How BDCs work
A BDC is a creature of statute. Congress created the category in 1980 by amending the Investment Company Act of 1940, in a bill called the Small Business Investment Incentive Act, on the theory that small companies could not get capital from banks and needed a listed vehicle to bring public money to them. At least 70% of a BDC's total assets must sit in qualifying investments in US private companies and certain small public ones.
The tax deal is the reason the structure exists. A BDC typically elects to be treated as a regulated investment company under Subchapter M, which lets it deduct what it pays out to shareholders and so avoid federal tax at the fund level — provided it distributes at least 90% of its investment company taxable income each year. That is why BDC yields are advertised in double digits, and why any interruption to interest income shows up in the dividend almost immediately. There is no retained-earnings buffer by design.
Leverage is capped by an asset-coverage test. The default under Section 61 was 200% coverage, roughly one dollar of debt per dollar of equity. The 2018 Small Business Credit Availability Act let BDCs elect a 150% floor instead — roughly two to one — with approval from a majority of independent directors, effective a year later, or from shareholders, effective the day after the vote. Most took it. Ares Capital's own investor FAQ states flatly that BDCs must maintain 150% coverage to borrow or pay dividends.
Most large BDCs are externally managed, and the fees are private-equity fees in a public wrapper. A base management fee of 1.0% to 1.5% of gross assets is standard. MidCap Financial Investment Corporation discloses 1.5% of average net assets, plus a 20% incentive fee on net investment income above a 7% annualised hurdle with a catch-up, plus 20% of realised and unrealised capital gains. The manager earns on assets. The shareholder earns on outcomes.
The retail perpetual-life boom
The version that grew fastest is the one that never lists. Since 2017, managers have marketed "perpetual-life" non-traded BDCs through wirehouses, private banks and RIA platforms: continuously offered, priced monthly at net asset value rather than by an exchange, no fixed liquidation date, and no need for the to ever sell anything to give money back. Continuous offering in, quarterly limited liquidity out, illiquid loans held in the middle.
That middle term is the whole product. Because the loans cannot be sold in ninety days, the funds cap quarterly repurchases at about 5% of NAV. If requests exceed the cap, everyone is filled pro rata and the rest is deferred. Managers describe this as a designed feature protecting remaining investors, which it is. It is also a gate, and it only matters on the days it binds.
The scale is now serious. One market count puts the whole BDC industry at $575bn of gross AUM in Q1 2026, up 21% year on year. A Boston Fed study finds the number of BDCs rose from an average of 105 per quarter in Q1 2022 to 166 in Q4 2025. Ares Capital, at about $28.3bn of total assets, is the largest listed one. Blackstone Fund, at about $73.8bn, is more than twice its size and is not listed at all.
Cracks in the portfolio
Non-accrual is the one number in this business that cannot be smoothed. It means the fund has stopped booking interest on a loan because it does not expect to be paid, and it flows straight into the income that funds the dividend. At the twenty largest listed BDCs, the median hit 2.8% of cost in Q2 2026, up from 2.0% at the end of March — the highest since 2017, on Solve data analysed by the FT. We wrote that up on 19 August, and again when we put it next to the share prices.
The top of the range is worse than the median. Goldman Sachs BDC reported 5.0% of amortised cost on non-accrual in Q2 2026, or 2.9% at fair value; the names it disclosed included two second-lien positions in Wine.com moved to non-accrual, with Chase Industries restructured back to accrual and Thrasio's first-lien resuming. FS KKR reported 3.8% at fair value in the same quarter, and 7.1% of its book as troubled measured on cost. Ares Capital reported 2.4% at cost, with $708m of loans on non-accrual, up from 2.1% at amortised cost in Q1 2026. Kayne Anderson's BDC moved 4over and the last-out tranche of Diverzify to non-accrual, taking its rate to 2.7%, and carries its first-lien loan to American Soccer Co. at 44% below cost.
Now the disagreement. Over one twenty-day window in August 2026 the listed BDCs were up — Ares Capital +5.7%, Blackstone Secured Lending +5.9%, Blue Owl Capital Corp +6.5%, FS KKR +10.9% — and the managers were up far more, with Blue Owl Asset Management +30.8% and Ares Management +19.3%. Yet the average public BDC traded at 0.75x NAV on 3 August 2026, down from 0.84x at the end of 2025 and 0.87x in January 2026. At the bottom, Investcorp Credit Management BDC was at 0.22x NAV, Prospect Capital 0.36x and OFS Capital 0.41x, all on the same date. Rising off the floor and priced at a quarter of stated book are not contradictory. Together they say the public market does not believe the marks and is buying the bounce anyway.
Why it matters to this crash
BDCs are where meets a member of the public, and they are the only part of a $2tn asset class that has to tell you anything on a schedule. Two of the asset class's load-bearing promises are being tested inside them at once, and neither is passing cleanly.
The first promise is that private credit always refinances. In August 2026, 's Sophos went looking for private credit backing for more than $2bn of loans, the effort faltered, and the company went back to its existing leveraged-loan lenders offering a higher coupon, amortisation and tighter , with the sponsor declining to put in fresh equity. We covered that on 19 August. No default, no downgrade, nothing in any index — just worse terms, which is how losses get taken in a market with no prices.
The second promise is liquidity, and this is the one with the number attached. Blue Owl Technology Income Corp received estimated repurchase requests equal to 40.7% of shares outstanding in Q1 2026 and filled them at the standard 5%, pro rata; in Q2 2026 requests were 38.1% and the 5% cap held again. Blue Owl Credit Income Corp saw 21.9% requested in Q1 and 18.8%, or $3.6bn, in Q2. Blackstone's BCRED reported Q2 requests at 10% of shares, Cliffwater's corporate lending fund 17%, Apollo Debt Solutions 16.8%; Reuters reported most funds repurchasing about 5% of NAV and rolling the rest forward. The $26bn Apollo Debt Solutions fund said in June it would redeem the customary maximum of 5% after investors asked for more.
Our own tally of the queue, from a secondary compilation rather than filings we have seen, put non-traded BDC requests at a record 12.4% of NAV in Q2 2026, with 38% of requests met and roughly $9.6bn left queued. Nobody has to default for this to happen and no mark has to . Once the cap binds, the rational move is to ask for more next quarter to get the same amount out, and the queue feeds itself.
What would make this dangerous
Watch the funding side first. The pitch has always been that this is patient, unlevered, unrunnable capital. In August 2026, a record month for US investment-grade issuance at $145.2bn, two of the twelve deals on a single Monday were private credit funds run by Blackstone and Blue Owl selling high-grade bonds. Funds with public bondholders have covenants and leverage ratios calculated off loans nobody marks daily. If that bid goes away while non-accruals keep climbing and the loan books keep shrinking — PitchBook LCD found repayments and sales outrunning new commitments at vehicles run by KKR, Blue Owl and Apollo's MidCap Financial — the funding is where the squeeze appears, not the credit.
The 2008 template is the specific thing to fear. NAVs fell sharply, distributions were cut as portfolio income weakened, and some BDCs raised equity at steep discounts to NAV to survive, diluting the holders they were raising from. American Capital and Allied Capital were the best-known casualties; American Capital was eventually sold to Ares in 2017. A BDC trading at 0.22x NAV cannot issue shares to repay debt without handing most of the remaining book to the new buyers. That is the trap, and three listed BDCs were inside a discount range like that as of 3 August 2026.
The third thing is the one with no observable trigger. What we watched in the First Brands filing was two institutions holding views of the same exposure that could not both be right, with no mechanism to force them to meet. In a public market that disagreement becomes a spread and both sides get marked. In a BDC portfolio it becomes a fair-value estimate made by the manager who earns fees on the assets. If gated redemptions ever force actual sales of loans into a market that has no price discovery, the first genuine transaction print is the event. Everything before it is an opinion.