Do Bank Systems With Large Foreign Exposures Suffer Bigger Real-Economy Shocks? Evidence From 2008-2025

International bank exposure did not produce the same outcome in every crisis. Claims had expanded to roughly $32 trillion before 2008 and then fell about 12% over seven quarters, whereas they were still 4.8% higher year on year by mid-2020 during the pandemic. The contrast shows why the scale of foreign lending has to be read together with the type of funding shock, banks’ loss-absorbing capacity and the policy response when explaining growth, employment and consumption outcomes. [2307] [2309] [2317]

Executive Dashboard

• ~$32tn – Cross-border claims before 2008: International claims had reached roughly $32 trillion and were growing at more than 20% year on year, indicating that global bank balance sheets and wholesale funding dependence had expanded rapidly just before the crisis. [2307]
• -$5.024tn / -12% – Seven quarters from the 2008 Q1 peak: A cumulative decline of $5.024 trillion, about 12% from the peak, shows that the crisis produced genuine balance-sheet deleveraging rather than a small valuation adjustment, with direct implications for credit supply. [2309]
• +4.8% y/y – Cross-border claims by mid-2020: Cross-border claims were still 4.8% higher year on year despite the severe pandemic output shock, providing strong evidence that foreign exposure is not mechanically equivalent to credit contraction when buffers and policy support are stronger. [2317]
• 51.0% – Nonbank share of global financial assets, 2024: Nonbank financial institutions held 51.0% of global financial assets, so a modern stress assessment that looks only at banks can miss important market-based funding, collateral and liquidity channels. [2300]

Figure 1. Placing the 12% post-2008 contraction, the sharp 2011 Q4 deleveraging and the 4.8% year-on-year increase by mid-2020 side by side shows that international bank credit did not move in the same direction in every crisis.

Key findings

1. In 2008, foreign exposure lined up with an observable credit-supply channel rather than merely a cross-country correlation. International claims had climbed to about $32 trillion before the crisis and then fell 12% over seven quarters from the 2008 Q1 peak, creating conditions for external funding withdrawal to spill into domestic lending. ~$32tn before crisis; -12% (-$5.024tn) over seven quarters [2307] [2309]
2. The 2011 Q4 euro-area deleveraging episode again showed how stress at globally active banks could be exported to borrowers abroad. Cross-border claims on advanced-economy borrowers fell by about $630 billion in one quarter as the sovereign-bank loop and funding pressure interacted. 2011 Q4: -$630bn [2310] [2311]
3. The pandemic is the strongest counterexample to a mechanical “more foreign exposure means a larger credit collapse” rule. Cross-border claims were still 4.8% higher year on year by mid-2020, showing that the origin of the shock and the strength of bank buffers can overturn the pre-2008 pattern. Mid-2020: +4.8% y/y [2317] [2306]
4. Capital and liquidity conditions materially changed outcomes for a given level of international exposure. The 2008 recapitalizations and the ECB’s 2020 capital and operational relief show that loss-absorbing capacity and policy support can prevent external stress from becoming a generalized domestic lending stop. US TARP up to $250bn; euro-area capital relief above €20bn [2302] [2303] [2306] [2335]
5. Since 2022, household consumption has been more directly tied to debt-service burdens and loan repricing than to foreign bank claims alone. Debt-service ratios and the share of adjustable-rate borrowing determine how quickly monetary tightening reaches household cash flow, creating different lags across countries. Post-2022: DSR and repricing dominate the household transmission channel [2316] [2325] [2339]
6. The modern international financial network extends well beyond banks. Nonbank financial assets grew 9.4% in 2024 and represented 51.0% of global financial assets, so future stress analysis must combine bank foreign claims with funds, collateral, margin calls and market-liquidity channels. 2024 nonbank assets +9.4%; 51.0% global share [2300] [2297]

Direct answer: foreign exposure changed the transmission mechanism, not the outcome by itself

Before the global financial crisis, international banking expanded at exceptional speed. BIS reporting banks held roughly $32 trillion in cross-border claims, with annual growth above 20%. From the 2008 Q1 peak, international claims then fell by about $5.024 trillion over seven quarters, a decline of roughly 12%. When the funding system itself broke, a large external balance sheet made the route from market stress to domestic credit much shorter. [2307] [2309]

That experience should not be turned into a timeless rule. During the first phase of the pandemic, global output contracted more sharply than in 2008, yet cross-border bank claims were still up 4.8% year on year by mid-2020. The difference was the starting condition of banks and the speed of policy support. Foreign exposure mattered, but capital, liquidity and the nature of the shock determined whether it became an amplifier or a buffer. [2317]

A useful distinction is between the level of exposure and the speed at which it was accumulated. A large but stable international book is not equivalent to a smaller book that expanded rapidly on fragile funding. For crisis analysis, the change over the eight to twelve quarters before the event can be as informative as the level itself.

Why foreign exposure amplified the 2008 shock

The pre-2008 international banking boom was not simply a rise in diversified assets. European banks built large dollar positions, including claims on US non-banks, while relying heavily on interbank and wholesale funding. Once market funding froze, institutions that depended on rolling short-term liabilities had to shrink assets quickly. In 2008 Q4, international bank balance sheets contracted on an unprecedented scale and claims on both banks and non-banks fell. [2308]

This is why a foreign-claims-to-GDP ratio cannot be read as a stand-alone risk score. A bank system funded by stable deposits and thick capital can carry a large international book with less pressure to cut domestic lending. A system with similar assets but fragile foreign-currency funding is more exposed to a run in wholesale markets. The same exposure number can therefore imply very different real-economy risks. [2321] [2322]

Recovery speed is another outcome that matters. Two countries can record a similar GDP trough yet follow very different paths if credit supply normalizes quickly in one and remains impaired in the other. Measuring the time back to trend in employment and household spending can reveal persistent banking damage that a one-year growth rate misses.

The euro crisis added a sovereign-bank feedback loop

Cross-border deleveraging returned during the euro-area sovereign debt crisis. In 2011 Q4, claims on advanced-economy bank and non-bank borrowers fell by $630 billion, the largest decline since late 2008, with euro-area banks at the center of the pullback. [2311]

Yet the euro crisis cannot be reduced to foreign lending. Falling sovereign bond prices weakened bank balance sheets, raised funding costs and tightened credit to the real economy. The ECB described this sovereign-bank nexus as a self-reinforcing channel. Two countries with similar external exposure could therefore experience very different outcomes if their domestic sovereign risk, deposit base or access to central-bank liquidity differed. [2310]

Direction matters too. Claims held by banks headquartered in a country are not the same as funding received by borrowers in that country from foreign banks. Keeping the home-bank and host-economy perspectives separate is essential for tracing who exports a shock and who imports it.

Figure 2. Setting the 2008 Q1 peak in international claims equal to 100 gives an endpoint near 88 seven quarters later. The cumulative $5.024 trillion, roughly 12% contraction captures the sustained balance-sheet deleveraging that distinguished the global financial crisis.

The pandemic broke the simple “more exposure, more damage” story

The clearest counterexample came in 2020. Cross-border claims did not collapse even though activity did. BIS research finds that better-capitalized banking systems lent more during the pandemic shock. International diversification, in some settings, can also stabilize lending when weakness is concentrated in one borrower country rather than in the bank’s home market. [2317] [2318]

Regulation and policy response also changed the starting point. The ECB allowed banks to draw down capital and liquidity buffers, while euro-area macroprudential releases created more than EUR 20 billion in capital capacity. Those measures were intended to absorb losses and keep credit flowing. A fall in a regulatory ratio during such a period therefore does not necessarily mean that the bank became structurally weaker; it may show that a buffer was used as designed. [2306] [2335]

For consumption, distribution may matter more than the national average. Adjustable-rate borrowers, fixed-rate borrowers, cash-rich households and liquidity-constrained households react differently to the same policy-rate move. Country averages should therefore be treated as summaries of heterogeneous transmission rather than as a complete household mechanism.

After 2022, household cash flow became a more direct consumption channel

Higher policy rates work through banks, but the household impact depends heavily on how debt is structured. BIS debt-service ratios measure principal and interest payments relative to income. IMF analysis emphasizes that the speed at which higher rates hit households depends on mortgage maturity, refinancing schedules and the prevalence of adjustable rates. [2316] [2339]

A country can therefore have a globally connected banking system and still avoid an immediate consumption shock if household leverage is moderate and debt reprices slowly. The opposite can also occur: a domestically oriented banking system may transmit a strong rate shock if households are highly indebted and loans reset quickly. Since 2022, those cash-flow differences have become a central part of the explanation.

In the high-rate period, the timing of loan repricing matters more for household spending than the headline stock of foreign bank claims. Adjustable-rate borrowers feel tightening quickly, while long fixed-rate borrowers may not face a cash-flow shock until refinancing. Countries with similar banking internationalization can therefore show very different consumption lags because household contracts transmit the same policy move at different speeds. [2316] [2325] [2339]

Credit shocks do not pass one-for-one into household consumption

Households can smooth temporary financing shocks. ECB micro evidence shows that households more exposed to bank funding stress reduced non-housing debt but used liquid assets to stabilize consumption. That finding helps explain why a lending shock may appear quickly in bank balance sheets but only later, or less strongly, in aggregate consumption. [2319]

The protection is uneven. High-saving households have room to draw down liquid assets, while low-income or highly leveraged households may not. Aggregate household saving rates can therefore conceal the distribution that matters for the spending response. A country-level study should treat saving capacity as a buffer, not as proof that all households can absorb a credit shock. [2313]

Household spending does not move one-for-one with bank lending. Credit-constrained households can react quickly to tighter loan supply, while households with liquid assets may smooth spending by drawing down savings. Measuring consumption transmission therefore requires household saving, labor-income and employment conditions in addition to aggregate loan volumes. [2313] [2319]

Figure 3. The matrix crosses high or low foreign exposure with strong or weak capital, liquidity and policy buffers. The most fragile combination is high exposure with weak buffers, while high exposure alone does not mechanically imply large real-economy losses.

Separating foreign exposure from domestic credit is essential

BIS international banking statistics track cross-border and foreign positions, while Total Credit data separate household and non-financial corporate borrowing. Using both avoids a common mistake: assuming that a fall in external claims must produce the same fall in domestic credit. [2315] [2321]

Domestic banks, public programs or alternative lenders may replace part of the withdrawn foreign funding. Conversely, domestic underwriting standards can tighten sharply even when external claims fall only modestly. This is why the more credible empirical design uses pre-crisis exposure and traces its interaction with a crisis shock, rather than treating post-crisis deleveraging as an exogenous cause. Network and home-host studies use related strategies to separate the direction of transmission. [2342] [2343]

Foreign bank exposure and domestic credit can move in different directions. A withdrawal by international banks may be offset by domestic deposits or central-bank liquidity, leaving total credit relatively stable, while a country with little external exposure can still experience a severe home-grown credit bust. International claims and domestic credit should therefore be modeled as separate channels rather than treated as interchangeable measures of leverage. [2315] [2321] [2322]

Capital, liquidity and asset quality determine whether exposure becomes a constraint

Bank capital is the first line of defense against losses, but cross-country capital ratios are not perfectly comparable. IMF FSI metadata show differences in consolidation, accounting and reporting practices. The safest interpretation is to combine the level before a crisis with within-country changes rather than ranking countries mechanically by one ratio. [2323] [2332]

Liquidity and non-performing loans matter for the same reason. A highly international bank with ample liquid assets can meet withdrawals without selling assets into a falling market. A weakly capitalized bank with rising NPLs may cut credit even if its foreign book is smaller. Foreign exposure is therefore best treated as one dimension in a balance-sheet system rather than as a single vulnerability index.

Buffers determine where losses are absorbed. A bank with ample common equity and liquid assets can take losses on foreign positions without immediately cutting domestic lending, whereas a thinly capitalized bank dependent on short-term market funding is more likely to shrink its balance sheet. That difference is why the same gross foreign exposure can produce very different real-economy outcomes. [2323] [2324] [2299]

Policy speed can reverse the apparent relationship between exposure and growth

The United States announced up to $250 billion for the TARP Capital Purchase Program in October 2008, while the United Kingdom announced at least GBP 50 billion of bank recapitalization combined with liquidity and guarantee measures. Both programs explicitly aimed to restore confidence and support credit supply. [2302] [2303]

This creates a major identification problem. The countries hit hardest may also deploy the strongest support, which can reduce the eventual GDP decline. Observing a highly exposed banking system with a modest recession does not prove that the exposure was harmless; it may mean policy offset the transmission. Crisis-by-crisis analysis should therefore account for recapitalization, liquidity support and regulatory relief.

The speed of intervention can change the observed relationship itself. Rapid recapitalization and liquidity support reduce the need for banks to fire-sell assets or cut domestic lending after an external funding shock. Delayed support can turn a similar initial loss into a larger credit contraction, so ex-post GDP comparisons need to account for the losses that policy successfully prevented. [2302] [2303] [2306]

Four episodes should not be compressed into one average coefficient

The 2008 crisis began inside the banking and market-funding system. The 2011-12 euro crisis added sovereign-bank feedback. The 2020 shock began in the real economy and was met with unusually forceful support. The post-2022 episode is primarily a rate and cash-flow adjustment. Because the origin of the shock differs, the role of foreign banking exposure must differ as well.

A stronger research design estimates each episode separately, interacting pre-crisis foreign exposure with bank capital, liquidity and household leverage. It also checks pre-trends and adjusts for valuation effects. Since BIS foreign claims are reported in US dollars, exchange-rate swings can change exposure-to-GDP ratios even without equivalent underlying deleveraging. [2324] [2325]

A single pooled coefficient can hide the mechanisms that differ across the four episodes. Bank funding stress dominated in 2008, sovereign-bank feedbacks mattered in the euro crisis, the pandemic combined a shutdown shock with exceptional support, and the post-2022 period has emphasized interest rates and household cash flow. Episode interactions are more informative than one average exposure effect. [2310] [2317] [2339]

Figure 4. Across the four episodes, foreign exposure mattered most in 2008 and the euro crisis, much less during the pandemic, and only conditionally after 2022 as household leverage and loan repricing became more important for consumption transmission.

Today, nonbank finance is too large to leave outside the picture

Post-crisis bank regulation strengthened capital and liquidity, but risk migrated and diversified across the financial system. The Financial Stability Board reports that nonbank financial assets grew 9.4% in 2024 and represented 51.0% of global financial assets. A bank-only exposure measure therefore captures less of the modern international network than it did before 2008. [2300]

At the end of 2025, large international banks generally showed stable risk-based capital, leverage and liquidity metrics, yet the IMF still highlighted pockets of consumer-credit deterioration and the risk of renewed financial tightening. Future stress tests should connect banks to nonbanks, market valuation losses, collateral calls and funding liquidity rather than stopping at foreign bank claims. [2297] [2299]

More leverage now sits outside traditional bank balance sheets. Funds and other nonbanks can sell assets rapidly when prices fall or collateral requirements rise, and those sales can feed back into banks through market funding and collateral values. Stable foreign bank claims therefore do not guarantee weak international transmission when stress begins in securities markets. [2300] [2297]

Counterevidence and alternative explanations

The strongest counterevidence comes from 2020. The real-economy shock was exceptionally large, yet cross-border bank claims were still 4.8% higher year on year by mid-year and the synchronized foreign-asset retrenchment seen after 2008 did not occur. That contrast implies that the origin of the shock and the strength of capital and liquidity buffers matter more than the stock of foreign claims by itself. [2317] [2306] [2335]

A second explanation is domestic vulnerability. An economy with modest foreign bank exposure can still suffer a severe consumption or credit downturn when household leverage is high, loans reprice quickly or bank asset quality deteriorates. Conversely, a highly international system can absorb an external pullback when domestic deposits and policy liquidity replace lost funding. Ranking countries on foreign claims alone would therefore confuse one transmission channel with overall fragility. [2316] [2323] [2325] [2339]

Current conditions, scenarios and practical implications

Today the relevant network is broader than international banks. Nonbank financial assets grew 9.4% in 2024 and represented 51.0% of global financial assets, while end-2025 bank capital and liquidity metrics were generally stable even as the IMF continued to flag pockets of consumer-credit deterioration and the risk of renewed tightening. A future shock that begins with market prices, collateral or margin calls could tighten credit even if cross-border bank claims themselves remain steady. [2300] [2297] [2299]

Three scenarios are especially useful for decision making. A simultaneous foreign-currency funding squeeze and capital loss resembles 2008 and creates high amplification risk. A large real shock with strong bank buffers and rapid policy support can resemble 2020 and preserve international lending. A stable banking system combined with persistently high rates and elevated household debt-service burdens can still deliver a slow consumption recovery. Stress tests and forecasts should therefore map each scenario to different indicators and policy tools. [2302] [2303] [2306] [2316] [2339]

Conclusion: identify when foreign exposure is an amplifier and when it is not

Three findings are well supported. International bank linkages and deleveraging transmitted the 2008 and euro-area crises. The pandemic showed that large international balance sheets can remain resilient when capital and policy buffers are strong. The high-rate period shows that household debt-service and repricing structures can dominate the consumption channel.

What remains conditional is the size of the effect: the same foreign-claims ratio can produce different outcomes depending on funding, capital, liquidity, domestic credit substitution and policy response. What remains unresolved is a single causal average across countries using fully harmonized long-run measures of capital, NPLs and household loan structures. The best answer to the research question is therefore not “large foreign lending always causes larger recessions,” but “foreign exposure amplifies real-economy shocks when international bank funding is at the center of the crisis and balance-sheet buffers are weak.”

The decisive question is not simply whether foreign exposure is large, but how that exposure is funded and capitalized and where the shock originates. When international funding closes at the same time that loss-absorbing capacity is weak, cross-border links amplify stress. When capital and liquidity are strong and policy rapidly supplies replacement funding, the same network can support diversification and continued lending. [2318] [2342] [2343]

Sources and method

The report combines BIS international banking statistics (CBS/LBS), IMF financial soundness and macro data, World Bank household consumption indicators, and OECD unemployment and saving measures. It does not invent country coefficients where harmonized observations were not supplied; numeric statements are limited to verified aggregate figures and findings in the research pack.

• [2297] International Monetary Fund. Global Financial Stability Report — April 2026. https://www.imf.org/en/publications/gfsr/issues/2026/04/14/global-financial-stability-report-april-2026
• [2299] Basel Committee on Banking Supervision. Basel III monitoring report — end-2025. https://www.bis.org/publications/202609-qis-basel-iii-monitoring-report
• [2300] Financial Stability Board. Global Monitoring Report on Nonbank Financial Intermediation 2025. https://www.fsb.org/2025/12/global-monitoring-report-on-nonbank-financial-intermediation-2025/
• [2302] U.S. Department of the Treasury. TARP Capital Purchase Program. https://home.treasury.gov/news/press-releases/hp1207
• [2303] Bank of England. Recapitalisation of the UK Banking System. https://www.bankofengland.co.uk/news/2008/october/recapitalisation-of-the-uk-banking-system
• [2306] European Central Bank. ECB Banking Supervision provides temporary capital and operational relief. https://www.ecb.europa.eu/press/pr/date/2020/html/ecb.pr200312~45417d8643.en.html
• [2307] Bank for International Settlements. BIS Quarterly Review — March 2008. https://www.bis.org/media-releases/20080303-quarterly-review-march-2008
• [2308] Bank for International Settlements. BIS Quarterly Review — June 2009. https://www.bis.org/media-releases/20090608-quarterly-review-june-2009
• [2309] Bank for International Settlements. BIS Quarterly Review — June 2010. https://www.bis.org/media-releases/20100614-latest-bis-quarterly-review-discusses-financial-turbulence
• [2310] European Central Bank. The euro area sovereign debt market: lessons from the crisis. https://www.ecb.europa.eu/press/key/date/2012/html/sp120628_1.et.html
• [2311] Bank for International Settlements. BIS Quarterly Review — June 2012. https://www.bis.org/media-releases/20120604-june-2012-quarterly-review-returning-doubts-drive-financial-market-volatility
• [2313] OECD. Household savings. https://www.oecd.org/en/data/indicators/household-savings.html
• [2315] Bank for International Settlements. Credit to the non-financial sector — overview. https://data.bis.org/topics/TOTAL_CREDIT?m=6_380_66
• [2316] Bank for International Settlements. Debt service ratios — overview. https://data.bis.org/topics/DSR?m=6_380_671
• [2317] Bank for International Settlements. International banking amidst Covid-19: resilience and drivers. https://www.bis.org/publications/qr-202012/international-banking-amidst-covid-19-resilience-and-drivers
• [2318] Bank for International Settlements. Bank loan supply during crises: the importance of geographic diversification. https://www.bis.org/publications/working-paper-827-bank-loan-supply-during-crises-importance-geographic-diversification
• [2319] European Central Bank. Banks’ financial distress, lending supply and consumption expenditure. https://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1687.pdf
• [2321] Bank for International Settlements. Consolidated banking statistics — overview and metadata. https://data.bis.org/topics/CBS
• [2322] Bank for International Settlements. Locational banking statistics — overview. https://data.bis.org/topics/LBS
• [2323] International Monetary Fund. 2019 Financial Soundness Indicators Compilation Guide. https://www.imf.org/en/data/statistics/fsi-guide
• [2324] World Bank DataBank / IMF FSI. Bank capital to total assets (%) — metadata. https://databank.worldbank.org/metadataglossary/global-financial-development/series/GFDD.SI.03
• [2325] Bank for International Settlements. Debt service ratios — data documentation. https://www.bis.org/statistics/dsr/dsr_doc.pdf
• [2332] International Monetary Fund. FSI Country Metadata Table 2. https://data.imf.org/en/datasets/IMF.STA%3AFSI_COUNTRY_METADATA_TABLE_2
• [2335] European Central Bank. ECB supports macroprudential policy actions in response to coronavirus. https://www.ecb.europa.eu/press/pr/date/2020/html/ecb.pr200415~96f622e255.es.html
• [2339] International Monetary Fund. Global Financial Stability Report — April 2023, Chapter 1. https://www.imf.org/-/media/files/publications/gfsr/2023/april/english/ch1.pdf
• [2342] International Monetary Fund. Crisis Transmission in the Global Banking Network. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Crisis-Transmission-in-the-Global-Banking-Network-43865
• [2343] Bank for International Settlements. Home sweet host: Prudential and monetary policy spillovers through global banks. https://www.bis.org/publications/working-paper-853-home-sweet-host-prudential-and-monetary-policy-spillovers-through-global-banks

Why recovery speed and statistical definitions matter

The trough is only one part of a financial shock. An economy that suffers a sharp one-year contraction and restores credit quickly may incur a smaller long-run cost than one with a milder initial decline but several years of impaired lending. Employment and household spending should therefore be evaluated through both their peak deterioration and the time needed to return toward pre-crisis conditions. [2312] [2314]

This changes how foreign exposure should be interpreted. A rapid withdrawal of cross-border funding can produce a visible short-run shock, yet domestic banks or public credit programs may replace it and accelerate recovery. Conversely, limited external withdrawal can still be followed by a slow recovery when NPLs stay high and recapitalization is delayed. Outcome variables should include both maximum damage and recovery duration.

CBS and LBS are both BIS international banking datasets, but they answer different questions. CBS follows consolidated banking groups by nationality, while LBS records positions according to the location of banking offices. Treating them as interchangeable can double-count positions or confuse home-country and host-country transmission. [2321] [2322]

Capital and NPL ratios are also not perfectly comparable across countries. Accounting rules, consolidation, intragroup netting and reporting frequency differ. It is safer to compare each country with its own history and use several balance-sheet indicators together than to build a precise ranking from small level differences. [2323] [2332]

Exchange rates add another technical problem. Foreign claims are often reported in US dollars, while GDP ratios require their own conversions. A crisis-period currency swing can move the numerator and denominator even when underlying lending changes less. Valuation effects, statistical breaks and expansions in reporting coverage should therefore be flagged before interpreting a change as true deleveraging.