War in the Gulf, a shut chokepoint, and an AI market that swings by the week have not stopped record sums flowing into developing economies. Emerging markets are no longer where investors run from in a crisis. They are increasingly where investors hide.
The rule that has governed global finance for forty years has been that when something breaks, money leaves the developing world. It happened in 1994, in 1997, in 2008, in 2013 when the US Federal Reserve merely hinted at slowing its bond purchases, and again after 2020, when the COVID pandemic shock pushed Zambia into default that November, and Sri Lanka and Ghana into default two years later.
By that rule, 2026 should have been a bloodbath. A war that began in late February has effectively closed the Strait of Hormuz, the passage that carried roughly a fifth of the world’s oil. Brent crude has traded near $89 a barrel in mid-August, up about a quarter on pre-war levels after touching $110 earlier in the year.
Fertiliser prices have followed, and behind them food. The yield on the 10-year US Treasury note, the number against which almost all developing country borrowing is priced, sits near 4.65%, against 3.97% before the fighting started. Traders have spent the summer arguing not about how quickly the Fed will cut rates, but whether it will have to raise them.
The rule did not hold. Institute of International Finance data shows foreign investors put $214.4 billion into emerging market debt in the first seven months of 2026, against $177.7 billion in the same stretch of 2025, the strongest run in more than two decades.
Governments have exploited the appetite. Roughly $19 billion of sovereign bonds were sold in July alone, about twice the average for that month over the past ten years, taking issuance for the year to a record $187 billion.
Something has changed, and it is worth being precise about what.
The comparison that flatters the developing world
Ask whether emerging economies have handled geopolitical disruption better than rich ones and the honest answer is that they have handled their own balance sheets better, which is not quite the same thing.
The IMF’s April 2026 Fiscal Monitor put global public debt just under 94% of GDP in 2025, on course to cross 100% by 2029, a year earlier than the fund projected only twelve months previously. The accumulation is driven overwhelmingly by the largest economies.
On the fund’s April World Economic Outlook database, US general government gross debt is projected at 126% of GDP this year and 142% by 2031, the biggest absolute increase in the advanced world. Japan sits above 200%. Twenty-three economies now carry gross debt above 100% of output, and the list is dominated by rich countries, not poor ones.
Set against that, the emerging market picture looks almost conservative. Central banks across Latin America, Asia and Africa spent the 2022 to 2024 inflation shock raising rates early and hard, which left them with real yields that are genuinely positive, and room to cut when their advanced counterparts have none.
Reserve buffers are thicker. Central bank independence, which was a slogan in the 1990s, is now closer to institutional fact in Brazil, Mexico and South Africa.
Ratings agencies have noticed. Pakistan, Ghana, Ecuador, Nigeria and Argentina have all collected upgrades. Oman and Azerbaijan have reached investment grade. Fund managers report the strongest upgrade momentum among lower rated sovereigns in over a decade, an unusual thing to say in a year of war.
Political risk, meanwhile, has migrated. The disorder that investors once priced into Latin American and African assets now shows up in a record US government shutdown, fractured European coalitions, and defence spending commitments that nobody has explained how to fund.
The growth arithmetic points the same way. The IMF’s July update expects emerging market and developing economies to grow 3.8% this year and 4.5% in 2027, against 1.7% and 1.8% for advanced economies. Global growth of 3.0% in 2026 recovering to 3.4% in 2027 is what the fund calls a V shaped path around the war shock.
The caveat matters, though, and it is a heavy one. The World Bank’s June Global Economic Prospects cut its global forecast to 2.5% for 2026, the weakest since the pandemic, and downgraded two-thirds of economies. South Asia, the fastest growing region, decelerates from 7% to 6.3%.
Low-income countries manage 5.4%, three-tenths lower than previously expected, with fertiliser driven food inflation doing much of the damage. Most sobering, the bank calculates that by 2028 developing economies other than China and India will have spent nearly a decade making no progress at all in closing the income gap with rich countries.
So, the financial resilience is real. The developmental resilience is not. Bond markets and living standards have decoupled, and anyone reading the inflow numbers as evidence of broad-based prosperity is reading them wrong.
Why the money is moving
For a decade-and-a-half, global portfolios were built on a single assumption, that American assets were the default, and everything else was a satellite allocation. That assumption is being quietly unwound.
The IMF has begun writing about the erosion of the US Treasury’s safety premium in its own fiscal surveillance. Investors who spent 2025 watching the dollar post its sharpest annual fall in eight years have concluded they are over allocated to one jurisdiction, and, in a fragmenting world, they want to be spread across many.
The mechanics reinforce the mood. For Japanese and other Asian institutions, hedging US Treasuries back into home currency now wipes out most of the yield, which makes local Asian bonds structurally more attractive than they were.
Emerging market debt was yielding around 6.9% in February, against roughly 4.2% for US bonds and 3.6% globally. Rising Treasury yields have narrowed that gap since, but not closed it.
The pull side is the story of a decade of quiet plumbing work. Emerging economies have built domestic capital pools deep enough to reduce their dependence on foreign money altogether.
Local currency sovereign bonds outstanding totalled roughly $13 trillion by the end of 2024, against about $1.4 trillion of international hard currency sovereign debt, according to research from JP Morgan and UBS. Large economies, such as Brazil and South Africa, now fund themselves overwhelmingly at home, in their own currency, from their own pension funds and insurers.
That changes the physics of a shock. When foreign investors sell, domestic institutions are on the other side of the trade. Fund managers describe the result as an absence of the liquidity crunches that used to define emerging market sell-offs. Prices fall, but the market does not gasp.
Positioning is the third leg. After what Bank of America’s head of emerging market fixed income strategy David Hauner, speaking to Reuters, called the ‘valley of tears’ running from roughly 2015 to 2025, a stretch of strong dollar, US exceptionalism, and serial defaults, global investors are still structurally underweight.
Emerging economies hold about 60% of the world’s population and produce around 40% of global output, yet account for barely a tenth of the MSCI All Country World Index. Several months of inflows barely dent a decade of under-investment.
There is a risk buried in the composition of the money, and the IMF flagged it in April. Portfolio flows to emerging markets have risen eightfold since the global financial crisis to about $4 trillion in cumulative terms.
Portfolio debt liabilities now average around 15% of GDP, against roughly 9% in 2006. About 80% of that capital comes from non-banks, twice the share of twenty years ago, and non-bank money is faster money.
Private credit in emerging markets, opaque by design, has grown fivefold in a decade to somewhere between $50 billion and $100 billion. Deeper markets have not abolished the sudden stop. They have changed who would cause one.
Safe haven, or simply the least crowded trade
The safe haven question deserves a careful answer, because the marketing departments have got ahead of the evidence.
A true safe haven does two things. It holds value when everything else falls, and it stays liquid when liquidity vanishes. Emerging market assets do neither reliably. What they have done in 2026 is something narrower and still significant. They have offered diversification at a moment when the traditional refuges look compromised.
Look at the split inside the flows. While $214.4 billion went into debt, roughly $86 billion came out of emerging market equities in the same seven months, nearly ten times the outflow at the same point in 2025. This is not a wall of money buying an asset class. It is a discriminating reallocation into yield, and away from concentrated technology risk.
The performance record is similarly mixed. The JP Morgan GBI-EM Global Diversified index of local currency debt lost 2.25% in the first quarter as the dollar strengthened on safe haven demand, then gained 3.85% in the second.
An index of inflation linked emerging market local currency government debt has returned 11.3% this year, against 1.5% for the broader local debt index, and a small loss for the Bloomberg Global Aggregate. The winners are specific, not general.
Currencies tell the same story. Emerging market currencies erased their 2026 gains by the start of July as speculation about higher US rates revived the dollar. Capital Economics’ aggregate currency risk indicator has nonetheless stayed near multi-year lows, which is the more interesting fact. Currencies weakened without anyone fearing a crisis.
The deepest evidence for a structural shift comes from official reserve managers rather than fund managers, though it needs reading carefully. The dollar’s share of global reserves has fallen from roughly 71% in 1999 to 57.1% in the first quarter of 2026.
That latest reading, however, was up from 56.4% three months earlier, and the IMF is at pains to point out that much of the recent movement reflects exchange rate valuation effects rather than central banks actively selling dollars.
Intent shows up more clearly in what reserve managers say and in what they buy instead. In the World Gold Council’s 2026 survey, 74% of central banks expected the dollar’s share to be moderately or significantly lower within five years.
Official gold buying ran at an estimated 244 tonnes in the first quarter, ahead of both the previous quarter and the five-year average, with Poland the largest single purchaser.
Central banks are not calling emerging markets a haven. They are calling the ‘Old Haven’ crowded, and looking for anything neutral. Emerging market debt is one beneficiary of that search. Gold is the bigger one.
Professional investors are behaving accordingly. Several large houses, BlackRock’s investment institute among them, have cooled on emerging market equities and hard currency debt even as flows continue.
Managers describe themselves as highly selective, ignoring benchmarks, avoiding countries with debt problems and skipping those where yields no longer compensate. That is not haven behaviour. It is careful, well-paid risk taking.
What the institutions are actually saying
The IMF’s July update describes an economy pulled by two crosscurrents, a war shock that punishes energy importers and vulnerable states, and an AI investment boom that lifts anyone plugged into the technology value chain.
Global disinflation has stalled. Headline inflation is expected to rise from 4.1% in 2025 to 4.7% in 2026 before easing to 3.9% in 2027, with energy and food doing the work.
The Fiscal Monitor adds the geopolitical arithmetic. IMF staff estimate that a one standard deviation shock to their geopolitical fragmentation index is associated with public debt ratios rising about 1.5 percentage points of GDP over the medium term. Fragmentation is not an abstraction. It has a price, it is paid in borrowing costs, and it is being paid now.
The World Bank supplies the development warning. Its June report is explicit that emerging economies unable to build the ecosystem and policy environment for wide AI adoption risk falling further behind, and that private investment growth in developing economies has been declining since the 2000s even as public balance sheets improve.
The US Energy Information Administration expects Brent to average $87 a barrel across 2026 and does not see Middle East production returning to near pre-conflict levels until early 2027.
The IEA has warned of the widest global supply deficit in five years. For oil importing developing economies, that is another eighteen months of imported inflation.
Fund managers add the risk nobody controls. The threats most often named to the flow story are not war headlines but food prices, fertiliser costs, and El Nino. A drought does more damage to a frontier sovereign’s fiscal position than a missile does.
The next shock is already priced, badly
If the world has survived geopolitics, artificial intelligence (AI) is the test that has not yet started. And the strange thing about emerging markets in 2026 is that they are simultaneously the most exposed and the least prepared.
Start with the index. As of July 31, information technology accounted for 40.8% of the MSCI Emerging Markets Index. Taiwan is the largest country weight at 26.6%, China 21.4%, and South Korea 20.3%.
Taiwan and Korea together are almost half the benchmark. In January those weights were 21% and 15.7%, with technology at 30.3%.
India, meanwhile, has slid from roughly 20% of the index in mid-2024 to under 12%. Its equities are down around 5% in local currency terms this year, with the Sensex at 77,728 on August 17, having lagged badly through the first half before a July rally that pulled about $1.6 billion of foreign money back in. Expensive crude and a weaker rupee did most of the damage.
The emerging market equity benchmark is now, to a first approximation, a leveraged bet on the AI semiconductor cycle.
That has been enormously profitable. It also means every boom and bust in AI sentiment transmits straight into an asset class marketed as diversification. Korea’s Kospi moving 3.7% in a single session on AI earnings is not an emerging market story at all. It is a Silicon Valley story with a Seoul postcode.
Then there is the labour market, where the exposure runs the other way. IMF research puts around 40% of global employment in occupations exposed to AI, rising to 60% in advanced economies but sitting at 40% in emerging markets and 28% in low-income countries.
The fund’s 2026 work on new job creation finds AI related skills appearing in almost 5% of US job postings by 2025, with incidence in emerging economies roughly half that.
The comfortable reading is that developing economies face less immediate disruption. The correct reading is that they face less immediate disruption because they have fewer of the cognitive jobs that AI both threatens and rewards, and they are much less equipped to capture the productivity gains.
The IMF’s AI Preparedness Index, which covered 174 economies when it was published in 2024, places India at 0.49 against 0.77 for the US and 0.80 for Singapore. Bangladesh scores 0.38.
For countries whose development model runs through services exports, business process outsourcing, back-office work, entry level coding and customer support, this is the central strategic question of the next decade, and it is barely being discussed in the same rooms where capital flows are celebrated. Cheap labour was the comparative advantage. AI attacks precisely the tasks that made it valuable.
The economies best placed are the ones already inside the hardware chain, Taiwan and Korea above all, along with the handful of middle-income economies drawing data centre investment on the strength of cheap power. The ones most at risk are populous middle-income countries with young workforces, thin digital infrastructure, and social safety nets designed for a different century.
What to watch out for
The bull case for emerging markets rests on three things holding. That the Fed does not have to raise rates. That Hormuz reopens before food inflation does structural damage to importing sovereigns. That the reallocation away from American assets is a strategic decision rather than a carry trade wearing a strategic costume.
The first two are out of the hands of finance ministries from Accra to Jakarta. The third is not. Governments that use this window to extend maturities, deepen domestic investor bases, and build the digital and educational infrastructure that AI adoption requires, will look, in five years, as though they earned something.
Those that simply enjoy the cheaper borrowing will find out that the oldest rule in global finance was not repealed in 2026. It was merely suspended.