In fifteen years private credit grew from under $400 billion into an asset class of roughly $2.6 trillion on broad definitions, and its funders shifted from pension plans and insurers to you and your financial adviser. This issue is not about whether to buy it. It takes apart three things: what this business actually earns, what the semi-liquid wrapper trades away, and why product innovation always outruns investor protection.
The Principle
Private credit is not a new asset but an old business under new ownership: middle-market lending moved off bank balance sheets and into funds. The excess return is a fee charged for three things — illiquidity, information asymmetry, and leverage.
Source · The Line
Benjamin Graham, The Intelligent Investor (1949), Ch. 1, drawing the line between investment and speculation.
"An investment operation is one which, upon thorough analysis, promises safety of principal and an adequate return. Operations not meeting these requirements are speculative."
— Benjamin Graham, The Intelligent Investor, Ch. 1 (1949)
Reading It Closely
Basel III raised capital requirements and the 2013 US leveraged lending guidance questioned deals above six times debt/EBITDA; banks withdrew from direct middle-market lending and funds took over, charging borrowers SOFR plus 450 to 650 basis points. Decompose that yield: risk-free rate + credit compensation + illiquidity compensation + fund-level leverage (most run 0.8 to 1.25 times). The first two are available in public high yield and can be sold any day; the extra fee buys the last two. And Graham's "thorough analysis" is unavailable here: the underlying borrowers disclose neither financials nor loan terms.
The Case
On the IMF's April 2024 Global Financial Stability Report definition, global private credit including undrawn commitments stood at about $2.1 trillion, three-quarters of it in the United States; around 2010 the figure was under $400 billion. Three things happened at once — the asset base swelled, rising base rates in 2022–23 made coupons look superb, and a cohort of funds that has never lived through a full default cycle began opening to individual channels. The first two produced the track record; the third is the reason it is being distributed.
Limits · Decision Checklist
This is not a fraud. From 2008 to 2021 default losses genuinely ran below comparably rated public high yield, because a single lender restructures faster and covenants are tighter. The limit is that this entire history sits inside zero rates and an expansion. The other failure mode is competition: too much money chasing the same borrowers dilutes covenants, and covenants were the edge.
- How much of my spread is credit compensation versus illiquidity compensation? Is the latter worth the price over my horizon?
- What share of the gross yield do management fees, performance fees, and leverage costs consume?
- What period does the track record cover? Does it include one full default cycle?
Essence · Reflection
You are not buying a higher return. You are trading away liquidity and information to take on a business someone else found inconvenient.
Compute the product's all-in fees, subtract them from the gross yield, and compare what is left with the net yield of a public high-yield ETF. That gap is the market's price for your inconvenience.
The Principle
In a semi-liquid product, liquidity is not a property but a contract term: true in normal times, pro-rated away precisely when everyone wants to use it — which is exactly what it was designed to do.
Source · The Line
Warren Buffett's letter to Berkshire shareholders (2001 annual report), on long-tail reinsurance risk.
"After all, you only find out who is swimming naked when the tide goes out."
— Warren Buffett, Berkshire Hathaway 2001 Shareholder Letter
Reading It Closely
An interval fund, under Rule 23c-3 of the Investment Company Act of 1940, must make repurchase offers at fixed intervals for 5% to 25% of outstanding shares; most choose 5% per quarter. Non-traded BDCs use quarterly tender offers on similar terms, with a board power to suspend. The mechanism that matters is proration: if 20% of shares want out in a quarter against a 5% cap, each holder gets a quarter of their request and the rest joins the next quarter's queue. But the gate protects those who stay — without it the manager must dump loans at a discount to fund redemptions, and the loss falls on those who did not redeem.
Listed BDC / closed-end fundexchange-traded
Exit: T+1, sellable any time
Pricing: exchange price, can sit ±20% from NAV for long stretches
Who decides: the market — the discount is your exit costNon-traded BDCnon-traded
Exit: quarterly tender, cap typically about 5% a quarter
Pricing: NAV as marked by the manager
Who decides: the board, which may cut or suspendInterval fundRule 23c-3
Exit: repurchase at fixed intervals, 5%–25%, usually 5% a quarter
Pricing: NAV as marked by the manager
Who decides: the rule's floor plus board discretionPrivate LP funddrawdown fund
Exit: lock-up plus capital calls, measured in years
Pricing: manager's mark; secondary transfers need a discount
Who decides: the partnership agreement — essentially no exitLiquidity is never simply present or absent. It is a question of who decides, under what conditions, how much you may take out.
The Case
BREIT (Blackstone Real Estate Income Trust) is the only full stress test this structure has yet had — the underlying assets are real estate rather than credit, but the mechanism is identical. It once exceeded $70 billion, with charter limits of 2% of shares a month and 5% a quarter. Requests breached the cap in late 2022: December was filled at roughly 43%, and against about $5.3 billion requested in January 2023 roughly a quarter was met. The gate ran intermittently for fourteen months, and only in February 2024 were requests met in full. Investors were trapped, but the fund was not forced to dump assets at a discount — the gate did its job, and the cost fell on those who wanted out.
Limits · Decision Checklist
The gate protects the fund, not your cash-flow plan; the better it works, the less you can leave. The subtler trap is reading "stable NAV" as "low risk": BREIT's reported value barely moved through the peak of redemptions while listed REIT indices fell roughly 25%. The difference was not in the risk but in the frequency of quotation.
- Have I read the actual text on repurchase caps, intervals, and the board's power to suspend?
- If I wanted everything back in a gated quarter, how many quarters would proration take?
- Can my liquid assets cover 24 months of cash needs without touching this?
Essence · Reflection
The moment liquidity is most precious is the moment the contract makes it disappear.
Skip the brochure and open the repurchase section of the prospectus. Can you state, in five minutes, how many quarters a worst-case exit takes?
The Principle
A private credit NAV is not a price but an appraisal — the manager models a value for loans that never traded. The low volatility and low correlation it reports contain something real and something manufactured by the method of measurement.
Source · The Line
Warren Buffett's letter to Berkshire shareholders (2002 annual report), on marking derivatives.
"In extreme cases, mark-to-model degenerates into what I would call mark-to-myth."
— Warren Buffett, Berkshire Hathaway 2002 Shareholder Letter
Reading It Closely
In accounting terms these assets sit in the third level of fair value (Level 3): no observable market inputs, so the manager's assumptions govern. The consequences are that reported volatility is smoothed and Sharpe ratios flattered, while correlation with public markets is understated and the diversification benefit overstated. The test is simple: through the rate rise of 2022–23, several listed BDCs traded 10% to 25% below their own reported NAV while non-traded BDCs reported NAVs that barely moved. Both numbers cannot be right at once.
The Case
Two bankruptcies in September 2025. Tricolor Holdings, a subprime auto lender, filed for liquidation; Fifth Third took a charge of roughly $200 million and JPMorgan wrote off about $170 million. On September 28, the auto parts maker First Brands Group filed for Chapter 11 with liabilities above $10 billion, exposing off-balance-sheet factoring and receivables pledged more than once; Jefferies disclosed roughly $715 million of exposure through its Point Bonita Capital. What matters is not the size of the losses but their timing: before they broke, neither risk had appeared in anyone's quarterly NAV.
Limits · Decision Checklist
Smoothing is not falsification. Valuing a loan off its hold-to-maturity cash flows is defensible accounting when you do not intend to sell, and the discount on a listed BDC carries sentiment as well as information. The problem is in the use: running a smoothed series through a volatility calculation, a mean-variance optimizer, or a correlation against equities systematically overstates the diversification gain — which is the very reason the product gets recommended into a portfolio.
- What share of assets is Level 3, marked by the manager? Can I find it in the disclosures?
- Does this manager run a listed vehicle, and what is its discount to NAV right now?
- Did my allocation model use this fund's historical volatility? If so, am I paying for smoothing?
Essence · Reflection
A smooth line may mean the risk is small. It may also mean nobody quotes a price each day.
Sort the low-volatility holdings in your portfolio into two columns: those whose steadiness comes from the asset, and those whose steadiness comes from not being priced daily. How heavy is the second column?
The Principle
The speed at which a product reaches retail channels is set by distribution incentives; the speed at which the rules for judging its suitability are updated is set by regulators. The two never move together, and the buyer carries the lag.
Source · The Line
Charlie Munger, The Psychology of Human Misjudgment (1995 talk, collected in Poor Charlie's Almanack).
"Never, ever, think about something else when you should be thinking about the power of incentives."
— Charlie Munger, The Psychology of Human Misjudgment (1995)
Reading It Closely
The central warning in CFA Institute's July 2026 report is not that the asset class will collapse, but that semi-liquid structures make an explicit trade among liquidity, transparency, and investor protection, and whether suitability standards keep pace with product innovation remains unresolved. The regulatory answer came on September 30, 2026: the SEC voted out a package of proposals — proposals, not effective rules — titled "Expanding Responsible Retailization of Private Markets," covering reform of interval fund repurchase rules, performance fees for retail-facing regulated funds, a rule framework for multiple share classes in closed-end funds, and a request for comment on qualifying as an accredited investor by credential or examination. Access is shifting from how much money you have toward how much you understand: more defensible in principle, and in practice it hands the judgment back to you.
The Case
The Woodford Equity Income Fund is the textbook case of a mismatched wrapper. This UK retail fund peaked at roughly £10.2 billion in June 2017, offered daily redemption, and held a great deal of unlisted and micro-cap stock; by the time dealing was suspended on June 3, 2019 it had shrunk to about £3.7 billion, it went into wind-down that October, and investors lost roughly a third. The mismatch was not in the assets but in the wrapper: a promise of daily redemption wrapped around assets that take years to realize. Private credit's gates are precisely the lesson being absorbed — the cost is written into the terms, and the terms are the part retail investors read least.
Limits · Decision Checklist
The opposite holds too: walling individuals out of private markets has its own cost. The number of US listed companies fell from roughly 8,000 in 1996 to just over 4,000 today, and a meaningful share of value creation now happens where nothing is published. The question is not whether to open access but on what terms. One further limit: proposals take a year or two to take effect and are usually amended, so do not allocate off a proposal.
- Who brought this to me? Is their pay tied to how long I hold, or only to this purchase?
- Do I meet an asset threshold, or do I actually understand the terms well enough to explain the risk to someone else?
- Does the wrapper's redemption frequency match how long the underlying assets take to realize? Who absorbs the mismatch?
Essence · Reflection
To judge a financial product, first ask whether you found it or it found you.
Sort the positions you added over the past year into two groups: those you bought after your own research, and those prompted by a recommendation or an advertisement. Did the two groups perform differently?
Going Deeper
Is private credit the next 2008?
Structurally it is different, so probably not a sudden collapse: most of these funds take no deposits, owe no daily redemption, and run 0.8 to 1.25 times leverage against the roughly 30 times of the 2008 investment banks. But the transmission path is more hidden — banks extend credit lines to these funds, insurers hold large share positions, and the same underlying risk can be counted at several layers. The likelier outcome is a long stretch of low returns and slowly acknowledged losses: no Lehman moment, just a dozen consecutive quarters of small NAV adjustments and a queue you cannot leave. The second is harder to notice.
What happens to this asset class when rates fall?
Almost all of it is floating rate, so coupons fall directly with the benchmark; and a large part of the past three years' handsome performance came from rates rather than credit selection skill. The worse squeeze is two-sided: spreads compress under competition while the benchmark declines. The capital-cycle framework from Day 58 applies directly here — the scale of capital a sector has absorbed over the past five years is itself a negative indicator of its future returns.
Can AI solve diligence on these products?
It can ease the processing of information, not its acquisition. The core obstacle is that loan terms and borrower financials are not publicly disclosed — the problem is not too much data to read but data you cannot obtain. AI's real gain lies in the disclosed part: reading several hundred pages of prospectus, comparing repurchase terms, computing how fees erode compounding, pulling the trend in Level 3 share out of the notes to quarterly reports — work that was previously too costly for anyone to do. But it will not turn an undisclosed borrower into an analyzable object: technology raises your efficiency at using disclosed information; it does not move the boundary of disclosure.