Private Credit Risk Timeline 2008–2026: SRTs, Leverage and the Shadow-Banking Shift

Ask “how big is private credit?” and the honest answer is “it depends what you count.” Strict private-credit funds and direct lending are roughly $1.5–$2 trillion. Broader definitions add investment-grade private lending, asset-backed finance, mortgages, infrastructure credit and other privately originated assets, which produces $30 trillion or $40 trillion-plus addressable estimates. Apollo’s CEO has described a $40 trillion universe, about 95% of it investment grade, with only around $2 trillion in the levered direct lending at the centre of most risk commentary. PIMCO has pushed back: it argues that popular addressable-market sizing lumps in every loan that could theoretically be refinanced privately, and puts a disciplined long-run estimate at $6–$8 trillion once loans already securitised, held by insurers or owned by existing funds are stripped out.

Every number needs a source, a date and a definition. FSB: $1.5–$2 trillion, assets, end-2024. Apollo: $40 trillion-plus, addressable / private investment-grade credit definition. Writing “the $41 trillion private-credit market” collapses those into one figure that no regulator uses. The accurate phrasing is “the $40–$41 trillion addressable credit market cited by some industry participants.”

A traditional corporate bond goes from a company, through the public bond market, to many investors. A bank loan goes from a company to a bank. Private credit goes from a company to a private fund or institutional lender, through a negotiated loan that is generally not publicly traded and whose terms can be bespoke.

There is no single “private credit”

Risk differs substantially across segments: direct lending, investment-grade private placements, asset-backed finance, distressed and special situations, mezzanine, real-estate debt and infrastructure debt. A senior investment-grade infrastructure loan and a second-lien loan to a leveraged software company are both “private credit,” and they are not the same asset. Private credit has also expanded from middle-market lending into large corporate deals and asset-backed finance — consumer loans, equipment, aircraft, data centres, royalties and receivables — a structural shift, not just growth.

“Shadow banking” is a reader-friendly historical term for non-bank financial intermediation (NBFI) — credit provided outside the regulated banking system. Private credit is one component. Others include money-market funds, hedge funds, finance companies and securitisation vehicles. Pension funds and insurers can invest in private credit without themselves becoming “shadow banks.” Regulators increasingly use “NBFI” precisely because “shadow banking” implies money vanishing into a separate system, when the reality is a network that stays tightly connected to banks.

Private credit is also not “unregulated.” Managers and vehicles are subject to securities law, fund rules, fiduciary duties, investor-protection requirements and, in some vehicles, leverage limits. What differs is prudential regulation: private-credit vehicles are less directly prudentially regulated than bank balance sheets, not exempt from oversight.

A public bond has a market price that updates continuously. A private loan may not trade at all, so its value can depend on manager marks, third-party valuation, models, comparable securities and borrower fundamentals. This can smooth reported volatility — but the economic risk still exists, and can show up later and more abruptly than in public markets. PIMCO and Apollo spent much of 2026 publicly disagreeing about how often private assets should be re-marked and how conservative those marks should be.

On 2 September 2026, Reuters published an analysis of regulatory filings from 44 US business development companies (BDCs). Their combined investments were worth $92.88 billion in fair value on 30 June against $95.19 billion of reported cost. The aggregate fair-value-to-cost ratio fell from 99.25% at end-December 2025 to 97.77% in the first quarter and 97.57% by mid-2026 — portfolio values moving below cost overall. Across 10 BDCs with comparable filings, loans on non-accrual — where a borrower is significantly behind or considered unlikely to pay, so interest is no longer recognised normally — rose from 2.5% of portfolio cost at end-2025 to about 3.4% by end-June. Non-accrual is not the same as total loss, but rising non-accruals are a stress indicator. Software loans were marked down more frequently than loans in other industries.

Private equity bought software companies; those companies borrowed, often from private-credit funds; higher rates increased the debt-service burden; and now AI may challenge some of those business models and valuations. Cash-flow pressure can then affect debt service. This is a specific sector-concentration test, not a claim that AI has caused every markdown. It also cuts both ways: private credit is financing large AI-infrastructure and data-centre debt while simultaneously absorbing disruption risk in the software borrowers it already lent to.

Traditional closed-end private-credit funds often have long-term committed capital, which reduces classic run risk. The liquidity story changed as wealth and retail-oriented products grew: the IMF estimates about 15%, or roughly $300 billion, of the ~$2 trillion direct-lending universe now sits in semi-liquid structures that permit periodic — often quarterly — redemptions. Many of these funds cap withdrawals. In June 2026, one large Apollo semi-liquid fund limited redemptions after quarterly exit requests reached around 17% of the fund, reviving debate about whether these vehicles promise more liquidity than their assets can support.

Redemption gates protect remaining investors and reduce forced asset sales, but they also mean investors cannot always withdraw everything immediately. This is not identical to a bank run: banks promise deposit liquidity on demand, while private funds have different redemption contracts. The accurate term is redemption pressure, and “bank run” should be used only as a clearly labelled analogy.

The myth is that private credit replaced banks. In reality, banks originate loans, finance private-credit funds, lend to private-credit borrowers, provide subscription lines, provide NAV loans, create securitisations, execute synthetic risk transfers and distribute private-credit products. The FSB’s May 2026 report captured around $220 billion of drawn and undrawn bank credit lines to private-credit funds across available member data; commercial estimates run to $270–$500 billion. The range is wide because definitions differ, reporting differs and granular data are missing — and that uncertainty is itself one of the regulatory concerns.

Direct exposure can look small while connections are complex. EU banking-sector data compiled by the European Banking Authority has put direct bank exposures to private-credit funds and related managers at roughly €150 billion in mid-2025 — a small share of assets overall, but concentrated in some institutions. A bank with limited direct fund exposure may still share borrowers with funds, provide their credit lines, hold SRT relationships and have insurer links. Headline direct exposure does not reveal the whole network.

A synthetic risk transfer (SRT), also called a significant risk transfer, lets a bank buy protection against some default losses on a portfolio of its loans. An investor agrees to absorb defined losses on a reference portfolio; the loans stay on the bank’s balance sheet; the credit risk is partially transferred; and if regulatory tests are met, the bank may obtain capital relief.

Two corrections matter. First: in a synthetic SRT the bank does not sell the loans to a fund — it transfers defined credit risk while keeping the assets. Second: “insurance” is a useful lay analogy but technically wrong. An SRT is a securitisation / credit-risk-transfer structure with an insurance-like economic effect, not an ordinary insurance policy. Structures split into funded (the investor posts collateral upfront, lowering counterparty-risk concern) and unfunded (the investor provides a guarantee, so collateral is not necessarily upfront and counterparty dependence is greater).

On 17 February 2026 — not September — the Basel Committee on Banking Supervision published an analysis of synthetic risk transfers. Across Canada, the euro area, the United States and the United Kingdom, protected bank assets were estimated at about €750 billion, roughly 1.1% of total bank assets, with a jurisdictional range of roughly 0.9–1.8%. SRTs have been especially popular in the EU, where new transactions more than tripled between 2016 and 2024 after regulatory changes gave some synthetic securitisations preferential treatment. The SRT investor base is dominated by private investment funds, creating a direct bank-to-nonbank credit-risk connection.

The Committee’s conclusion is the sentence that matters: SRT-related risks — including banks’ increased dependence on non-bank financial intermediaries — appear modest at present, are to some extent actively managed, but merit continued monitoring by supervisors as the market grows and structures become more complex. That is “a fast-growing market needs closer monitoring,” not “Basel sounds a crisis alarm.” In its July 2026 Financial Stability Report, the Bank of England noted UK banks exploring riskier unfunded SRTs and said the Prudential Regulation Authority is testing complex or less robust structures; SRT-protected loans still represent only a small share of major UK banks’ loan books.

Did the bank really transfer the risk?

Partly. A bank may finance the SRT investor, have other exposures to the same fund, share the same underlying borrowers, or depend on nonbank protection. One regulatory question is circularity: if a bank finances the same nonbank that is selling it credit protection, some of that protection may be economically less independent than it looks. This does not mean all SRTs are circular — but risk transfer can coexist with new interconnections.

A NAV loan is borrowing by a fund against the value and cash flows of its portfolio assets, used to provide liquidity or finance distributions, new investments or refinancing. As of April 2026, Moody’s reported the global fund-finance market had passed $1 trillion, with private-credit expansion an important driver. That $1 trillion is fund finance overall — subscription facilities, NAV loans and hybrids — not $1 trillion of NAV loans alone.

NAV loans are not automatically dangerous: they can bridge timing mismatches and reduce pressure to sell assets. Risk depends on loan-to-value, asset quality, covenants, maturity, diversification and leverage already present. The concern is layered leverage: a portfolio company may be leveraged, the fund may borrow, the fund’s investor may also use leverage, and a bank may finance one or more layers. Moody’s has also flagged NAV facilities’ exposure to payment-in-kind (PIK) loans.

PIK and “amend and extend”

With PIK, instead of paying all interest in cash, a borrower adds some interest to the loan balance — temporary breathing room, but a rising debt balance. PIK is not automatically a default, though rising PIK use can be a stress indicator. “Amend and extend” — a lender extending maturity or modifying terms — can avoid an immediate default but can also delay recognition of borrower stress. This is why reported private-credit default rates vary so much: whether the number counts only payment defaults, or also selective defaults, distressed exchanges and PIK modifications, changes the answer.

There are real echoes: opaque credit, securitisation and risk transfer, leverage and complex connections. There are also real differences. Private-credit funds often have longer-duration, committed capital and do not rely on demand deposits, so the classic run dynamic is weaker. Post-2008 bank regulation is stronger, and SRTs receive greater prudential scrutiny than many pre-crisis structures. The market has also not yet been tested through a prolonged, severe downturn. The analogy helps frame the research question about correlated automated behaviour — it does not describe a documented repeat of the global financial crisis.