Appendix — Data, Models and Sources

Supporting detail for the report above. Nothing here is required to act on the calls; it's here so you can check my work and disagree precisely.

Appendix A · Market data snapshot

All prices my own pulls at approximately 12:00 ET, Friday August 28, 2026.

A.1 Rates

September hike odds: 46% per CME FedWatch via Reuters, 55.7% per CNBC, captured at different points in the session; ~35% Thursday.

A.2 Equities and crypto

Note the alt managers rose on a hawkish day — higher rates help their insurance and annuity spread businesses. That's a genuine offset and it's why the ARES short is labelled speculative rather than high conviction.

Appendix B · The two models

B.1 BDC net investment income sensitivity (OBDC)

Inputs, from OBDC's June 30, 2026 results: NAV $14.26/share; net leverage 1.11x; weighted average yield on debt investments 9.9%; spread over base ~5.6%; 96% of assets floating rate.

Derived per share: total assets ≈ $30.09 (NAV $14.26 + debt $15.83); floating assets ≈ $28.90.

Key assumption: ~40% of the debt stack floats (revolver and SPV facilities), ~60% is fixed-coupon unsecured notes issued in the 2021–22 window. This is the assumption most worth challenging — sensitivity is roughly ±$0.02/share per 100bp for every 10 percentage points of mix.

Context: the base dividend runs $1.24/year annualised and was covered 109% last quarter. An extra $0.19 is 15% of incremental dividend headroom.

B.2 Borrower interest coverage grid

Inputs: SOFR ~3.75%, typical direct lending spread ~560bp, all-in coupon ~9.35%. Coverage = 1 ÷ (leverage × coupon). Assumes no EBITDA growth and no addback normalisation — both of which would make the picture worse, not better.

Below ~1.25x is where sponsors stop injecting equity and start negotiating. The escalation order, per Benefit Street's Anant Kumar: maturity extension → PIK → sponsor cheque → covenant relief. His framing: "the fourth amendment on the same name is not a bridge to recovery, it's deferral."

B.3 Deriving the 2-percentage-point break-even

At a 65% recovery assumption, each incremental 1.0pp of the gross book going permanently non-accrual costs:

• 1.0pp × 35% loss severity = 0.35% of gross assets
• At 1.1x leverage = ~0.74% of NAV = ~$0.105/share for OBDC

Set against +$0.186/share of NII from a 100bp hike: +$0.186 ÷ $0.105 ≈ 1.8, call it ~2 percentage points of incremental permanent non-accrual before the hike stops paying for itself.

Where this could be wrong: the 65% recovery assumption is the load-bearing input. At 50% recovery the break-even tightens to ~1.3pp and more of the sector falls on the wrong side. At 80% it widens to ~3.5pp and only the genuinely impaired books fail.

Appendix C · Warsh speech — primary source extracts

From "In Our Time," Chairman Kevin Warsh, Jackson Hole, August 28, 2026. Read in full at federalreserve.gov.

C.1 The hiking bias — closing of the "Economy Today" section

"We must be confident that underlying inflation is moving to our objective, clearly and at sufficient speed. Otherwise, we have work to do."

"I stand here today committed to a discipline, not to a decision."

C.2 The credit passage — the buried lede

"Credit spreads on corporate bonds and leveraged loans are near the low ends of their historical ranges, and issuance volumes in these markets have been quite strong this year. Looking beyond fixed-income markets to the banking business, in the July Senior Loan Officer Opinion Survey on Bank Lending Practices, banks tell us that standards for commercial and industrial loans are on the easier end of their historical range. That helps explain the growth we've seen this year in those loans. Credit and loan markets are showing few signs of policy restraint."

"Certain sectors — like housing and agriculture — are showing strains. But, on balance, I would be hard pressed to describe broad financial conditions as restrictive."

C.3 The inflation disaggregation

• PCE inflation: 3.7% over twelve months, 4.1% over six months
• Of the 199 PCE components: 54% rose more than 3% over twelve months, 49% over six months, versus a pre-pandemic norm of 32%
• Warsh attributes "65 months of sustained, elevated inflation" squarely to the central bank
• Unemployment 4.1%, described as consistent with full employment; weak payroll gains attributed to flat labour supply rather than weak demand

C.4 The principles that imply balance-sheet action

Fifth principle: short-term interest rates are the predominant tool. Sixth principle: "money matters… We should pay attention to money created by the central bank and money that comes from the banking and financial systems."

Annex Wealth's Brian Jacobsen flagged the live scenario: Warsh may push to stop expanding the balance sheet before hiking, potentially producing the first Chair-in-the-minority dissent since Eccles in 1939.

Appendix D · BDC fundamentals

Sell-side markers: JP Morgan holds FSK at neutral with an $11 target raised August 19 — below the current $12.25. Oppenheimer raised ARES to $162 on August 27.

D.1 OBDC insider transactions — trailing twelve months

Every filing is a purchase. There are no sales in the record.

For comparison, ARCC insiders bought a cluster in February 2026 at $19.13–19.29: Markowicz 15,000, Schnabel 12,500, Lem 5,186, Henson 4,000.

Appendix E · Sector damage data

E.1 Defaults and non-accruals

• Fitch US private credit default rate: 6.1% trailing twelve months through July 2026 — an all-time high. At 65% recovery that alone is ~2.1% of book value per year, or ~4.4% of NAV per year at 1.1x leverage. The record was set before any hike.
• Median non-accruals across the 20 largest listed BDCs: 2.8% of cost in Q2, up from 2.0% at March-end — 80bp in a single quarter, back to 2017 levels. The second derivative is the problem: four more quarters at that pace puts the sector median at 6%.
• Morningstar / PitchBook LCD adjusted series — counts the whole debt stack of any borrower with one impaired tranche: 5.95% across the top-10 BDCs (from 5.41%), spanning 101 borrowers. Broader $516bn universe: 1.9% reported versus 3.3% adjusted.
• $772m of BDC interest income at risk = 204bp of yield on $38bn of cash interest income, dragging sector cash yield from 8.3% to 8.1%. Worth noting: 204bp is roughly two full 100bp hikes' worth of NII benefit, already gone before the Fed moves.

E.2 PIK and loan quality

• Lincoln International: PIK features now in more than 10% of direct lending loans, up from 7% in late 2022
• The "bad PIK" cohort — 6% of loans — has seen loan-to-value climb from 49% to 86%. Against a 90-cent clearing market, that leaves roughly four points of cushion.

E.3 The redemption channel

• Robert A. Stanger: Q2 redemption requests 12.4% of NAV — the highest ever recorded; only 38% fulfilled; $9.6bn backlog
• Across the 26 private-placement BDCs on regular tender schedules: requests 6.7% of NAV versus 5.5% in Q1; redemptions exceeded new capital by nearly 3x
• Non-traded BDC fundraising: −82% year over year to $2bn in Q2 — the lowest since 2020

E.4 The clearing price

~90 cents on the manager's own marks, for a diligenced first-lien direct lending book. Established by Pantheon's August 4 purchase out of BlackRock TCP Capital — headline 95% of December 31 gross fair value, but CFO Erik Cuellar put the effective discount at "about 10%" on the call, 10.4% including expenses. Corroborated by LP-led credit secondaries printing 90–92% of NAV through 2026. Measure every discount against 90, not against par and not against 95.

E.5 The listed-versus-non-traded gap

Monthly NAV trajectory worth tracking, because it shows a sign flip between 2025 and 2026 — consistent with AI impairing the legacy software loan book:

E.6 Capital formation — the anomaly

• Global private credit final closes: $190bn in 1H 2026, +53% year over year, already ~80% of full-year 2025. Direct lending $73bn in Q2 alone.
• New-issue spreads remain tighter than exit spreads by roughly 150bp. A market defaulting at a record 6.1% is writing new paper cheaper than the paper it is exiting. The 2026 vintage will be the worst in the asset class's history, and it is being written right now.
• NEPC: no direct-lending-financed LBO larger than $2bn since March 2026. Sponsor-backed middle-market was ~57% of US PE activity in Q2.
• KBRA: 27% of private credit borrowers had interest coverage below 1.0x in 2024.

Appendix F · Second- and third-order effects

F.1 Alt managers lose the pipeline, not the credit

Apollo, Ares, Blackstone, Blue Owl and KKR don't eat the loan losses — their limited partners do. What they eat is deployment. Fee-related earnings are a function of fee-earning AUM, and fee-earning AUM growth requires deployment, not just fundraising. Apollo posted record Q2 FRE of $785m, +25% YoY at a 58.5% margin — but that describes a Q1–Q2 environment. With no >$2bn direct-lending LBO since March, the FRE growth rate is what compresses, and these stocks are priced on the growth rate.

Levels: ARES below $120 or APO below $118 is where the market stops paying for FRE growth and starts paying for FRE.

F.2 Quantitative tightening removes the cheapest leverage in the system

The marginal funding for middle-market private credit is CLO liabilities. TCPC demonstrated in August that the retained equity of a middle-market CLO issued in May 2026 could transfer within ten weeks at a price implying ~28% loss on that equity. Widen CLO liability spreads with balance-sheet runoff on top and you remove the cheapest leverage in the system: BDC returns on equity fall structurally, originations slow, and new-money loan spreads finally widen — resolving the spread anomaly, but in 2027, not 2026.

F.3 The consumer-facing borrower is squeezed from both ends

Affirm's CEO noted the national average gasoline price at $4.09/gallon, last below $3.00 on March 2. Consumer-services and retail-adjacent private credit borrowers face rising input costs, weak pricing power and a rising coupon simultaneously — Kumar's "operating costs and financing costs rise but revenue fails to keep pace."

F.4 Cross-asset read-through

Appendix G · Historical analog — and why it only half applies

The BDC sector traded at ~0.83x book in September 2015 and ~0.81x at the January 2016 trough on energy-driven credit stress, recovering to ~1.00x by August 2016. Today's sector multiple is comparable. Two differences cut in opposite directions:

Argues for a faster recovery: in 2015–16 the problem was concentrated, identifiable and commodity-priced — energy loans marked against an observable oil strip. Price discovery was fast, so recovery was fast.

Argues against: today's problem is software and business services — borrowers underwritten at 6–8x ARR or EBITDA in 2020–21, whose end markets are being disintermediated by AI, and whose loans have no observable price at all. And the 2015–16 recovery happened while the Fed hiked once in fifteen months into a disinflationary world. This time the Fed would be hiking into 4.1% six-month core PCE.

Appendix H · Sources

• Chairman Kevin Warsh, "In Our Time," Jackson Hole keynote, August 28, 2026 — full text; the "Economy Today" and "Key Principles" sections
• Blue Owl Capital Corporation — June 30, 2026 results — NAV, leverage, floating-rate mix, portfolio yield
• OBDC Form 10-Q, quarter ended June 30, 2026 (accession 0001655888-26-000056) — Schedule of Investments, non-accrual detail
• CNBC — Warsh expresses concern about inflation, advocates a "quieter" central bank
• CNBC — 2-year Treasury yield jumps as Warsh says Fed may "have work to do"
• CNBC — "Nobody underwrote for that": private credit faces a key test as higher rates squeeze borrowers
• The Week in Private Credit, August 21, 2026 — fundraising, non-accrual medians, Stanger redemption data, NEPC LBO data
• Morningstar / PitchBook LCD — private credit is showing more signs of distress — the adjusted non-accrual series
• Yardeni Research — Private Credit Monitor
• Axios — Warsh: rate increases in play if inflation doesn't fall
• Reuters via GV Wire — rate-hike expectations rise on Warsh speech — strategist reaction, FedWatch odds