Antonio Fatás, Professor of Economics at INSEAD, hosts one of the most popular webinars on the global economy every year. This year, over 3,200 participants logged on to hear his Economic Outlook for 2026.
Here are the key points:
1. Economic decoupling is real
The big surprise in 2025 was that the economy performed better than expected despite ‘Liberation Day’ and massive tariff announcements from the US. Actual tariff rates ended up at around 10% – less than half of the announced 25%. Many companies and sectors were granted exemptions, which significantly reduced the real economic impact.
But the figures also reveal a more important trend: the US and China are undergoing a real economic decoupling. Chinese exports to the US have fallen significantly, while countries such as Vietnam are experiencing explosive growth, partly because production is shifting and partly because goods are being redirected through other markets.
Economic decoupling is no longer just rhetorical. It is real, measurable and will accelerate in 2026.
2. 2025 marks a turning point for the global order
‘2025 is an inflection point,’ Fatás emphasised. ‘We will not return to where we were before.’
The US-centric world order is undergoing fundamental change. While the US is using ‘strategic coercion’ through tariff threats, new alliances are forming elsewhere: the MERCOSUR agreement between Latin America and the EU has been signed. China and Canada have initiated surprising talks. China, South Korea and Japan are discussing joint economic initiatives – unthinkable just a few years ago. At the same time, China has increased investment in the Belt and Road Initiative to record levels.
This is not about one president or one election. Confidence in the US as a reliable partner has fundamentally changed.
3. The AI boom: High expectations meet modest results
IT investments contributed to US GDP growth at the same level as during the internet boom in the 1990s. Combined with a 12% return on the stock market, this kept the economy going through uncertainty.
But does AI live up to the expectations baked into stock market valuations?
The S&P 500’s price-earnings ratio is around 40. That level has only been seen once before: in 1998, the year before the dot-com bubble burst. Even adjusted for current interest rates, valuations resemble those of 2007 – the year before the financial crisis.
And there is one crucial difference from the 1990s: the risk is extremely concentrated in the ‘Magnificent Seven’ tech companies. ‘We have put all our eggs in a few baskets,’ warned Fatás. ‘These companies must deliver – otherwise, an adjustment will be needed.’ Conversely, these companies have the robustness and the will to ensure that AI investments succeed in the long term.
4. Productivity gains from AI are not keeping pace
To justify these valuations, AI must deliver massive productivity gains. The figures are not yet convincing.
US productivity growth has risen from 1.5% annually before 2020 to around 2.1% since. That is positive, but far from revolutionary. By comparison, productivity accelerated to nearly 3% annually during the internet boom of 1995-2004.
In Europe, the picture is even more mixed. If AI were truly the engine, we would see similar patterns across developed economies – we do not.
We are betting that AI will be more transformative than the internet. But the evidence is not there yet.
5. History warns: High valuations signal low future returns
Fatás showed data from Robert Shiller: There is an extremely strong correlation between high P/E ratios and low 10-year returns. An overvalued stock market usually results in poor long-term returns.
There was one exception around 2015, when returns were higher than P/E ratios indicated – this was solely due to real interest rates falling from around 6% to almost zero. We cannot repeat that effect.
As long as interest rates remain stable, history’s warning applies: current valuations promise lower future returns.
Will the crash come in 2026?
The risk of recession does not come from the length of the expansion, central banks or fiscal policy, which is actually expansionary in the US, the EU and China.
The risk lies in market valuations and the narrative about AI. If productivity gains fail to materialise – or come slower than expected – valuations could collapse. The concentration in a few tech stocks dramatically amplifies the risk.
‘The probability of a correction is higher today than it was two years ago,’ concluded Fatás. ‘That doesn’t mean it will happen tomorrow. During the internet boom, overvaluations lasted much longer than anyone thought possible. But the ingredients for a significant adjustment are there.’
So there will be an adjustment. The question is how deep it will be and whether it will happen as early as 2026
—
Scandinavian Executive Institute has collaborated with INSEAD for over a decade to deliver board programmes and leadership programmes of the highest caliber. In a world marked by geopolitical upheaval, technological leaps and economic uncertainty, it is crucial that you, as a senior executive, have access to the latest research and strategic tools to navigate the complexity. Our programmes give you just that, as well as a network of like-minded people facing the same challenges.
