Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Rapid severe AI disruption has become a rolling theme among equity types, who have been in full reverse-ferret mode for some months now around their penchant for capital-light firms.
With equity traders increasingly turning to speculative fiction to get ahead of the pack, downside risk scenarios are now getting more attention among credit types. Matthew Mish, UBS credit strategist, in a research note last night:
Investors increasingly want to talk about AI disruption and our tail risk scenario: a rapid, severe AI disruption. This is not our baseline; however, at the risk of Monday morning quarterbacking, over 80% of what is written below was published back in November. What is new: a clearer catalyst – rapid, severe AI disruption
What does this downside scenario look like?
assuming contagion impacts not modelled in our earlier note, we anticipate US HY, LL and PC [high yield, leveraged loan and private credit] defaults could rise to 3-6%, 8-10% and 14-15%, respectively.
Holy moly. These are big global financial crisis-like — or at least dotcom-bust-like — numbers. While we haven’t spotted a time horizon for the forecast, we’ve whacked them onto this chart of historic default rates to give you a sense as to where they sit versus the last couple of decades’ experience:
Junk bonds 🤷♂️, levered loans 👀, but private credit 🤮
Could private credit default rates really jump all the way to 15 per cent? While this looks like catastrophising for the sake of catastrophising, UBS have decent grounds for their ‘yes’ (while reiterating that this is their risk case and not their base case).
The argument centres on an examination of sector-level default rates in the US high-yield market over the past couple of decades. Around a fifth of US high-yield energy companies defaulted around 2016 and again in 2020. Around a quarter of high-yield basic industry firms defaulted in the dotcom bust and again around the GFC. More than half high-yield finance companies defaulted in the GFC. And a full four-fifths of high-yield telecom companies defaulted in the dotcom bust. While any individual borrower can default for whatever reason, default waves tend to come for whole sectors at a time. As such, diversification isn’t just a nice-to-have for credit investors. It’s a thing that stops you losing your shirt.
And, in case you missed it, private credit and leveraged loans have an extraordinarily concentrated exposure to a couple of large sectors — software and business services. These sectors are precisely the ones in the line of prospective AI fire.
Moreover, even before the threat of AI-mageddon, private credit showed signs of strain, like the share of interest paid-in-kind reaching post-pandemic highs and leverage in some sectors reaching 7.5-8.0x debt/ebitda. Furthermore, weakening covenants, aggressive earnings adjustments and opaque valuations make the prospect for recoveries in the event of default look meagre.
What might this mean for valuations?
in this extreme case, regression analysis suggests US IG, HY and LL spreads could trade to 160-170, 575-675 and 800-900bp, respectively.
For public market credit, this would be a substantial spread widening that would pummel excess returns. But nothing we haven’t seen a fair number of times over the past couple of decades.
UBS don’t offer a view on where spreads on private credit loans would shake out in this scenario. Perhaps because a) private credit loans don’t trade, and; b) the scenario calls for a near evisceration of new issuance:
credit availability will tighten materially, particularly for firms dependent on leveraged finance markets – PC and LL issuance could decline 50-75% YoY.
OK, we’re going to call it. The UBS downside scenario would be a full-blown credit crunch. And while it’s primarily a nonbank financial institution credit crunch, nonbank financials are large. And these things have a habit of metastasising into financials. As a major bank, the point is not lost on UBS:
Lastly, financials will be impacted through several channels. One is NFBI loans, currently $2.5tn including all undrawn commitments. In a tail scenario, we estimate ~$1.6-1.8tn in total drawn exposures, of which about 30-40% is to private equity/credit/BDCs or SPVs/CLOs/ABS and we would consider higher risk.
Quite how AI hyperscalers will be able to finance their planned data centre build-out in this scenario is something we’ll leave to another post. But it’s intriguing to note that investors’ worries that AI hyperscalers will win too much might actually bork the financing of AI’s commercial rollout.
The pivot in investor concern from AI-might-disappoint to AI-might-win has seen trillions wiped off the market capitalisation of capital-light firms over the past few weeks. So far, index-level public market credit spreads remain super-tight. An absence of economic recession, some middling inflation, and supportive monetary and fiscal policy should make for a constructive investment environment.
But that’s as long as the rapid, severe AI disruption increasingly priced by stock jockeys doesn’t rip through business models. If it does, UBS’s tail risk scenario doesn’t actually look particularly panicky.
Further reading:
— What the leveraged loan market can tell us about the software sell-off (FTAV)
— Will software eat the creditors? (FTAV)
— US investment grade credit markets care about the tech wreck, just not very much (FTAV)
— It’s not a default if you never ask for cash — right? (FTAV)
— How bad could private credit default rates get? (FTAV)
— Inside the big boom in ‘business development companies’ (FTAV)
— Inside the private equity-insurance nexus (FTAV)
Read the full article here