Postcard from Bangkok #1: The drivers of resilience are under strain

The mechanism can potentially become self-reinforcing: rising equity prices make asset-owning households wealthier, stimulate spending, support corporate earnings and sustain valuations. It can also work in reverse: a major market correction could simultaneously weaken consumption, profits and investment, turning a financial shock into an economic slowdown.

McKinsey estimates global household net worth at $570tn (£427tn) in 2025 – nearly five times global GDP. More strikingly, only 20% of its recent growth is thought to have come from actual capital formation; the remainder mainly reflects the revaluation of existing assets and inflation.

Labour income remains essential, of course, and wealth effects are very unevenly distributed. Yet the traditional distinction between the real economy and financial markets is becoming less relevant: markets no longer merely anticipate the economic cycle – they are increasingly helping to shape it.

* AI could transform the economy on a lasting basis, without necessarily guaranteeing the profitability of all the investments it is generating: This is a scepticism I am hearing more and more, both here and back in Paris. In this context, McKinsey’s comparison with Global Crossing and fibreoptics in the late 1990s is illuminating.

Back then, operators correctly anticipated the future explosion in demand for data transmission – but some investments were made too early, with debt maturities falling due before revenues materialised. Global Crossing went bankrupt in 2002, even though demand subsequently grew spectacularly.

AI shares certain similarities. According to estimates presented by J.P. Morgan, capital expenditure by six major technology players could approach $900bn in 2026, then reach between $1.25tn and $2tn in 2027. Yet future revenues remain difficult to assess, as does the pace of technology adoption. An innovation enabling high-performing models to operate with far less computing power could even reduce the profitability of existing infrastructure.

This is compounded by the gap between the speed of technological progress and the time needed to build energy infrastructure. Above all, McKinsey reminds us that major technological revolutions have historically created value less by eliminating jobs than by creating new products and markets. The key question is therefore as much about the timing of productivity gains as it is about how they will be distributed between technology producers and the companies using the technology.

* Producing AI and benefiting from its productivity gains are two different things: During a discussion on Asia, a Google representative noted that a country’s potential should not be measured solely by its ability to build data-centres, manufacture semiconductors or develop its own models. Economic gains may depend more on how quickly businesses and households adopt these technologies.

Thailand illustrates this adoption potential. According to data presented by Google, 73% of Thai Gemini users access AI through their mobile phones, while 25% are over the age of 55. This uptake beyond younger, technologically skilled populations points to potential productivity gains across many sectors, including small businesses and traditional industries.

That said, AI could also challenge some of the development models that have underpinned Asia’s success – India and the Philippines, for example, have developed significant service-export industries by leveraging lower labour costs than in Western countries. US and European companies have offshored call-centres, accounting, administrative functions and part of their IT services to these countries.

These activities have created millions of skilled jobs and supported the emergence of a middle class – yet these are precisely the kinds of tasks AI can now partly automate, potentially reducing the economic advantage of offshoring them.

The paradox, therefore, is that Asian countries could be among the main beneficiaries of AI adoption while seeing some of their historical competitive advantages called into question. Their ability to transform their businesses and develop new, higher value-added activities will be decisive in turning this technological revolution into sustainable growth.

Globalisation is redistributing its balances

* China may have the means to reduce its imbalances but may choose not to do so: Brad Setser challenges the idea that Chinese debt prevents a more ambitious policy to support households. He distinguishes the financial difficulties of local governments from the position of the central government, whose direct debt is estimated at around 30% of GDP (depending on the scope used) and which owns substantial public assets.

In his view, Beijing could increase certain pensions or reduce social-security contributions in order to lower precautionary savings. Economic policy continues to prioritise productive capacity, however, while domestic consumption remains insufficient to absorb output. Setser therefore estimates China’s trade surplus could approach $1.5tn in 2026, with a currency he considers to be undervalued by 20% to 30%.

In the automotive sector, production capacity could reach 50 to 55 million vehicles per year, for domestic demand well below that level. This second China shock is no longer limited to labour-intensive goods – it now affects automotive manufacturing, machine tools and sophisticated equipment. These are sectors in which China has become a direct competitor to advanced economies – not out of necessity but by choice.

* Geopolitical fragmentation does not necessarily mean the end of globalisation, but it could fundamentally change its economic cost: A panel on US-China relations produced a telling expression – ‘stability without trust’. The two powers are seeking to avoid open confrontation while continuing their technological rivalry. China’s share of US imports reportedly fell from around 21/22% in 2017 to 8/9% in 2025, according to J.P. Morgan’s China economist.

This decline in direct trade does not mean that interdependencies have disappeared, however. They are shifting towards critical raw materials, advanced technologies and the security of trade routes. US restrictions on semiconductors and Chinese controls on rare earths are reciprocal tools of pressure that, paradoxically, make dialogue indispensable.

Mistrust is nevertheless pushing each power to develop industrial capabilities that already exist elsewhere, in order to reduce strategic dependency. While this may improve national economic security, it raises global investment needs without proportionately increasing overall productive capacity. It therefore creates an additional source of competition for capital.

* Europe is not declining uniformly – its economic and strategic centre of gravity is shifting northward and eastward: Sander Tordoir of the Centre for European Reform encourages us to move beyond a uniform reading of Europe’s slowdown. Weak aggregate figures conceal diverging national trajectories – Denmark, the Netherlands and Croatia have at times delivered growth performance closer to that of the US than a comparison of the two regions would suggest.

Rearmament could reinforce these differences. According to Tordoir, European NATO members could, within two years, spend as much on military equipment as the US does today. This effort would be driven especially by Scandinavia, Germany, the Netherlands and Eastern Europe, which combine – albeit to varying degrees – a perception of the Russian threat and the necessary fiscal capacity.

At the same time, Germany is facing the second Chinese industrial shock – its companies are losing markets in China, while their former customers are becoming competitors in international markets. According to Setser and Tordoir, net exports have cumulatively subtracted around three percentage points from German growth since the end of 2023, although this deterioration cannot be entirely attributed to China.

A German stimulus package could therefore benefit foreign producers more than in the past if domestic supply does not respond sufficiently to the additional demand. Finally, Tordoir highlights a contrast: Europe is regaining the ability to act in defence and trade protection, but productivity-enhancing reforms – the Single Market, the Capital Markets Union and the Draghi report – are progressing much more slowly. Europe now appears more capable of mobilising capital to protect itself than of reorganising its economy to produce more.

Vulnerabilities of the new regime

* Changes in the investor base could become an increasingly important driver of bond markets: Brad Setser points out that Asia’s historically high external surpluses are no longer being recycled as automatically as in the past into official reserves and US treasuries. A growing share of these savings appears to be moving into equities and private investments.

This shift is taking place at a time when the US is running a public deficit of around 6% of GDP, while hyperscalers could issue approximately $350bn of debt in 2026 and $450bn in 2027. Jahangir Aziz notes that central banks and official investors – traditionally less sensitive to yields – are leaving more room for private investors, who adjust their allocations according to prices and risks.

This supply effect may force the market to find a new equilibrium through higher yields or risk premia. Despite a correction in the bond bear market, long-term rates could therefore remain under pressure independently of central-bank decisions alone.

This competition for capital could also penalise weaker borrowers, including US households, certain companies with floating-rate debt and emerging markets dependent on the dollar. Bond risk therefore depends not only on the level of public debt, but also on investors’ capacity and willingness to absorb it.

* The renewed strength of European banks is not, by itself, sufficient to measure the resilience of the financial system: ECB representatives described European banks as an island of stability, thanks to renewed profitability, high capital levels and improved asset quality.

Ed O’Brien nevertheless pointed to a less obvious consequence: banks are selling their impaired loans more quickly than in the past, improving their balance sheets but reducing their signalling function regarding economic stress.

At the same time, certain riskier activities have developed in investment funds, hedge funds and private credit. The Global Financial Stability Report estimates hedge funds’ gross assets at around $13tn – more than three times their 2013 level. These players provide liquidity in normal times, but their leveraged positions can amplify a correction when margin calls trigger simultaneous sales.

J.P. Morgan also refers to $200bn to $250bn of particularly fragile private debt – notably linked to private-equity transactions carried out during the low-rate period – for which refinancing could become challenging in 2028/29. If rates remain high, extending maturities may not be sufficient to resolve borrowers’ difficulties.

Lastly, the ECB highlights the interdependencies created by certain circular AI financing structures – as well as the risk that a cyber incident occurring during a period of financial stress could amplify its effects. Stability therefore depends as much on the strength of institutions as on their interconnectedness and their collective ability to withstand a liquidity shock.

* Energy reminds us the constraints facing the global economy are not only financial: Natasha Kaneva highlighted two interesting paradoxes. Despite disruptions in the Middle East, oil flows from the region are estimated to have amounted to around 20 million barrels per day, compared with 23 million before the conflict.

The availability of crude oil does not, however, guarantee the availability of refined products – the world may have relatively abundant crude supplies while still facing pressure on diesel, petrol or jet-fuel prices because of insufficient refining capacity.

This is compounded by governments’ caution in using strategic reserves. Several prefer to preserve their stocks precisely because they fear the conflict will last. While individually rational, this behaviour may collectively limit the capacity to cushion the current shock.

The World Economic Outlook illustrates a similar logic: fiscal measures that neutralise energy-price signals can require far more resources than targeted, temporary transfers. Immediate resilience may therefore come at a cost to the capacity to absorb future disruptions.

What are the initial investment implications?

One observation from Dubravko summarises the implications of these discussions: “Quality, quality, quality.” His point is not to abandon risky assets but, rather, to favour companies with strong balance sheets, robust cashflow generation and the ability to finance their investments sustainably.

Ultimately, these conferences lead, at this stage, to the same question: will the world run short of capital – or is capital becoming increasingly concentrated in a few geographies, sectors and asset classes? At the other end of the spectrum, we are witnessing US hypergrowth that is more than offsetting current vulnerabilities. Will this period of hypergrowth last long enough for returns on investment to materialise?

Michaël Nizard is global head of asset allocation research at Edmond de Rothschild Asset Management