Video summary

Banks, Private Credit, and the Future of Risk | Global Conference 2026

Main summary

Key takeaways

Finance

Finance-focused summary (direct lending / private credit conference panel)

Topic

Whether direct lending within private credit is facing a serious problem now—and how risks could evolve (defaults, liquidity/redemptions, AI disruption, vehicle structure, and portfolio mark-to-model practices).

Key risk drivers discussed

Panelists framed the debate around at least five core concerns:

  • Borrower creditworthiness
  • Underwriting quality / “lack of discipline”
  • Leverage (including potentially “hidden” leverage)
  • AI’s disruptive impact (especially for enterprise software/business models)
  • Investment vehicle structure (liquidity, gates/redemptions, and marking/valuation mechanics)

Additional macro/market pressures mentioned:

  • Oil prices (gasoline/diesel costs)
  • Inflation
  • Economy
  • Interest rates

Is there a problem / is it systemic?

Consensus leaning: There is a problem in parts of direct lending, but not a systemic crisis comparable to the 2008 Global Financial Crisis (GFC).

Panel views (by participant)

  • Brad (Rogoff)

    • Sees issues but “not big enough” to be systemic.
    • Points to more aggressive lending, industry concentration (notably software), and higher leverage.
    • Notes possible asset-liability mismatch, but not at systemic scale.

  • Chris (Cross)

    • Also “not systemic,” with emphasis on narrative risk from “set-and-forget” and mark-to-model practices that can suppress visible volatility.
    • Expects cracks to emerge, potentially crystallizing via liquidity events.

  • Mona (Duffey / Smalley Duffey)

    • Not systemic; argues today’s risk is more distributed than in 2008.

  • Jeffrey Gundlach

    • Offers a broader credit-cycle analogy (dot-com, mortgage, and CDO era).
    • Attributes issues to asset/liability mismatch, opacity, and trust erosion driven by valuation practices and redemption dynamics.
    • Key datapoint: Q1 2026 showed “increasingly large redemption requests,” including “41%” in one instance, framed as an early warning of liquidity stress.

  • Vivek Batra

    • Emphasizes non-systemic nature and stresses differences between BDC/retail exposure versus institutional drawdown exposure.
    • Quantifies composition: ~25% of direct lending tied to BDCs (retail-ish) and ~75% tied to institutional drawdown funds (long-term locked capital).
    • Argues liquidity gates can mitigate runs in BDCs, while the rest typically lacks interim liquidity.

Quantitative metrics explicitly cited

Default / non-accrual / loss benchmarks

  • Public credit default rates: about 1.5% (and ~3–4% after adding “liability management exchanges” in referenced datasets)
  • BDC non-accrual rate: about 2% (average; some managers wider)
  • GFC default peak: 10.84% (used as a “distance from peak stress” comparison)
  • S&P 500 peak-to-trough in GFC: -50%

BDC redemption / loss timing (2026)

  • Q1 2026: “41%” redemption request cited (fund not named)

Valuation / marking dispersion (transparency debate)

Using PitchBook LCD-type analysis (per Vivek Batra):

  • For loans with 3+ managers: median mark dispersion ~54 cents (last quarter)
  • For broadly syndicated loan bid-ask: ~97 cents
  • For private credit loans marked below 70 (about ~1.5% of the market): median dispersion ~8.75 points
  • In public markets: about ~10% of broadly syndicated loans saw 10-point or more price movement during the quarter (framed as worse volatility/dispersion)

AI / valuation framing

  • “Rule of 40” software company multiples referenced:
    • ~20x free cash flow for rule-of-40 companies
    • ~15x for non-rule-of-40 (public comps comparison)
  • Software index down ~25% (used as an overall equity credit proxy)
  • Best performing bull investment grade software names down <1%
  • Worst performers down ~15 points or more (relative performance dispersion)

Refinancing / leverage math (portfolio risk)

  • Example mechanics: if growth fails, leverage can remain high; as software valuations fall, “equity cushion” erodes.
  • In a scenario described by panelists:
    • Buying at 16x–20x (deal multiples from private equity)
    • Applying leverage that produces “sub-50% LTV”
    • But if valuations decline, effective LTV can rise materially (panelists cited ~70–75% LTV in at least one example)

Performance metrics / comparisons used

  • Volatility and mark behavior (private credit marks versus public loan/bid-ask movements)
  • Default / non-accrual versus historical stress levels (with GFC comparisons)
  • Focus on dispersion across managers, with the expectation that dispersion increases if recession/turn occurs

Disclosures / cautions mentioned implicitly

  • Regulatory scrutiny referenced:
    • Claim that the SEC is investigating “a couple of private credit firms” for fraud (no details or outcome provided)
  • Multiple panelists stressed:
    • This is not “risk-free”
    • Losses and dispersion are expected if the economy turns
    • Marketing/nuance issues (e.g., “semi-liquid” and “investment grade” framing) may mislead investors

Recurrent framing: “Not a systemic GFC,” but still acknowledging meaningful losses can occur.

Methodology / frameworks referenced (as discussed)

1) Systemic risk vs credit losses framework

  • Separate risk by vehicle structure and liquidity:
    • BDC / retail portion: redemption/gates exist
    • Institutional drawdown portion: capital is generally locked longer-term
  • Compare default/non-accrual levels versus historical peaks (e.g., 10.84% GFC default peak)
  • Assess contagion risk through interconnectivity, leverage, and mismatch

2) Marking / valuation transparency & dispersion framework

  • Compare mark-to-model dispersion and bid-ask across:
    • Private credit (manager-level marks)
    • Public broadly syndicated loans
  • Core idea: marks can look stable in “normal” times, but dispersion widens in stressed/distressed credits.

3) Software credit vulnerability framework (AI + growth)

  • More resilient software characteristics:
    • “System of record”
    • Proprietary data / network effects
    • Regulatory overlay
    • Switching costs
  • More at-risk characteristics:
    • “Point solutions” with horizontal exposure
    • Lower differentiation
  • Link disruption to credit outcomes via:
    • Growth assumptions failing → deleveraging not happening → higher effective LTV when multiples compress

4) Private equity / private credit linkage logic

  • Private credit can enable LBOs at high entry multiples.
  • If valuations fall and growth fails:
    • Debt service capacity weakens
    • Liability management/default risk can follow

Explicit recommendations / strategic implications

  • No direct buy/sell directives.
  • Emphasized takeaways:
    • Differentiation and dispersion matter: expect performance to diverge by manager underwriting and borrower fundamentals.
    • Be skeptical of marketing claims about liquidity and mark stability (with criticism of “semi-liquid” terminology).
    • Due diligence should focus on:
      • Loan documents/structural terms
      • True downside / “left-tail” outcomes
      • Refinancing wall exposure (especially in software)

Tickers / assets / instruments mentioned

No single-company tickers were provided.

References and instruments/index mentions:

  • S&P 500
  • High yield / investment grade credit
  • Broadly syndicated loans
  • Private credit / direct lending
  • BDCs (Business Development Companies)
  • CDOs (Collateralized Debt Obligations)
  • Mortgage-backed securities (MBS), including subprime MBS
  • Treasuries and Ginnie Mae (mentioned in SVB analogy)
  • Interval funds and gating mechanisms
  • LBOs (leveraged buyouts)
  • ARR (mentioned in software lending context)
  • Crypto: none mentioned

Presenter / source list (as named)

  • Brad Rogoff
  • Christopher Cross
  • Smalley Duffey (panelist name appears as “Smalley Duffey” / “Mona” in subtitles)
  • Jeffrey Gundlach
  • Vivek Batra
  • Howard Marks (referenced via Oaktree memo: “What’s Going On in Private Credit”)

(No additional sources besides the above were explicitly credited.)

Original video