Crisis on Markets: Speculative frenzies - Part 1
From the archive, #20 (Sept 5, very tedious 14 minute read)
This week: one topic (speculative episodes) and an update:
Speculative episodes
- Long run efficiency vs short run irrationalism
- Variable reward systems and comparisons to gambling
- Embedded debt
- Recency heuristics
- Macro importance of the supply side (liquidity)
- Passive pyramid schemes
- Arbitrage mechanism breakdowns
Quarantine egg update
Thoughts on speculative bubbles:
Following along from last week's discussion of Tesla, I have been prodded to discuss my observations about speculative bubbles, which would be my choice of Ph.D topic were I ever to pursue that avenue, while being fully cognizant that my bias towards empirical data and my lack of reverence for existing models would surely doom my candidacy from the outset.
My direct involvement with financial bubbles goes back to my time in Japan in the mid to late 80s, but continued with the EM/Asia boom in the mid-90s, the tech boom in the late 90s when I was head of Asian e-commerce, the emergence of China in the mid-2000s, and the Bitcoin run in 2016-17. In all of these I was trading at some level or another, and since the mid-90s my research role was usually focused on divining market direction during those ebullient times, or their aftermaths. I have participated less in real estate expansions, although I did buy an apartment in Tokyo in 2004 near the bottom of that 12 year cycle, and in the US after the 2008 crash, but my involvement there was oriented more towards research and market prediction than first person involvement.
As I have mentioned, I am enamored by bubbles and the crises that inevitably follow. The outlier events of human history are tremendously revealing, a crucible in which to observe human nature at its strangest and most extreme. While I focus on equity bubbles in this note, much of this applies to markets like cryptocurrencies and real estate, though not bonds, which have the built-in convergence mechanism of redemption -- whereas most other assets are perpetual/non-redeemed in nature.
In this note I will discuss frameworks which I think are useful for the discussion of speculation, and use those to develop a hypothesis for how the current frenzy for tech may unfold. This post is part 1, with a focus on enabling structures; hopefully I can finish up next week with a discussion of inflection points, and downside catalysts.
[Writing this after the tech stocks' mini-crash on Thursday/Friday has been a useful reminder that these periods are indeed rather risky.]
Long run efficiency, but periods of short run inefficiency
In the long run, earnings should be correlated with market capitalization; bankrupt companies will go to zero, companies with large earnings will do well. In the old (Graham and Dodd) days, stocks used to trade at 10X current earnings and had dividend yields above government bond yields. Clearly the market would be much lower if that was the prevailing valuation mechanism, but that valuation paradigm has been long left behind as fund managers decided they care more about relative performance rather than absolute capital preservation.
As the market has become more growthy, both because of successful tech companies delivering high growth and ascending to the top of the market cap ranks, as well as the advent of the next generation of similar high growth companies, investors have to go out further into the future to value companies, which implies taking the risks (execution, macro, etc) that go along with more time. Once a market is dominated by growth stocks, investment performance becomes a combination of estimating future earnings but also guessing *how far into the future the rest of the market will be looking for those earnings*, something which seems to inch forward but jump backwards. Uncomfortable periods of seeming inefficiency may prevail, as the market ascribes a high probability to the success of many growth stocks, even if only a handful will eventually become successful.
An equity bubble is often the function of an increase in a market's expected growth duration (and overly optimistic probabilities for growth companies in aggregate), seen from the eyes of a value manager whose relative underperformance has catalyzed fund outflows, creating job insecurity. Since the demise of the Nifty 50, value bubbles are easier to refute and thus harder to initiate. The fact that the FAAMG group has delivered / exceeded their earnings targets supports the idea that valuations for some growth companies can be more than justified.
The macro importance of the supply side
We have discussed the influence of the Fed on liquidity in previous crises, but as I think about this cycle, it's almost as if they've removed government bonds from consideration as an investment alternative, by driving their yields so low that there is little remaining upside. I used to make a joke that portfolio theory should include rocks as an asset class, because they diversify standard portfolios and have uncorrelated returns (zero); maybe bonds have actually become rocks. Assets must have some upside potential to justify the capital risk of holding them, otherwise investors will default to demand deposits (which is what they have done).
I'm inclined to believe that bubble periods tend to converge around policy changes in liquidity, because a sudden difference in the investment environment creates large flows and momentum over short periods of time, compared to normal economic expansions when earnings and corporate events occur more gradually. I would include large changes in debt policy as well, such as the Asian countries pegging their currencies to the US dollar, allowing domestic companies to source much cheaper USD funding.
Variable reward systems and comparisons to gambling
Speculation has the entertaining aspects of variable reward systems (like gambling) which derive their power from the uncertainty of an outcome but with an observable average result which keeps the participant engaged.
That said, reducing speculation to mere gambling is both silly and fundamentally wrong, even if they are both variable reward systems. In particular:
- Gambling is a negative sum game, the house always wins and thus aggregate wealth is not increased. Financial markets are usually positive sum games, because of the wealth effect -- everyone who owns stock becomes wealthier when the market rises. There are some losers (people who are short), but the winners dominate the losers. In gambling, my loss is the casino's gain and vice versa. Index returns essentially broadcast the daily wealth effect; in a casino, an index would decline at the rate the casino was making money, which would be bad for business. To be fair, the odds of the equity market can turn around rather quickly, and the negative wealth effects are at times very extreme.
- In equities at least, the ability to short (play the other side) means that investors have two ways to make money, they are not wedded to a particular side (even if most are long).
- The underlying stories of assets like equities (and crypto) change over time, as do the market's reaction to those changes, whereas the odds and fundamentals of gambling remain largely constant. Some might argue that speculators therefore are very incentivized to take an interest in all factors which affect asset returns, which is probably a good thing.
- Speculation enables companies like startups to pursue risky ventures, some of which work out, and some which do not. Gambling makes casinos and bookies richer -- although one might argue it is entertaining and makes things like sports more exciting.
- Because of its long term wealth effect, most investment portfolios hold equities, which are affected by periodic speculative activity. There is tension between long term prudence-value and short term exuberance-adrenalin playing in the same sandbox, and is asymmetric -- the speculators probably mess up the investors' sand castles more than the other way around.
Embedded debt
A discussion of speculative bubbles requires some consideration of the debt environment. Although casino gambling has some degree of credit (markers for whales, for instance), most mass market gambling has low levels of associated debt (I assume the inveterate gamblers who finance their habit from home equity loans do not dominate). In the equity market, margin loans are the most visible debt mechanism, and one which helps explain reflexive spirals (stocks fall, margin calls are not met, broker sells position, driving market down more, creating more margin calls). Brokers make healthy profits from margin loans, and therefore encourage their use, especially now that retail commissions are close to zero.
Debt is not an absolute requirement of a bubble/crash cycle -- as far as I am aware, the cryptocurrency run in 2016-17 was more a function of a rapidly widening base, rather than particularly debt-fueled -- but most bubbles rely on increasing amounts of debt, and deep crashes are usually debt-related.
Recency heuristics
Many investors overweight recent risk/reward relationships in spite of long term caveats. When I was at INSEAD in the early 90s, a (Finnish) finance professor told his students he put all his savings into Nokia, rather than have it diversified as per conventional portfolio theory. Although we may snicker now, Nokia had a fantastic run from 1990-2000, although it endured 50% retracements in both 1994 and 1996. I mention this story because it illustrates how recent data can overpower long term relationships and theoretical knowledge, which in this case was compounded by a sense of national exceptionalism.
To try to characterize this in the current environment, I looked at rolling 1M returns and 1M volatility for Tesla (unprofitable tech representative), Apple (profitable tech), JP Morgan (cyclical), and MSCI All-Countries World Index (all listed equities). The Sharpe ratio, essentially a measure for how much return you receive per unit of risk taken, in the long run tends to about +1.0. Since mid-April, Tesla's Sharpe ratio has oscillated between -0.5 and +5.0 (ie 5X return vs risk); Apple has been between +0.2 and +4.0; JP Morgan -1.6 to +2.0, and MSCI World between 0.0 and +3.0. Of course, their monthly volatilities are currently very different: MSCI World 3%, JPM 6%, Apple 9%, Tesla 20%.
The statistics above make good recency arguments for Apple and Tesla, because they give you returns which are usually a multiple of the risk you take, their relative risk is high, and drawdowns have been rare (hence few periods of negative Sharpe ratios). The heavily diversified MSCI World has no drawdown periods (in this statistic), but both the volatility and Sharpe ratios are lower, so the returns are a different order of magnitude compared to Apple and Tesla (which are among the highlights of the MSCI World return). The cyclical JPM, with periods of Sharpe ratios below -1.0, makes little sense for a speculator unless they believe their market timing with respect to cyclicals is impeccable.
The point of all this is to say that the recency heuristics make sense, we all tend to overweight our own data and observations (and the stories of colleagues making money on Tesla or Apple!) and that will tend to skew our view of the risks and rewards of the market 'casino', while assuming that, as someone of clearly above-average trading acumen, we will be able to get out before the odds become too skewed to the downside.
Passive pyramid structures:
One idea I have been playing with is the emergence of self-organized (spontaneous, passive) pyramid structures. The hypothesis is that during a speculative period (usually driven by an increase in liquidity policy), there will be some sectors or stocks which will (somewhat randomly) emerge with a set of interrelated "blue sky" criteria:
- strong price momentum
- compelling narrative / concept
- some history of execution delivery
- some history of defying the "experts"
- widespread financial news coverage
In the case of Apple and Tesla (and Zoom), it helps to have a consumer product which is well-liked and revolutionary/futuristic. In order to maintain price momentum, a widening of the investor base is required, which is why news coverage is so important, until the investor base can grow no more (because earlier investors are encouraged by extreme prices and slowing momentum to exit or trim positions).
To me, the simultaneous stock splits of Tesla and Apple are about access, or rather the *promise* of greater access -- even if the marginal effect of a stock split isn't large, it supports the *narrative* of a widening investor base. The innovation of fractional share ownership for retail brokers is similar. The rumors of Tesla being added to the SP500 (which didn't come to pass, we found out on Friday), is a similar phenomenon -- there are few ways to force institutional investors to buy an overvalued stock -- but inclusion into the SP500 is a compelling one..
What I'm hypothesizing here is that certain stocks will spontaneously emerge as centers of speculative activity, and that these will become passive pyramid schemes, drawing in more and more participants until they eventually collapse (because price momentum goes negative, without any valuation levels to support the price). And even if they aren't active pyramid schemes like in the late 1800s or early 1900s, the principles are similar, and they will expire when they run out of additional fuel sources, something Tesla seems to have reached (for the time being, anyway).
Arbitrage mechanisms
It is often said that "the market can remain irrational longer than you can remain solvent" and during extreme periods, market correction mechanisms like shorting and arbitrage can be suspended if the underlying trend is overwhelming. Once the momentum wanes, however, one should expect these mechanisms to return in force.
In the early 1990s, the investment banks were having a ball doing Nikkei 225 stock-futures arbitrage. Since futures traded at about a 1-1.5% premium (quarterly), this meant 4-6% nearly risk free annual returns, and since we were levered 20X, this produced aggregate annual returns on capital of about 80-120% with minimal risk and effort. The Japanese government, which was looking anywhere else but their own policies as a scapegoat for the market decline, decided to support the market by buying up futures, further increasing the futures premium. The investment banks duly increased their balance sheets, but at some point (I think it was 1992), we literally ran out of capital in our Japanese entities, and the futures premium jumped again, sustaining absolute premia levels above 1.5%, which was unprecedented -- because arbitrageurs could no longer keep the index and futures connected any more, the greater inefficiency regime persisted until we could get more capital.
At least in that case, we were making money. In the case for hedge funds looking to profit on Tesla shorts, the vast majority of those positions have blown up so even if they can make a strong case for overvaluation, the history of that trade (similar to the misery of the decades-long short JGB trade) probably keeps most of that activity on the sidelines.
Additional areas:
This is already longer than I would like it, but other topics I thought relevant, and might discuss in the future include:
- Moral hazard: The Fed/Greenspan put and the strategic side-effects of well-meaning tactical measures
- Government signaling: short sale rules, emergency policy announcements
- The widespread error of using deterministic frameworks on non-deterministic systems
Part 2 next week will discuss some thoughts on inflection points, and a selected examination of historical turning points. Hopefully it will not be too late to say something useful about the current tech enthusiasm.
My sense is that Tesla will underperform both the NASDAQ and Apple next week.
Quarantine update (day 8/14):
The egg competition is entering its final week. I have earned four eggs of 18 allocated (2 egg underperformance), and have used them all -- but only one was consumed directly (loco moco). The other three were ingredients in something made for collective consumption (veal meatballs, carnitas fried rice, tarte tatin). My housemate put four eggs into an amazing two-layer chocolate flan, which I am tempted to make once I have access to an unbounded egg supply.
We have been watching Succession, which improves on re-viewing; the dialogue simply sparkles. Since we are on a bit of an egg theme:
"If you want to make a Tom-let, you have to break some Gregs"
Six more days until I am free to roam the beaches and bowling alleys of the Big Island, the challenge for the weekend is the assembly of a very large Lego model of a Star Wars Tantive IV spaceship, fueled by grilled lemon lemonade (and alcoholic derivatives) and Mexican-style chicken fried in manteca.