Structural Reads · Public Series

Private Credit Under the Total Portfolio Approach

A structural reading of the private-credit system, and of the allocation mechanism that could change its regime.

Wim Van Laere · Integration Capacity Analysis · integrationcapacity.org · 10 July 2026

Solvency clock

Looks sound. The built layer is real: a $1.5–2.0 trillion asset class, deep committed capital, low headline default rates, investment-grade ratings on much of the paper.

Coherence clock

Watch. The measures that move first are moving: defaults including quiet restructurings at a record, payment-in-kind income rising, public prices standing far below private marks.

Exposed flank

Refinancing that depends on continued institutional inflow — and a new allocation method, the Total Portfolio Approach, that changes how fast that inflow can turn.

How to read this document. A directional, structural read of where a gap sits and how it behaves under pressure — not a forecast of prices, defaults, or any dated event. It structures a thesis published by Mark Farrington and credits it as such. Hand read on public data; no model run. Research-grade, under live forward test — consistency, not validation.

1 · Read this first

This report examines the private-credit system — the funds, the borrowers, the capital providers behind them — with the structural lens used in the earlier public reads in this series (maritime trade, European energy). The question the lens asks is always the same: not whether the system looks solvent, but whether it still functions as one purposeful whole, and what happens at the moment real pressure arrives.

The thesis this report structures is not ours. In 2026 Mark Farrington, a macro portfolio manager with three decades of experience managing currency and macro risk for large institutional allocators, published the argument that the adoption of the Total Portfolio Approach (TPA) by pension funds, insurers and sovereign wealth funds creates a specific vulnerability in private credit: the same institutions that supply, in his estimate, around seventy percent of the capital to this asset class are changing the machinery by which they allocate it, from slow benchmark-driven processes to fast, centralised, model-driven reallocation. His conclusion is that a buyers' strike — a pause in new institutional allocations during stress — would meet the market's refinancing needs at the worst possible moment, and that the synthetic hedges TPA managers place on illiquid books carry a basis risk that shows itself exactly then. The full argument is in his paper, listed in the sources; the reader should go to it.

What this report adds is a skeleton. It places his argument inside a structure that has been applied to other systems before, states which parts of the system sit on which clock, names the reserves being drawn, and ends in a watchlist of observable measures that would confirm or weaken the reading. It complements his analysis; it does not call a trade, and it makes no forecast.

2 · The system, briefly

Private credit is direct lending outside banks and public bond markets: funds raise committed capital from institutions and lend it, mostly at floating rates, mostly to companies backed by private-equity sponsors. The Financial Stability Board, in its report of 6 May 2026, sizes the asset class at $1.5–2.0 trillion at end-2024, concentrated in a few jurisdictions and, by sector, in technology, healthcare and services. The same report states that borrowers in this market typically carry lower credit quality and higher leverage than comparable public-market borrowers.

The capital comes overwhelmingly from institutions. Farrington puts the share supplied by pensions, insurers and sovereign wealth funds at about seventy percent; publicly traded business development companies (BDCs) — the retail window into the asset class — account for perhaps ten to fifteen percent of the market and are the only part of it that is priced daily. Around the funds stands a second circle: banks extending credit lines to the funds themselves (the FSB counts roughly $220 billion of drawn and undrawn bank lines, and notes commercial estimates ranging from $270 to $500 billion), and insurers holding rated private-credit paper on their balance sheets.

One historical fact matters for everything that follows: this market, at this size, has never been through a full credit cycle. The 2008 crisis predates it; the 2020 shock preceded the widespread adoption of TPA; the 2022–2024 inflation period came with sustained growth and fiscal support; the 2023 bank failures were a duration event, not a credit event. The system's behaviour under a real credit downturn, under its present owners and their present methods, is untested.

3 · The two clocks

The reading instrument is the same used in the earlier reads of this series. Every durable system runs on two clocks. The solvency clock counts what is built and banked: assets, committed capital, contractual yield, ratings. The coherence clock counts whether the system still holds together as a purposeful whole: whether its stated purpose matches its conduct, whether its information is honest, whether its reserves are being renewed or consumed. The two clocks can diverge for a long time, because reserves can keep a system looking healthy while its coherence is spent. The work of a structural read is to catch that divergence while the surface is still calm.

Farrington's own classification maps onto this cleanly, and the correspondence is his to claim: what he calls the perpetual, mean-reverting layer of a market is what the solvency clock measures, and what he calls the transient, breakout-prone layer — where regime change gathers — is what the coherence clock measures. His trading question, whether the surface is pricing mean-reversion while a breakout builds underneath, is the same question as whether the two clocks are diverging.

On the solvency clock, private credit looks sound, and the reading here is that this soundness is real, not fake. The committed capital is real; the floating-rate contractual yield, a pickup of several percentage points over public markets, is real; a large share of institutional portfolios is now allocated to the asset class, and reporting through May 2026 shows pension funds still adding to it. None of this is disputed in this report.

The coherence clock is the subject of the next three sections: the gap between what the system says about itself and what it does (section 4), the reserves it is drawing on (section 5), and the mechanism that could flip the direction of those reserves (section 6).

4 · The gap between the marks and the prices

The private-credit system reports its own health through two channels it largely controls: default statistics and net asset values. Both channels currently say calm. Both are contradicted by channels the system does not control.

Consider defaults. Headline default rates in private credit are commonly reported in the low single digits. But when Fitch Ratings counts defaults including liability-management exercises — the quiet restructurings in which a loan's terms are amended, covenants reset, cash interest converted to payment-in-kind — the United States private-credit default rate reached a record of about six percent in April 2026. The difference between the two numbers is not noise; it is a measurement choice. An amend-and-extend keeps the headline clean while the loss is deferred and compounded. Payment-in-kind tells the same story from another side: PIK income has been rising across BDC books, and reporting through late 2025 and early 2026 (Bloomberg, drawing on rating-agency data) describes a growing pile of bad PIK — interest converted to paper claims on borrowers already under stress — as pointing toward future defaults.

Consider valuations. Net asset values in private credit are model marks, set by managers, tested infrequently. They are smoother than public prices because they are tested less often — not because the risk is lower. Where the same assets meet a real price, the gap is visible. Publicly traded BDCs have traded at deep discounts to their stated NAV: Blue Owl Capital Corporation, one of the largest, moved from roughly a twenty percent discount in November 2025 to roughly twenty-five percent by early April 2026. In April 2026 Saba Capital — a fund known for arbitraging exactly this kind of gap — announced a strategy to buy stakes in private BDCs and interval funds from investors who need liquidity, at discounts it targets at thirty to forty percent below NAV. Mercer Capital's reading of this evidence is direct: if an investor cannot readily sell at NAV, then NAV is not fair value, and current public prices suggest that private marks are lagging the price signals of the public markets.

In the vocabulary of this series, this is a say-do gap made unusually visible: the system's statement about itself (its marks, its headline defaults) and its conduct where conduct can be observed (real transactions, real restructurings) have moved apart, and the size of the discount is close to a direct measurement of the distance. In April 2026 the chief executive of JPMorgan used his annual letter to warn that losses in private credit may prove larger than expected and that not all marks are conservative. The system's calm, in other words, is not evidence of safety; it is evidence that the parts of the system that would register stress are the parts tested least often.

5 · The reserves being drawn

In this series, a system's endurance is read through the reserves it can draw on. Two are active here; a third is not applicable and is said so plainly — this is a financial system, and no reserve of nature enters the read.

The reserve of other people's capital. The deepest reserve in private credit is the standing inflow from institutions: pension funds, insurers, sovereign wealth funds, and increasingly retail savers through non-traded BDCs and interval funds. As long as the inflow continues, maturing loans are refinanced, stressed borrowers are amended and extended, and the marks are never tested by forced sales. Reporting through May 2026 shows this reserve still filling — pension funds adding to private-credit allocations despite the visible cracks. But this reserve has one property that distinguishes it from capital already committed: it is governed by allocation processes, and allocation processes can change. That is the subject of section 6.

The reserve of the future. The second reserve is time borrowed against later repayment. Every payment-in-kind conversion moves cash interest into a larger claim on a future that must then perform; every amend-and-extend moves a maturity into a heavier refinancing later. This reserve is being drawn at a rising rate — the record six percent default measure of section 4 is, in large part, a measurement of exactly these deferrals — and it concentrates the system's obligations into the refinancing window now open. In adjacent markets the same window is crowded: roughly $875 billion of US property debt comes due in 2026 alone (Bloomberg, February 2026), competing for the same institutional capital. Farrington's own framing of 2026 is that rollover demand builds through the year and must find new institutional money — at exactly the moment the owners of that money are re-examining how they allocate it.

A system drawing simultaneously on inflow and on deferral can look healthy for a long time. The structural point is narrow: the calm is financed, not earned, and both sources of financing depend on the same group of institutions continuing to behave as they have behaved.

6 · The mechanism: what TPA changes

The Total Portfolio Approach is an allocation method. In the traditional model, an institution sets a strategic asset allocation — so many percent to each asset class — and specialist teams fill the buckets against benchmarks; reallocation between buckets is slow, deliberate, and rare. Under TPA, the asset-class silos are dissolved: a central team manages the whole portfolio against a single risk model, compares every position with every other as competing uses of the same risk budget, and reallocates when the model's view changes. In November 2025 CalPERS, the largest public pension fund in the United States, became the first of its kind to adopt the method, explicitly to gain speed — in its own words, to seize market opportunities. In 2026 Korea Investment Corporation confirmed TPA as an organisational priority and began piloting it on a share of its assets. The direction of travel among large allocators is one way.

What this changes for private credit is not the stock of capital but the behaviour of the reserve described in section 5. A bucket-filling allocator is structurally patient: private credit has a bucket, the bucket refills, and illiquidity is tolerated because nothing competes with the bucket directly. A total-portfolio allocator is structurally comparative: every quarter, private credit must defend its place against everything else in the risk model — and in stress, when the model's volatility and correlation inputs jump, the positions that can be adjusted are the liquid ones, while the illiquid book sits fixed, its risk rising in the model precisely when it cannot be sold. Farrington's central claim is that the rational output of that machinery in a stress scenario is a buyers' strike: not selling — there is little ability to sell — but a pause in new commitments. In a market whose refinancing depends on continued inflow, a pause in new commitments is not a neutral act; it is the withdrawal of the reserve at the moment of maximum need, transmitted to borrowers as scarce refinancing credit, and to the marks as the arrival of real prices.

His second observation, credited here explicitly because it is his: TPA managers do hedge the illiquid book — with total-return swaps, sector credit-default swaps, index overlays. But these hedges are written on liquid instruments, and the basis between a liquid overlay and an illiquid book is itself a risk that appears exactly in the scenarios the hedge is bought for. A hedge whose liquid leg moves while the illiquid leg is marked smooth does not reduce the gap of section 4; it monetises one side of it and defers the other. This is a second say-do gap: protection is declared; whether it holds when the illiquid leg finally reprices has not been exercised.

Honesty requires the counter-case, and it is strong enough to state in full. TPA is symmetric machinery. An allocator that compares everything against everything can also rebalance into private credit — when public markets fall first, the model may show the illiquid book underweight, and CalPERS itself frames the method as a way to move faster toward opportunity, not only away from risk. Which face of the machinery shows in the first real stress window is, today, an open empirical question — and it is precisely the question the watchlist in section 8 is built to answer. The reading here does not assume the strike; it names the strike as the mechanism to which the system is newly exposed, and watches for it.

7 · The contagion flank

A repricing of private credit would not stay inside private credit; the connecting tissue is documented by the institutions responsible for watching it. The Financial Stability Board's May 2026 report counts the bank credit lines to private-credit funds noted in section 2, warns that valuation opacity and reliance on private ratings can amplify strains, and points at the growing class of funds offering redemption options to investors — semi-liquid vehicles whose redemption promises are procyclical, generous in calm and crowded in stress. The IMF's April 2026 Global Financial Stability Report raised a parallel alarm about the insurance channel: insurers holding rated private-credit paper, often through structures whose ratings sit far from any market test, face outsized losses in stress scenarios. And the warning from inside the banking system, cited in section 4, points in the same direction. None of these institutions forecasts a crisis; each names the same three couplings — banks to funds, insurers to paper, savers to semi-liquid vehicles — as the paths a private repricing would take into the wider system.

8 · Regime classification and watchlist

Stated in the vocabulary of the thesis this report structures: on the solvency clock, private credit is mean-reverting — the built layer is real, the yield is contractual, and nothing in this read implies it must break. On the coherence clock, private credit is breakout-prone — the say-do gap is wide and measurable, the reserves that finance the calm are being drawn at a rising rate, and the machinery governing the largest reserve is being rebuilt in a direction that makes it faster in both directions. A breakout is not predicted; the conditions under which one gathers are present and, unusually, publicly measurable. The confidence of this read is that of a hand read on public data — the tier is stated on the cover, and the misses table in section 10 is part of the read.

One paragraph on the wider frame, and no more. The institutions that own this asset class — public pension funds, insurers, sovereign wealth funds — are not private actors only; they are coupled to states, and their allocation behaviour imports the wider political and geopolitical environment into this market. In this read they are treated simply as the capital boundary of the system. A separate read in this series examines that boundary from the other side.

The watchlist below is the operational output. Each row is public, monitorable, and directional; movement on these rows is what would change the classification — in either direction.

Two triggers added after review. The final two rows were contributed by Mark Farrington (July 2026) in response to an earlier draft of this report, and are published here with his framing preserved. Both describe flows arising from the novelty of the Total Portfolio Approach rather than from any change in the asset class: as he puts it, they "don't really tell you much about cyclical health of the asset class, and yet they can have material effect." That is precisely the divergence this report is built to watch — movement on the coherence clock while the solvency clock reads unchanged — and neither trigger was visible from the outside-in structure without a practitioner's inside-out view.

9 · How this reading was produced

This is a hand read on public data — public reports, regulatory publications, rating-agency measures, market prices and dated news records. No proprietary data was used and no model or engine was run. The discipline followed is simple to state. What the system says about itself — its marks, its headline statistics, its stated purpose — is read against what it does where conduct can be observed: real transactions, real restructurings, real redemption behaviour, real allocation decisions. Where the two diverge, conduct governs. Where conduct has not yet been exercised — and in this young asset class much has not — the question is treated as open, not as reassuring. The result is a structural classification and a watchlist, deliberately not a forecast: this instrument is research-grade and under live forward test, and its previous reads, with their misses, stand on the public record. Consistency with the evidence is claimed; validation is not.

The thesis structured here is Mark Farrington's; the structure, and any errors in it, are the author's.

10 · Misses and open questions

11 · Sources

Integration Capacity Analysis (ICA) — a structural reading of institutions, economies and states · Wim Van Laere · integrationcapacity.org · contact@integrationcapacity.org · research-grade, under live forward test — consistency, not validation · a programme of The Great Homecoming