// analysis
The risk that goes in circles
European banks use synthetic risk transfers to cut capital on retained loans. Trace investors, financing links and when risk circles back.
A bank can keep a loan on its balance sheet while transferring part of its credit risk to an investor. If the supervisor recognises that the transfer is sufficiently significant, the capital requirement attached to those exposures can fall. The mechanism is legitimate, supervised and, at present, authorities do not describe SRTs as an imminent systemic threat. But the European market is expanding quickly. A second question follows: when the funds absorbing the risk are themselves financed by banks, how far has that risk actually left the banking system?
One detail changes the entire story: the loan does not move.
In a synthetic securitisation, the bank keeps the asset, continues to manage the borrower relationship and continues to receive the loan’s cash flows. What it transfers is a defined layer of potential credit losses on a reference portfolio.
The Bank for International Settlements describes three main structures: a credit-linked note, or CLN, issued directly by the bank; a financial guarantee or credit derivative with an investor; or a CLN issued by a special-purpose vehicle that itself provides the protection to the bank.
The acronym SRT requires one clarification. In this article it means synthetic risk transfer. In Europe, the same initials are also used for significant risk transfer, the prudential recognition that a synthetic or traditional securitisation has transferred enough risk to justify capital relief. The concepts often overlap, but they are not identical.
That matters because the bank does not decide on its own that the risk has disappeared.
In the European Union, the supervisor must recognise the significant transfer before the bank can take the prudential benefit. The ECB assesses both the individual transaction and the bank’s broader risk-management framework.
What the bank is actually buying
The attraction becomes clear once risk-weighted assets, or RWA, enter the calculation.
The BIS gives a deliberately simplified example. A bank holds a €1 billion loan portfolio with a 65% risk weight. Before the transfer, that produces €650 million of RWA. The portfolio is split into three tranches: 1% junior, 7% mezzanine and 92% senior. The bank retains the junior and senior tranches while hedging the 7% mezzanine tranche through a CLN.
In the BIS illustration, RWA then falls from roughly €650 million to €263 million. Assuming a 12.5% CET1 ratio, required capital falls from roughly €82 million to €33 million.
This is not an estimate of the average return generated by an actual SRT. Nor has the bank created €49 million of new capital. It is a stylised example showing why paying for protection can be preferable to selling loans, shrinking the balance sheet or issuing new equity.
This optimisation is not inherently suspect. It can reduce risk concentration, diversify who absorbs losses and allow a bank to preserve a lending relationship it does not want to sell.
The useful question therefore lies elsewhere: who bears the losses after the transaction, how is that investor funded, and can it keep bearing them when the cycle turns?
Europe has accelerated
European numbers require some discipline because several different quantities are often mixed together.
In May 2026, the ECB reported that the significant institutions it supervises had originated synthetic securitisations covering €258 billion of underlying portfolios in 2025, up from €175 billion in 2024. That was a 47% year-on-year increase and a 90% increase between 2022 and 2025.
Within the same perimeter, the stock of underlying exposures rose from €223 billion at the end of 2022 to €570 billion at the end of 2025.
Those figures measure the notional value of the reference portfolios. A €258 billion reference pool does not mean €258 billion of risk was transferred, nor €258 billion of CLNs was issued. Only selected tranches are protected.
Internationally, the BIS estimated that SRTs protected loan portfolios approaching €800 billion at the end of 2024. Using a different scope and methodology, the Basel Committee estimated roughly €750 billion of protected assets in Canada, the euro area, the United States and the United Kingdom, equal to around 1.1% of total bank assets in those jurisdictions.
The two numbers should not be added together. Their difference is informative: the BIS notes that there is still no global data repository or consistent regulatory reporting covering SRT issuance, pricing and credit performance across jurisdictions.
The BIS provides another useful benchmark. In its sample of 44 issuing banks at end-2024, SRTs protected around 5% of loans on average. Estimated capital relief was around 43 basis points of CET1 on average, with a few cases above 100 basis points.
The market has therefore become meaningful without yet dominating aggregate bank balance sheets. The ten largest issuers represented 64% of the outstanding amount in the BIS sample, while 90% of protected assets were wholesale exposures, mostly corporate loans.
The more interesting part begins with the investor
The transfer is economically useful when the investor taking the tranche can actually absorb the losses.
Buyers are mostly non-banks: credit funds, asset managers, hedge funds, pension funds, insurers and, in some jurisdictions, public institutions. Modern structures are often funded or collateralised upfront, materially reducing the risk that an investor promises protection it cannot subsequently deliver.
In May 2026, the ECB said unfunded protection provided by counterparties other than governments and development banks represented only 11% of outstanding protected tranches. That is an important resilience feature and belongs in any serious assessment.
An investor can nevertheless finance its purchase.
A CLN can be pledged as collateral in repo. The fund contributes some capital and borrows the remainder from a bank. The BIS reports that such financing is typically subject to substantial haircuts, often 40-60%, daily margining and cross-collateralisation arrangements.
Another distinction is essential: the available evidence does not justify writing that banks generally finance their own SRTs.
The BIS says the bank financing the investor is generally different from the bank that originated the transaction. This is an interconnected banking system, not necessarily a closed loop inside one institution.
That changes the risk without removing it.
What the ECB is still trying to measure
On 24 March 2026, ECB Supervisory Board member Pedro Machado disclosed a new survey of a broad set of banks on SRT investor financing. Its scope explicitly includes funding provided by significant banks to investors buying securitisations originated by other banks.
Two months later, the ECB framed the problem more directly. As volumes rise, bank-nonbank interconnections deepen in ways that are not always fully mapped.
It also said data gaps and data-quality issues still prevent a fully reliable assessment of spillover risks, particularly regarding exposures to investors, investor characteristics and post-transaction performance of reference portfolios.
That is where the real risk angle sits.
The market is not unknown to supervisors. Individual transactions are supervised. Protection structures are documented. What remains incomplete is the cross-sectional map, bank by bank, fund by fund and jurisdiction by jurisdiction.
The BIS reaches the same diagnosis. It uses the expression “circles of risk” for situations in which risk transferred from a bank to a fund can return indirectly to banking because another bank finances the fund’s purchase.
It immediately adds that available evidence suggests these loops remain modest in scale.
That second sentence is what prevents a serious subject from becoming an artificial crisis narrative.
An ECB study adds three harder questions
An ECB Working Paper published in March 2026, by Alex Osberghaus and Glenn Schepens, uses euro-area transaction-level data to study SRT use.
The authors report three results worth watching.
First, banks tend to transfer loans that are relatively capital-intensive compared with their estimated economic risk. That is consistent with RWA optimisation being a central use of the instrument.
Second, their analysis finds weaker internal borrower monitoring after a synthetic transfer.
Third, banks are more likely to sell protection to non-bank investors with which they already have a lending relationship.
The paper matters because it uses micro data that are not available to the public. Its status must also remain explicit: it is a research Working Paper, not an official ECB policy position. Its findings complement prudential evidence; they do not by themselves establish a systemic vulnerability.
The ECB itself noted in March that observed leverage among these investors remains modest on average and that system-wide round-tripping risks appear contained for now.
Private credit makes the map wider
SRTs sit inside a much broader web of bank and non-bank financial relationships.
In its June 2026 Risk Assessment Report, the European Banking Authority identified almost €150 billion of EU/EEA bank exposures to private-credit funds and related asset managers in June 2025. Those exposures were spread across 79 banks in 13 Member States and represented 0.6% of total assets on average.
That figure does not measure SRT financing.
It includes several kinds of exposure to the private-credit ecosystem, and the EBA explicitly calls it an indicative measure constrained by reporting thresholds and available data.
Its relevance is different. A fund buying an SRT tranche can have several other relationships with banks: credit lines, collateralised financing, lending to common borrowers, vehicle financing or links through asset managers.
Systemic risk therefore depends less on the legal label attached to one transaction than on the sum of dependencies connecting participants.
A credible stress scenario
The useful scenario is not “SRTs cause the next financial crisis”.
Primary sources available as of 18 August 2026 do not support that claim.
A more rigorous sequence looks like this:
- a recession or sector shock raises defaults across several reference portfolios;
- protected tranches begin absorbing the losses defined in their contracts;
- CLNs pledged as collateral may lose value or become harder to price;
- banks financing some investors may demand more collateral or reduce credit;
- investor appetite for new SRT protection may weaken;
- banks that had incorporated renewal of those protections into capital planning must adapt by retaining more RWA, raising capital, selling assets or slowing some new lending.
The BIS therefore presents SRTs less as a likely source of the original shock than as a possible transmission amplifier, capable under stress of reinforcing both tighter bank lending and adverse bank-NBFI feedback loops.
The ECB watches three concrete channels: rollover risk, counterparty risk in unfunded structures and flowback risk, when capital requirements on retained senior tranches rise again.
Its May 2026 assessment remains reassuring: it had found no material maturity wall, no acute concentration of unfunded counterparty exposures and no immediate system-wide flowback vulnerability.
That is exactly why the issue is worth mapping before stress makes the exercise far more costly.
Europe wants securitisation to grow at the same time
The apparent contradiction is not really one.
The European Commission wants to revive securitisation as a risk-sharing and financing tool. Its reform package presented in June 2025 aims to simplify parts of the framework and make prudential treatment more risk-sensitive. The Council adopted its negotiating position in December 2025.
The ECB supports a sound securitisation market while arguing that simplification must not weaken transparency or the quality of the risk transfer.
Since January 2026 it has also operated a fast-track process for sufficiently simple and standardised SRT transactions. In the first four months of 2026, only two transactions had used that route, both assessed in eight working days.
Europe’s objective is therefore to expand a tool supervisors consider useful while improving visibility into the chains that form around it.
The policy challenge is not to prevent risk transfer. It is to make sure that microprudential capital relief at one bank does not, through accumulated cross-exposures, become a macroprudential blind spot.
What we know, and what we do not
Several conclusions are sufficiently documented to state firmly.
The market is growing rapidly, particularly in Europe. The loans remain on bank balance sheets. A tranche of credit risk is genuinely transferred and, in most modern structures, protection is funded or collateralised. Capital relief is real when recognised by supervisors. Some funds use bank financing for SRT positions. Banks and the investors buying protection are interconnected.
Several stronger claims would go beyond the evidence.
Public data do not show that SRTs are currently a crisis in waiting. They do not show that banks generally finance their own risk transfers. They do not allow us to quantify precisely how much transferred risk returns indirectly to banking. And the hundreds of billions of “protected” portfolios are neither expected losses nor the amount of investor capital actually committed.
That boundary between what can be measured and what remains only partially mapped may be the most important signal in the entire market.
The risk is not mysterious. It changes owner, legal form and sometimes funding source.
The question is whether, when a shock arrives, the institution that agreed to absorb it still has the resources to do so without relying on the same banking system that wanted to shed it.
Method and limitations
Data cut-off: 18 August 2026.
ECB, BIS and Basel Committee amounts are not added together because their scopes, dates and methodologies differ. “Protected” or “securitised” portfolio amounts refer to underlying reference assets, not to the size of the risk tranche transferred.
This article prioritises supervisors, public institutions and primary research publications. Research papers are explicitly labelled as research. Secondary or paywalled figures used in the previous version of this article were removed whenever a primary source could establish the underlying fact.
Primary sources
- Bank for International Settlements, The rise and risks of synthetic risk transfers, 16 March 2026
- Basel Committee on Banking Supervision, Synthetic risk transfers, 17 February 2026
- ECB, Pedro Machado, Changing the tune but not the tone: synthetic risk transfers in Europe, 24 March 2026
- ECB, Pedro Machado, Strengthening the supervisory grip on securitisation: reading the data, anticipating the risks, 14 May 2026
- ECB Working Paper Series No. 3210, Synthetic, but how much risk transfer?, 27 March 2026
- European Banking Authority, Risk Assessment Report, June 2026
- European Commission, Securitisation policy file
- Council of the European Union, position on reviving the EU securitisation market, 19 December 2025
For the broader migration of credit risk towards non-bank finance, see The migration of credit risk.
This analysis is not investment advice.
// cite this analysis
l0g, “The risk that goes in circles”, l0g.fr, published July 28, 2026, updated August 18, 2026, https://l0g.fr/en/analysis/the-risk-that-goes-in-circles/
$ cd ../analysis