IMF Global Financial Stability Report — October 2026

Source. International Monetary Fund, Global Financial Stability Report, October 2026: Resilience, Risk, and Transformation (published 13 October 2026). All IMF findings and figures cited below are taken from the report; interpretation, scenarios and the monitoring framework are our own and are identified as such. Original report

Executive Summary

The IMF's central finding is that financial markets have been resilient since the April 2026 report, but that financial-stability risks remain elevated. Strong corporate earnings, solid growth and heavy investment in artificial intelligence (AI) have kept financial conditions accommodative on net, despite geopolitical tension, an energy shock and policy uncertainty.

The report separates risks, the plausible near-term catalysts for market turmoil, from vulnerabilities, the structural features that determine whether a shock stays contained or becomes systemic. It selects three salient risks: an AI-related market correction, a sharp rise in sovereign bond yields, and cyber threats. It judges the sovereign-yield and cyber risks to be getting worse, and the AI risk to be about the same.

The vulnerabilities that could amplify them are concentrated outside the core of the regulated banking system: leverage in the AI ecosystem, rollover risk in sovereign debt, liquidity mismatches at insurers and pension funds, contingent liquidity commitments in private credit, hedge-fund leverage, and eroding buffers in weaker emerging markets. The banking system remains broadly resilient in the IMF's stress test, but a larger share of banking assets is classified as weak than a year ago.

Our reading is that the report's most useful contribution is analytical rather than predictive. The systemic risk may lie less in any individual vulnerability than in the interaction between vulnerabilities, through leverage, collateral, funding markets and interconnected balance sheets. This review develops that idea as a framework for understanding transmission, not as a forecast of a crisis.

1. Financial Resilience versus Systemic Resilience

Market resilience describes asset prices that absorb shocks without disruption. Systemic resilience describes the capacity of the financial system to keep providing credit, payments and liquidity when several vulnerabilities are active at once. The two can diverge.

The IMF's own framework makes the distinction explicit. Instability arises when risks are amplified by vulnerabilities in ways that impair key institutions and markets and disrupt credit or financial services. A shock alone is not the test. The test is what the balance sheets it reaches are able to absorb.

The current backdrop shows why the distinction matters. Since the second quarter of 2026, 14 of 33 major central banks have raised policy rates, against two in the first quarter. Financial conditions tightened early in the year, but rising corporate valuations more than offset higher rates. The IMF therefore describes conditions as accommodative on net.

This is the point we would stress. Accommodative conditions are partly produced by the same asset prices that would adjust in a correction. Resilience that rests on valuations is conditional resilience. Strong prices reveal little about leverage or liquidity until those are tested.

Individually, the IMF notes, the three risks do not necessarily threaten financial stability. They become systemic when underlying vulnerabilities impair intermediation and credit provision.

2. Artificial Intelligence and the New Investment Cycle

The economic case for AI and the financial case for AI-related assets are different questions. A technology can raise productivity over decades while the capital committed in anticipation of it is mispriced over a shorter horizon. Railways and the internet both illustrate that pattern. The IMF's concern is the second question.

What the IMF finds. Core firms in the AI ecosystem have been profitable so far, which eases pressure on valuations. The risk is that weaker or less durable AI investment and profitability trigger a correction, and that the correction is amplified by rising leverage, concentrated exposures and circular financing. In the World Economic Outlook scenario, an unwinding of AI investment lowers global GDP, investment and consumption by as much as 0.4, 1.3 and 0.6 per cent respectively by 2028.

The report documents the financing build-up in specific terms:

• Hyperscalers' share of global investment-grade bond issuance has risen to 10 per cent from under 1 per cent since 2024.
• Hedge funds' gross equity positions plus equity margin loans have grown from 10 per cent to about 15 per cent of advanced-economy GDP.
• Single stocks now account for 28 per cent of leveraged exchange-traded funds, up from 16 per cent.
• Fourteen publicly disclosed circular-financing transactions since 2023 involve roughly $300 billion of equity investment led by hyperscalers in financially weaker AI firms, about $650 billion of debt, and $1.4 trillion of future revenue commitments. Around $320 billion of that debt sits with the investee firms, including guarantee-backed, contingent and off-balance-sheet obligations.

Why it matters. The first lesson is that AI exposure has to be traced along the whole financing chain. It spans equity valuations, investment-grade credit, leveraged retail and derivative products, margin lending, and revenue commitments between related parties. The IMF's circular-financing figures show how a single expectation can be financed repeatedly: an investor funds a customer who contracts to buy the investor's services.

The second lesson concerns transmission across borders. The IMF expects advanced Asian equity markets with high betas to global AI equity returns to suffer most in a sharp slowdown. It also notes that many jurisdictions have increased their holdings of US equities since 2022, in part through larger allocations to hyperscalers and other AI-related firms. Stress could therefore reach investors with low direct AI betas.

Our interpretation. AI-related debt issuance is also a rate story. The IMF cites it as an additional source of pressure on bond markets, because it may compete with sovereign supply for investor demand. A correction in AI valuations and a rise in long-term yields are therefore not independent events. We return to this in Section 9.

3. Sovereign Debt and the Repricing of Long-Term Capital

The sovereign bond market is the pricing base for most other assets. The IMF rates the risk of a sharp yield rise as worse than in April, and judges the potential impact could be severe given the foundational importance of sovereign bonds.

Decomposing the move. The IMF attributes the rise in 10-year yields between 27 February and 30 September 2026 to two forces. One is higher-than-expected short-term rates. The other is higher term premiums. Expected short rates drove yields in most advanced economies, while term premiums played the larger role in emerging markets. Term premiums are now at the upper end of historical ranges, especially for advanced-economy bonds. The report links the rise in term premiums to concerns about persistent fiscal deficits, high debt and the associated increase in bond supply.

This decomposition is the right way to read the yield curve. A higher long-term yield can reflect any combination of:

• a higher expected policy-rate path;
• higher expected inflation, or a higher inflation risk premium;
• higher real rates;
• compensation for fiscal risk;
• a larger term premium, driven by supply and weaker demand from price-insensitive buyers.

Each carries a different policy and market implication. A rise driven by stronger expected growth is benign for debt sustainability. A rise driven by fiscal risk or inflation uncertainty is not, because it raises the cost of financing at the same time as it weakens the case for easing.

Where the fragility lies. The IMF describes a mixed picture. Some fragilities have improved. Hedge-fund repo borrowing against sovereign bonds in the euro area and the United States has partly unwound without notable repo pressure, and liquidity metrics have improved in the euro area and the United Kingdom. US Treasury liquidity indicators look favourable. Liquidity in Japanese government bonds has continued to deteriorate, with wider bid-ask spreads, especially at long maturities.

Others have worsened. Rollover risk remains high, and could be exacerbated where debt managers replace long-term debt with short-term debt at scale. Historically high debt and persistent deficits in major economies remain unresolved.

The IMF identifies three downside propagation channels: rollover risk for governments that have shortened maturities; selling of long-dated bonds by price-sensitive investors, including foreign investors, investment funds and households; and a deterioration in dealer-intermediated liquidity when volatility rises.

It also flags an unusual valuation configuration. Historically compressed sovereign credit spreads alongside a 10-year US Treasury yield above 4.5 per cent last occurred in 2007.

Transmission. Higher long-term yields reach the real economy through four routes: public-finance costs as debt is refinanced, benchmark pricing for corporate credit and mortgages, discount-rate effects on equities and long-duration assets, and the collateral channel, where sovereign bonds that back repo and margin lose value. The IMF's warning is explicit that a sharp rise in yields would depreciate asset valuations, including AI-related assets, and erode the value of government bonds used as collateral.

The IMF recommends putting public debt on a sustainable path, building a more durable maturity profile, extending central clearing of bonds and repo, and ensuring central-bank liquidity facilities are operationally ready. It also stresses clear communication and central-bank independence to anchor inflation expectations.

4. The Sovereign–Bank–Non-Bank Nexus

Financial stability is increasingly a system-wide balance-sheet problem rather than a question of bank capital ratios alone. The sequence we find most useful runs as follows.

A rise in sovereign yields marks down bond holdings. It lowers the value of the same bonds as collateral. Leveraged investors face margin calls and sell what is liquid, which moves prices further. Banks that finance those investors face counterparty exposure and falling collateral values. Meanwhile, higher benchmark yields raise borrowing costs for corporates and households. Each step is individually rational and collectively amplifying.

The IMF's evidence shows where the links are thickening. In the United States, bank loans to non-bank financial intermediaries (NBFIs) rose 18 per cent in 2026, and exposures to private equity and private credit firms rose 30 per cent. The report concludes that linkages between some banks and NBFIs have deepened. We would add that these exposures are less visible in conventional capital ratios than direct lending to the real economy.

The same architecture explains why solvency and liquidity need to be separated. An institution can hold fundamentally sound assets and still become a source of pressure if funding is withdrawn or collateral is called. Forced sales then turn a liquidity problem into a valuation problem for everyone holding similar assets. Monitoring that looks only at solvency will miss the stage at which a stress becomes systemic.

5. Non-Bank Financial Intermediation

The IMF treats NBFIs differentially, and so should analysts. Different institutions are vulnerable in different ways, and most perform useful functions.

Hedge funds. Chapter 2 finds that hedge funds now hold about $13 trillion in assets, more than three times their 2013 level. In calm conditions they support market liquidity, price discovery and risk sharing. In stress, leverage, crowded positions and correlated redemptions can trigger synchronised deleveraging, strain liquidity and spill over to prime brokers, especially in markets with large hedge-fund footprints and thin liquidity. The IMF calls for better data and for prudential measures calibrated to the source of vulnerability, together with credible central-bank backstops for severe dysfunction in core markets.

Insurers and pension funds. Their share of illiquid investments has nearly doubled over the past decade. They hold more than a quarter of publicly traded corporate bonds in the United States and the euro area. The report also notes that the three largest managers could now account for more than half of institutional investment in private equity and private credit. Concentration along the investment chain raises the chance that institutions respond to liquidity pressure in similar ways. A long horizon is a genuine strength, but it does not remove the need to meet cash-flow and collateral obligations.

Private credit. Direct lending now accounts for nearly $2 trillion of loans globally. Default rates have fallen, but borrower leverage has risen, notably in software and services. Semi-liquid vehicles accounted for more than half of new US fundraising in 2025, before the recent redemption wave. Among 10 business development companies with $300 billion in assets, four, holding about $90 billion, have commitments exceeding their available liquidity buffers. In stress, borrowers could draw on contingent facilities just as fund liquidity is eroded by lower collateral values, which could convert liquidity pressure into a contraction in credit supply.

Our interpretation. Private valuations are updated less frequently than traded prices. That does not remove credit risk. It can delay its recognition. The IMF also describes private credit as a sector that has not yet faced a test of tighter conditions.

The review should not be read as treating non-banks as inherently dangerous. The policy question is whether liquidity promises, redemption terms and contingent commitments align with the liquidity of the underlying assets.

Tokenised assets. Chapter 3 adds a forward-looking note. Tokenised markets remain small and may bring efficiency gains. The IMF cautions that they can create new channels for leverage and for liquidity and maturity mismatches, in addition to technology and infrastructure dependencies.

6. Banking-System Resilience and Stress Testing

The IMF's 2026 Global Bank Stress Test applies an adverse scenario with five components: a reassessment of AI productivity gains and a sharp correction in technology valuations; short- and long-term yields up by 150 and 300 basis points; global real GDP growth about 3 percentage points below baseline in the first two years; global unemployment up 3.5 percentage points; and inflation more than 4 percentage points higher by 2027.

The headline result is reassuring. The aggregate Common Equity Tier 1 (CET1) ratio of the global banking system falls by about 150 basis points to approximately 11.5 per cent, driven mainly by credit impairments, partly offset by stronger net interest income.

The qualification matters more. Banks classified as weak, meaning those with post-stress CET1 ratios below 7 per cent, represent 22 per cent of global banking assets, up from 18 per cent in the 2025 test. The drivers differ by region. In North America the increase comes from market risk, as sharp rises in long-term government yields cause valuation losses, particularly at a small number of large banks. In Asia it reflects capital depletion in oil- and gas-importing economies. Most other regions, particularly emerging markets, show more limited depletion.

The IMF also tests exposure to private assets. If bank losses equal 20 per cent of their private-asset exposures, aggregate CET1 capital in the US sample falls by about 2 percentage points, and banks representing less than 1.6 per cent of sample assets would see Tier 1 ratios fall below 7 per cent. In the EU and UK, the same shock costs about 1 percentage point of CET1 and no bank falls below the threshold. The IMF concludes that stand-alone NBFI and private-asset events appear absorbable, but that vulnerabilities could be amplified if such losses coincide with a severe downturn.

How to read these numbers. The 22 per cent figure is a share of assets held by banks that fall below a stress-test threshold under a hypothetical scenario. It is not a forecast of losses, and it does not mean those banks would be insolvent. The relevant distinctions are between scenario, vulnerability, potential loss, realised loss and insolvency. What the result does show is the distribution: aggregate resilience coexists with pockets of weakness, and those pockets are linked to bond-market risk, energy exposure and NBFI interconnections.

7. Cyber Risk and Financial Infrastructure

Cyber risk belongs in a financial-stability report because it attacks the system's functions rather than its balance sheets. A major incident affecting payments, trading or access to deposits could cause immediate losses and undermine confidence. The IMF rates the cyber risk as worse than in April.

Three trends drive that judgement. Digitalisation has widened the attack surface through cloud services, third-party providers and real-time payment and settlement systems. Reliance on a small number of critical technology providers and complex dependency chains creates concentration risk and common points of failure. And threats have become more sophisticated, including frontier AI models with strong network-attack capabilities. The defensive picture is also deteriorating: in more than 70 per cent of cases so far in 2026, software providers had almost no time to patch vulnerabilities before exploitation began, 20 percentage points higher than a year ago. The IMF notes that frontier AI is both an enabler of attacks and an important defensive tool.

Our interpretation. A well-capitalised and liquid institution can still be exposed to an operational disruption at a shared provider. Conventional balance-sheet analysis does not capture that concentration. The combination with market stress is the part we would watch: an outage during a period of volatility, when liquidity is already thin, would be more damaging than the same outage in calm conditions. The IMF's priorities are continuity of critical services, incident reporting, third-party oversight and resilience testing, with emerging markets building foundational cyber hygiene and supervisory capacity first.

8. Emerging-Market Vulnerabilities

The IMF finds emerging and frontier economies relatively resilient to recent shocks. Risk premiums that widened when the war in the Middle East began have retraced to historically compressed levels, and authorities used targeted fiscal, monetary, currency and capital-flow measures to cushion the energy shock.

The cost shows up in buffers. The measures weighed on fiscal space and reserves, leaving less room if energy pressures intensify or global conditions tighten. Reserve adequacy is comfortable in aggregate but below pre-war levels across most emerging markets. The median ratio as of June 2026 is 135 per cent for investment-grade countries and 117 per cent for below-investment-grade countries. For the weakest group, those both below investment grade and with ratios under 100 per cent, the median fell from 72 per cent in the first quarter to 63 per cent in the second.

Frontier economies show a different pattern. Spreads widened by about 60 basis points at the outbreak of the war and have more than retraced. Frontier issuers made up close to one-fifth of emerging-market sovereign supply in the first half of 2026, including several first-time borrowers with lower ratings. The IMF also notes increasing use of less transparent financing, including sovereign total return swaps, which can create vulnerabilities when borrowing costs rise and are publicly reported in only a few economies.

The IMF also finds that economies relying solely on foreign-exchange intervention saw little meaningful currency recovery, while those that combined intervention with monetary tightening generally recovered more.

Our interpretation. Emerging markets are not a single group. Reserve cover, rating category, debt composition, the investor base and the transparency of financing differ widely. Compressed spreads reflect current risk appetite, not the capacity to finance under less favourable global conditions.

9. An Integrated Macro-Financial Interpretation

The IMF presents three risks and several vulnerabilities in parallel. The analytical question is how they connect. The following is our own interpretation, not an IMF conclusion.

The risks are correlated through rates and collateral. A correction in AI valuations lowers equity wealth and may weaken investment. A rise in long-term yields raises discount rates, which pressures the same valuations. Hyperscaler bond issuance competes with sovereign supply for the same investors. The IMF describes these links individually. We would stress that they run in both directions, so a shock to one market can transmit to the other through the cost of capital.

Leverage determines whether a correction is orderly. Equity margin loans, leveraged ETFs and hedge-fund positions are the transmission devices in the AI chain. Sovereign repo and prime-brokerage financing are the equivalents in the bond market. When both are hit at once, margin calls arrive while liquidity is thinnest.

Funding liquidity is the second stage. A modest initial price decline can become larger if holders must sell. Funds with semi-liquid structures face redemptions while contingent credit lines are drawn. Insurers and pension funds with illiquid portfolios may sell what remains liquid, which is often sovereign and investment-grade corporate bonds, adding to yield pressure.

An illustrative sequence (hypothetical, not a forecast). Suppose inflation uncertainty and heavy issuance push term premiums higher while AI profit expectations are revised down. Equity valuations fall as discount rates rise. Leveraged investors face collateral calls on both equities and bonds. Some NBFIs raise cash by selling liquid assets. Banks financing those investors face higher counterparty risk and valuation losses, concentrated, as the stress test suggests, in the institutions most exposed to bond-market risk. Credit conditions tighten for the real economy. The outcome would depend on capital buffers, liquidity availability, the distribution of exposures and the policy response.

Which channels deserve the closest monitoring. In our view: the term premium and the rollover profile of sovereign debt; leverage in AI-linked equity financing; contingent liquidity in private credit; the liquidity of dealer-intermediated bond markets; and bank-NBFI financing links. Cyber disruption is a separate catalyst that can arrive at any point in the sequence.

Limits of the framework. The IMF is clear that its risks are plausible catalysts, not predictions, and that stress-test outcomes are conditional on scenarios. Measurement is also uneven outside the regulated banking sector, and the IMF repeatedly calls for better data on leverage and contingent liquidity. Any analysis of NBFI transmission therefore rests on partial information and behavioural assumptions. Some risks are also moving in a more favourable direction: the IMF notes that household debt burdens have eased and that some bond-market leverage has unwound. The framework is for understanding mechanisms, and it should not be read as a view that stress is imminent.

10. Implications for Institutional Investors

This section describes how the IMF's findings can be incorporated into a macro-risk framework. It does not provide investment recommendations.

Duration and sovereign exposure. Long-dated assets are sensitive to real yields, inflation risk premia and term premia, not only to the policy-rate path. Sovereign analysis should combine fiscal fundamentals with market-financing conditions such as supply, maturity structure and the investor base.

Credit. Refinancing risk can rise without a deterioration in current earnings. Maturity walls, interest coverage and sensitivity to higher base rates deserve weight alongside ratings. The IMF's evidence on rising leverage among direct-lending borrowers is an example.

Equity concentration and AI exposure. AI exposure should be assessed through the financing chain, including capital expenditure relative to operating cash flow, debt maturity, reliance on related-party revenue commitments, and the leverage of the investor base.

Liquidity and leverage. Portfolio stress tests should include collateral calls, redemptions and reduced market depth, not only price declines. The relevant question is whether an asset can be held through stress, and not only what it is worth.

Counterparty and operational risk. Concentration in prime brokers, custodians, cloud providers and critical technology vendors is a portfolio risk as well as a firm-level one.

Cross-asset correlations. Diversification that relies on historical correlations may weaken when equities, credit and duration respond to the same shock, such as a sharp rise in real yields or a fall in risk appetite. Historical diversification patterns are a hypothesis to be tested, and should not be treated as a guarantee.

11. What Should Institutional Investors Monitor?

The table below is our own monitoring framework, not an IMF product. The IMF supplies the diagnosis; the indicators and the descriptions of deterioration are our judgement. We have deliberately avoided numerical thresholds, which depend on each investor's mandate and liabilities.

No single indicator is decisive. The signal is a combination: rising term premiums and rising leverage and falling liquidity at the same time.

Conclusion

The October 2026 GFSR shows a financial system that has been resilient so far and whose resilience is partly conditional on favourable valuations. The IMF does not predict a crisis. It identifies three catalysts, a set of amplifying vulnerabilities, and a banking system that is broadly sound in aggregate with a growing share of weaker institutions.

Our reading is that the sovereign bond market is the node where the other risks meet: it sets the discount rate for AI valuations, supplies the collateral for leveraged funding, anchors bank balance sheets and prices long-term capital for governments, firms and households. Whether a shock stays contained depends less on its origin than on leverage, liquidity and interconnection at the moment it arrives.

Global financial markets may be resilient today, but resilience should not be confused with the absence of vulnerability. The practical task for investors and policymakers is to understand how those vulnerabilities interact, and to monitor the points where they connect.

References: International Monetary Fund, Global Financial Stability Report, October 2026, Executive Summary, Chapter 1 (Beneath the Resilience: Rising Risks, Unresolved Vulnerabilities), Chapter 2 (hedge funds) and Chapter 3 (tokenisation), available at imf.org. IMF Blog, 12 October 2026, on the report's key risks (imf.org/en/blogs/articles/2026/10/12/global-financial-markets-face-rising-risks-from-ai-bond-yields-and-cyberattacks).

This review is an independent analytical interpretation of the IMF publication. It is intended for research and educational purposes only and does not constitute investment advice or a recommendation to buy or sell financial instruments.