DEAD RECKONING (2)
Part 2 of a three-part dispatch from someone who has seen this film before
The Plumbing
Financial crises do not, as a rule, announce themselves from the top of the system. They build from the bottom, in the granular mechanics of individual transactions, and travel upward through a chain of vehicles whose interconnections are only fully understood once they start to fail. The 2008 crisis was visible to Tanta long before it was visible to Ben Bernanke, not because she was smarter than the Chairman of the Federal Reserve, but because she was reading different documents. She was reading loan files. He was reading aggregate data. The aggregate data said one thing. The loan files said another. The loan files were right.
This instalment looks at the loan files, or at least as close to them as the opacity of private credit permits, and attempts to describe accurately and without embellishment how a multi-trillion dollar market is actually constructed, where the stress is accumulating, and why the reported numbers may be telling a materially different story to the one embedded in the underlying assets.
* * *
How It Is Built
Private credit did not emerge from nowhere. It is, in large part, a consequence of the 2008 crisis itself. Post-crisis banking regulation, principally the Basel III framework, required banks to hold significantly more capital against leveraged and middle-market loans. The economics of that lending deteriorated sharply. Private credit funds, which are not subject to the same capital requirements, moved into the gap. Borrowers who could no longer access bank lending found a new source of capital. Investors searching for yield above what public markets offered found a new asset class. The arrangement suited everyone, at least for a decade of low rates and benign credit conditions.
The structure that resulted is layered in ways that matter when things go wrong. At the base are the loans themselves, typically extended to private equity-backed companies in the middle market. Above those loans sit the private credit funds, the Apollos and Areses and Blue Owls of the world. Those funds borrow from banks to amplify returns, a practice known as back-leverage. According to figures cited by Moody’s, Wall Street banks had provided approximately $300 billion in such financing to private credit funds by mid-2025, a figure that JPMorgan itself disclosed had tripled from roughly $50 billion in 2018 to $160 billion in 2025 for its own book alone.
Above the private credit funds sit the distribution vehicles through which their exposure is sold to end investors. Non-traded Business Development Companies, BDCs, have grown from essentially zero in 2021 to more than $200 billion in assets, marketed primarily to high-net-worth individuals seeking yield. Alongside these sit CLOs, insurance company balance sheets, and pension fund allocations. By the time the risk has travelled from the original loan to the end investor, it has passed through three, four, or five separate vehicles, each with its own fee structure, leverage ratio, and liquidity profile.
The critical point is that most private credit vehicles borrow heavily to buy more loans, and that same leverage magnifies losses when price marks move against them. BlackRock TCP Capital illustrated this in January 2026. When it disclosed a 19 per cent NAV decline, the underlying portfolio loss was significantly smaller. Leverage had amplified valuation adjustments into a severe net asset value drop. This is not fraud or mismanagement. It is arithmetic. But it is arithmetic that end investors in non-traded BDCs were not prominently shown when they allocated capital.
* * *
The Software Thesis and Its Collapse
The loans at the base of this structure are not uniform in their risk. They concentrate in sectors. And the single largest concentration, representing roughly 25 per cent of all private credit lending, is software.
The logic of software as a private credit borrower was compelling for many years. Subscription-based recurring revenue was predictable. Margins were high. Customer churn was low. The businesses were, in the parlance of private equity, asset-light and cash-generative. Robert Smith of Vista Equity Partners captured the prevailing consensus in 2018 when he argued that software contracts were better than first-lien debt, because no business could operate without its software.
That thesis is under structural attack from a direction that credit models built on macroeconomic cycles were not designed to anticipate. AI tools are beginning to replace the workflow software whose subscription revenues funded debt service. The software sector, which had been the model private credit borrower, is repricing from a 2030 disruption risk to a nearer-term reality. The marks, however, have been slower to move than the underlying economics. The underlying deterioration of leveraged software companies is happening on a timeline measured in months, while the marks that determine their stated value update on a timeline measured in quarters. That gap is the story.
The 2021 and 2022 loan vintages are where the exposure concentrates. These were the years of peak private equity software acquisition activity, financed at peak valuations, with peak leverage, in a near-zero interest rate environment. Approximately 91 per cent of BlackRock TCP Capital’s NAV reduction in Q4 2025 was tied to investments underwritten in 2021 or earlier. The firm’s own filing noted that certain of these businesses had benefited from elevated pandemic-era demand that has since softened, and that because the investments were originated in a low rate environment, several had struggled to adapt to sustained higher interest rates. AI disruption sits on top of both of those pressures.
The exposure is not abstract. By early 2026, specific names were appearing across the portfolios of multiple funds facing redemption pressure. PowerSchool, an education software company originally developed in 1997 and acquired by private equity, carries debt held by at least four funds that have suspended or limited redemptions in 2026, including Ares, HPS, Blue Owl, and Blackstone. Anaplan, a financial forecasting software firm taken private by Thoma Bravo in 2022 at a 41 per cent premium to its public market price with the help of $3 billion in private credit, was losing $200 million annually on $565 million in revenue during its last year as a public company. The question of how such a business is expected to service a quarter of a billion dollars in annual interest expense is not one that requires sophisticated modelling to answer. The Congressional Research Service, in an Insight paper published in April 2026, estimated private credit exposure to software-as-a-service companies at approximately $500 billion.
* * *
The NAV Problem
The BlackRock TCP Capital case is worth examining in detail, not because it is the worst example, but because, as a publicly traded BDC, it was required to disclose. Most private credit vehicles are not. It is also not necessarily representative of the broader BDC universe: TCP’s concentrated exposure to a small number of severely impaired positions makes it an outlier in terms of magnitude. But as a demonstration of the mechanics, specifically how leverage amplifies loss and how quarterly marks can fail to signal deterioration until it arrives in a single devastating update, it is instructive.
The 19 per cent NAV write-down was largely driven by significant fair value reductions across six company investments. According to analysis by Octus, many of those positions had been marked above 90 per cent of par as recently as September 2025. By December, they were effectively worthless or close to it. The marks had given investors minimal warning.
This is the private credit equivalent of the CDO rating problem. In 2006 and 2007, mortgage-backed securities retained investment-grade ratings until the moment they did not. The ratings were the product of models whose assumptions no longer matched reality. Private credit NAVs are the product of manager models tested at most quarterly, using methodology that is proprietary and opaque to end investors. The Department of Justice has publicly warned about creative marks and divergent valuation practices in private portfolios, while an SEC inquiry into Egan-Jones Ratings has placed the integrity of private credit ratings under a spotlight.
* * *
PIK as the Shadow Default
The most instructive single number in the current private credit landscape is not the headline default rate. It is the PIK rate.
Payment-in-kind interest allows a borrower to defer cash interest by rolling it into the loan principal. When included at origination, PIK is a legitimate structural feature. When a borrower negotiates it later, after the loan has closed, it is almost always a signal that cash flow is insufficient to service the debt. PIK income in the direct lending market averaged roughly 4.2 per cent pre-pandemic, rose to 7.4 per cent post-pandemic, and reached approximately 8.8 per cent in the third quarter of 2025, according to data from Keefe, Bruyette & Woods cited by Vanguard.
The distinction that matters is between what analysts call good PIK and bad PIK. Good PIK is planned at origination. Bad PIK is what happens when a borrower who was supposed to pay cash cannot. According to data from T. Rowe Price’s OHA platform, approximately 6.1 per cent of the private credit loan base had bad PIK as of Q3 2025, up from approximately 2.6 per cent in 2021. Separate analysis by Lincoln International found that as of Q1 2025, 56 per cent of investments with PIK interest had no PIK at underwriting, implying a shadow default rate closer to 6 per cent, significantly higher than the 2.1 per cent default rate reported by KBRA. Among companies with bad PIK, loan-to-value ratios had risen substantially from inception, with one study by Chronograph citing an increase from 45 per cent at inception to 83 per cent. That trajectory represents a fundamental erosion of lender security, occurring quietly, without triggering formal default definitions, and therefore without appearing in headline statistics.
The headline private credit default rate has been reported at sub-2 per cent for several years. Adjusted for PIK conversions, maturity extensions, and what the industry calls liability management exercises, multiple credible sources place the true distress rate significantly higher. Fitch Ratings reported elevated distress levels by the end of 2025. According to S&P Global Ratings, of 35 companies executing 38 liability management exercises between mid-2017 and August 2024, only five, or approximately 14 per cent, avoided subsequent bankruptcy or default. They are delay mechanisms, not cures.
Charlie Munger, who had no patience for financial euphemism, described EBITDA, the earnings metric on which most private credit leverage ratios are calculated, as bullshit earnings. The EBITDA add-back problem, where companies exclude costs from their calculations with lender blessing to make leverage appear lower than it is, means that loans reported at five times leverage may in reality sit at six and a half or seven times. This is not illegal. It is endemic.
* * *
Who Is Reading the Loan Files?
By early 2026, the interest coverage ratios for many mid-market firms had plummeted below 1.0x, meaning they were no longer generating enough cash to cover their own interest payments. Amendment activity increased 13 per cent quarter on quarter in Q4 2025. Maturity extensions and covenant holiday activity were both up 14 per cent. Sponsor cash infusions rose 31 per cent. These are the numbers that appear in aggregate data. They are the financial equivalent of the rising delinquency rates that Tanta was tracking in individual mortgage pools in late 2006, before the word subprime had entered common usage. The increase in sponsor cash infusions is, it should be noted, a genuine structural buffer. Private equity sponsors have capital to deploy, and their willingness to inject equity distinguishes this cycle from 2006 and 2007, when mortgage borrowers had no such backstop. However, sponsor capital is finite, and a 31 per cent increase in the need for it could equally be read as evidence of stress rather than resilience.
But Tanta could read the loan files. She had processed them. She understood, from direct experience, what the documentation said and what it concealed. She knew that the automated underwriting systems were approving loans that a human underwriter, reading the same file with professional scepticism, would have declined.
The equivalent knowledge today sits with the analysts inside private credit funds who are reviewing borrower performance packages, processing covenant waiver requests, and negotiating PIK toggle elections. They are seeing, transaction by transaction, the gap between what the marks say and what the businesses can actually sustain. Some of them know that the software thesis is under structural attack that does not appear in the quarterly NAV. Some of them are aware that the 2021 and 2022 vintage loans in their portfolios carry a weight that current interest rates and current AI disruption are making progressively harder to bear.
They are not writing Substacks.
In 2006, the warning was available and largely ignored. Tanta wrote it. You could read it. The system chose not to. In 2026, the equivalent warning may not even have found its voice yet, because the people who could give it are inside the machine, collecting fees, hoping the cycle turns, and staying quiet. It is also possible, of course, that the silence reflects a less alarming reality: that the stresses are being managed, that the problems remain idiosyncratic rather than systemic, and that the cycle will turn before the cracks widen. That interpretation cannot be excluded. But the aggregate data does not support complacency, and the people best placed to give reassurance are not giving it publicly.
That silence is data.
* * *
References: Part Two
1. BlackRock TCP Capital Corp., Form 8-K filed with the SEC, 23 January 2026. Preliminary Q4 2025 results, 19 per cent NAV decline, six portfolio companies driving approximately two thirds of the reduction, 91 per cent tied to investments underwritten in 2021 or earlier.
2. Octus (formerly Reorg), “BlackRock TCP 19% NAV Reduction Driven by Loans Previously Marked Near Par,” 28 January 2026. Analysis of fair value marks prior to write-down.
3. Vanguard, “Private Credit’s Third Act: Reconnecting with Banks,” 19 February 2026, citing Keefe, Bruyette & Woods data as of 30 September 2025. PIK income trajectory: 4.2 per cent pre-pandemic, 7.4 per cent post-pandemic, approximately 8.8 per cent Q3 2025.
4. T. Rowe Price / OHA, “PIKing Your Spots in Private Credit,” 27 January 2026. Bad PIK by amendment rising from 2.6 per cent in 2021 to 6.1 per cent in Q3 2025.
5. Lincoln International, cited in TCW, “The Big PIK-ture,” 18 August 2025. 11 per cent of investments with some PIK interest; 56 per cent with no PIK at underwriting; implied shadow default rate of approximately 6 per cent versus 2.1 per cent reported by KBRA.
6. Chronograph, “How Direct Lending Competition Is Impacting Private Credit Deal Terms,” 2 December 2025, citing Lincoln International data. LTV for companies with bad PIK rising from 45 per cent at inception to 83 per cent.
7. S&P Global Ratings, cited in Ankura, “Liability Management Exercises: Bandage or Bridge to Recovery?”, 11 February 2026. Of 35 companies executing 38 LMEs (mid-2017 through August 2024), only five (14 per cent) avoided subsequent bankruptcy or default.
8. PIMCO, “Private Credit’s Other Lanes Still Offer Value,” March 2026. BDCs trading at largest discount to book value since post-COVID recovery; rising PIK share; software sector concentration.
9. Department of Justice, public statements on valuation practices in private portfolios. SEC inquiry into Egan-Jones Ratings.
10. With Intelligence, “Private Credit Outlook 2026,” March 2026. Non-traded BDCs growing from zero in 2021 to more than $200 billion.
11. Congressional Research Service, “Private Credit Funds Redemption Restrictions: Market Context and Policy Issues,” IN12674, 2 April 2026. SaaS exposure estimated at approximately $500 billion; table of funds with redemption restrictions.
12. Maureen Tkacik, “The Private Credit Cartels,” The American Prospect, 6 April 2026. Named software holdings including PowerSchool (Ares, HPS, Blue Owl, Blackstone) and Anaplan (Thoma Bravo, $3 billion private credit). ---
This document is an analytical commentary. It does not constitute financial advice, a financial promotion, or a recommendation to buy, sell, or hold any security, financial instrument, or investment product. The author is not a regulated financial adviser. The author holds no positions, long or short, in any of the securities, instruments, or companies discussed.