The biases liable to cause you the most problems tend to be the ones you don’t know you have.
The human mind strives to make sense of the world, and so it abhors true randomness–where important outcomes fluctuate wildly for reasons we don’t understand. It is one thing to not be able to forecast the future, much as we may want to, but even more humiliating is it to be unable to adequately explain the past. This leaves one totally at the mercy of the universe, making it difficult to even take reasonable precautions.
It does not seem to be a coincidence that many early societies ascribed weather events to the moods and caprices of supernatural beings. Much more than today, primitive man was at the mercy of weather fluctuations. Why, he would have thought, did the drought kill my crops, or the floods destroy my home? To be left with no answer at all would mean totally powerlessness–better instead to say the gods must have been angry. One can at least strive to understand their will, so they don’t get angry again.
This is because agency, even malign agency, is a relief. At least something is in control, even if we do not like it or fully understand it. True chaos, where nobody and nothing is in control, is far more frightening.
Of course, in our modern era, fewer people in the public square profess to believe in divine explanations for the sociological phenomena we observe in our lives.
But the same urge persists. We must make sense out of the chaos. Something must have agency. Something must be in control.
Without the gods, that something must be a person, or people.
And the ugly chaos to be explained, which has the power to ruin our lives, is not primarily weather. For sure, there are natural disasters that can ruin particular areas. But what has the power to instantly impoverish us on a massive scale in ways we don’t understand?
The answer is financial markets.
One of the strange things in the aftermath of the 2008 financial crisis was the fact that nearly all explanations had the structure of a morality play. There was, in other words, a villain, whose malignancy (either through stupidity, selfishness, or bad incentives) was the ultimate reason for the collapse in asset prices.
And this was an oddly bipartisan urge. The villains differed between the explanations, of course. But the existence of a villain was almost universal.
And as always, people prefer villains that reinforce their existing narrative of the world.
In the leftist re-telling, the villains were corporate greed, capitalism, and lack of regulation. Short-termism in Wall Street meant that financial executives had enormous incentives to issue dubious mortgages that they could then package into asset-backed securities (often while lying about the underlying assets’ content) and sell to gullible clients. Deregulation, such as the repeal of the Glass-Steagal Act, meant that the investment banking arms of banks were no longer separate from the commercial banking arm. This entailed that risk-taking behavior in certain exotic securities by the investment bank division would threaten the safety of ordinary deposits of individuals in the commercial banking division.
In the rightist re-telling, the villains were government meddling in the housing and banking markets. The Community Reinvestment Act forced banks to lend to uncreditworthy minorities in order to avoid government fines, thus inflating the housing bubble. Meanwhile, implicit government guarantees of the banks (which became explicit during the crisis, at least if your name wasn’t Lehman) incentivized the banks to take on excess risk, knowing they’d profit if prices went up and get bailed out if they went down.
And then there were the less explicitly political motivations that came down to human stupidity. Bankers optimistically thought housing markets would always keep going up and were over-leveraged on this assumption. Even though senior managers had enormous incentives to get the answer right by virtue of their large holdings of their own company’s stock (most notably Lehman CEO Dick Fuld), they still screwed it up. This happened through a combination of not understanding credit risk and not knowing how much the problem could spread through connected banks. Meanwhile, gullible and greedy homeowners took out mortgages they couldn’t afford hoping to flip houses, even if it meant falsifying their income to obtain the mortgage in the first place and then later whining when the price later dropped, leaving them underwater.
The motivations didn’t even need to always be coherent for people to like them and nod along. Lots of people enjoyed Michael Lewis’ book and movie “The Big Short”. Fewer seemed to notice that the movie couldn’t decide if bankers were stupid or evil and decided to have a bet either way. But it requires a certain schizophrenia to say that bankers were stupid because they didn’t know the obvious consequences of their actions and evil because they did know those consequences and figured out that they’d privately benefit.
Several things should be immediately apparent from the long list above.
First, the set of potential causes is sufficiently large that while it’s possible to show that some factor may have contributed, it is difficult to rule out every other potential cause. And indeed, it seems almost certain that the answer ultimately is a combination of at least several of the above causes.
In particular, the full explanation needs to account the time series as well – the “when”, not just the “what”. Explanations that fail this are obviously incomplete, particularly for claimed causes like “Wall Street Greed”. Did Wall Street only suddenly get greedy starting in the early 2000s? Did this replace the halcyon days of Wall Street philosopher kings during the 90s tech boom, or the 1980s bull market, personified in the character Gordon Ghekko from the movie Wall Street?
But the time series also causes problems for conservatives who want to blame everything on the Community Reinvestment Act forcing banks to lend to unqualified borrowers. While this certainly can’t have helped, this law was passed in 1977. If this is the main driver, why did it take until 2003 to really kick into action?
When there is a massive engineering or organizational failure, there are nearly always a considerable number of places where if someone had acted differently, the disaster would have been averted. The same seems likely here. A lot of the above explanations were indeed significantly involved in some form or another. Moreover, it would be foolish to write off the explanations above simply based on the fact that they have a particular narrative structure that seems overly convenient.
And yet, there is another perspective that didn’t seem to get much emphasis, either at the time, or since.
Which is the following.
In any financial crisis, there will be both a proximate cause and an ultimate cause.
The proximate cause asks: “Why did this particular financial crisis occur?”
The ultimate cause asks: “Why do financial crises occur in general?”
A related way of phrasing the distinction, then, is to ask the following:
If there hadn’t been the sub-prime mortgage crisis of 2008, would we have just had the state municipal bond crisis of 2018, or the commercial mortgage-backed securities crisis of 2020, or the Spanish Debt Default Crisis of 2021?
In other words, was the 2008 crisis just one particular example of a class of periodically recurring events?
One of the striking things about financial panics and crises is that they have been going on approximately as long as financial markets themselves, which means if your explanation is something unique to 2008 and not something that existed in the Dutch Tulip Bubble, the South Sea Bubble, the crash of 1929, etc., you are implicitly saying that the current crisis is sui generis – it bears no ultimate resemblance to these other crises. And the individual circumstances of each of these varied greatly. There was very little governmental meddling in financial markets, for instance, in Holland in the 1600s. There was very little securitization in New York in the 1920s.
Rather, it seems an entirely reasonable position to take that periodic bubbles, crashes, and panics are simply an emergent property of financial markets.
Now, the phrase “emergent property” is something of a weasel word expression, a way to say approximately that “we’re pretty sure this relationship holds, but we don’t really know why”, and to sound sophisticated while doing it.
But (other than the pretentiousness of highfalutin but hand-wavy phrases) sometimes that is indeed the appropriate attitude to take. And there is ample evidence suggesting that, in this case, it seems to be true.
If you’re disinclined to take my word for it, let me discuss two papers, both co-authored by famous economists, that are fairly easy to follow, and in my opinion, grossly under-studied in asset pricing.
The first of these is Smith, Suchanek and Williams (1988) – “Bubbles, crashes, and endogenous expectations in experimental spot asset markets”.
Asset pricing in real life is notoriously difficult, because the financial world is such a messy and complicated place. To be able to say what a security is truly worth requires an ability to forecast distributions of cash flows far into the future, to know how discount rates should be calculated given risk exposures and the price of risk, to understand the role of the macro economy, and many other complicated things. Which means that it’s usually very hard to say for sure exactly what an asset should be worth, and hence whether it’s correctly priced or not.
Vernon Smith, who went on to win the Nobel Prize in Economics for his work, took an interesting approach to this problem. He said, effectively, let’s strip away everything that makes asset pricing hard, and create experimental asset markets in a laboratory. Instead of stocks lasting forever, let’s make them only last a set number of periods, say 20 rounds. Instead of dividends depending on the complicated macroeconomy, let’s make them depend on a dice roll. So in this case, there’s no “news” – if the die landed on a 5 instead of 3 this round and dividends were higher as a result, it shouldn’t change what you expect the stock to be worth next period, because the next roll is still the same in expectation. And instead of the massive uncertainty and information asymmetry that makes it hard for investors to know what’s going on, or what other traders know, let’s make everything in the above setup common knowledge, so everyone knows what’s going on, and everyone knows that everyone knows what’s going on.
This is, in other words, an environment designed to explicitly exclude every single one of the explanations above that might cause financial instability and financial crises.
So what does standard academic finance predict should happen?
Finance, of course, is a very broad field, so it’s risky to make claims about what the entire discipline predicts. But if we think of traditional asset pricing models like the CAPM or something similar, there are two main predictions.
First, the price should equal discounted cash flows. It is straightforward to calculate the expectation of payoffs from however many dice rolls are still to go, so expected cash flows are easy to figure out. As for discount rates, there is a strong argument that these should be zero. In particular, for an experiment carried out over 20 minutes, the time value of money is small enough to ignore. Meanwhile, the experiment is carried out for a small amount of money relative to participants’ wealth, so they should be relatively risk-neutral. Perhaps more importantly, to the extent that the die imposes risk, it is a risk that is uncorrelated with everything else in the world, because it is completely random. As a result, the price of the stock at each point should equal its expected cash flows.*
Second, the No-Trade Theorem of Milgrom and Stokey (1982) should hold, and the trading volume in the market should be zero. Put simply, the market lacks any of the reasons for participants to trade in the first place. They don’t have idiosyncratic exposures that they need to hedge, like labor income or asset endowments. They don’t have information asymmetry, which would rationally make them think they had an advantage, or be at a disadvantage. They don’t have liquidity needs, to provide for a child’s college fund. As a result, we all know that the fair price is $5, none of us is willing to sell for a penny less than that, none of us is willing to buy for a penny more than that, nobody should expect to make any profit by trading, so trade shouldn’t take place.
So, without any of the complications of the real world, do these predictions hold?
Not even vaguely.
What you instead find is that there usually are bubbles and crashes as the value of the stock initially rises considerably above its fundamental value and only declines slowly somewhere towards the end of the experiment. Volume is high as well, though exactly what “high” means when the benchmark is zero is open to debate – volume is very reliably non-zero. Asset volatility is also very high due to all of the above, despite the complete absence of any economic news in this market.
In other words, the whole thing looks just like regular equity markets.
And in the midst of the replication crisis in psychology, this experiment has been reproduced literally hundreds if not thousands of times.
The responses to this have been approximately twofold.
Behavioral finance types, eager to play the role of the maligned but vindicated prophets, reply “Aha! The whole edifice of finance is broken! We must begin from scratch with psychology,” which they then proceed to do by designing hundreds of variants of the experiment to find out what makes the bubble bigger and smaller to hope to understand mechanism.
I am a big fan of psychology when done well, don’t get me wrong.
But there is one fact that I think is hard to avoid.
30 years later, we still don’t really know exactly what problem the people in the lab think they’re solving, or how they’re doing it.
Whatever they’re doing, it’s not the standard finance problem that we’re currently writing down. And you can say that they’re just morons, but this is deeply unsatisfactory. “Crazy” is not a hypothesis, and “stupid” is not much better. “Crazy” is hearing voices in your head. “Stupid” is looking at a calculus problem and turning in a blank page on a test because you have no idea what the answer is. Neither of these remotely describes the reliable, replicable effects we’re seeing. You have to explain in what way people are biased, or how they’re misunderstanding the nature of the problem, and how this produces the result in question.
The efficient markets types tend to have two rejoinders.
Those who actually know the experimental literature reply that if you run the experiment multiple times, people do eventually figure it out, and things tend to finally settle down to something that looks vaguely like asset pricing.
To which I reply, “Great! Our closed form solutions are in fact a specific outcome of a generalized learning process that we don’t fully understand. We should try to model that.”
Because remember, this is the simplest possible way to design a version of the problem (other than perhaps asking people “how much would you pay for a $5 bill?”), and it still takes them a while to get it. In the vastly more complicated real world, there are far more ways to screw up the problem, and it’s possible to remain confused or mistaken for much longer.
The second rejoinder, which is usually said with a sneer, is that these are just stupid college students who don’t know jack. Financial markets select for competence and understanding because the stakes are so much higher. The smart get richer, the dumb get poorer, and markets eventually correct themselves. This is the “market selection” argument of Milton Friedman in his book “Essays on Positive Economics”.
Even if the critique is sometimes said dismissively, it is a serious one, and should not be brushed off lightly. There are two ways one could respond to this. One could point to the large numbers of experimental studies done on financial professionals that seem to replicate many of the biases of college students. But that ends up mostly leading to a game of whack-a-mole over the remaining differences between the setups, the incentives, and the participants.
It’s far more convincing to go to the source. We want to find evidence from real-world financial markets that looks like the points we were making above. Moreover, we want evidence that’s highly difficult to explain under standard models.
So let’s instead consider what I consider to be perhaps the most under-appreciated paper in all of asset pricing, even more so than the Smith et al. experimental asset market work. It’s Cutler, Poterba and Summers (1989) – “What Moves Stock Prices?” (gated published version here, ungated working paper with the main results here).
In the wake of the 1987 stock market crash, the authors did something very simple, but very insightful. They went back through history and looked at the largest one-day movements (positive and negative) on the New York Stock Exchange. And then they asked the question “What happened in the world on that day to justify these price movements?” How do you answer this question? Well, you read the paper that day and see what they said.
And the answer, most of the time, is “Nothing of any major importance.”
My favorite example of this was on September 3rd, 1946, when a 6.73% one-day decrease was discussed in the New York Times with the sentiment that there was, and I quote word-for-word, “…no basic reason for the assault on prices.”
This tells you, in other words, that October 19th, 1987, is the rule, not the exception.
But to ram the point home further, the authors then went back and examined stock market movements on days of large geopolitical and economic significance. Pearl Harbor. The Cuban Missile Crisis. The Assassination of JFK. How much did these important things move the price?
Answer: Not as much as you’d think. In particular, the markets dropped -4.73%, -2.67% and -2.81% respectively.
To recap – Cuban Missile Crisis: -2.67%.
“No basic reason for the assault on prices”: -6.73%.
Which is to say, it’s not that there are some movements that occur on days without any major news.
It’s not that there are some large movements that occur on days without any major news.
No, it’s that most of the largest movements occurred on days without any major news.
How many models explain that fact? How often do you hear this being discussed by academics?
Not very often. Like many inconvenient facts, it got discussed for a while, then mostly ignored.
And this is about as far away from satisfying stories of human agency as its possible to get. Large, value-destroying movements of financial markets happen without any real underlying economic reason. They happen in ways that leave the participants themselves confused as to what’s going on. Whatever the cause is, it doesn’t look like a simple morality play, and it’s not clear that concepts like “blame” are an especially useful way of improving our understanding of what’s at play.
Because the sad truth is that despite a few models in the follow-up to the Cutler, Poterba and Summers (1989) paper, we haven’t really advanced a great deal since then in our understanding of why large market movements happen without any real economic news. And this isn’t something that just causes headaches for standard economics models either. Even psychology doesn’t get us that far. If people are biased and reacting incorrectly, what are they reacting incorrectly to? What bias causes them all to change their mind at once at times that are hard to predict?
Neither Cutler, Poterba and Summers (1989) nor Smith at al (1988) are considered as stylized facts that new models have to be able to explain before they’re able to be taken seriously.
Which to me is a great shame, because these events are enormously useful ways to distinguish between models.
There are hundreds of models that can potentially explain July, August or September 1987.
There are far fewer that can also explain October 1987.
It’s for this reason that I used the highly unsatisfactory description “emergent property”. If the biggest movements happen without any real news, it seems to imply that there is something inherently unstable about people’s expectations or trading strategies that cause them to all suddenly change their minds at once, as if everyone is watching everyone else and trying to decide when to jump for the exits. Keynes described this aspect quite well when he characterized the stock market as being like a beauty contest:
[P]rofessional investment may be likened to those newspaper contests in which the competitors have to pick out the six prettiest faces from a hundred photographs, the prize being awarded to the competitor whose choice most nearly corresponds to the average preferences of the competitors as a whole; so that each competitor has to pick, not those faces which he himself finds prettiest, but those which he thinks likeliest to catch the fancy of the other competitors, all of whom are looking at the problem from the same point of view.
It is not a case of choosing those which, to the best of one’s judgement, are the really the prettiest, nor even those which average opinion genuinely thinks the prettiest. We have reached the third degree where we devote our intelligences to anticipating what average opinion expects average opinion to be. And there are some, I believe, who practice the fourth, fifth and higher degrees.
You may well have gotten this far in the hopes that the end of this essay will give you the final, satisfying answer of what’s going on. Alas, I don’t have the full answer. If I did, I’d tell you (or the Economics Nobel Prize committee).
If you want a hint at what a model might look like that would get you at least part of the way there, try here.
But there are a few things we can learn.
First, you ought to read every article about financial catastrophe as if it began: “Financial markets throughout history are known to be subject to periodic instability, regardless of size, time period or setting. This happens in ways that seem unrelated to obvious underlying economic events, and the cause is still highly unclear.”
This will put you in the appropriately caveated frame of mind when you read the latest story of outrage about how someone or other is creating chaos in financial markets.
It’s not that there aren’t considerable parts of finance where human agency is an important, if not the most important, way of understanding events. It’s just that there are non-trivially large and important areas where that doesn’t seem to be the case.
Second, if you’re trying to develop your own model, my hunch (and it is nothing more than a hunch, I admit) is the following: if your model of the world can’t explain Smith et al. and Cutler et al., it’s probably missing something important.
It’s no shame if your model can’t explain it, frankly. I know of very few that can. But assuredly, experimental asset bubbles are not caused by (to pick a few examples at random) greedy bankers, fiat currency, or the government.
Third, you should assume that regardless of whatever government policy is enacted today, periodic instability will continue to recur. The reason is obvious – I’m quite certain the government doesn’t know why periodic financial instability occurs, and so any legislation that it passes which eliminates financial panics will probably do so largely by accident.
You can, however, take precautions assuming that these panics and crises will continue to occur. Chief among these is “take on less risk, and especially less leverage, than you think you can get away with”. Another related one is to hold risky assets in the proportion you’re comfortable with if they in fact melt down from time to time. This has somewhat of the flavor of Nicholas Taleb’s argument in “Fooled by Randomness” and “The Black Swan”, but the underlying reasoning is different. In Taleb’s view, crashes represent periodic “Black Swan” risk that we can’t really hope to understand. But this doesn’t explain why bubbles and crashes are quite predictable and unsurprising events in experimental markets. You may think that real-world crashes will always be fundamentally unpredictable and insoluble, but the same can hardly be said about the experiments. I personally am hesitant to assert the inherent inscrutability of the former phenomenon until I understand the latter phenomenon more.
And finally, beware of responding too strongly to the unsatisfactory feeling that an explanation is false because it’s morally unsatisfying, or feeling that having a potentially wrong explanation is always better than no explanation at all.
That’s exactly the hunter gatherer instinct.
In the short run, it may feel better to invent stories that the gods have maliciously caused the flood that has washed away your retirement savings. But doing this gets you further away from understanding weather patterns, not closer.
Be wary of making the same mistake about finance.
*If you’re a strong form experimentalist, you might object that participants in the Smith experiment may still be risk averse, because they tend to turn down positive expected value gambles in the lab. But recall, this should make the value of the stock less than its expected cash flows, when in fact it usually trades at more than its value early in the experiment.