How will the value be distributed across those layers, and how do the prospects of different parts of the world depend on which layer gets the value?

Luis

That’s a great question. So I’ve been arguing that Europe could try to get the value of the AI. We’re not going to get it from the lower layers. We could get it from the implementation. So the idea for Europe could be if we managed to make the other layers competitive, interoperable, then we could get value on implementation. So let me split it bit by bit.

So on the hardware layer it’s clear that China and the US are capturing the value. So on the hardware layer, if the hardware layer is where the value is, then that’s clearly going to benefit the US. And I think that it looks like learning curves are very steep. Look at Intel. It had a competitive advantage in PC hardware for four or five decades until just this last generation where they got hammered. But they’ve had decades. It’s very hard. The learning curves are very steep. You need to keep it clean. You need to print it carefully. You need to design very complicated things. It’s very hard to enter. I don’t think there is really an entry possibility. So a lot of capture will get into the hardware layer. I think the evidence is pretty strong regardless of what happens upstream. Profits will go in there.

For cloud computing, I think there could be big switching costs, moving your data from one cloud to another. Europe is really trying to avoid that, and it’s really trying to make sure that the data is yours and you can move it. But the cloud players can add features to make sure that you want to stay and that if you move it, you lose some value. So there could be quite a bit of switching costs. I think we need to make sure that in cloud computing. the data is encrypted and it remains on servers that are located geographically here in Europe, so that not all the value going back to the US. But I think both geopolitically and economically, the risk is clear that also on the cloud layer, the US will have extraterritorial reach because those are American companies.

On the LLM layer, on the foundation model layer, it seems to me that what we are observing is that there is very strong competition and it’s very hard to obtain a competitive advantage. It seems to me that what we see is all the time one company gets some feature. We all love it for three months, and then we suddenly start trying the other one because it just got a little bit better feature. I am basically between Gemini, Claude and OpenAI. I’m switching between all of those. It seems very hard to get an advantage.

Also, there is a big open architecture possibility. Llama was trying. Mistral is building on that. All those weights are out in the open. So we’re going to be able to have some, at least some of the applications that are more energy efficient or smaller. They can go on the old systems, and they can actually enjoy the fact that these are open architectures. So I would think that that layer remains quite competitive, with one caveat, which is the introduction of switching costs through memory. If the system starts to remember you and starts to know who you are, then switching systems is going to be costly.

I think that all of that data, we should make sure we should do our best to make sure that the data is yours, and that it’s very easy to port. I think the portability is crucial. Think of the example with social media. In social media, there’s no portability. The data of my graph and of my everything about me belongs to—

Andrei

Meta.

Luis

Or Meta or to Twitter. I will never say X. This is my one principled objection to Twitter. And if you are in disagreement with them, you start again from zero. Okay, I have what, a hundred and few thousand followers on Twitter. If I want to abandon them and start somewhere else, that’s my problem. But if not, then I stay there. So imagine a world where I send a message and everybody who likes me can follow me from any platform, but it’s completely interoperable. Market power would change radically, right?

And I think the regulation you were talking about, optimal regulation, would do its best to make sure that interoperability exists, that we don’t fall into the same trap that we fell with social media on some of those verticals, to appropriate quite a bit of the value at the European level. Now, how do you do that and avoid extraction by all those other upstream players that we have been talking about, from hardware to infrastructure to the LLMs. Well, we have to move fast, which we’re not doing, and we have to keep the markets competitive. We have to do our best to keep those markets competitive through interoperability and all these other demands that the data can be moved and the clouds are not proprietary, etc. I think that it’s possible, but it’s tricky because the truth of the matter is, if you don’t have the hardware, everything else flows downstream.

Andrei

So you think it’s kind of one of the key points here seems to be that you think the EU should be using the leverage it has to move as much of the value to the implementation layer, because that’s the layer where Europe is strong.

Luis

Yes. I’ve been pushing a “smart second mover” strategy for Europe, which is a strategy that basically has Europe trying to free ride on this gigantic investment boom in the development and the data center development that is already taking place. We take it for granted. We’re not going to try to imitate because we’re too far behind. And let’s use all our scarce resources in securing the autonomy, encrypting the data, having the data centers locally based, but mainly in developing a strong implementation layer.

Andrei

And in that context, would you worry that Europe would have some of the same problems it has had with regulating the tech giants in the past? Because I’m guessing this becomes a geopolitical game pretty quickly.

Luis

It does become a geopolitical game quickly, that’s the problem. The US government is really throwing its weight behind these big giants. And it’s going to be very hard for Europe to insist on level playing fields and interoperability, etc. And we are seeing it now. Of course, there was a digital tax, there was the OECD pillar two that we were going to harmonize aspects of corporate taxation. Trump said that’s off the table. I mean, it’s going to be very difficult to do certain things that rely on mutual acceptance. The US is going to throw its power, and we’re going to have to just basically swallow it.

And consider the Turnberry “agreement” between Donald Trump and Ursula von der Leyen this summer. That was a trade dispute. In every trade dispute until now, the way they work is okay, you put the tariffs, I’m going to reply with the same here. Europe comes out of the room saying, “huge victory. He’s putting all these tariffs and we’re not doing anything.”

“Sorry. What’s the victory?”

“They’re not going to do any other things.”

“Where does it say that Donald Trump is not going to do any other thing?”

“No, no. They’ve promised not to do any other thing with our cars.”

And of course, there’s no promise. So we accept the tariffs. We don’t do anything in retaliation. And on top of that, we didn’t really get any commitment not to have any further actions from the US. The truth of the matter is that geopolitically, we are very dependent. And the Ukraine war, which would take us in other directions, is part of the reason we need the US defensive umbrella. And we are going to be struggling a lot to get that defensive umbrella to continue.

Anson

Given that, I’m curious how you think about economic security because I think a lot of the reasons for this smart second mover strategy is that it’s a lot harder to build out huge amounts of energy infrastructure and data centers. But then very common in these kinds of discussions about data centers is to think that we want to have some kind of sovereign compute. If this is so important to the economy, then we want to make sure that we have our own ability to have data centers in the EU. And if people need to use AI, then we need data centers there. How do you think about that?

Luis

I don’t think the public investment in this is going to be the big solution. So the EU has two sets of rules. One is the factories and programs and the gigafactories. The gigafactories are five big factories that are AI data centers. But the level of investment that is being put now into these one gigawatt plus centers which are really extremely costly—we are going to have one of those in Portugal. It’s private sector investment. So this is going to be one data center that will be local.

We’re going to have more local infrastructure. So basically Portugal, Spain and northern countries because of energy issues are getting some big data center investments in Spain by the Ebro River. I don’t know what the Ebro is in English. The Ebro, the big river that comes through the north below the Pyrenees, taking all the Pyrenees’ water. There’s going to be two big investments there. So we will have kind of “sovereign data centers”. But these are not truly sovereign because in some sense the ones in Spain are basically Google and Amazon. They’re Azure and Amazon Web Services investments. But if they are local we get some control.

I mean ideally eventually there will be some local European companies doing this. I don’t think that public investment is the solution because the numbers that we’re talking about—we’re talking about hundreds of billions of build up per year, up to a trillion for 2030. These are numbers that are really very large. And of course, public investment is not at that level. All of these companies — Amazon, Microsoft, Apple, etc. — they’re all spending in R&D more than any government, just one company in Europe. So it’s not going to be possible to keep up through public investment. The private sector has to want to do it. And in order for the private sector to want to do it, regulation is crucial both in terms of permitting, but also in terms of all these regulatory obstacles that we seem to be throwing all over the place.

Andrei

So on net from this sort of game… because I think a lot of people in Europe are upset by the relatively aggressive stance that the US government is taking on a number of these issues. On net, is this good or bad for Europe? Because on one level, I guess aggressive US government action means that we’re less likely to be able to move value to the layer that’s convenient to Europe. But maybe this sort of aggressive action would also make it less likely that we will be too risk averse, right? Because, you know, we want to — our instinct is to stop a lot of things that the tech giants don’t want us to stop and that the US government might not want to stop. So will the US government save us from ourselves?

Luis

So that’s — I mean, I think that consequence would be welcome, or at least to some extent welcome. We had a year of wake up calls. We said wake up call number one, Trump gets elected. Wake up call number two, the sofa scene where Vance and Trump ambushed Zelensky, the Ukrainian president. All the time it’s like this is a wake up call for Europe. We cannot trust our old ally, the US. We need to act together. And then we go back to sleep. People have these wake up calls, and every time you’re going back to sleep. They wake up calls don’t seem to be waking us up at all.

To some extent, what has happened this year should have unleashed a wave of “we’re going to invest in AI and destroy all this legislation”. And one post that I wrote on this I mentioned before on the Silicon Content blog was exactly asking, why is it so difficult to undo this thing? And Europe doesn’t have a very easy error correction mechanism. In Europe, the same European Commission that had this explosion of legislation, that was the Green Deal and the digital legislation over the five years between 2019 and 2024, is now tasked — the same president, Ursula von der Leyen — is now tasked with undoing it. “Oh, we went too far. Let’s undo it.”

Well you know, the rapporteurs, the people who wrote the legislation in Parliament and in the Council, the people in the Commission who pushed it, all of these three institutions — the governments, the parliament and the European government, which is the Commission — all of these institutions are going to be tasked with undoing a lot of rules that they themselves pushed. And they saw big victories when they put them. So now we’re saying, “oh, you know what? We thought the Act was great. But now that we realize it’s going to slow us down and Trump is going to be a risk, let’s continue.” That is very hard to happen.

The coalition that runs Europe involves the center right, the center, the center left and the Greens. All of those parties basically were the same ones that passed the first legislation, and they’re the same ones that have to now undo it. And there are many differences inside that coalition as to what can happen. The very first piece of legislation should have been removed, which had to do with excessive corporate reporting and paperwork. It was guaranteed to pass. Everybody thought it was going to pass, and then Parliament turned it down because a lot of people were invested in the existence of that legislation. So I hope Trump partly saves us from ourselves or the US partly saves us from ourselves in some of these aspects. But I am not very hopeful.

AI, interest rates & fiscal future [01:08:17]

Andrei

So one direction I was also hoping to bring back into this discussion is a bit more of the macro finance angle, right? There’s been quite a bit of discussion about the potential impact of AI on things like interest rates and things like that. How do we think about that in the context of fiscal sustainability, macro financial stability? These are hot topics in general and hot topics in the European Union in particular. Any thoughts on that?

Luis

Yes. So I wrote a post that I titled R Without G, talking about how the European Union could get high interest rates and not growth. Let me unpack this a little bit, but not do it for Europe yet and then apply to Europe. So there was a very recent paper by Auclert and some co-authors that was presented at NBER this summer. People can get a link. Maybe we can post the links to the papers that I mentioned. The paper provided a very simple demand and supply framework and applied to AI.

So basically they talk about the price of assets as being the result of a demand and supply equation. And when there is a lot of demand, prices go up. The tricky thing is prices go up — everybody has to remember in our audience that means interest rates go down. So those two things go in opposite directions. So they argue that over the last 40 years, the demand has greatly outstripped supply. And so the prices go up, and the interest rates go down. So we have a very long secular drop in interest rates.

And they basically say in their calculation the asset demand has multiplied by four. So a big increase in asset demand because of all of these things that have to do with slow growth, with demographic change, people need assets for when they retire, and whenthey are old they need safe assets in particular. All of that has led to a very big drop in interest rates. It’s been a godsend for everybody who was in debt, particularly countries that were in trouble — they could issue debt for free!

But they argue that AI is going to change this, and that AI is going to increase interest rates. First, because of the impact of what we have been discussing of higher productivity. High productivity growth means the supply is going to increase. They’re going to supply equity. They’re going to have to raise equity. That’s a supply. They’re going to have to raise equity to pay for the AI investments for all the AI labs, for all this. And at the same time, the demand might go down because younger workers think, oh, well, the economy is growing a lot, so I don’t really need assets because the economy is going to grow so much. So they say this is going to lead to a drop in price and an increase in interest rates. And that’s their argument — so a drop in asset prices, and as a result an increase in rates.

So basically their view is that we will have bigger G, higher growth rates and bigger R, but the growth rates will be higher than R. So no problems for fiscal sustainability. Remember that sustainability depends on R minus G. R is how much do you have to pay when you issue debt. And G is how much is the pie growing with which you pay. So if R grows a lot, but growth doesn’t go up, each time I have to be paying more and more and more, and I’m squeezed. If my growth rate is growing a lot more than what I used to pay the debt, I can afford it.

So they say the growth rate probably grows a lot and it grows more than R. So overall that’s sustainable. So what I worry about Europe is that you are going to have the bad part of having to pay higher rates without having the good part of having higher growth rates. If you’re putting obstacles in the way of how you are adopting the AI — the taxi drivers oppose self-driving cars. The legal profession opposes AI in the law, and the doctors don’t want AI. And you get these human bottlenecks everywhere. Then you’re not going to have increases in growth rates, but you will have to pay the global, higher global interest rates that everybody is facing because of the AI revolution and the higher productivity of capital that comes with that and higher investment boom and all that.

So as a result, what you could have is that you make much worse debt sustainability problems which plague our welfare state in the European Union. So we have countries that have not only high explicit debt, almost 120% of GDP in France. And they also have high implicit pension debt, 3 or 4 times GDP, probably more in some countries. And all of this has to be financed with the G, and you have to pay all these increasing R on it. And if you don’t get a G and you get a higher R, you’re going to be in big trouble in terms of sustainability.

So you asked before whether Trump would be somebody who would wake up Europe. This is another reason to wake up. We have a problem demographically. And this is not just continental Europe, it’s also the UK. And we need more growth, and we need to take a much more aggressive pro-growth stance, much more aggressive pro-growth stance.

Andrei

So in terms of this overall problem and that piece in particular, I’m a little bit surprised by the fact that in the economics profession, there seems to be, if not a consensus, a very strong majority view that AI will lead to an increase in interest rates. But couldn’t you make an equally strong case that it could lead to a decline in interest? I mean, precautionary saving. You know, I get this AI thing—

Luis

I’m a bit scared.

Andrei

I’m very scared. Right? And I think in the Valley, people are having this discussion about, you know, I need to do well in the next five years. Otherwise I’ll be a serf forever or something along those lines. Couldn’t people just save because they want some exposure to the companies that will own the economy? And that’s kind of a first order thing. So they really want to buy assets now because their human capital will depreciate.

Luis

So I think that it’s not impossible that we have that—you’re right that now the world seems very uncertain. Also, it’s possible that inequality grows very much. And that’s going to go in the opposite direction as well. Indeed, because the rich people save more, they don’t consume as much. I mean, at some point, Elon Musk is not going to consume his $1 trillion package if that happens. So yes, there are a couple of forces. The precautionary saving is one. And the other one is the inequality increase that could push in the other direction. I would side with the consensus of the economic profession. But you’re right that there is a question mark over it.

As a first order approximation, the slowdown in growth over all of these years led to this drop in R. And an acceleration in growth, if that’s what we think is going to happen as a first order effect, I think will increase the return on capital and will lead to an increase in R.

Andrei

So empirically, those things go together as opposed to—

Luis

I would expect that. But as we said, we are peering into the unknown and we have to be all modest and humble.

Andrei

And the other thing that surprised me a little bit was the fact that you were tying this increase in R with problematic implications for Europe in particular, and the reason why I was a bit surprised by that is that I think of Europe as a continent of creditors. We run huge net surpluses with the rest of the world. So it would seem to me, okay, we’re creditors. R will go up.

Luis

So we are exposed to these gains.

Andrei

We will get richer. So in some sense, we will get richer. Our governments will have more problems, but we’ll get richer. So as long as the government finds ways to tax—

Luis

So let me unpack that. That’s a great point. So it’s true that we are net savers. And that means that as a continent we should get some exposure to the good side of the AI. So it’s true that—

Andrei

Even if it happens in the rest of the world, right?

Luis

And Enrico Letta was writing a report about how Europe is doing badly, and he came up with this expression that European savers are exporting their savings into American companies that are employing the European workers and entrepreneurs who cannot make their — so this is a lot happening. You are down in the West Coast and you see all these Europeans running, Indians and all these other nationalities.

So it’s true that the savings should capture some of these additional. We should be benefiting. We should be on the good side. Now, the distributional impact is a bit tricky, because who is going to benefit from those savings rates? For example, Holland has big pension funds with big exposure to interest rates. But in places like Spain and France, essentially the state is doing the whole pension through a pay-as-you-go system. So the overwhelming majority of the population has zero financial wealth. They have housing which could also go up. And so if you’re long on housing, you’re going to benefit probably from this run-up. But they’re not exposed to financial assets. Only the very top 3, 4 or 5% of the population will have significant exposure to these financial assets. So the distributional issues are not obvious. But you’re right that there is a net saver income effect that is positive income effect, meaning Europeans are wealthier, as you put it, when R goes up.

Andrei

One thing you were hinting at when discussing the macrofinance question was demographics. And I must say that I worry about demographics quite a lot in the European context and in the global context as well. So should demographic change change our view of the trade off between the benefits and the risks of AI? I think a lot of people are in this mindset that, at least in the rich West, things are pretty good as they are. So we can kind of afford continuing with things as they are and be quite risk averse when we write down AI regulation and do things like that. But aren’t we actually on a burning platform? Aren’t things going to get worse unless something turns up like AI?

Luis

So you’re very right. And my colleague and co-author and friend Jesus Fernández-Villaverde has been sounding the alarm about the fact that demography is turning. Total fertility rates are plummeting, not just in the developed world like we expected, but actually in the developing world. Colombia, Tunisia, Turkey — they’re seeing collapsing rates, which is very strange because normally… they’re saying they’re going to grow old before they get rich. Normally the opposite happens. The countries get rich and then they start aging. So the demographic collapse is really problematic. And it’s true that you will have a need of a care economy. For example, in the scenario, in the positive scenario that you were describing, where AI does many of the tasks that we think are human, care is something that maybe can be helped by AI. And so replacing some humans in those professions, which are going to be where the enormous share of population that will be old, will be useful.

And I remember discussing this with Joshua Gans and I said, “oh, people are not going to want a robot to care for them.” He said, “are you kidding? I would much rather have a robot take care of my needs — cleaning or showering me or whatever it is — than a human.” I was like, “oh, actually, maybe that makes sense. That robot is gentle. Maybe you can do it.” So he was arguing that the robots will have a big value as carers and potentially that people will want them. I don’t know, we’ll have to see. But if not in caring, in many other things. And again, there is evidence that people are doing therapy with AIs. So maybe there is more and more range. But if not in therapy, in many other things, we need growth. We need the labor that we’re not going to get from this lack of fertility. And so let’s have a more AI positive posture. Absolutely.

Of course, it could be that AI leads people to want to have AI companions. And I don’t know if—

Andrei

That makes the fertility crisis worse!

Luis

It makes the crisis worse! But okay, that’s a consumption choice that we cannot predict how that will play out. But it does seem like people like to have AI friends. I think that that is happening.

Andrei

So I think the question, one of the things I had in mind when I asked the question was about stuff like R&D and things like that, right? So you know, R&D, we think of R&D as being done by relatively young people, although I appreciate that it’s changing. It’s not so much AI helping us with the care economy, although that’s important as well. You know, in semi-endogenous growth models we need population growth to get any growth. And I guess with fertility declines that’s problematic. So at the very least you need to be able to shift more humans into R&D.

Luis

And not only humans, but AIs too. But AI as well. Some of the work by Jones and the paper by Philippe Aghion — the recent Nobel Prize winner — and Ben Jones and Chad Jones argues that basically, to the extent that AI is just capital, it’s not going to make a big deal. Where it really makes a big impact is in R&D. To the extent that AI can accelerate the production of ideas, then AI can really accelerate growth. That’s, I think, the scenario where you will see the big growth acceleration, having taken into account all the caveats that I put about, you need regulatory approval and so on.

But I agree with you that in terms of generating ideas, which is really the driver of growth, if we don’t have the scientists, we better have the AI generating ideas or we need to move more people into scientific production. I really am pretty optimistic about how AI will help in the production of ideas. Somebody like Terence Tao —he was arguing that AI was helping him collaborate with many collaborators. So basically he says, look, you always have small teams of mathematicians because you need to trust each other, because you don’t know if one key step in the proof wasn’t well done. But now with AI, we use Lean to do the little bits of the proofs. We decentralize it, and then we can check each other’s work and somehow we have bigger teams.

Other mathematicians are saying that AI is already helping them to make a proposition, or not quite yet. I don’t think there’s an AI theorem, but I think there are some results already. So combinatorics, the protein folding Nobel Prize — we do see some impact of AI in accelerating research, which could be crucial. Indeed, given our demographics, we need to have the research sector produce somehow.

Anson

All right. I think that’s a great place to end. Thank you, Luis and Andrei, for coming onto the podcast.

Luis

It was a lot of fun. I appreciate it.

References

I. Silicon Continent Posts

• On EU Digital Regulation:

  • Why Is the EU AI Act so Difficult to Kill?

  • The Strange Kakfa World of the EU AI Act.

  • Is GDPR undermining innovation in Europe

• On Macroeconomics & Interest Rates:

  • R Without G

  • Can AI solve Europe’s Problems?

• On AI and labour markets:

  • The AI Becker problem

  • How to think about the economic impact of AI

• On Europe’s Second Mover Strategy

  • The Smart Second Mover

II. Academic Papers & Research Mentioned

Labor, Productivity, and The “Training Ladder”

• The “Chatbot” Paper (Customer Support):

  • Reference: Brynjolfsson, E., Li, D., & Raymond, L. R. (2023/2025). “Generative AI at Work”. NBER Working Paper / QJE.

• Software Developers Field Experiment:

  • Cui, A., et al. (2025). The Effects of Generative AI on High Skilled Work: Evidence from Three Field Expermients from Software Developers. MIT Feb 2025

• Aggregate evidence of substitution

  • Brynjolfsson, Erik, Bharat Chandar, and Ruyu Chen. (2025). “Canaries in the coal mine? Six facts about the recent employment effects of artificial intelligence.” Digital Economy Stanford.

  • Lichtinger, Guy, and Seyed Mahdi Hosseini Maasoum. (2025). “Generative AI as seniority-biased technological change: Evidence from us resume and job posting data.” Available at SSRN.

• Autonomous vs. Non-Autonomous AI:

  • Reference: Ide, E., & Talamas, E. (2025). “Artificial Intelligence in the Knowledge Economy”. Journal of Political Economy

• Superstar Effects:

  • Rosen, S. (1981). The Economics of Superstars. American Economic Review.

The The AI-Becker Problem:

• Apprenticeships: Garicano, L., & Rayo, L. (2017). “Relational Knowledge Transfers”. American Economic Review.

• Garicano L and L Rayo. (2025) “Training in the Age of AI: A Theory of Apprenticeship Viability”. CEPR, also here.

Macroeconomics of AI & Growth:

• Asset Pricing and Interest Rates:

  • Reference: Auclert, A., Malmberg, H., Rognlie, M. and Straub (2025). “The Race Between Asset Supply and Asset Demand”. SSRN

• AI and Economic Growth (R&D):

  • Aghion, P., Jones, B. F., & Jones, C. I. (2017/2024). “Artificial Intelligence and Economic Growth”. NBER.

  • Bottlenecks/ORings: Jones, B. “Artificial Intelligence in Research and Development”. NBER. (2025)

  • Acemoglu, Daron. (2025). “The simple macroeconomics of AI.” Economic Policy.

  • Johnson, Simon, and Daron Acemoglu. Power and Progress: Our Thousand-Year Struggle Over Technology and Prosperity. Hachette UK, 2023.

III. Reports & Open Letters

• Future of Life Institute Letter:

  • Title: Pause Giant AI Experiments: An Open Letter.

  • Date: March 2023.

  • Future of Life Institute

• The Letta Report:

  • Title: Much More Than a Market.

  • Author: Enrico Letta (April 2024).

  • European Council Report

• The Draghi Report :

  • Draghi Report on European Competitiveness (Sept 2024),

Visit EpochAi podcast hub to to see all episodes, subscribe and more.

Not only is the coalition broad, but encompasses huge national and regional diversity. In Brussels, the Parliament has 705 members from roughly 200 national parties. The Council represents 27 sovereign governments with conflicting interests. A law faces a double hurdle, where a qualified majority of member states and of members of parliament must support it. The system should produce gridlock, more still than the paralysis commonly associated with the American federal government. Yet it works fast and produces a lot, both good and bad. The reason lies in the incentives: every actor in the system is rewarded for producing legislation, and not for exercising their vetoes.

Understanding the incentives

The Commission initiates legislation, but it has no reason to be reticent. It cannot make policy by announcing new spending commitments and investments, as the budget is tiny, around one percent of GDP, and what little money it has is mostly earmarked for agriculture (one-third) and regional aid (one-third). In Brussels, policy equals legislation. Unlike national civil servants and politicians, civil servants and politicians who work in Brussels have one main path to build a career: passing legislation.

Legislation is valuable to the Commission, as new laws expand Commission competences, create precedent, employ more staff, and justify larger budgets. The Commission, which is indirectly elected and faces little pressure from voters, has no institutional interest in concluding that EU action is unnecessary, that existing national rules suffice, or that a country already has a great solution and others should simply learn from it.

The formal legislative process was designed to work through public disagreement, with each institution’s amendments debated and voted on in open session. The Commission proposes the text. Parliament debates and amends it in public. The Council reviews it and can force changes. If they disagree, the text bounces back and forth. If the deadlock persists, a joint committee attempts to force a compromise before a final vote. Each stage requires a full majority. Contentious laws took years.

This slow process changed in stages. The Amsterdam Treaty (1999) allowed Parliament and Council to adopt laws at the First Reading if an agreement was reached early. Initially, this was exceptional, but by the 2008 financial crisis, speed became a priority. The Barroso Commission argued that EU survival required rapid response, and it deemed sequential public readings too slow.

The trilogues became the solution after a formal “declaration” in 2007, though the Treaties never mention them. Instead of public debate, representatives from the Parliament, Council, and Commission meet privately to agree on the text. They work from a “four-column document.” The first three columns list the starting positions of each institution, the fourth column contains the emerging law. The Commission acts as the “pen-holder” for this fourth column. This gives them immense power: by controlling the wording of the compromise, they can subtly exclude options they dislike.

Because these meetings are informal, they lack rules on duration or conduct. Negotiators often work in “marathon” sessions that stretch until dawn to force a deal. The final meeting for the AI Act, for instance, lasted nearly 38 hours. This physical exhaustion leads to drafting errors. Ministers and MEPs, desperate to finish, agree to complex details at 4:00 a.m. that they have not properly read. By the time the legislation reaches the chamber floor, the deal is done, errors and all.2

The European Parliament is the institution that is accountable to the voters. But it is the parliamentary committees, and their ideology, that matter, not the plenary or the political parties to which MEPs belong. Those who join EMPL, which covers labor laws, want stronger social protections. Those who join ENVI want tougher climate rules.

The committee coordinator for each political group appoints one MEP to handle the legislative file: the Rapporteur for the lead group, Shadow Rapporteurs for the others. These five to seven people negotiate the law among themselves, nominally on behalf of their groups. In practice, no one outside the committee has any say.

When the negotiating team reaches an agreement (normally, a grand coalition of the centrist groups), they return to the full committee. The committee in turn usually backs the deal, given that the rapporteurs who made it represent a majority in the committee, and the committee self-selects based on ideology.

Crucially, the rapporteurs then present the deal to their political groups as inevitable, based on the tenuous majority of the centrist coalition that governs Europe. “This is the best compromise we can get,” the rapporteur invariably announces. “Any amendment will cause the EPP/Greens/S&D/Renew to drop the deal.”

Groups face pressure for a simple up-or-down vote, and often prefer to claim a deal than doing nothing. MEPs who refuse to support the deal may be branded as troublemakers and risk losing support on their own files in the future.

Often just a couple of weeks after the committee vote, the legislation reaches the full Parliament to obtain a mandate authorizing trilogue negotiations, with little time for the remaining MEPs to grasp what is happening.

The dynamic empowers a small committee majority to drive major policy change. For example, in May 2022, the ENVI committee (by just 6 votes) approved a mandate to cut by 100% CO₂ emissions from new cars by 2035. De facto, this bans new petrol and diesel cars from that date.

Less than four weeks later, in June 2022, Parliament rubber stamped that position as its official negotiating mandate, with a “Ferrari” exception for niche sports cars. This four weeks left almost no time to debate, consult national delegations, or reconsider the committee’s position. From that slim committee vote, the EU proceeded toward an historic shift to electric vehicles continent-wide.

Similarly, the EMPL committee approved, in November 2021, the Directive on Adequate Minimum Wages, even though Article 153(5) of the Treaty on the Functioning of the EU explicitly excludes “pay” from the EU’s social policy competences. Co-Rapporteurs Dennis Radtke (center-right EPP) and Agnes Jongerius (center-left S&D) formed a tight alliance and gained a majority in committee, sidelining fierce opposition from countries like Denmark and Sweden that wished to protect their national wage-bargaining systems.

The committee’s text was rushed to plenary and adopted as Parliament’s position fourteen days later (in late November). The system let a committee majority deliver a law the Court of Justice ruled partially illegal in November 2025 precisely at the request of the Nordic states, striking down Article 5(2) on criteria for setting minimum wages.

The player you’d expect to check any excesses is the Council of Ministers from the member states, which represents national governments. But the way the Council participates in the drafting dilutes this check. The Council is represented by the country holding the rotating Presidency, which changes every six months. Each Presidency comes in with a political agenda and a strong incentive to succeed during its short tenure. With a 13-year wait before that member state will hold it again, the Presidency is under pressure to close deals quickly, especially on its priority files, to claim credit. This can make the Council side surprisingly eager to compromise and wrap things up, even at the cost of making more concessions than some member states would ideally like.

The Commission presents itself as a neutral broker during the trilogue process. It is not. By controlling the wording of the draft agreement (“Column four”), the Commission can subtly exclude options misaligned with its preferences. It knows the dossiers inside out and can use its institutional memory to its advantage. Commission services analyze positions of key MEPs and Council delegations in advance, triangulating deals that preserve core objectives.

The Commission also exploits the six-month presidency rotation. Research shows it strategically delays proposals until a Member State with similar preferences takes over.3 As the six-month Presidency clock winds down, the Council’s willingness to make concessions often increases. No country wants to hand off an unfinished file to the next country, if it can avoid it. The Commission, aware of this, often pushes for marathon trilogues right before deadlines or the end of a Presidency to extract the final compromises.

As legislation has grown more technical, elected officials have grown more reliant on their staff. Accredited Parliamentary Assistants (APAs) to MEPs, as well as political group advisers and Council attachés, play a large role. These staffers have become primary drafters of amendments and key negotiators representing their bosses in “technical trilogues”, where substantial political decisions are often disguised as technical adjustments.4

COVID-19 accelerated this. Physical closure increased reliance on written exchanges and remote connections, favoring APAs and the permanent secretariats of Commission, Parliament, and Council. The pandemic created a “Zoom Parliament” where corridor conversations, crucial to coalition-building among MEPs, disappeared. In my experience, they did not fully return after the pandemic. This again greatly strengthened the hand of the Commission.

Quantity without quality

The result of this volume bias in the system is an onslaught of low-quality legislation. Compliance is often impossible. A BusinessEurope analysis cited by the Draghi report looked at just 13 pieces of EU legislation and found 169 cases where different laws impose requirements on the same issue. In almost a third of these overlaps, the detailed requirements were different, and in about one in ten they were outright contradictory.

Part of the problem is the lack of feedback loops and impact assessment at the aggregate level. The Commission’s Standard Cost Model for calculating regulatory burdens varies in application across files. Amendments introduced by Parliament or Council are never subject to cost-benefit analysis. No single methodology assesses EU legislation once transposed nationally. Only a few Member States systematically measure a transposed law’s impact. The EU has few institutionalized mechanisms to evaluate whether a given piece of legislation actually achieved its objectives. Instead, the Brussels machinery tends to simply move on to the next legislative project.

Brussels’ amazing productivity doesn’t make sense if you look at how the treaties are written, but it is obvious once you understand the informal incentives facing every relevant player in the process. Formally, the EU is a multi-actor system with many veto points (Commission, Parliament, Council, national governments, etc.), which should require broad agreement and hence slow decision making. In practice, consensus is manufactured in advance rather than reached through deliberation.

By the time any proposal comes up for an official vote, most alternatives have been eliminated behind closed doors. A small team of rapporteurs agrees among themselves; the committee endorses their bargain; the plenary, in turn, ratifies the committee deal; and the Council Presidency, pressed for time, accepts the compromise (with both Council and Parliament influenced along the way by the Commission’s mediation and drafting). Each actor can thus claim a victory and no one’s incentive is to apply the brakes.

This “trilogue system” has proven far more effective at expanding the scope of EU law than a truly pluralistic, many-veto-player system would be. In the EU’s political economy, every success and every failure leads to “more law,” and the system is finely tuned to deliver it.

Share

We can build back greener, without so much as a hair shirt in sight. ... [We can] fly guilt-free, and our homes will be heated by cheap reliable power... And everywhere you look... there will be jobs. Good jobs, green jobs, well-paid jobs, leveling up our country while squashing down our carbon emissions.

Boris Johnson, 19.10.2021.1

Adam Brzezinski and I analyzed in a new paper tens of thousands of speeches from the 9^th European Parliament (2019-2024) to understand how politicians talk about trade-offs. We focused on the shift in climate rhetoric before and after the Russian invasion of Ukraine. We found that when a policy becomes expensive, politicians do not just argue about the economic cost. They rewrite reality.

Climate policy has effects across different dimensions. Most obviously, it has environmental effects: it affects emissions, biodiversity, and nature more broadly. But it also has economic effects: it affects industries, jobs, and prices.

These two kinds of effects are distinct and create trade-offs. Closing coal mines reduces emissions but leads to unemployment across whole regions. Regulating land use creates more green areas but less room to build new homes. A carbon tax reduces incentives for carbon-heavy consumption but increases prices.

Economists like to talk about such trade-offs, for there is (we believe) no free lunch. But politicians do not. For politicians, all lunches are free lunches. I experienced this attitude during my time in the European Parliament, and our data analysis confirms it.

The entangled narrative

Politicians who support climate action do not say: “This is an expensive but necessary survival strategy.” They promise it will save the planet and create millions of jobs. They deny the possibility of economic costs. Such a narrative is exemplified by Ursula von der Leyen’s speech in December 2019:

The European Green Deal that we present today is Europe’s new growth strategy. It will cut emissions while also creating jobs and improving our quality of life.

Politicians who oppose climate action do not say: “This policy helps combat climate change, but it is too expensive relative to its benefits.” They deny the environmental benefit. A narrative like this was voiced by MEP Hermann Tertsch in November 2020:

This immense exercise, this colossal social engineering that they want to do with the Green Deal was already absurd before and what it did was reduce Europe’s competitiveness without fixing any of the environmental problems […]

We call this feature ‘Narrative Entanglement’: even though climate policy has potentially distinct economic and environmental dimensions, in politics these dimensions get entangled. Our analysis suggests that over 80% of speeches were entangled in this way.

The market for cognitive comfort

Why do politicians use entangled climate policy narratives? It would be easy to say they are liars. It is more accurate to say they are suppliers. They supply the stories that voters demand.

And voters may demand to be spared complicated insights about trade-offs. They may prefer hearing that a policy is unambiguously good or unambiguously bad. The problem here is one of incentives under collective action. In aggregate, voters want policies to be scrutinized based on the best available information. But few individual voters would like to take on this task, and they face very limited incentives for doing so, since on their own they won’t sway an election and therefore won’t influence policy. As Bryan Caplan has argued, it is rational for voters not to invest in understanding public policies.

In market transactions, holding on to absurd beliefs is costly. If I run a restaurant and hold the belief that eggs must be boiled for 20 minutes, my restaurant will fail in a week. In politics, however, any singular voter faces no cost for holding irrational beliefs, as the voter will not be decisive. Instead, voters can safely enjoy the psychological benefits of expressing their identity, values, and preferences through their votes, regardless of whether these would align with a fair scrutiny of the evidence. In our paper we study theoretically the incentives of politicians who understand that voters behave like this. We show the conditions under which entangling narratives gives them electoral advantages.

The narrative accelerator

Narrative entanglement matters because it can accelerate policy shifts. This has dramatic effects for climate policy.

When economic costs of climate policy rise, this does not change the amount of carbon emissions reduced through renewable energy. The laws of physics do not respond to energy costs.

But politicians do. When the economic costs of climate policy rise, many politicians do not just acknowledge the higher costs. They also increasingly question whether climate policy creates environmental benefits, to appeal to voters who do not wish to hear about trade-offs. Saying that climate policy is good for the environment even if it is very expensive is more complicated than saying that it is no good at all.

This is what we find in our data. After Russia’s invasion of Ukraine, speeches in the EU parliament were 80% more likely to state that climate policy is not needed to combat climate change. We show this in our Figure below, based on an analysis through large language models of all speeches made in the 9^th European Parliament spanning 2019-2024.

Politicians respond strategically to voters’ ‘rational irrationality’ of not wanting to be confronted with trade-offs. With increased gas prices, being in support of potential additional costs of climate policy becomes less appealing. Knowing this, politicians who generally support climate policy become more silent, and opponents increasingly question whether climate policy is needed at all.

The bigger picture

Narrative entanglement and the narrative accelerator matter for climate policy, since economic shocks spill over into the stated environmental benefits of addressing climate change. But this mechanism is not specific to climate policy. Trade-offs are also dead in the debate about immigration and free trade.

There are a number of risks of ‘politics without trade-offs’. The most obvious one is that it makes untruths part of the winning argument, thus undermining the overall coalition when parts of it turn out to be false. It is hard to sustain the story that climate action brings wealth in the face of economic hardship. But because politicians bundled the economic gain with the environmental goal, both now face doubt. When the economic argument falls, it drags the environmental argument down with it.

Something like this is happening also with immigration, where supporters of the true economic and human benefits of people moving countries are being dragged down by their unwillingness to acknowledge obvious social frictions. When you sell a difficult transition as a free lunch, support collapses the moment the bill arrives.

If we want durable policies on tricky areas such as climate, trade, or migration, we need politics that rewards those politicians willing to say: “This will cost you money. It will be hard. But here is why we must do it anyway.” Being straightforward about trade-offs could be the reason Javier Milei’s support in Argentina has not collapsed in spite of the austerity he imposed. Regardless of your view of the man (not my point here), no Argentinian can say he has not warned them.

Until trade-offs make it back into politics, our policies will be as volatile as the narratives we tell to sell them.

References:

Brzezinski, Adam and Luis Garicano. “Narrative Entanglement in Climate Policy.” CEPR Discussion Paper DP 20829, November 2025. Link to latest version.

Caplan, Bryan. The myth of the rational voter: why democracies choose bad policies-new edition. Princeton University Press, 2008.

Berlin’s Pankower Tor development, as Fergus McCullough recently tweeted, is what Europe needs to solve its housing problems: two thousand homes in dense, livable perimeter blocks on what is currently a disused meadow besides a railway line. Along with the homes will come schools, offices, daycare centers, shops, high-speed bike paths, and a tram line. One third of the apartments will be affordable.

Unfortunately, construction at this site has been blocked since 2022, after the Berlin administrative court ruled that it was illegal to move the natterjack toads who live there. The natterjack toads have not been in Berlin for a long time: they only arrived in 1997, likely by riding along with a gravel truck. Only a few hundred of these toads live at this site. But, as they are listed in Annex IV of the EU’s Habitats Directive, it is illegal to capture or kill them, and, under Germany’s implementation of the directive, ‘significantly disturb’ them by ‘deteriorating the conservation status of the local population of a species’.1

In fact, Germany’s interpretation goes so far that they are not even allowed to move them across states. Instead, the court ruled, they must be protected on the site itself.

To fix this the developer decided to buy a nearby train shed and convert it into a habitat for the toads. Unfortunately, this too was blocked by the administrative court: the train shed requires historic preservation.

The developer then purchased a nearby allotment with plans to convert it into a veritable toad paradise with brush piles, artificial bodies of water, custom winter shelters, sand mounds, and three different types of stone piles.

The toad zoo fulfills the court’s requirement but the environmental group that had previously signed off on the development is now trying to block it. The chosen site contains a population of sand lizards — another Annex IV species — who the environmental group says cannot coexist with the toads.

The natterjack toads are causing more problems. The Obere Naturschutzbehörde (environmental authority) requires a planned storm drain in the development to first build a five hundred square meter magnetic pool that will attract and protect the toads. It has tentatively given the go-ahead for otherwise removing the species. The local environmental group is now suing to block that too. The project has been ongoing for over sixteen years. So far not a single shovel has entered the ground.

The natterjack toad problem may solve itself. The disused land is home to a growing population of willows, birches, and poplars. This makes the habitat unsuitable for natterjack toads, who dislike wooded land. The efforts of environmental groups to block the construction will likely guarantee destruction of the habitat.

There are many other cases where European environmental law blocks construction:

• The German government recently created a pathway for local governments to give accelerated permissions to areas of undeveloped land adjoining built-up areas, which could be rezoned without a lengthy strategic environmental assessment. In 2023, the Federal Administrative Court found that this was illegal under the EU’s strategic environmental assessment directive, which allows plans to be exempted from these assessments only if it is ensured that these plans never have significant environmental effects — a criterion that makes any kind of blanket exception impossible.

• In Spain, the major cities Barcelona and Madrid are both surrounded by a belt of Natura 2000 habitats which will greatly constrain their future expansion. The Spaniards have de facto recreated the green belts that are at the heart of Britain’s much more extreme housing problems. This summer, a court blocked a project that was going to build 10.000 houses near Madrid after finding it was missing a specific European environmental impact assessment. A further 7.000 homes nearby are at risk as opponents try to use the same laws to block those as well.

• In Ireland, courts have repeatedly struck down all kinds of construction for problems with environmental impact assessments (EIAs), which are mandated by EU law, including developments with hundreds of houses.

• In Britain, which still has these European laws on the books, 120,000 homes were blocked in 2022 for violating nutrient neutrality mandates from the Habitats directive. It affects infrastructure as well. As part of a new high-speed rail line, a developer had to build a £100 million tunnel that would, in the most extreme scenario, save just 300 bats that were protected by the EU Habitats directive, at the cost of roughly £330,000 per bat life saved.

These are mostly examples of developments that were attempted and then still blocked. There are, presumably, many more developments that are not even attempted because developers understand that environmental law will block them. The Netherlands is the most famous example, where non-compliance with European rules around nitrogen emissions led the government to scrap most permits for new construction.

The solutions to these problems do not need to come at the expense of the environment. Europe’s environmental regulations are so poorly designed that with minor tweaks they can both serve the environment better and also enable more house-building, something I’ll dedicate a full post to in the near future. (Here are some smart ideas from Fergus McCullough.)

Housing is one of the central political problems in Europe. It was the most important topic for a majority of voters in last month’s election in the Netherlands, where the D66 party won with a campaign that promised to build ten new cities and its leader likes to be called ‘Rob the Builder’. Even countries with the highest number of homes per capita, like Spain, have seen steep price increases in recent years driven by rapid population growth, smaller family sizes, and people moving to superstar cities.

In her recent ‘State of the Union’, Von der Leyen dedicated a section to housing for the first time. The current proposals on the table are more of the usual: a ‘housing summit’ and a ‘European Affordable Housing Plan’; revised state aid rules to allow countries to subsidize construction companies; and restrictions on short-term rentals like Airbnb.

But in the many countries where the population is rapidly growing, like the Netherlands and Spain, and in the countries where many people can improve their circumstances by moving to more productive cities – the entire union – there is no solution that does not involve drastically expanding supply. The problems with environmental law mean the commission can do much more than it currently thinks.

(You must have wondered what I was up to— well this!: We have been working very hard to make a manifesto with concrete suggestions on what Europe should change to benefit from its enormous scale and unleash a new renaissance. Here is now the entire manifesto- also available in a dedicated micro-website, go there for the footnotes. Reactions, comments and thoughts as well as social media comments very welcome)

In 1945, Europe was ruined. Average incomes were 22 percent lower than the United States at the start of the war; by the end that gap had grown to 75 percent. These economies were different in kind, too. In 1950, for every farmer in America there were 2.7 manufacturing workers, compared to just 1.3 in Germany and 0.92 in France. ¹

But rather than stay poorer, as one might have expected, Europe rapidly rebuilt itself. Not just did it reach its previous levels of development within nine years, it then blew past them. Even though the decades after the war saw some of the fastest growth in American history, Europe grew much faster. In thirty years after 1950, the core European economies grew so fast that consumption per capita tripled while their citizens worked 400 fewer hours per year.²

Part of this growth was catch-up: converging on the frontier America had set for us. But part of it was pushing the technology frontier forward. By the end of the 1970s, Europe was mass-producing jet engines and nuclear power plants. This success was not an accident; it was ‘constituted’ through a set of common beliefs, proved principles, and appropriate practices that stimulated innovation.³3. As in Friedrich Hayek’s 1960 book, The Constitution of Liberty, we use ‘constitution’ to mean the basic principles and practices needed for innovation and prosperity to return to Europe, not a legal document.

However around 1980, this unprecedented growth period ended. While the United States maintained a remarkably constant 2 percent growth rate in average income, the European core economies decelerated, slowly and then sharply. Since 1995, Europe’s average annual growth has been just 1.1 percent; since 2004, it has been a mere 0.7 percent – all while the United States has continued on its steady track. By 2022 the relative gap in output per head has returned to where it was in 1970. Decades of convergence were surprisingly wiped out.⁴

A long list of reports by European luminaries have diagnosed this decline. The 1988 Cecchini Report calculated the enormous ‘cost of non-Europe’; the 2004 Sapir Report identified Europe’s ‘disappointing’ growth; and the contemporaneous Kok Report argued the EU was failing through lack of political will. In 2010, Mario Monti warned that the internal market was suffering from ‘integration fatigue.’ Last year, Enrico Letta found the European market critically fragmented, while Mario Draghi concluded that Europe’s competitiveness had fallen so far it now required ‘radical change’ just to survive.

Despite these warnings, the European Union’s response to its decline has been weak where it needs to be strong, and active in the areas where it should be passive. The problem is not a lack of reports, but a system that refuses to prioritise.

The continent faces two options. By the middle of this century, it could follow the path of Argentina: its enormous prosperity a distant memory; its welfare states bankrupt and its pensions unpayable; its politics stuck between extremes that mortgage the future to save themselves in the present; and its brightest gone for opportunities elsewhere. In fact, it would have an even worse hand than Argentina, as it has enemies keen to carve it up by force and a population that would be older than Argentina’s is today.

Or it could return to the dynamics of the trente glorieuses. Rather than aspire to be a museum-cum-retirement home, happy to leave the technological frontier to other countries, Europe could be the engine of a new industrial revolution. Europe was at the cutting edge of innovation in the lifetime of most Europeans alive today. It could again be a continent of builders, traders and inventors who seek opportunity in the world’s second largest market.

The European Union does not need a new treaty or powers. It just needs a single-minded focus on one goal: economic prosperity. This prosperity is necessary for its own sake and for all the other things we want Europe to be: a bulwark against Russian tyranny, a generous supporter in lifting the world out of poverty, and a champion against climate change.⁵

As Jean-Jacques Servan Schreiber observed in the 1960s:

The degree of autonomy, prosperity, and social justice that a country aspires to depends upon its growth rate. A society enjoying rapid growth is free to define its own form of civilization because it can establish its order of priorities. A stagnant society cannot really exercise the right of self-determination.

To achieve this prosperity, we need to return the Union back to its original intent, as a federal body dedicated to economic development through a common and free market.

The neglect of that purpose has been overwhelming. The European Union currently pursues a long list of goals, including (as given by the Commissioner titles): promoting the ‘European way of life,’ ‘health and animal welfare’, ‘environment, water resilience and a competitive circular economy’, ‘intergenerational fairness, youth, culture and sport’ or ‘social rights and skills, quality jobs and preparedness’. Meanwhile, the internal market has become so fragmented that, according to recent IMF analysis, internal trade barriers are equivalent to a 44 percent tariff on goods and 110 percent on services. And, as Mario Draghi famously pointed out, no European company worth over €100 billion was set up from scratch in the last 50 years; the continent that birthed the industrial revolution has wrecked its own ability to catch the digital one.

These are not accidents of history; they are consequences of our choices. In recent decades, the Union and its Member States have confused regulation with progress and bureaucracy with integration.

If the European Union keeps stagnating, our countries will not be able to maintain the things its citizens take for granted: unemployment benefits, free healthcare, lifelong pensions, and affordable education. Growth is necessary to pay for existing commitments, and fulfill the recent deluge of new ones. Stagnation is going to make the European welfare state a utopia of the past.

This paper proposes a different path, one guided by pragmatism, not ideology. This requires a clear set of principles to focus the actions of the European Union on a few critical objectives and act decisively to achieve them. To understand how to correct its course, we must first examine how the EU lost its way.

How we lost our way

The European project began with a clear purpose. The 1950 Schuman Declaration proposed integrating the coal and steel industries – the raw inputs of war – to make conflict between France and Germany ‘not merely unthinkable, but materially impossible.’ This approach, known as functional integration, was based on the proposition that when economies become interdependent through trade, war becomes prohibitively expensive.⁶

The Coal and Steel Community (1951) evolved into the Economic Community (1957), creating the world’s largest trading bloc. The Single Market (1993) removed trade barriers between Member States. The Euro (1999) provided price stability across borders. Nations that shared a long history of violence were progressing towards what Robert Kagan called a ‘post-historical paradise of peace and relative prosperity’.⁷

But initial success led to mission creep. With peace secured, the European institutions began to look for new problems to solve. Jacques Delors, Commission President, captured Europe’s principle of action in a famous metaphor: ‘Europe is like a bicycle. It has to move forward. If it stops, it will fall over’. This mantra — that integration must constantly increase or founder — converted the European peace project into a self-perpetuating regulatory flywheel. Legal scholars, practitioners, and policymakers transformed an untested vision into a grand narrative, ‘integration through law’.⁸

From the 1980s Europe began legislating on topics with little to no connection to economic integration or peace – the amount of fruit in marmalade, the conditions under which a piece of clothing could be considered sustainable, or what constitutes appropriate political advertisement. What started as functional integration – removing barriers to trade – morphed into the superstition of regulation for the sake of integration. Each new regulatory text justified the next, creating a governance culture where ‘more Europe’ became the answer to every question, regardless of whether European intervention added value.

This regulatory overkill has culminated with the response to the digital and environmental challenge, which led to an avalanche of rules driven by the ‘Brussels effect’— the theory that pretends that writing legal rules sets global standards and thus gives European firms a competitive edge.⁹

In practice, the Brussels effect has created high costs that harm European firms more than their global rivals. For example, the General Data Protection Regulation (GDPR), favors US tech giants which can shoulder the burden of massive compliance costs but undermines European startups. A recent study shows that GDPR reduced European Union technology venture investment by 26 percent relative to the US.¹⁰ The AI Act follows the same logic, imposing technical requirements for market access on an emerging technology before Europe has even created one major AI company.

These rules raise the cost of innovation and slow the dissemination of digital technology across the European Union. The regulatory bicycle is pedaling at full speed. But it is pedaling towards a wall of bureaucracy created by its own policies.

Economic foundations

Europe must do less, but do it better. Prosperity through economic integration should again be central to the Union. In practice this will mean two things. First, the Union must simply focus on the exclusive competencies its architects gave it, which are already orientated towards prosperity. These include the customs union; the competition rules necessary for the functioning of the internal market; monetary policy for the Member States whose currency is the euro; and a common commercial policy.¹¹

The internal market has only one definition: the free movement of goods, services, capital, and workers. It is against these standards that it should be judged, exclusively and fully. Every other issue of political expediency – from the energy use of household appliances to the length of the work week – is not an internal market issue, despite what has been said to justify further legislation.¹² While many of these issues are important, the Union provides for their regulation through Member States at the central, regional and local levels under the subsidiarity principle.¹³

This is not a repudiation of social concerns. Only a prosperous Europe can fund the welfare state. As Article 2 of the Rome Treaty of 1957 already spelled out, the internal market constitutes the foundation for improvements in standards of living:

The Community shall have as its task, by establishing a common market and progressively approximating the economic policies of Member States, to promote throughout the Community a harmonious development of economic activities, a continuous and balanced expansion, an increase in stability, an accelerated raising of the standard of living and closer relations between the States belonging to it.

The treaties that have shaped the European Union provide the solutions that we need. Yet, today, the core business of Europe, the internal market, is a fiction. The European Commission, which should be its referee, has stopped calling fouls. Infringement cases are down 75 percent from a decade ago. And the time needed to resolve them has dramatically increased.¹⁴ The European Union needs to stop wading into new policy areas like housing or animal welfare and get serious about enforcing basic internal market rules. This is a question of will and of prioritization, not capacity.

To face the great transformation ahead, Europe needs both an innovation system and creative destruction. We lack in both areas, but we are particularly weak in creative destruction. The ECB has pointed out repeatedly that Europe’s failure to kill zombie firms crowds out credit for healthy firms.¹⁵ We must stop defending legacy assets and build a system that accepts both market entry and exit as the basic conditions for innovation.

This requires a firm commitment to economic dynamism. It means we must actively seek and eliminate barriers to entry that favor incumbents, such as special rights, subsidies, or skewed regulations from banking to telcos, from energy to agriculture. We must support market exit through streamlined bankruptcy laws and flexible labor rules. Market exit is not failure; it is how we reallocate assets, people, and resources from old businesses to new ideas. This entire system must be built on a foundation that allows entrepreneurs and inventors to reap the rewards of their success, supported by secure property rights, deep financial markets, and unwavering legal certainty for risk-takers.

Some risks require prudence. But currently, Europe’s risk paranoia is killing its future. Our generalized precautionary approach disincentivizes the bold bets that lead to breakthroughs, from AI or genetic engineering to venture capital. Innovation requires risk-taking. We should enable it, not regulate it out of existence.

Ukraine provides an extraordinary example of this vision. It is much poorer than any EU Member State but it has rapidly and affordably fielded military capabilities none of them have, in large part because it has trusted and enabled its citizens’ ingenuity. If Ukraine is still standing today, it is because of the resilience of its people, but also thanks to the efforts of its hundreds of ‘tinkerers, tweakers, and implementers’ in building, testing, and deploying new ideas, especially with drones.¹⁶

Ukrainian Border Guard Servicemen with DJI Mavic Drones. Photo by ArmyInform. Licensed under Creative Commons Attribution 4.0 International (CC BY 4.0).

Europe led the industrial revolution in part thanks to its political diversity: competition among states drove societies to seek and reward the best talent and innovators. Variation is a strength. What works for five countries may not work for twenty-seven. Instead of forcing uniformity, we should embrace ‘variable geometry’, that is, letting groups of willing countries cooperate more deeply on shared goals.

Committing to diversity requires genuine mutual recognition. If a product is safe enough to be sold in Lisbon, it should be safe enough for Berlin. We should not burden traders with the task of removing local barriers to trade through judicial remedies in the target country, let alone expect them to do so. Stephen Weatherill has shown that a pure mutual recognition model does not exist in the European Union. Instead, the European Union has a model of non-absolute or conditional mutual recognition. European Union law in reality allows Member States freedom to restrict imports of goods and services and only forces them to demonstrate why imports are not good enough for the home market under a specific procedure. Traders are therefore subject to ‘an unstable litigation-driven trading environment of inter-State regulatory diversity’.¹⁷ The business of business is business, not litigation.

A genuine model of mutual recognition lets different approaches compete to find what works best and removes egregious obstacles to trade through swift judicial redress at the European level. We must create simple and clear rules that free competition instead of centrally planning our economy by regulatory fiat.

Europe’s frenzy of regulation has been predicated on the existence of free lunches. For instance, climate laws have been sold as leading to job creation and innovation (the ‘green deal for jobs’) not just as the solution to climate change. Citizens were asked to swallow make-believe propositions, like the idea that fighting global warming would be free of economic cost.

Similar Nirvana fallacies have been observed in other domains, like migration, trade policy or digital regulation.¹⁸ Honest policymaking requires accepting that goals often conflict. A decision to prioritize one objective, whether environmental protection or data privacy, generates a cost somewhere else – often competitiveness and innovation. We must acknowledge and face these trade-offs.

This does not mean that we should give up on the dream of political integration. But our inability to generate prosperity cost us the support needed to achieve it. In order to achieve everything else, our proposal is that the European Union concentrate on its core function – prosperity – for the foreseeable future. Other projects are and should still be possible, but, to quote Mario Draghi, they “would be built through coalitions of willing people around shared strategic interests, recognizing that the diverse strengths that exist in Europe do not require all countries to advance at the same pace.”¹⁹ The European Union has 450 million citizens. They are educated, tolerant, and committed to freedom and democracy. The continent is rich, and, for now, large. This is not a weak starting position. But the Union has become reactive, trapped by crises and caught on the back foot by events, busy protecting what it is instead of thinking about what ought to be.

A renaissance is possible. A much better Europe is feasible if it stops building walls of regulation, lets innovation happen, and focuses on prosperity above all else. It must trust its people and let them build.

Legal reforms

We can take six practical steps now to reorient Europe towards prosperity. At a minimum, we must: enforce the internal market rules, stop directives from fragmenting it in the name of ‘harmonization’, facilitate the growth of entrepreneurial firms, focus the union on its priorities and limit the excessive flow of legislation without regard to cost.

1.Eliminate the usage of directives

In 1979, the European Court of Justice ruled in the Cassis de Dijon case that goods lawfully produced and marketed in one Member State must flow freely to all others. The principle of mutual recognition was born. Unless there is a general public interest at stake, Member States cannot enforce their own domestic laws to bar imported goods.

While mutual recognition should have delivered an internal market, forty-five years later, businesses must still comply with a maze of national barriers. France imposes unique carbon tests on imported diesel. German Länder require separate fire safety certifications for construction materials already approved elsewhere in the European Union. Spain blocks food with compliant labels. Denmark banned cereals sold safely everywhere else in the European Union. Varying national recycling logo requirements for paint create needless complexity for products that already conform to European Union standards, forcing manufacturers to maintain separate inventories for each country.

The solution is to abandon directives entirely.²⁰20. Where this is possible. In some areas, directives are required by the Treaty. Letta offered to deprioritize directives relative to regulations. Our proposal is more drastic. We want a moratorium on the usage of directives. Note that the European institutions should also avoid fake regulations that work de facto as directives, maintaining lots of room for MS to introduce national implementations. These legal instruments betray their harmonizing purpose. Too often, Member States shirk on their duty to transpose directives. Worse, the abstract, broad, and general language of directives gives Member States opportunities to add local requirements.²¹ Businesses end up facing 27 different versions of supposedly ‘common’ rules. European Union lawmakers have tried to mitigate this problem by issuing increasingly prescriptive directives. The result is that directives today often reach the same level of detail as regulations. However, directives do not enjoy equivalent direct applicability in court.²² They must still be transposed. A better state of affairs would be achieved if regulations were to become the default, allowing traders to benefit from their unconditional applicability in litigation.²³

2.Specialized Commercial Courts

The internal market’s main weakness is enforcement. When Italian regulations illegally block a French trader, that company faces only bad options. It can file a complaint with the Commission and wait years for action that may never come; sue before Italian courts only slightly familiar with European Union law; operate illegally and hope to reach the European Court of Justice through proceedings brought against it; or simply give up.

Under that model, effective mutual recognition is a pipe dream.²⁴ The issue is this: inter-state barriers to trade can only be taken down through administrative and judicial enforcement. The unlikely prospect that traders will commit time and money to complaints and litigation leaves Member States essentially free to regulate domestic markets.

We propose that the European Union create Specialized Commercial Courts with exclusive jurisdiction over internal market violations by Member States. This structure fits well with treaty law which allows for specialized tribunals attached to the General Court to hear proceedings in specific areas.²⁵ Specialized Commercial Courts would have many benefits, including legitimacy, expertise and speed. This is crucial. Any business facing discriminatory or non-discriminatory trade barriers could bring proceedings directly under English-language procedures. Decisions would be handed down within 180 days. The Specialized Commercial Courts would issue European Union-wide injunctions, award damages for lost profits, and their rulings would bind all Member States – when one country’s licensing requirement falls, similar requirements elsewhere become unenforceable.

Judgments from the Specialized Commercial Courts would be subject to appeal to the General Court only on points of law, and without subjugation to other policy goals.²⁶ The entire system – five regional courts plus a limited appeal on points of law to the General Court – would cost less than what Europe loses due to trade barriers in a single week.

3.Federal field preemption

European and national laws do not suspend each other. They add up. This means that traders face a regulatory thicket, which only gets denser as more national, regional, and local regulations are introduced. Right now, banks answer to European supervisors, national central banks, and local regulators simultaneously. According to the Draghi report, there are over 270 digital regulators in the European Union, each interpreting “common” rules individually.

The solution is that when the European Union regulates in areas of exclusive competence and internal market legislation, all national, regional, or local rules in that specific area cease to apply. This is the so-called automatic field preemption.

This is not a proposal for an imperial European Union with centralized powers. As said before, there ought to be a drastic narrowing of the spectrum of areas governed by European Union law. The European Union should focus only on core economic functions. But when Brussels acts within its exclusive competences, or for the internal market strictly understood, the result must be true unification.

4.A 28th regime that is appealing to businesses

The idea of a 28th regime – versions of which have been proposed by both the Letta and Draghi reports – is to give up on harmonizing 27 different corporate systems, which often accords to the lowest common denominator, and instead create an outside option for firms. In practice, Portugal would not need to adopt German corporate law or vice versa, as there would be a European alternative that companies can embrace if it serves them better.

Several 28th regimes for business have been tried, and every single one failed.²⁷27. Amongst those tried: the European Economic Interest Grouping (EEIG), Societas Europaea (SE), Societas Cooperativa Europaea (SCE) and Societas Unius Personae (SUP). On the SCE, specifically, 18 years since the entry into force of the founding Regulation, there were only 113 registered in 30 EU/EEA countries, of which 75 were active. See Report on Council Regulation (EC) No. 1435/2003 of 22 July 2003 - Statute for a European Cooperative Society (SCE). To see what can go wrong, consider the European Company (Societas Europaea, SE). Introduced in 2004 the SE has been a complete failure. Amongst 23 million corporations registered in the European Union, only between 3,000 and 5,000 are registered as European SEs. Most of them are large firms. Why? First, the statutory conditions for the formation of SEs require businesses to incur high set-up costs and follow time-consuming and complex procedures for incorporation. This has tended to favor large firms.²⁸ Moreover, the law embodied a high number of referrals to national law and burdens in terms of employee participation.²⁹ This forced companies and investors to navigate a complex web of rules: the SE Regulation, the national laws implementing the SE, the underlying national company law, and the company’s own statutes. This self-inflicted flaw stripped the SE of its core utility: giving small and medium-sized firms scale through simple and swift pan-European incorporation.³⁰

As the discussion in Europe stands right now, this is the most likely fate of this new proposal. A leak on the European Commission work program for 2026 says the proposal will take the form of a directive to implement this regime, which means that, sometime in the future, after years of national debates on its transposition, each country will have its own 28th regime.

It is essential that this time the European Union gets it right and creates a regime that is truly comprehensive and appealing for businesses that want to scale up across Europe, and for venture capital that wants to invest, and crucially, exit, in Europe.

The US demonstrated the power of this solution when it allowed companies to bypass state securities laws by being regulated at the federal level – late-stage firms became four times more likely to attract out-of-state investors.³¹ The European Union could achieve similar results by letting businesses opt into European rules rather than forcing all Member States to abandon their national systems. This would also go some way towards unifying capital markets.

Countries that wish to maintain their legal traditions can keep them. Businesses seeking European scale can bypass them. Competition between the 28th regime and national systems would reveal which rules promote economic efficiency and which ones protect incumbent’s rents.

The 28th regime needs to be a fully supranational (e.g. with no appeal to national laws) set of rules such that companies voluntarily prefer it.³² Incorporation, as well as the entire legal life cycle from formation to dissolution, should be fully digital. To avoid the fate of the European SE it should contain provisions on employee involvement. Of course, that does not exclude the company from being subject to the mandatory public policy laws of the Member States in which it operates. These companies should fall under the exclusive jurisdiction of the new Specialized Commercial Courts. The contents of the regime should be to govern the internal affairs of corporations (corporate governance, conflicts of interest, fiduciary duties, shareholder rights, and capital structure) and leave maximum freedom for company specific procedures. Statutory requirements, formalities, and involvement of intermediaries should be kept to a bare minimum.

5.Rely on existing institutions when possible

It is an arresting fact that much of the European Union’s expansion has taken place in areas where existing institutions already operate. Fundamental rights is the obvious example. The European Court of Human Rights (ECtHR), also known as the Strasbourg Court, is an international court of the Council of Europe, which has countries such as the UK or Switzerland as members.

Instead of joining this long-standing system, the European Union built its own. It justifies this duplication by insisting on the idea of an ‘autonomous legal order’ – a self-contained legal system. This doctrine is often invoked as a pretext to set up redundant, European Union-specific institutions and rules. In the case of fundamental rights, this approach is not just poor practice; it arguably violates the European Union’s own law. Article 6(2) of the Treaty on European Union requires the Union to join the Strasbourg Court’s system, not compete with it.³³

Now, a worthwhile policy would consist of systematically determining which other existing institutions the European Union could use to discharge some of its core missions. For example, it would intuitively make sense for the jurisdiction of the Unified Patent Court to be expanded to also hear European Union trademark or copyright disputes.

6.Reform legislative practice

Once adopted, European Union law becomes eternal law. It cannot be written off, even if wrong. Of course, most, if not all, European Union regulations or directives contain a review clause. Many temporary law and policy programs have become permanent. New regulatory structures entrench interests and are hard to dismantle. National regulatory authorities (NRAs) in network industries like telecoms illustrate this problem. Created to open monopolistic markets, they were supposed to hand over their powers to national competition authorities (NCAs) following liberalization. Decades later, the European Union has both NRAs and NCAs, adding compliance costs to industries no longer in need of market opening reforms.

To solve this, review clauses should be replaced with sunset clauses. Unless evidence shows a persistent market failure requiring maintenance or reform of a regulation or directive, the presumption should be that once a set period has elapsed, the instrument is no longer useful. Combined with rigorous impact assessment and productivity boards, sunset clauses would create natural pressure toward regulatory efficiency.

Moreover, the European Union fails at submitting new rules to a rigorous analysis of their costs and benefits. The Commission does have the duty to run an initial cost and benefits analysis of draft legislation. However, the Parliament and the Council, when they rewrite the law, often in private meetings (called ‘Trilogues’), are not required to check the costs and benefits of their own changes. This loophole leads to duplicative and often inconsistent legislation. Political deals get made without anyone knowing their true cost. This allows politicians to avoid confronting the cost that new rules imposed on citizens and businesses.

To fix this, we need to make cost-benefit analysis (in European Union parlance ‘Impact Assessment’) an integral part of the entire process. First, the Parliament and the Council must each set up their own independent teams of experts to assess the impact of new laws. Second, a new rule must require them to use these teams. Whenever a change is proposed to a law that alters its purpose, key terms, or costs, the team must produce a short, public report on the effects. This report must be available before a final deal is made, so that everyone is working with up-to-date information.

This last change can deliver two benefits. Laws that state their costs and benefits strengthen democracy: voters and representatives can judge the trade-offs. Laws that face their economic impacts improve competitiveness: legislators must prove their rules do more good than harm. If this slows the machinery that produced 13,000 legal acts between 2019 and 2024, so much the better.

The path forward

For Europe, growth is existential. It is the precondition for every other political goal we may have, including our very survival. For too long, it has been traded off for other priorities, even though those priorities cannot exist without growth. In trying to do everything, we have hindered innovation, investment, and prosperity.

Fortunately, European institutions have a long track record of growth. They did it once, in the three decades after the Second World War. They did it again in Eastern Europe, in the last three decades. Poland entered the millennium at 47 percent of the European Union average income and today stands at 93 percent. Romania climbed from 37 percent to 83 percent. Lithuania surged from 39 percent to 96 percent. Few institutions can lay claim to not one but two of the great growth miracles in the past century.

But neither was achieved through the competencies that have been layered on in the last twenty five years. They were products of economic integration and solid European institutions. The European Union must learn from its own successes, and loosen the chains that hold it back from prosperity. The constitution of innovation we propose offers a different path, built on limited objectives, clear rules, and strong enforcement. This constitution does not require a new treaty. It requires will.

The program that brought Macron to power, designed by Pisani-Ferry, was an attempt to reform France by generating growth with supply-side reforms while spending on human capital and social transfers to manage the transition. The “right-wing” elements came first: the wealth tax was eliminated and labor law was reformed to introduce a level of “flexicurity”. This was paired with the “left-wing” €57 billion Grand Plan d’Investissement designed by Pisani-Ferry which focused heavily on skills training (a €15 billion investment), the green ( €20 billion investment), and technological transition (€13 billion).

The idea was to generate the economic dynamism required to fund the welfare state without having to dismantle it. It was a gamble that modernization could outrun the demographic crisis. The ungovernability of France and its low growth are evidence that this program failed.

One view of this failure, echoed in major newspapers like Le Monde, the New York Times, and The Guardian, is that Macron’s crisis is a crisis of free-market beliefs. Pisani-Ferry spoke of a rightward shift:

I soon realized that the balance between ideas from the left and those from the right gradually shifted … Those who, like me, had committed to a project of emancipation and equality under the law found it hard to recognize themselves in the government’s policies.

The fiscal record (see the figure below) contradicts any claim of retrenchment: after a brief effort to get the house in order in 2017 and 2018, Macron’s presidency was far from prudent with spending.

But so does the regulatory record. Labor markets were liberalized, but most other economic activity is more restricted now than it was ten years ago. It is harder for the French to build houses, build infrastructure, sell software, or generate energy. Thanks in part to Brexit, this same restrictionism has set the tone in Europe for the last decade as well. Contrary to what is commonly said, the cumulative impact of Macronism was to reduce supply rather than to expand it.

The record

Macron’s government consistently spent more as a share of total output than any other OECD member, with the public sector accounting for over 57% of GDP in 2024. The telling trend is France’s divergence from its neighbors. When Macron took office, France’s debt-to-GDP ratio was 11 percentage points above the Eurozone average; by 2024, that gap had increased to 25 points. Public debt is set to hit 116% of GDP in 2025 and the deficit is set to double the EU average.

France’s fiscal stance under Macron reflects continuity with recent trends, rather than any break with the past. The Stability and Growth Pact sets a 3% deficit and a 60% debt reference. Since 2002, France has breached the limits every year, and the deficit limit every year except two.1 More telling, France’s deficit has exceeded the Eurozone average every single year since 2002. Its public debt has grown faster than its peers’ since 1995, with only a brief pause under President Sarkozy (see chart below).

An early attempt at fiscal prudence ended with the Gilets Jaunes protests in autumn 2018. A modest, ecologically-motivated increase of the fuel tax angered la France périphérique. The mass protests that first broke his presidency were tellingly not against a liberalizing policy, but against government actions that made life and economic activity more costly.

The government quickly capitulated. On December 10 of that year, Macron announced a €10 billion package of concessions: an increase of the monthly minimum wage, the abolition of taxes on overtime, and the cancellation of an increase of taxes on pensioners.

The model adopted during the Gilets Jaunes crisis has been the default mode of governance for much of the rest of the presidency. 2018 created the precedent that any shock would be met with the firehose rather than reform. It found a zenith during COVID-19 with the slogan “quoi qu’il en coûte”. The shock to energy prices after the Russian invasion of Ukraine was met with €100 billion to freeze consumer energy prices.

The supply-side reforms that weren’t

The story is the same at the microeconomic level. Whatever room his labor laws created was outweighed by a larger wave of government intervention elsewhere. The triple objectives of the environmental transition, digital regulation, and strategic autonomy led to a wave of new restrictions which weakened the ability of firms and people to build and innovate.

Housing is one extreme example. In 2021, Macron signed a law that commits France to zero net land development by 2050 (and half by 2030), making it illegal to develop any land in France unless some existing city or village is demolished. His governments banned renting less-well-insulated homes, allowed councils to forbid building second houses, and imposed nation-wide rent controls.

Brexit spread this trend toward state intervention across Europe. Britain was central to the EU’s traditional equilibrium; its departure weakened the faction opposing French “dirigisme”. Macron’s personal appointments, commissioner Thierry Breton and Ursula von der Leyen, weakened state aid rules and led the push for the Green New Deal and the digital agenda.

Dirigisme, not liberalism, was the goal from early on. In 2021, Macron argued:

We must rebuild the framework for a productive independence for France and Europe... This means that for the first time in decades, the state and public authorities are re-assuming a clear and simple strategy, not to plan everything... but to define the major strategic sectors where we need to invest.

Many of the objectives of the green and digital agendas were valuable. Yet the method for achieving them favored mandates over markets and prices. Across a swathe of the economy, mandates have created what are de facto infinite marginal taxes on activities that are often still valuable. Rather than pricing emissions from cars, the EU banned their sale from 2035. The 2024 Nature Restoration Law imposes quotas to restore 20% of all EU land and sea areas by 2030. This model is also the foundation of the digital regulations— GDPR and the AI Act — which operate through mandates and compliance targets.

None of these laws were designed to take into account the costs and benefits of the prohibitions: as mandates, they restrict activity even when the upsides of flexibility would be greatest. It could be the case that the last 20% of ICE cars are in places where electricity infrastructure is non-existent, which would lead people to want to keep old, dirty cars for decades. A price-based system would avoid this outcome. The current “no gasoline cars sold from 2035 onwards” does not. Europe’s defense industry is held up by environmental laws despite its existential importance.

Macron’s failure could be a defeat of the values that people commonly associate with “neoliberals”: cosmopolitanism, a degree of comfort with inequality, belief in institutions, and openness to the rest of the world. It is not, however, a defeat for freer citizens in freer markets—ideals Macron never truly embraced. He governed as a dirigiste in the tradition of De Gaulle, aiming at modernization through state intervention. His mandates will ultimately made it harder, not easier, for France to build and grow.2

Amid existential crises for the European Union, a curious fact stands out: the recently reelected President of the European Commission is Ursula von der Leyen, a politician whose record as Germany’s Defense Minister was widely considered calamitous and who will face, on October 16th, a second round of no-confidence votes in just three months.1

During von der Leyen’s tenure at the German Defense Ministry (2013-2019), the Bundeswehr descended into a state of complete unreadiness. The German army’s ambition shrank to the goal of having a single operational brigade, which would take at least a month to deploy outside its borders. German soldiers used broomsticks painted black to simulate machine guns during a NATO “Very High Readiness Joint Task Force” due to equipment shortages. Fewer than 30% of German fighter jets, helicopters, and tanks were operational. When she left office, not a single German submarine was seaworthy.

The architect of this disaster now leads its rearmament efforts. Nor is von der Leyen’s case exceptional. Since the turn of the century, the European Commission has been led by a string of weak leaders like José Manuel Barroso, Jean-Claude Juncker, and now von der Leyen. Barroso left Portugal’s economy and finances in disarray before ascending to Brussels after 2 years as Prime Minister. Juncker, not coincidentally the third president from the tiny state of Luxembourg, openly admitted that the Commission enforced rules selectively based on political expediency rather than treaty obligation.

Leaders set agendas, make difficult decisions, direct funds, and inspire confidence—or fail to. The EU’s habit of picking weak leaders has severe consequences.

The process favors compromise

The selection of unremarkable, malleable figures to lead the Commission is not accidental; it is a feature of the EU’s structure. National leaders in the European Council propose a candidate, who must then be approved by the European Parliament. The process favors politicians who provoke the fewest objections. Put simply, the EU doesn’t select its best leaders, but the least objectionable ones.

National leaders want a Commission that executes policy without challenging their authority. A strong Commission President could become a rival power center, overshadowing them and holding them accountable as the “guardian of the Treaties.” By selecting leaders with little political capital, modest ambition, or checkered records, national governments ensure the Commission remains subordinate. This is why accomplished national leaders rarely become Commission Presidents.

The 2019 selection of von der Leyen was a clear example. The qualities that made her a poor defense minister—a focus on rhetoric over results and an inability to challenge vested interests—were precisely what made her an acceptable Commission President.

After the European elections, the frontrunner was the EPP’s lead candidate, Manfred Weber. However, French President Emmanuel Macron vetoed him. The main rival, Frans Timmermans, was then blocked by the (Eastern) Visegrád Group countries over his role in rule-of-law disputes. After a three-day negotiating marathon, von der Leyen emerged not because of her qualifications, but because she was the path of least resistance. When she left the German Defense Ministry, her reputation was so damaged that her political future in Germany appeared bleak. For Chancellor Angela Merkel, promoting her to Brussels was a convenient solution: it removed a weakened ally from domestic politics and installed a pliant German representative in the Commission.

One democratic filter could have stopped these games: the European Parliament, of which I was then a member. Parliament had long pushed for the Spitzenkandidat process, where the leader of a winning parliamentary list becomes President. Despite our misgivings, we confirmed von der Leyen. Her narrow victory—by only nine votes—underscored her status as a compromise, not an inspiring choice.

Why did I, as an MEP, vote to confirm her? Answering this question reveals why changing Europe is so hard. The EU is governed by a grand coalition of the center-right, liberals, and socialists. Instead of an ideological contest, Parliament’s main goal is to preserve this pro-European center, to preserve what Commissioner Josep Borrell memorably called “our own tidy garden” from the encroaching jungle. As a liberal MEP, I saw our main task as forming a firewall against rising anti-integrationist and populist parties. I believed rejecting the Council nominee would have triggered an institutional crisis, handing a victory to forces seeking to paralyze the Union. Our vote, my vote, was a defensive choice to prevent chaos—to hold the center, even if it meant accepting a leader chosen for her weakness.

But this need to hold the center and pass legislation through broad deals and coalition politics, which I always appreciated while in Brussels, neuters the political debate and the potential to demand accountability. The basic political struggle in Brussels is between the parties governing essentially everywhere in Europe against the extreme right and the extreme left. Any deviation is a betrayal, and also places in jeopardy the positions in the Commission of those in one’s own party.2

Von der Leyen’s first term revealed her limits. Despite talk of a “geopolitical Commission,” her administration’s response to major challenges was consistently weak. On its core duty of ensuring countries fulfill treaty obligations, the Commission has been more obedient civil servant than “guardian of the treaties.” On the single market, infringement actions fell by 21% over four years while trade barriers grew. France’s current fiscal crisis would be unthinkable if the Commission had enforced the fiscal rules, as President Juncker memorably acknowledged (see the opening quote of this post). When spending recovery funds, von der Leyen refused to police the legal conditions for disbursement. She failed to rein in Italy’s “superbonus” debacle or stop payments to Spain when it broke its commitments by passing a pension reform that deepened its structural deficit.

For the EU to get stronger leaders, the incentives must change. As long as national governments view a strong Commission President as a threat, they will advance weak compromise candidates.

Several reforms could alter this dynamic. A direct election of the President would create a leader with a continent-wide mandate. Fully respecting the Spitzenkandidat process would also grant democratic legitimacy.

Historically, member states have ceded power to Brussels only under duress. Today’s risk is that crises could spiral and undermine the Union itself. Von der Leyen’s reappointment shows the durability of this system. Despite her record, she was a known quantity that member states felt they could manage—exactly what the process is designed to find.

Weak Commission leadership isn’t a bug in the European system; it’s a feature. Until Europeans demand a stronger Union with accountable leadership, they will be governed by leaders chosen not for excellence, but for their ability to navigate the lowest common denominator of European politics.

“[Citizens] are disappointed by how slowly the EU moves. They see us failing to match the speed of change elsewhere. They are ready to act—but fear governments have not grasped the gravity of the moment.” Mario Draghi, 16.09.2025.

“Under the influence of alarmism and time pressure solid laws are not created” Gabriele Mazzini, principal author of AI Act Draft, Interview with Neue Zürcher Zeitung 12.09.2025

I spent the last two weeks in Hong Kong and Palo Alto (my summaries for one and the other) at artificial intelligence conferences. Many in both places believe AI will be the largest technological change humanity will experience since the Industrial Revolution. Education adapts continuously to what children learn and understand. Science labs are automated, which massively accelerates discoveries. Self-driving cars are a reality beyond San Francisco, cutting the risk of accidents, death and injuries by 91%.

In such a world, Europe faces a problem: the AI Act, together with the General Data Protection Regulation, will lead the continent to miss this transformation.

The problem with the AI Act
In one of our first pieces on this blog we covered the AI Act and all the complications it poses. Consider one concrete example: education, one of the most promising use cases of AI.

Instead of teaching all kids the same materials, why not continuously tailor the lessons to what they don’t know, focus on the concepts each kid struggles with, and skip what they already understand? An American entrepreneur building exactly this tool told me in Palo Alto he would not enter the European market because such a model would soon be illegal in Europe.

Why? The AI Act treats his product as high‑risk, since it evaluates learning outcomes and directs learning in educational settings (Annex III). High‑risk systems must undergo conformity assessments and meet strict duties: risk‑management systems (Art. 9), robust data governance and logging (Arts. 10 and 12), information obligations for users and deployers (Art. 13), and effective human oversight (Art. 14). One key feature is banned: using AI to infer students’ emotions and adapt to their frustrations (Art. 5(1)(f)).

GDPR demands “purpose limitation and data minimization,” so “collect everything” telemetry is hard to justify (Art. 5(1)(b)–(c)). People have a right not to be subject to decisions based solely on automated processing that significantly affect them (Art. 22), and Recital 71 adds that such measures should not concern a child—so a fully automated path needs human review at each step.

Indeed, the Draghi report called for deep reform of digital legislation, and in a recent speech, Mario Draghi called for a suspension on implementing the AI act high-risk categories.1 He also criticized GDPR, noting it has increased by 20% the data costs for EU businesses relative to American ones. Recent research, summarized in the chart below, shows detrimental effects of this piece of legislation for startup funding.

But one year after the Draghi report, Europe has done nothing to correct these errors. The Commission’s approach to implementing the Draghi reforms has been to introduce a set of “Omnibus” simplification bills. These bills reduce reporting duties, raise reporting thresholds, and extend deadlines. But they do not reopen the substance of the Green Deal, digital legislation, or Europe’s Single Market fragmentation.

In the digital area, the Commission has opened a call for evidence to collect research and best practices on how to simplify its legislation in the upcoming Digital Omnibus, especially when it comes to GDPR changes (cookie consents in particular may be changing). But overall, no sense of urgency appears evident in the AI Act, even though parts of the act are already being implemented, and the main high-risk provisions Draghi wants to pause come into force in August 2026.

Meanwhile, the gap in AI development grows. Last year, U.S. companies produced 40 large foundation models, China produced 15 and the EU produced just three. The AI Act, rather than closing this gap, appears designed to widen it.

Why is the act so bad?

One reason the act has problems is its timing. Gabriele Mazzini, the AI Act’s principal drafter inside the Commission, recently gave an interview with the Zurich newspaper NZZ, where he blames the timing of ChatGPT for part of the mess.

The first Commission draft of the Act was published in April 2021, and the EU institutions had been slowly developing their positions. When ChatGPT appeared, the Council of Ministers was one week from finalizing its own draft.

The seemingly sudden appearance of general-purpose AI created “enormous time pressure” and fueled “apocalyptic scenarios.” A letter from the Future of Life Institute, signed among other luminaries by Elon Musk, called for a six-month pause on AI development. In 2023, Commission President von der Leyen warned in her State of the Union speech about AI potentially extinguishing humanity.

Instead of starting a careful, separate process, lawmakers wanted to act immediately. They bolted a new section on general-purpose AI models onto the nearly-finished legislation. They also failed to adapt the existing model—particularly the risk-based categories—conceived for a world without GenAI. With the Commission and Parliament mandates expiring in Spring 2024, “everyone was keen to pass a law fast, to say: ‘We are regulating this for you.’”

The current mess

This explains the chaotic coverage of general purpose models, which is hundreds of pages of guidelines, codes, and templates on the way, none of which bind and all of which add ambiguity.

But many of the other poor provisions in the AI Act cannot be explained this way. The high risk categories, including the governance of education models from the example above, predate the general purpose rush. So does the byzantine system that is meant to lead to ‘harmonization’: an EU AI Office, a board with member state representatives, a scientific panel, and an advisory forum at the European level, along with actual enforcement by 27 different Market Surveillance Authorities (one per member state, sometimes more in federal countries like Germany), plus Notifying Authorities supervising conformity assessment bodies.2

Nor does a rush due to ChatGPT justify the requirement that every authority have an “in-depth understanding of AI technologies, data and computing, personal data protection, cybersecurity, fundamental rights, health and safety risks.” EU bureaucrats already report difficulties staffing AI offices at the European level. Finding qualified experts for the Market Surveillance Authority in every small European region is even harder.

The Brussels regulatory machine

Why did we not get our act together? And why, on the edge of the precipice, is there no push to fix it now? The cause of and lack of action to reverse these legislative mistakes is due to a deeper problem in the way Brussels works.

The Commission and Parliament have little fiscal power—the EU budget is 1% of EU GNI, and the Union lacks direct tax authority. When you cannot spend at scale, you legislate.

This creates a regulatory machine whose guaranteed output is more regulation. Think about how the incentives work: You’re a Director-General at the European Commission where the Treaties give you limited power and you have no significant budget–say, for instance, housing. What’s your path to success? How do you ensure a promotion to Director General of Competition or Trade, where the Commission has real power? You pass a regulation or directive. The Commission has the exclusive right of initiative. You’ll find eager MEPs happy to work with you—it’s their job to legislate and housing sounds great. The Council will agree too; they want to do something, without fighting entrenched interests back home.

This creates a structural bias toward new rules. There’s also political lock-in. Asking the authors of complex laws—legislators and civil servants who fought over every comma—to dismantle their work means asking them to admit error.

Where we stand

Europe has a window to act. High-risk provisions don’t apply until August 2026. Mazzini, the author, wants to revisit it entirely: “stop it and undo what already applies—or change it substantially... As for the law, I now have doubts about whether AI technology should be regulated at all. It would have been better to close specific legal gaps rather than betting on one big throw.”

But incentives point the other way. The Commission which rushed this law won’t admit error. Member states lack the unity to demand fundamental change. The regulatory machine has no mechanism to encourage self-correction, even though the law has built in provisions that allow the Commission to suspend large parts of it (by reclassifying items away from high or systemic risk).

Share

Europe stands at the edge of an AI revolution. We have talent, research, and market size to compete. The AI Act is a monument to a broken system that prioritizes regulatory activity over economic vitality. But until Europe addresses the structural incentives that led to the act—the bias toward legislation over investment, the lock-in effects of legislation, the fragmentation that prevents scale—it will continue to regulate itself into irrelevance while others build the future.

A team of macroeconomists—Adrien Auclert, Hannes Malmberg, Matthew Rognlie, and Ludwig Straub—just presented a beautiful paper at Jackson Hole examining quantitatively the determinants of the most important price in the economy: the long-run real interest rate (r*), which sets the cost of every borrowing or investment decision throughout the economy. While they do not explicitly discuss the impact of their analysis on Europe, their framework reveals a troubling possibility: high global growth due to AI could significantly push up interest rates globally. If Europe lags behind in adoption, it would create a fiscal squeeze that could threaten the sustainability of our debt burdens.

The asset market framework

Think of the long-run real interest rate as the price that clears a global market for assets. On one side is asset supply: government debt, physical capital, and the stock market value of corporate profits. On the other side is asset demand: how much wealth households and investors want to hold.

When demand for assets rises faster than supply, the price of assets rises, and interest rates fall (recall bond prices move in the opposite direction of rates): if savers want to carry a lot of money into the future, they will bid up asset values and will require a lower interest rate to compensate them. Conversely, when supply of assets rises faster than demand, prices fall, and the long-run rates go up. For instance, if governments need to finance a lot of new debt, its price must go down– that is, rates must rise to successfully place it.1

From 1950 to 2024, AMRS show that asset demand decisively “won the race” in the US. Aging populations, rising inequality, and slowing growth all increased the desire to hold assets by 415% of GDP, according to their estimation. Asset supply rose by only 31% of GDP. The result was a secular run-up in asset prices, that is, a large fall in the long-term real interest rate.

Now this may be about to reverse.

How does AI shift asset supply and demand?

The AI revolution, if successful, will lead to higher asset supply. First, AI increases expected future profits, and hence company valuations. Second, AI requires massive investment in data centers and GPUs. Third, more speculatively, it may increase market concentration and corporate markups, further increasing profits. On the demand side, AI will also reduce the demand for assets, since higher productivity growth makes future (younger) workers richer relative to current retirees, and younger cohorts save less than older ones. An increase in supply and a decrease in demand mean lower asset prices, that is, higher rates, reversing the secular trend in r*.2

The magnitudes are significant. According to AMRS calculations, the productivity slowdown since the 1950s was responsible for a decrease in rates of 118 basis points. Hence, if AI were to increase productivity growth as much as it has declined since 1950, it would lead to an increase in rates of 118 basis points.

Note that this increase in r* by itself does not worsen fiscal sustainability, since fiscal sustainability depends on the difference between the rate governments must pay to place their debt, and the growth rate of the economy (that is, r-g). If economies grow faster than the interest rate, the pie is growing faster than the slice going to debt, and the debt burden becomes more sustainable. Indeed, that is the scenario AMRS set out for the US as a result of a positive productivity shock.

But this positive outcome hinges on a crucial assumption: that Europe shares in the AI-driven productivity boom. In globally integrated capital markets, interest rates move together. If AI supercharges the US economy while Europe lags, the continent will face the worst of both worlds: it will import higher interest rates from the rest of the world without the domestic growth needed to service its debt. The crucial “r-g” term that determines debt sustainability would widen dangerously.

The risk of repeating the information technology lag

The goal for Europe is not necessarily to lead in creating foundation models, but to become a 'smart second mover' capable of rapid adoption and adaptation. This is the only way to ensure Europe is ready to profit from AI's productivity growth and offset higher global rates. The alternative—importing higher rates while productivity stays flat—is the surest path to fiscal crisis.

For instance, for France, with 113% debt-to-GDP, a 1% increase in rates would necessitate permanent spending cuts or tax increases of 1% of GDP. For Italy, with 135% debt-to-GDP, the same shock requires 1.3% of GDP in permanent fiscal adjustment.

Worryingly, the risk that Europe will lag in adoption—creating this dangerous “r without g” scenario—is substantial, driven by both our overly cautious regulatory approach and deep-seated institutional rigidities.

The historical experience with the adoption of information and communication technologies (ICT) should serve as a warning. In recent work just presented at the ECB’s 2025 Sintra meeting, Benjamin Schoefer (2025) highlights evidence demonstrating that ICT-intensive industries demand higher rates of worker reallocation and restructuring (Bartelsman et al., 2016). Europe’s institutional rigidity stopped the necessary adjustment. Research suggests that firms in high employment protection environments specialized away from fast-growing, disruptive sectors like ICT to avoid high adjustment costs, favoring instead incremental innovation and leading to sluggish ICT diffusion (Samaniego, 2006; Saint-Paul, 2002). Furthermore, the high “cost of failure” (see our post here) imposed by these rigidities constrains the venture capital investment crucial for scaling innovative tech ventures (Bozkaya and Kerr, 2014).

In sum, Europe's institutional setup appears ill-suited for the rapid transformation demanded by AI. Without addressing these regulatory frictions and, crucially, increasing labor market fluidity, Europe risks repeating its experience with ICT. This inertia underscores the profound danger of the scenario described above: facing the fiscal pressure of higher global interest rates driven by AI growth elsewhere, without reaping the domestic productivity benefits necessary to offset them.

I thank Ludwig Straub and Pierre-Olivier Gourinchas for their comments and conversation.

References

Auclert, Adrien, Hannes Malmberg, Matthew Rognlie, and Ludwig Straub. “The race between asset supply and asset demand”. Jackson Hole Economic Symposium, August 2025.

Bartelsman, Eric, Pieter Gautier, and Joris De Wind. “Employment protection, technology choice, and worker allocation.” International Economic Review 57.3 (2016): 787-826.

Bozkaya, Ant, and William Kerr. “Labor regulations and European venture capital.” Journal of Economics & Management Strategy 23.4 (2014): 776-810.

Saint-Paul, Gilles. “Employment protection, international specialization, and innovation.” European Economic Review 46.2 (2002): 375-395.

Samaniego, Roberto M. “Employment protection and high-tech aversion.” Review of Economic Dynamics 9.2 (2006): 224-241.

Schoefer, Benjamin. “Eurosclerosis at 40: labor market institutions, dynamism, and European competitiveness.” Sintra ECB Forum on Central Banking, July 2025.

But nuclear power is more competitive than it seems. The truth is that nuclear costs, like so many things in Europe, are a result of policy choices. The last plants built in Europe took nearly two decades to complete. In South Korea, it takes an average of five years. Across the continent, nuclear power is stuck in a doom loop. Safety rules make projects slow and expensive. Because they are slow and expensive, few get built. This breaks the positive learning curves seen in earlier programs. France's nuclear expansion in the 1970s and 80s, for example, showed that as more reactors of a standard design were built, costs steadily decreased—by as much as 23% for a given series. With so few projects today, the workforce and supply chains lose the skills to build them well.

Two of Europe’s most recent projects are Finland’s Olkiluoto 3 and France’s Flamanville 3 reactors. Both were treated as one-off experiments. Olkiluoto 3 was finished about 14 years late in 2023 and cost nearly four times its original budget, rising from €3 billion to around €11 billion. Flamanville 3 connected to the grid in December 2024, 12 years behind schedule; its cost had quadrupled from €3.3 billion to over €13 billion in 2015 euros according to EDF (France Cour des Comptes reports 24 bn in current euros including financing). On a consistent 2010-$ per-kWe basis, Olkiluoto-3 cost about $6,600/kWe and Flamanville-3 about $9,800/kWe ex-financing (≈$14,100/kWe including financing) — two to four times the standardized French builds of the 1970s–80s.1

Making nuclear truly competitive is critical, because a fair market comparison must also account for reliability. On that front, wind and solar cannot be the whole solution to decarbonization. As we have argued before, a dark, calm week in winter—a ‘Dunkelflaute’— can stall renewable output across much of the continent. Batteries cannot fill this gap; Germany’s planned storage for 2030, for example, would cover less than one percent of the energy needed for such an event. Nor can interconnections do it, given the high correlation between weather patterns across the continent.

The UK case

To understand why, we can look at Britain, where a government task force has made proposals that could lead to faster, cheaper plants. The problems it describes are common across Europe. The primary culprit behind expensive nuclear is the “As Low As Reasonably Practicable” (ALARP) principle, which EURATOM, the EU’s nuclear body, adopted as a standard in 1980. This approach requires extra measures unless their cost is grossly disproportionate to the safety gain. While this principle comes from a seemingly laudable goal of maximizing safety, the taskforce finds it drives delay and expense without commensurate safety gains. A profitable operator is seen as leaving cash on the table that could have been used to mitigate radiation, no matter how tiny the benefit.

The results are absurd. One UK plant builder had to redesign a reactor — already used safely across Japan—to cut radiation emissions by 0.0001 millisieverts (mSv). This is the dose you get from eating a single banana. The UK's public dose limit from nuclear plants is 1 mSv/year. The average person already receives about 2.7 mSv/year from natural sources and medical scans.

Right now, none of the (safe!) designs used abroad can be used to avoid requirements in the UK. Even when a component or design has operated safely for years abroad, UK regulators commonly require new testing and a fresh decision, which “effectively leads to double regulation.”

By contrast, countries that build nuclear energy efficiently, like South Korea, rely on a fleet approach, treating the tenth reactor like the tenth plane off a production line.

While Europe struggles with two-decade projects, South Korea builds identical plants on predictable schedules, finishing them in under five years. Their reactors run more reliably than those in the US, France, or Japan — delivering an average 96.5% of full capacity. When South Korea built four of its reactors in the UAE, the project was delivered roughly on time (2020-2024) and on budget, in sharp contrast to the European experience.

The cost of delay

The problem with European nuclear regulators is that they measure the cost of not regulating, but not the opportunity cost of regulating. A delay on a major nuclear plant means years of burning more fossil fuels, higher electricity prices, and greater import dependency. These are real, quantifiable costs to society and the economy, yet they are largely absent from the regulatory equation. This extends to the workforce, where we're likely to see a "cliff-edge drop in capability due to retirement" as long project timelines deter a new generation of engineers.

As Britain’s report puts it, “indirect costs are often not adequately considered by regulators in their analysis of what measures are proportionate to reduce the risk".

Olkiluoto 3 now produces 14% of Finland's electricity. Had it opened on schedule in 2009, Finland would have avoided 14 years of extra carbon emissions and dependence on foreign fossil fuels. Regulators should have to quantify those foregone benefits when deciding if another micro-reduction in dose is “reasonable”.

The path forward is to fix this regulatory dysfunction. Today, regulators face no penalty for delays that have wrecked European electricity markets. Governments must change these incentives. This means ending the absurd practice of re-evaluating reactor designs that have run safely for decades in allied nations like the US or South Korea. It means committing to a fleet approach, building the tenth reactor like the tenth car off an assembly line. These changes would defang nuclear’s smartest critics: clean baseload power does not need to take decades and costs tens of billions. Europe could have abundant, cheap, clean electricity if its regulators let it.

Here are some takeaways from the Turnberry accord.

1. Turnberry is not a deal, but a surrender. We accepted 15% tariffs on our key exports, promised $750 billion in US energy purchases, and $600 billion in American investments, and drastically reduced our own auto tariffs—all while getting literally nothing in return. When you hear boasting about the deal by a EU Commissioner, translate it as “be happy it is not worse.”

2. Contrary to the Commission’s assertion, this deal does not even buy “certainty in uncertain times”. The fine print of the deal is vague for a reason. Pharmaceutical and semiconductor tariffs remain unresolved with “all options on the table.” Investment terms lack specifics. Energy purchase mechanisms are not defined. The strategic ambiguity is designed to keep Europe constantly off-balance and negotiating from weakness, never knowing when the next demand will arrive. Knowing Trump, higher demands are sure to arrive.

3. Almost certainly, neither this deal nor the one with the UK was GATT/WTO-compliant. They violate the WTO’s Most Favored Nation (MFN) rule, which requires countries to offer the same trade terms—like tariffs—to all WTO members, not just select partners. Turnberry marks the death, for the EU, of the rules-based trading system on which our prosperity was built. Remarkably, Trump managed to destroy the global trading system and raise tariffs by more than the infamous 1930s Smoot-Hawley Act without starting a trade war.

4. Credible threats win negotiations. Trump’s unhinged behaviour paradoxically made his threats credible. Beyond that, the threat to impose 30–50% tariffs was believable because it was plausible that the economic and especially political damage would have hurt us far more than them. Europe’s export-dependent economies are already stagnant.

5. If the other side cannot hit back, economic theory shows that a big buyer like the US can gain by taxing imports. Put a 15 % levy on a €60,000 German car. To stay in the market the exporter reduces the factory-gate price to €55,000. Customs collects 15 % of that (€8,250) and the American buyer now pays €63,250—€3,250 more than before. The Treasury pockets the full €8,250, so the net gains to the US are €5,000 per car. Home producers also profit from the higher prices. This strategy works as long as the US has market power and the other side can't retaliate effectively—which Europe, due to its internal divisions, will not. Nor will the UK, Japan or Canada.1

But caveats do apply to this basic argument, and they mean most likely the US economy will not benefit from the Trump tariffs:

• These high, broad-based tariffs raise input costs to manufacturers and make them less competitive.

• Also, the US is no longer a manufacturing economy—as Richard Baldwin has remarked, 90% of its people work in services.

• Most importantly the growing regulatory complexity and arbitrariness of the tariff regime provides rents to those connected with power, not to innovators. It is a recipe for the biggest enemy of growth: regulatory overkill and crony capitalism. Consider the example of importing a can of beer from Belgium into the US. There is a 10% country-specific tariff on the entire value of the product. On top of this, the aluminium can itself is treated as a completely separate product, subject to its own additional tariff of up to 50%. The level of this tariff is based on the nearly untraceable origin of the raw metal—the country where the aluminium it was "smelted and cast." The tariff rises to 200% if the country is unknown. This forces an importer to research the obscure global supply chain of a minor component and apply multiple, overlapping tax rates to a single, everyday item.

6. You cannot win a trade war against the army that protects you. Our weakness is Russia. With a threatening Russia on our doorstep, and the Ukraine war ongoing, Europe cannot afford any break with the US. This makes us vulnerable to coercion. This dependency was the key truth in the room at Turnberry. Faced with the choice, Europe has to accept whatever terms Trump offers.

7. The EU's central budget of around €160 billion equals just 1% of its GDP, making it 36 times smaller than the US's federal budget. Just the US defense budget is four times the budget of the EU. When Trump threatens tariffs, Brussels cannot compensate affected industries without asking member states for money. There is no fiscal capacity to respond to coercion. The Commission's latest budget proposal maintains this structure through 2034.

8. Technology dependence creates huge leverage for the US. The talk about service retaliation was not credible. Europe runs on American cloud infrastructure, AI platforms, and software, creating a €300 billion annual deficit in digital services. We now face 15% tariffs on semiconductors we import, while our premier technology company, ASML, must provide its essential lithography equipment tariff-free to American manufacturers. We are paying more for the finished product while giving away the tools to make it.

9. Fixing the European single market would render the damage from any loss of American trade moot. The IMF estimates that internal EU trade barriers are equivalent to a 45% tariff on goods and a staggering 110% tariff on services. Today, the supposed 440-million-consumer market is actually 27 fragmented fiefdoms. The lack of a banking and capital markets union is a key flaw. The proposed “28th regime” would create voluntary EU-wide corporate law. Completing the banking union would enable continent-scale financial institutions. Real capital markets would channel Europe's massive private savings productively.

While von der Leyen proclaimed a “geopolitical Commission,” enforcement of single market rules — the Commission's core treaty obligation —collapsed. During her tenure, enforcement actions fell by 80%. Only 173 new infringement cases were opened last year, compared to over 700 a decade ago. The Commission busied itself expanding into housing, defense, and climate policy while abandoning its primary duty: ensuring goods and services flow freely across borders. It must now focus on its core business.

10. The actual deal is still not Von der Leyen’s fault. The French PM cried “soumission” on seeing the deal. But France (and the other member states) are responsible for the state of affairs. They continue to drag out, sabotage and fail to ratify Mercosur, blocking the best free trade alternative available. France (together with other member states) is responsible for slowing the single market of services and the capital market union at every turn in the name of national interest.

11. Europe needs to find ways to reduce Trump’s leverage despite its fiscal constraints. One option is improving the single market, which costs nothing and is good for growth. The other is defence integration. The proposed European Defence Mechanism, which would pool procurement and create a defence single market through an intergovernmental treaty, avoiding the EU's unanimity requirements, is such an option.

Share