Stock Market Leverage Higher than at Any Time in History | Michael Green

The scale of this has now meaningfully exceeded the scale of the housing crash. The share of US wealth—household wealth—that's held in equities has risen well above real estate. It has actually exceeded all other forms, including private business equity, by a wide margin. The participation rates of US households in the US stock market are extraordinarily high. And I don't know how else to say this other than: this is being driven by how we choose to invest rather than a thoughtful application of investment principles under a Ben Graham-type format. I know that people are screaming at the screen, 'Shut up, you boomer, you don't understand the new world.' I would just challenge you—challenge those individuals—that what I'm describing is not a sentiment, as much as I'm describing the math of the products themselves and the implications associated with that.

It's a pleasure to be here, Anthony. Thank you for having me.

Yeah. As you know, I operate off of a model that suggests that the majority of the impact that we're seeing in markets is largely a mechanical function of the way that we've chosen to invest—whether that is passively or through levered ETFs, which are actually the subject of a lot of the research I'm doing right now. There's very little thought that is being applied to the market, and I don't mean that in the context of people are doing stupid things. I literally mean it in the sense of: if you send a buy order in for the 2x levered Nvidia fund, it is going to buy two times as much Nvidia as you actually sent dollars. There's not going to be a debate within the fund about whether or not you should be buying this, you should not be buying this, etc. It's simply responding to the mechanical flows, and those flows themselves, particularly in levered vehicles, actually can create endogenous liquidity because of the way that they lever themselves on a daily basis. All of my work is suggesting that this is actually the majority of what we're seeing. Unfortunately, it's similar to the behavior of meme stocks, in which option positions created conditions under which forced buying from those hedging their underlying market exposures, in an attempt to preserve a neutral directional component, powered moves significantly higher in stocks like GameStop or AMC.

That same phenomenon is now unfortunately playing out at a much larger scale in some of the largest companies in the index, and is powering a lot of the behavior that we're seeing. I'm not arguing that the fundamentals for memory stocks are not correct. What I'm arguing is that the multiple of flow that is occurring here has actually destabilized systems that were used for very predictable reasons in the institutional space. The 2x levered ETF complex is a phenomenal source of profits for what's called volatility harvesting, in which you pair a long and a short position—the 2x levered long and the 2x levered short. It's part of the reason why these are always introduced in pairs, and you can create a performance calendar that really just looks like what's called volatility harvesting. Just to link the math directly behind that: imagine you have a stock that is up 10%, down 10% the next day. Many academics will study that and say, well, we need to average those two, and therefore there's zero gain associated with that stock. But in reality, because returns are linked—the first day it goes from 1 to 1.1, the next day it is multiplied by 0.9—that means you're left with 0.99. There's actually been a loss of 1%. That's what's referred to as volatility drag, and the strategy to capture that is called volatility harvesting. If you introduce that in a 2x levered version, it's not twice as much gain; it's actually four times as much gain, because now you have 1.2 times 0.8, which gives you 0.96.

That has been used by the institutional space for an extended period of time. The rapid proliferation of these types of ETFs in very illiquid single stocks has actually created conditions under which that behavior, which manifested itself as what's called an elasticity between the two shares that historically printed like minus 0.85 to minus 0.9—which indicates that they are broadly perfectly matched in terms of their behaviors—actually shifted significantly as retail began to use these levered ETFs to build long-term positions as a substitute for margin debt, which became very expensive and had recourse associated with it. And unfortunately, it's a very bad strategy, because the returns that are required in the underlying stocks to break even are very high. But it has created conditions under which price behavior is largely being determined by these mechanical aspects. That elasticity, just to put an actual number on it, has actually turned positive in some situations, suggesting that people are actually dollar-cost averaging into these strategies, which is a phenomenal way to lose money over time, but can actually power the stocks to extraordinary levels.

And that increase in the stock price has of course created a narrative that the commodity cycle in memory is over, that semiconductors will remain extraordinarily high profit-margin products, which is unfortunately, I think, creating both policy behaviors, management behaviors, etc., and so rolling over and actually impacting the real world—and George Soros's reflexivity concept. But it's creating a story that people are buying into, that everything is fantastic and AI is the solution, and the work that I'm doing right now actually suggests roughly two-thirds of this move is just tied to this mechanical component. So sure, Micron should have appreciated on the increase in earnings, but the multiple should have contracted significantly, and significantly more than it has. On a fundamental basis, I would highlight that the sanctions against China have created conditions under which China is aggressively ramping up its semiconductor production and producing in many of the areas, attempting certainly to produce in areas it has historically been unable to do. Anytime China has put its sights on a market, whether it's solar or steel or aluminum, etc., it has come to dominate that in the low-cost framework. That suggests to me that actually the commodity cycle may be much worse in terms of memory and chips than people are currently thinking about. That's my hypothesis—I can't be correct, I can't articulate that I'm correct on that, but I am correct on the math at this point: that it is the mechanical components that are driving these prices to extraordinary levels and creating the volatility that we're experiencing. And unfortunately, that suggests that much of this money is going to evaporate into that loss volatility and continuous releveraging framework. I think it's an unfortunate tragedy that is playing out before our eyes, even as people think that they're getting much wealthier.

Yeah, I mean, just to put that in context: 3% of the Korean population apparently has received a margin call in the last couple of weeks. To scale that to the United States, that would be roughly the equivalent of 10 million people experiencing a sudden loss in the market, tied to a leveraging feature that most people don't understand. And so Korea has actually only shut its market eight times in history, including the Asian financial crisis. It has now shut it six times this year. They have banned 2x levered and levered ETFs. They've expressed regret that they ever approved them. That suggests, as I think is the case, that the regulators don't understand the products that they're actually introducing and how those are being marketed to retail. And it is creating conditions, unfortunately, under which the wealth that is flowing into this space is likely to evaporate very quickly, with adverse consequences for the economy.

Well, unfortunately, when you say 'blow up,' part of the process of blowing up is actually just that the capital itself is lost. And so that influence of the flows into these vehicles is going to be a temporary phenomenon, but it's going to have long-term lasting consequences. I liken it to Tesla being included in the S&P 500, which caused a fantastic appreciation of Tesla, which has largely gone sideways to down in the period since. But unfortunately, it has maintained itself at a very, very high valuation even as the fundamentals have deteriorated. I think this is going to play out in a very similar fashion, and it ultimately means that the profits of the S&P 500 are likely to become far more cyclical and far more sensitive to the technology sector, primarily because we've dramatically increased the representation of hardware within the index. I understand that that is contrary to the narrative that most people are experiencing, but it is what the math suggests.

Yeah, I mean, to answer your question directly: at one point the levered ETF complex had grown to about 200 billion dollars in AUM. Multiply that by approximately two to two and a half, given the relative leverage across the products, and you're suggesting suddenly somewhere in the neighborhood of half a trillion dollars in capital that has flowed into this space, that ultimately is likely to evaporate over time due to this volatility drag feature. That's a real loss for the economy. It also suggests that a lot of capital is unfortunately being misdirected under the narratives that have emerged. Many of these companies—I would highlight SpaceX's very first action after getting public was to go out and borrow capital. That borrowing is underwritten not against the cash flow of SpaceX, which is fantastically negative, and nobody in their right mind would lend that money to SpaceX in the absence of a multi-trillion dollar valuation.

But that credit ultimately has to be repaid either through cash flow or through share issuance, and if the share prices collapse, then that leverage of course will evaporate alongside it. We're seeing this across the semiconductor space with the construction of data centers. An incredible amount of leverage is being added into the investment-grade community—the debt that can least afford any form of credit risk—and it's largely being underwritten against the equity values that we're seeing posted on the screen rather than the cash flows themselves, which, as I noted, have deteriorated and in many cases are quite negative. So we've created a replay in many ways of what happened in the housing market in 2006, in which credit underwriting was effectively written against the home prices, not necessarily the capacity of the underlying borrower to pay back the loan. This looks very similar, unfortunately.

I mean, the most nefarious component of it is what are called special purpose vehicles, where effectively the underwriting comes from the high-quality supplier, and they are not liable for it because their customers have actually done the borrowing—unless of course the customer can't pay, and then that debt can flow onto the balance sheet. And we see this across the MAG 7 at this point, and across the AI space. People should be concerned about this, but as I said, that is contra the narrative that is out there, and most people will not believe it until they actually see it.

I mean, the quick answer is: it depends. It depends on how much capital continues to flow into the levered space. As I mentioned, we've seen a radical change in behavior. Historically, that negative elasticity would suggest that people were harvesting out of these products. Today, unfortunately, the evidence is that people are dollar-cost averaging into these products—in other words, contributing additional capital to offset their losses in an attempt to quote-unquote pick the bottom in this correction. I just—unfortunately, the math suggests that the returns that are required in these securities to offset that volatility drag is impossibly high. In the case of a company like Micron, for example, which is currently printing a realized vol in excess of 100, the returns that need to be generated are well above 100-plus percent on an annual basis in Micron stock for people to break even that are dollar-cost averaging into this. It's possible, right? We could get bailed out by this, but I think it is unlikely. And the net impact of it is that the lending that is occurring on a regular basis is actually showing up now in the credit markets, as I mentioned, both for project financing, but also what's called non-depository financial lending, in which the banks are actually extending credit to financial companies—in this case fund entities in the form of an ETF, which is an individual company.

And because of the capital treatment of that, this is viewed as relatively low-risk lending. After all, they're only borrowing 50% of the assets. But that is a very different picture. When you have a 100-vol underlying asset, that's actually an incredibly risky lending scenario—to have 50% NAV against a 100-plus vol instrument. You have raised the risk in the financial sector quite meaningfully with this, and much of the narrative around the credit growth in that area has shifted from private credit, which is increasingly shut out of the market. And as they are shut out, we are seeing defaults and delinquencies continue to accelerate in that space. But the new form of lending is, hey, let's lend to levered financial ETFs. How much better could this possibly get?

Against that, as you point out, there are other strategies—for example, covered call overwriting, etc. Those strategies themselves are actually methods of underwriting debt. If you think about what happens to the covered call: you have very little upside, you receive the vast majority of your return in the form of the call premium that is being received, but you're exposed to 100% of the downside because you continue to hold the underlying security. If you do the payoff structure on that, that's just a form of corporate credit. And as a result, you have individuals underwriting corporate credit without any underwriting skills associated with it. This is—again, I'm not picking on retail for this. These are well-accepted and long-standing strategies. The scale to which they are being pursued, and the scale to which they are being relied upon for income, I think are broadly misunderstood by the user base.

Well, we obviously have to wait and see, but the scale of this has now meaningfully exceeded the scale of the housing crash. The share of US wealth—household wealth—that's held in equities has risen well above real estate. It has actually exceeded all other forms, including private business equity, by a wide margin. The participation rates of US households in the US stock market are extraordinarily high. And I don't know how else to say this other than: this is being driven by how we choose to invest rather than a thoughtful application of investment principles under a Ben Graham-type format. I know that people are screaming at the screen, 'Shut up, you boomer, you don't understand the new world.' I would just challenge you—challenge those individuals—that what I'm describing is not a sentiment, as much as I'm describing the math of the products themselves and the implications associated with that.

Right, and meanwhile the Bitcoin miners themselves have recognized that their primary asset is not the mining of Bitcoin, but the power purchasing agreements that they entered into when energy prices were low, and the offtake agreements they have associated with that, which they can now repurpose into sales to data centers. And as that occurs, the capacity and hash rate in the Bitcoin universe is actually falling, lowering the aggregate security of the asset itself, on top of all the other shenanigans that we were talking about in terms of levered Bitcoin holding vehicles like MicroStrategy, which of course has now turned into, hey, we will sell Bitcoin after all if we need the cash. This is unfortunately the reality of credit: if you have to repay debt, you have to get those dollars.

Yeah, I think there's an element of that. We certainly saw sentiment take a downturn associated with that. I think the bigger issue is the impact it has on the decision-making process of companies themselves, who are trying to justify the valuations with the expenditures that they see rewarded in the market. As I've mentioned many, many times in other podcasts, this is articulated most succinctly in a 2005 paper written by a guy by the name of Michael Jensen, called 'The Agency Cost of Overvalued Equities.' And he points out that in a fantastically undervalued equity, change agents such as vulture investors or activist investors can step in and buy an undervalued security and use the leverage associated with the voting of those shares to change management, change strategy, etc. There's nothing that can really be done against an overvalued security other than attempting to short it. And in the environment that I was describing, where the mechanical process and inflation of these bids that's occurring through the passive complex has wiped out the short seller community, there's really no capacity for any form of reform. And as a result, management teams increasingly become exceptionally spendthrift in order to justify the valuations that they're trading at. That in turn weakens their balance sheet, and ultimately, if their investments turn out to be driven by these frameworks as compared to pure economics—which you're hearing more and more people start to question—it creates conditions under which a very sudden deceleration can actually occur as stock prices decline.

This is occurring against a framework in which the economic growth is decelerating already and has significantly decelerated. GDP growth has declined in both real and unit terms consistently. Full-time jobs have not grown now for nearly four years. This is creating conditions under which the economy is weakening, but an administration that is very focused on the stock market is convinced that nothing has ever been better. And as a result, we're increasingly entertaining narratives like tightening financial conditions as if it's desired, as compared to recognizing that this is basically the gas that we are now consuming in order to keep ourselves going.

Yeah, I think there's some evidence of that. After nearly four years—or three and a half years—of recession in the industrial sector of our economy, we've now started to have a cyclical recovery associated with that. Part of it is tied to the simple reality that very few things go down forever, and we are beginning to see an impact on pricing as capacity has been reduced in many areas of the economy. Consolidation that has occurred has allowed production to be decreased, and at the same time the introduction of things like tariffs, etc., created a relative bid for some segments of the economy, but the broader industrial base continues to deteriorate. If we look at auto sales, housing sales, etc., those all suggest that the weakness is quite profound and quite real.

And it unfortunately, as I said, is being camouflaged by gains in equity and an extraordinarily high real interest rate that is actually compensating those who have significant quantities of cash for really doing nothing with that cash other than putting it into bank accounts that are now earning relatively significant income for them. That in turn is being used to fund the financial speculation. Perversely, if we were to cut interest rates, I think the reality is that it would ultimately prove contractionary, as that income is replaced and spending from the upper decile in terms of capacity, because of the interest income falling with a cut in interest rates, would likely begin to retreat, adding additional weakness to the economy. It's a really sobering message, and you are catching it at the end of a very long and tiring day, and so I'm delivering it with a bit more sobriety than it requires. But this is unfortunately what we see developing at this point.

Well, again, I think it's important to distinguish between what's called a pro-social bubble, in which investment is creating opportunities for those assets to return to the market under distress conditions under which their cost structure is radically lowered. And I ultimately think that we will see that in the AI space—that we'll ultimately discover, much like internet search, it simply becomes advertiser supported and it's available to all individuals. Unless of course we choose to embrace the national security narrative that this needs to be protected in some manner, and the unique champions that are pursuing their proprietary frontier models are, you know, too big to fail before they even get public. But certainly a possibility with this administration.

Yeah. I mean, it's remarkable. This is a president who ran under the campaign that he was the only president who hadn't started a war in the past x number of years. And now we've seen him start multiple wars, some of which were extraordinarily short-lived, like Venezuela. Others are proving to be more protracted and more negative, candidly, for the observation that the US has functioned as the global policeman. Ultimately I think that we will win against Iraq—or Iran, I'm sorry—but the cost of the loss of face and the loss of perceived capability is certainly emboldening countries like China to more aggressively pursue their own strategies, and unfortunately raises the risk or probability that a wider conflict ultimately develops. If you see the police respond to a home invasion ineffectively, it raises the probability that there will be future home invasions in the area. It also raises the probability of a broader increase in antisocial behavior. I think we're seeing that unfortunately on a global basis, and policy makers seem unwilling to examine their priors.