Morgan Stanley’s Mike Wilson spends a considerable amount of time in his weeklies addressing “pushback” — so, ostensible counterpoints to whatever his current outlook happens to be for US equities.
Typically, he brings up the pushback only to refute it in terms that suggest it had little merit in the first place. That’s entertaining, even as it sometimes feels like Wilson’s conviction is such that no pushback, irrespective of its merit, would be sufficient to dissuade him.
This week, he took on the notion that while the equity risk premium might be extreme (in the “death zone,” as he put it) it’s misleading given the dominance of mega-cap companies in US benchmarks.
As a reminder, Wilson is convinced (and he’s hardly alone) that the situation illustrated in the simple figure above constitutes “strong evidence that the current setup is quite risky, particularly when combined with the poor earnings environment,” as he put it.
There’s a sense in which that assessment is difficult to refute. Extreme readings on key, fundamental metrics just “are what they are” to a degree, where that means that all caveats aside, and all else equal, the return asymmetry is skewed to the downside.
This is probably one of those cases, but be that as it may, Wilson did see some utility in addressing the argument that it’s a cap-weighted thing, to lapse briefly into the colloquial. That contention was easy enough for him to refute.
“We calculated the equity risk premium for the S&P 500 using the equal weighted forward earnings yield and found that we are still at the lowest levels seen since the Financial Crisis,” he wrote, before spelling it out: “The low risk premium is not simply a function of expensive, large cap growth stocks, but is a broader issue that could have far reaching impacts on the index.”
For what it’s worth (which is actually a lot here), he also noted that at the sector level, the only place where rates-relative valuation isn’t extreme is in Energy.
When considered with the “triple threat” posed+ by stretched valuations (at least when assessed in the context of US reals, which are roughly back to where they were at the beginning of the year), the challenging outlook for corporate profit growth and a Fed that’s highly unlikely to adopt anything that even resembles a conciliatory stance in the absence of an utter meltdown in the growth and labor data, it’s quite difficult to rationalize equities.
Underscoring the point was Bloomberg’s Ven Ram. “Given where we are on the current Fed rate — let alone where we may be over the next few months — the case to invest the marginal dollar in US stocks is wafer-thin,” he wrote, adding that at 5.54%, the estimated earnings yield on the S&P “is not too far off the two-year yield [while] the average yield on companies rated A by Moody’s is a pretty handsome 5.30%, so the compensation to hold higher-risk stocks is far from compelling.”
The figures below (from Wilson) tell a familiar story, but they’re nevertheless worth highlighting. On the left, he shows how the difference between YoY earnings growth and expectations as they stood a year previous widened steadily last year, culminating in Q4’s reported profit decline, which represented a stark contrast with the 13% profit growth analysts saw for the same period one year prior.
The figure on the right speaks for itself. The idea that margins are simply going to inflect this quarter and climb thereafter to levels that would’ve constituted quarterly records prior to the pandemic (note consensus sees profitability matching the height of the Trump tax cut boost by Q3 of next year) seems to suppose extraordinarily deft cost management, and across-the-board.
Even if that turns out to be true, that kind of execution in defense of the bottom line would surely entail layoffs, and not just in tech. Workers are consumers when they aren’t working. And someone has to buy the products, otherwise there’s no revenue from which to squeeze more profits.
Of course, just because it’s difficult to make a rational case for stocks doesn’t mean there isn’t a case to be made. Indeed, the most famous market adage of all posits a battle between a levelheaded investor and an irrational market. Not to spoil the suspense, but the levelheaded investor ends up broke.
Has Wilson refuted the impact of generative AI yet? Generative AI will be one of the biggest productivity unlocks in a long time. I don’t expect it to impact margins this year, but maybe the market is already skating to where the puck is going? As I’ve said before, we might just be witnessing the effects of a right-tail event that’s counteracting what we might otherwise expect to happen under these market conditions. No reason generative AI can’t be the source of the next big bubble.
That being said, I’d encourage people to read this thread on the latest OpenAI product appropriately named Foundy: https://twitter.com/labenz/status/1630284912853917697?s=46&t=1y9VJgfxNRYk4rj2pgbc_g
I’m not super technical, but it’s pretty clear that the next generation of generative AI is already upon us and it’s likely going to be a significant upgrade over an already incredible GPT3. Don’t be surprised if customer service jobs start to disappear rapidly much like telephone operators of yore.
We are likely years away from Generative AI having even small P&L impacts for publicly traded companies and likely 5-10 years from meaningful impacts. The leading company’s product isn’t that good for more than playing around at the moment. They don’t even know how to monetize it yet. Adoption curves take time.
In the immediacy, we’re more likely to see small start-ups leverage Generative AI for meaningful use cases which minimize cash burn. Publicly traded companies will learn from these use cases and/or hire from these start-ups.
You do have a good thesis overall, but your timeframe seems unreasonable to me when compared to historic adoption curves.
Admittedly, I am a bit of a tech optimist, but I do think the customer service use case is one that isn’t very far away. Then again, I’m used to working for smaller tech companies where you can move a bit faster. Overall though, I do think we’ll move along the adoption curve much quicker than past innovations. I work with quite a few vendors in the SaaS space and many of them have generative AI functionality in the pipeline for the next two quarters. Monetization models and IP questions (among other societal questions) might take a while to settle, but that’s no different than the internet in its early days.
As an aside that I’d be remiss not to mention, I have no doubt these models will have more than enough Heisenberg material to continue penning missives long past the point where H is ready to hang up his digital quill. For all we know, he’s already pawned off much of his work on our new overlord (and not just the artwork), and spends his days searching for that perfect place to isolate himself 🙂
I have to agree with Hopium Dealer. In 1999 I was in the middle, trying to help a couple of major engineering firms understand and put the internet to good business use. Many of those early goals/uses have yet to be achieved.
It never occurred to me that H is using generative AI to produce the amazing artwork for these articles. The artwork has always been amazing even before the availability of the new AI.
If he is using AI for the artwork, you have here a great example of a productivity enhancing / monetizing use
Thanks
Completely off topic. H have you seen Jeffrey Sachs interview in the New yorker?
GenerativeAI is cool, but mostly hype. It doesn’t know truth from fiction. There will be content generated in order to trick ChatGPT into thinking what’s not true is true.
Employers are hoarding labor because labor was so hard to hire in 2021-2022. That’s why labor productivity growth is lagging and that is why unemployment is staying low. If employer tastes change, you would probably see a 6% unemployment rate. But tastes take some time to revert if at all