It's like 1999, but without the good vibes

Just another value investor's unsolicited opinion on the bubble question.

I know, I know - you’ve heard enough value investors complaining about the valuation of US stocks. They were doing it in 2019 at 3000, they were doing it in 2021 at 4500, and now we’re all doing it in 2024/25 at 6000. I’ll probably look really stupid in another couple years when it’s at 8000.

I’m also strongly of the opinion that as an individual value investor - particularly one investing in primarily non-US, non-tech small/microcaps like myself - you should really not give a shit what the Mag 7 are doing right now, nor whether the S&P 500 is in a bubble.

But hey, it’s fun to write about - and I think I have a couple things to say that aren’t being said that much.

Multiples

I’m old school - I think the best measurement of overall market valuation is a good old P/E ratio. I know people like to quote Buffett saying the market cap to GDP ratio “is probably the best single measure of where valuations stand at any given moment” but… well I just don’t agree with the fella. It doesn’t give heed to the corporate earnings attributable to privately-held companies, and more importantly, it assumes corporate earnings to GDP is a mean-reverting measure - which does not seem to be the case. You can go back and read fund managers in 2013 warning that the market is too expensive, and a crash imminent, because that so-called ‘Buffett Indicator’ is at the same level as in 1998. Some even destroyed their performance because of it, and are only now realising their follies - I’ll follow Buffett’s practice here of criticising by practice, not by name.

But anyway, back to PEs. People are throwing a round a P/E multiple of 22 right now, and when I think dot-com I think ~40 - those aren’t remotely similar. But I don’t think that’s a fair comparison.

First, while the P/E multiple did actually peak above 40 back then, that was a couple of years later in 2002, when earnings were compressed - partially by non-cash impairments. At the peak in March 2000, the multiple was only 30.

The 22 number also has issues. First, it’s a forward number, not a trailing P/E. That matters, because for some reason, we seem to be expecting a 20% YOY increase in earnings in 2025. Seriously. All the numbers can be downloaded straight from S&P here (they’re on a bottom-up basis - as in, calculated by summing analyst expectations for each individual company). Now, if that actually materialises, the forward P/E is a very sensible number to use - but in 2000 people were expecting 10-15% earnings growth for 2001, and what they got was -20%. I think it being a bottom-up estimate distorts it somewhat as well, as I doubt any sensible macroeconomist is expecting 20% growth in overall corporate earnings.

Second, it’s an adjusted number - with things like impairments, restructuring and other one-time gains/losses removed. That usually makes sense to do on a single-company basis, but on a country-wide basis, those are very real, very recurring costs. And the 30x from March 2000 is unadjusted.

If we instead use the unadjusted trailing numbers, the ratio is currently 29. To state the obvious - 29 is pretty close to 30.

(If you prefer to use the adjusted numbers, in March 2000 it was 28x and currently it’s 26x - so a similar story.)

2021 is another relevant comparison for a period we all now recognise as a loose-money-fuelled bubble. At the peak in December 2021, the unadjusted multiple was only 24, and the adjusted multiple 22.

Shiller P/E

Another incarnation of the P/E multiple I’m fond of is the Shiller P/E (a.k.a CAPE). It uses for the denominator the inflation-adjusted 10-year average earnings, smoothing out all the volatility in earnings to give a cyclically-adjusted view. Because corporate earnings tend to grow 3-4% per year in real terms, it averages about 15-20% higher than the actual P/E ratio.

It does have its critics - in times of sustained unusually strong earnings growth, it will suggest the market is slightly more expensive than it really is - but I think the smoothing effect more than compensates for this flaw.

This is what it has looked like over the past 150 years. In the dot-com bubble, it reached an all-time high of 44x; in 2021, it peaked at 39x, and currently, it sits at 38x. It’s really quite startling how little time we have spent above the current level.

Granted, the last 10 years have undeniably been a period of unusually strong earnings growth - but then so were the 10 years to both 2000 and 2021.

Pockets

In both the dot-com bubble and the 2021 tech bubble, valuations were certainly stretched altogether, but in and of itself, 30x isn’t a wild number. What really made these periods bubbles, in my view, was that there existed certain pockets of the market where valuations were not just optimistic but frankly inexplicable; and any risks or bearish sentiments were simply disregarded. All that mattered was what the next guy would pay.

I think we tick that box too. To give a few prominent examples:

• Quantum Computing Stocks - I won’t pretend to be an expert on the technology, but from what I understand, consensus is it’s no less than 10-20 years away from being technologically ready for commercial use - and it’s not particularly clear what that use will actually be (currently its main party trick is breaking encryption - great for wreaking havoc, not so much for generating revenue). QUBT (Quantum Computing Inc) probably represents the height of the degeneracy here. Have a read of this short report (or this one) - the level of deception is insane. Faking revenue via contracts with undisclosed related parties, lying about NASA contracts, pretending their office building is a cutting edge production facility about to start mass production. All that is publicly available info. And yet, it’s up 1600% in the last 6 months.

• Microstrategy - do I even need to explain this one? A bitcoin proxy trading at several times the underlying assets, which sells shares to buy bitcoin and boasts about how they’re increasing the bitcoin per share by doing so. Madness.

• SoundHound AI - I’m sure there are more absurd examples of shitty AI companies with uncompetitive products and crazy valuations, but this is the one that sticks in my mind. For a $5b market cap ($9b just a few weeks ago), you can have $67m in revenue (some of which is customer contract cancellation fees, recorded mysteriously as “product royalties” - just a typo I’m sure…) and a -$118m loss, on an AI voice agent that actually just scrapes Wikipedia and news sites. Oh, and it comes complete with its own delusional stock subreddit. See short reports here, here and here.

• Palantir - a great business, as far as I can tell. Maybe one of the best - read this piece from an ex-employee if you’re interested. But they currently trade at 60x sales, 420x EBIT (Cisco, anyone?). Some reverse engineering reveals how absurd this is. If the company keeps growing at the 22% CAGR it’s achieved since 2021 for another 24 years, and EBIT margins expand from 15.5% today to 35% (keeping in mind this is a very people-heavy business, which provides tailored solutions to each client - it’s never gonna be a Visa on margins), your annual return over the 24 years would be just 8%. That kind of sustained growth over that long a period is almost unprecedented - it would require zero meaningful competition paired with near-bottomless demand. I don’t think an 8% return compensates you for the risk that one of those assumptions is not the case.

• Tesla - I think this one is kinda hilarious, because in 2021, when it was trading at 20x sales as a car company (a reminder you can buy other automakers for 3-6x EARNINGS), you could have said “China will produce much cheaper EVs; demand will slow down to the point that unit sales actually decrease, Musk will politically alienate half of their customer base, the timeline for FSD will continue to be pushed back, and other AV companies will come out with superior technology - I’m shorting the stock” - and you would have been right on every single count, and to date you would have lost money. Nuts.

• Apple - they haven’t grown unit sales for a decade, they haven’t grown the top or bottom line for 5 years - and yet they trade at 38x earnings.

• Cava - I know, they do good food. And the unit economics are great. But 240x EBIT? Come on. If everything goes perfectly, they might be fairly valued - but is that a reasonable basis by which to value things?

• Nvidia - this one may be the most controversial. The forward P/E is only about 30 -that doesn’t seem too crazy on the surface, but think about what is going into that E. The other tech giants are spending insane amounts on the AI build-out - Microsoft, for example, plans to spend $80b on datacenters in 2025; Meta another $40b. To justify a 30x multiple, these companies don’t just need to maintain that spending - which seems to me more like temporary build-out spend than a recurring expense - they need to substantially increase it. Microsoft ‘only’ earned $90 billion in 2024 - I don’t think it’s being priced like the $80b datacenter spend will be a recurring (nay, growing) cost. Which is it, Mr. Market?

Some euphoric valuations, to be sure. But outside of these pockets, what does the sentiment feel like to you?

Vibes

The dot-com bubble was a fucking wild time in America - from what I’ve heard. Wall Street had gone mad, transitioning from earnings to revenue then to clicks and eyeballs as the predominant valuation methods - whatever it took to sell the new garbage IPO (though it’s clear the banks were drinking their own Kool-Aid by the end of it). The media followed suit, bringing coked-up tech CEOs onto their TV shows to pump their own worthless stocks. Sticking “e” at the start or “.com” at the end of your company’s name could double its valuation overnight. For the average person, risk was out the window - the only concern was whether your neighbour was making more money day-trading than you were. Expressing bearish sentiment would have had you laughed out of the room.

The 2021 tech bubble was a bit more restrained, but still a fairly euphoric time in the stock market. Fair value for anything tech seemed to be based on the best conceivable scenario - plus another 20 or 30 percent. All sorts of garbage was going public via SPAC, most of which subsequently fell 80-100%. Profitability was a bad thing for software companies, because it reminded investors that valuations are supposed to be based on earnings and cash flows (10x sales sounded more reasonable than 100x earnings). Most importantly, bearish sentiment was once again laughed out of the room - don’t you know stocks only go up? The fed’s never gonna let the market crash.

But to me (outside of the aforementioned pockets), today just doesn’t have the same vibe. I don’t see the same excitement that permeated those two periods (granted, only one of which I lived through). I’m probably a bit insulated, being (a) a Brit and (b) a value investor who mostly listens to other value investors - but it seems to me like bearish sentiments on US stocks are pretty much the norm among individual investors. Even the big banks (who make more money if their customers believe the future is bright) aren’t particularly bullish - Goldman caused some chatter a couple of months ago when they predicted that US stocks will return 3% annually over the next 10 years. CNN’s Fear and Green Index - which is not a brilliant indicator in my opinion but is at least another datapoint - seems to agree with my read, with the current score of 38 being well within fear territory.

In a way, today’s prices - outside of cutting-edge tech - feel more driven by a begrudging acceptance that US stocks simply outperform; and if you buy the S&P500, it doesn’t really matter what the current valuation is - you’ll do well if you just have patience.

A Footnote - Mag 7 Earnings

(I couldn’t find a good place to slot this in above, so I’m just dropping it at the end.)

Here’s something I think a lot of investors don’t appreciate - the enormous investment that’s currently flowing to Nvidia from the rest of the Mag 7 is essentially causing earnings to be overstated. When Nvidia sells a GPU to Microsoft, Nvidia immediately recognises the profit associated with it. Microsoft, on the other hand, capitalises it and depreciates it over its 3-7 year life. So in year 1, an aggregate profit is recorded, regardless of whether the GPU ends up actually producing any value.

Of course, that’s the same for any investment anywhere - but two things make this different. The first is that this intra-M7 investment has ramped up incredibly quickly - with typical ongoing capex, depreciation on the last few years’ worth of capex mostly offsets the effect. The second is that, unlike most investments companies make, it’s entirely unsure if these GPUs will even come close to paying for themselves. If they do not, and the tech giants subsequently scale back datacenter investment significantly, you get a double whammy: not only does this positive effect disappear, but you now have the depreciation expenses of the last couple years of investment further reducing accounting profits, with no offsetting benefit.

This effect matters when the Mag 7 accounted for almost 75% of the S&P500’s earnings growth in 2024. And it will continue mattering in 2025, as they are expected to account for a third of earnings growth. Is it a ticking accounting time-bomb which is soon to blow the market to shreds? No, far from it. But it is an effect that makes today’s M7 valuations that bit more extreme, and it’s worth being aware that it may reverse.

Caveats

I’m adding this section after-the-fact, because I think I probably came off a little more bearish than intended in this post. While I do think the current state of the market bears resemblance to 1999, I also think it’s pretty clearly not as extreme - both in terms of overall valuation and in the absurdity of those pockets.

While I can’t help but note that every time valuations have gotten this high, it’s been followed by a meaningful decline, I’m also not forecasting a massive crash around the corner. There’s a long history of value investors doing so, and looking very silly in hindsight, and I have no desire to join them.

It could well be the case that AI does deliver the productivity gains we’re hoping for (and these benefits accrue mostly to corporations, not workers) - such that earnings come up to meet prices at a reasonable multiple, rather than vice versa. Though this would probably also require a period of mediocre performance, rather than the excellent annual returns that have come to be seen as normal.

Conclusions

Howard Marks describes the market’s mood as a pendulum, which swings between the two extremes - euphoria/greed (where investors hardly care about risk at all), and depression/fear (where investors care about nothing but risk) - spending remarkably little time in the middle.

His view is that, while investors should not try to predict where the pendulum is heading, we should at least take stock of where it currently is, every now and then. That’s what this was.

He also recommends adjusting your positioning - the aggressiveness of your investments - based on that read. Personally, I’m very comfortable investing in the UK market (I have a better understanding of the culture and politics this side of the Atlantic, and can rely on my network for scuttlebutt where necessary). And I’m fortunate that while the US is near the euphoria end of the spectrum, the UK is more towards the depression end - so I’m just being aggressive in the UK instead (ask me how that’s going…).

But for those that prefer to stick to the US market, I think taking a more defensive stance - prioritising downside protection over upside potential, and earnings yield over earnings growth - may be a smart move right now. There are plenty of opportunities which should allow you to match the long-term return of the S&P500 if the bull market continues, and avoid significant loss if it does (REITs being a particularly common hiding place) - I don’t think there’s a need to be massively exposed to the possibility that “this time it’s different.”

That’s all from me.

---

If you liked this post, and appreciate getting this content for free, the single best thing you can do to help me is sharing it with a friend or fellow investor who you think may also enjoy my writing. (Liking the post doesn’t hurt either.)

Cheers!