Did the Banks Go Crazy?

Whatever economists might think, rationality and efficiency don’t always go together.

The great financial crisis of 2008 has provoked an extraordinary round of soul searching among economists. The reasons for this are not difficult to find. Not only did most members of the profession fail to predict the impending catastrophe, but many actively contributed to it, by aggressively rationalizing the very practices and institutional arrangements that gave rise to the collapse of the U.S. investment banking system.

In the background was the assumption, widely shared among economists, that contractual arrangements entered into by private parties were efficient until proven otherwise—so that if a particular financial arrangement (say, a bank making a huge loan to a person with no money or job) looked like it was crazy, that was only because it had not been well enough understood. Look more carefully, and you can find the hidden rationale, the secret genius of the market at work.

Thus buckets of ink (or terabytes of keystrokes) were wasted, essentially intellectualizing the work of financiers and bankers (much the way academics in the humanities intellectualize the work of artists and writers). The implosion on Wall Street was therefore a source of considerable embarrassment. Imagine an art historian, invited to offer impressions of a long-lost work by Duchamp. After waxing poetically for several minutes about its “transgressive” and “post-auratic” qualities, the historian is informed that a mistake has been made, and that the work in question is actually just a urinal. This is basically the situation that many academic economists found themselves in last fall.

The repercussions have been swiftly felt. While the public debate has been dominated by pointing fingers at a bewildering range of suspects, the debate in the economics profession has become quite narrowly focused on two rival theories.

Until recently, the dominant view of markets was that they were rational and efficient. These two virtues had always seemed to go together, like faith and hope. And yet in the fall of 2008 it became clear that financial markets had failed to perform efficiently. Two obvious explanations presented themselves. Some people, among them former U.S. Federal Reserve chair Alan Greenspan, suggested that banks had simply gone crazy. According to this view, rationality and efficiency are still an inseparable duo; it is just that banks and investors failed to act rationally.

The other possibility is that no one went crazy, but that rationality and efficiency do not actually go together quite as neatly as many had assumed. Perhaps individually rational action, even in reasonably well-structured markets, is able to produce outcomes that are collectively self-defeating.

The first theory is the easier sell. It is the theory that everyone wants to believe. Unfortunately, it is the second theory that is correct.


Why does anyone start a small business? According to Statistics Canada, one half of all new small businesses fail before the end of their third year, and only one in five survives a decade or more. In other words, starting a small business and thinking you are going to make money is far less plausible than getting married and thinking you will not get divorced. So why does anyone do it?

Self-deception is one answer. In the case of marriage, it is a fairly persuasive explanation. You are in love. You’re not thinking straight. Emotions are clouding your judgement. But what about small businesses? Perhaps entrepreneurs are just incurable romantics. Yet isn’t it strange to think that our economic system might depend upon this sort of irrationality?

John Maynard Keynes was not one to shrink from this conclusion. He argued that an essential ingredient of entrepreneurship was what he called the “animal spirits” of investors. The typical business plan contains about as much wishful thinking as the prospectus for “an expedition to the South Pole,” he said. “Individual initiative will only be adequate when reasonable calculation is supplemented and supported by animal spirits, so that the thought of ultimate loss which often overtakes pioneers … is put aside as a healthy man puts aside the expectation of death.”

The phenomenon that Keynes put his finger on is by now quite well documented. It is referred to as optimism bias. People tend to paint a rosy picture of things. To say that we all suffer from this tendency would be misleading, simply because “suffering from it” may well be an essential component of psychological well-being.

According to Keynes, this makes markets vulnerable to mood swings. Much of everyday economic activity depends upon our willingness to play along with the illusion that everything will work out well in the end. When something happens that challenges this assumption, it can have an arresting effect upon the economic system as a whole.

One can see this moodiness on display in Garth Turner’s recent book, After the Crash: How to Guard Your Money in These Turbulent Times, where the one-time investment guru now recommends not just withdrawal from the market, but from most of western civilization. In this delightfully nutty jeremiad, Turner walks the reader step by step through the process of withdrawing all money from the bank, getting the house off the grid, stockpiling toilet paper and otherwise preparing for the collapse of civilization. Animal spirits indeed. The book could just as well have been called “How to Make Things Worse.”

Of course, the reference to animal spirits was something of a throwaway line for Keynes, not central to his diagnosis of the crisis tendencies of capitalism. But it has since become a catchphrase for a movement in contemporary economics that seeks to dislodge rationality from its traditional place of honour, in favour of a more accurate picture of human psychology with all of its foibles and flaws. For partisans of this new “behavioural” approach to economics, such as George Akerlof and Robert Shiller, authors of Animal Spirits: How Human Psychology Drives the Economy, and Why It Matters for Global Capitalism, the unexpectedness of the financial crisis reveals a fundamental methodological flaw in the way that economists have approached their subject. Far from being perfectly rational, people are subject to “changing confidence, temptations, envy, resentments and illusions.” What triggered the market collapse? Essentially a mood swing.


Economists never really believed that people are perfectly rational. After all, they get up in the morning and go to work in the same world as you and I do. What they believed—and what many continue to believe—is that irrationality can safely be ignored when it comes to making predictions about the behaviour of the economy. Why?

Historically, many have been comforted by the thought that irrationality would produce mainly noise, or random deviation from the dominant tendency. If this were true, then irrationality could safely be disregarded whenever one was aggregating across a large enough number of people. Consider an analogy: when voting, some people no doubt make a mistake and tick off the wrong box on their ballot. But is this worth worrying about? Should we revise our assumption that when people vote, they are expressing their true preferences? Of course not, because not only are these mistakes likely to be uncommon, but they are also likely to cancel each other out.

When it comes to the economy, the same thing applies. Sure, there are some bozos out there. But bozos, it was claimed, precisely due to the bozoness of their choices, tend to cancel each other out.

Furthermore, you cannot count on people making mistakes. Think about playing chess. Naturally, your opponent will sometimes make mistakes. But you can’t build a strategy on that. You can’t expect the person to overlook an obvious opportunity, despite the fact that people often do precisely that. When you are playing chess you have to treat your opponent as rational and not overly prone to error.

The same thing goes for the economy. Money, like chess, tends to focus the mind. People do not like losing it. So when you’re entering into an adversarial market relationship—where your gain is someone else’s loss—you should treat it like a chess game. You should assume the best about people. This in turn will make market behaviour as a whole more rational.

Of course the idea that deviations from the ideal of rationality would be random has taken something of a beating over the past few decades. What studies of optimism bias and other such psychological quirks have revealed is a set of extremely common errors that are not random, but typically push in the same direction. People get far more upset about losses than about foregone gains. People discount the future in a way that assigns exaggerated significance to the near term. People underestimate the probability of boring events, and overestimate the probability of exciting ones. And so on.

This suggests that we cannot afford to ignore irrationality.

But here is the killer counter-argument, and the lynchpin of the much-derided efficient markets hypothesis. If people exhibit systematic biases in their reasoning, then they will be predictably irrational. If their irrationality is predictable, then it will be easy to make money off them. Irrationality, as they say on Wall Street, creates arbitrage opportunities.

For example, Lisa Kramer, writing in The Finance Crisis and Rescue: What Went Wrong? Why? What Lessons Can Be Learned? (an instabook put together by professors at the University of Toronto’s Rotman School of Management), argues that the 2008 crisis exposes the need for a new, behavioural approach to understanding finance. She offers as an example the fact that investors suffer from seasonal affective disorder, and so tend to ditch high-risk stocks as the number of daylight hours declines. But if this were true, the obvious recommendation would be to buy these stocks in the late fall or winter, not the spring or summer. Yet if people started doing this, the trend that Kramer claims to observe would disappear.

This observation points to a fundamental asymmetry between rationality and irrationality. The fact that some people are rational does not create a buying opportunity for the irrational, whereas the fact that some people are irrational does create such opportunities. Thus irrationality summons up countervailing forces that tend to press for its elimination. This countervailing force will have two effects—first, it will have a tendency to adjust prices to their “correct” level, and, second, it will drive irrational players out of the market.

Thus the idea that economic actors will behave rationally is not just a fantasy dreamed up by ivory tower intellectuals, or a methodological bias imposed by the desire of economists to use fancy mathematics. Even if we are irrational, the discovery that we are will tend to correct the situation. The only stable outcome is one in which everyone is rational.


Economists are not mistaken in thinking that when we take any particular instance of seemingly irrational behaviour, we can often find a rational explanation for it. To take a paradigm instance on financial markets, consider the case of a bank run. This occurs when a large number of depositors try to withdraw their money from a bank simultaneously. Of course, this is lemming-like behaviour, since the attempt at simultaneous withdrawal guarantees the failure of the bank and therefore makes withdrawal impossible. (One may recall the famous run-on-the-bank scene in Mary Poppins.)

One way of understanding banks is to think of them as firms whose core business model consists of borrowing short and lending long. Richard Posner, in A Failure of Capitalism: The Crisis of ’08 and the Descent into Depression, provides a lucid exposition of the basic structure. Banks make their money from the interest rate spread, since borrowers are willing to pay more for long-term than for short-term loans. How is this a sustainable business? Because of the law of large numbers. Banks borrow short term from many, many people (the depositors) and make long-term loans to a smaller number of people (e.g., businesses, homeowners). While every day some depositors will need to be paid back, the number is likely to be small, and in any case is likely to be offset by new deposits.

Thus the greatest enemy of banking is the enemy of all risk-pooling arrangements—correlation. If withdrawals are independent of one another, then the bank is stable. But if one person’s withdrawal prompts someone else to withdraw, which prompts someone else to withdraw, then the two events are no longer independent. They are correlated. Once that happens, the fundamental business model of a bank simply does not work.

This is why, back in the salad days of laissez-faire capitalism, the stability of banks was hostage to public sentiment. Rumours were able to generate extraordinary displays of crowd behaviour. The mere whiff of trouble at a bank could bring a stampede of depositors, clamouring for their money back. Furthermore, bank runs often gave rise to bank panics, when depositors, knowing that some banks were failing but not knowing which ones, tried to withdraw their money from all banks simultaneously. Compared to bank runs, bank panics were positively bacchanalian.

Akerlof and Shiller duly present bank runs as a paradigm instance of irrational mass behaviour. Indeed, it seems like an easy target. But it does not serve very well as an illustration of the madness of crowds. On the contrary, what is striking about the actions of depositors in the case of a bank run is that their actions are perfectly rational. While it may be the case that simultaneous withdrawal of deposits has the effect of destroying the bank, the fact remains that those who get there first do succeed in redeeming their deposits. So if your bank is going to fail, the best thing to do is to run down there and get your money out as soon as possible, devil take the hindmost.

In other words, there is a purely rational explanation for bank runs; one need not appeal to animal spirits. Two things make bank runs seem irrational. First, because what each person does depends upon what that person thinks that everyone else is going to do, the interaction is subject to very dramatic tipping-point effects. Second, once the run gets started it is collectively self-defeating, or inefficient. The best course of action for each individual generates an outcome that is worse for everyone.

But neither of these two features adds up to genuine irrationality. Indeed, if bank runs were irrational it is not obvious what one could do to fix them. And yet the introduction of deposit insurance during the New Deal era all but eliminated bank runs in the conventional banking sector. It did so by changing the incentives faced by depositors (so that there was no particular disadvantage associated with being among the hindmost).

One can see here the problem with the animal spirits view. It caters to popular prejudice in an unhelpful way, by distracting attention away from the structure of incentives that individuals face. It suggests that our problems are a consequence of human nature, rather than of human institutions.


The animal spirits theory of the financial crisis is, so far, the one that has made the greatest headway. It appeals to the outrage of the public and the desire to condemn the eggheads who got us into this mess. But it also satisfies the desire of the eggheads to redirect blame. One need only consider Greenspan’s remarks to the House Committee on Oversight and Government Reform this past October to see the self-exculpatory dimension of the view. His only mistake, he said, was trusting the banks to advance their own interests.

This all sounds plausible, until one stops to wonder how the Federal Reserve could possibly operate without that assumption. Indeed, if the problem is that banks periodically go crazy, how could the situation ever be corrected? What would a regulatory solution look like? We design roads based on the assumption that drivers are trying not to run into one another. Similarly, we design the system of financial regulations based on the assumption that firms want to stay solvent. It is not clear how we could approach either design question in the absence of these assumptions.

The model of the bank run—individually rational, collectively inefficient—provides a much better set of tools for thinking about the current crisis. At the heart of the subprime mortgage debacle were two significant innovations, both of which were entirely non-crazy.

The first was the securitization of mortgage loans. Banks are by nature conservative when it comes to making loans. They have to be, not only because of their underlying business model, but because they lend on fixed terms. Thus the upside of mortgage loans is not all that great—you get repaid your money with interest, at a fairly modest rate, and you pocket the difference between that and what you are paying your depositors. The downside, however, is enormous. Not only may you never see any of the interest you are owed, but you can also lose a substantial portion of the principal you lent out.

Equity lenders, on the other hand, such as shareholders, have a much greater upside to their investments. If the value of the asset goes up, they get a share of the profits. This is why investors in the stock market are willing to take much bigger risks than traditional banks. Unfortunately, for the average homeowner, going to the stock market to raise money to purchase is not possible.

This suggests that there is a group of people out there who want loans, and another group of people who would be willing to make them those loans. Banks are in a position to bring these two groups together, by making the loans, then bundling them up and selling the right to collect the payments to investors. This is the essence of securitization, and it is an idea that, in the case of mortgage loans, makes a certain amount of sense.

The second non-crazy innovation involved the creation of the now infamous collateralized debt obligations. Once you start bundling up a whole bunch of mortgage loans and reselling them to investors, a question arises about how defaults should be handled. When a single bank holds a mortgage, any failure to repay simply becomes a loss for the bank. But when the mortgage is held by a group of investors, the loss can be divided up in all sorts of different ways. The brilliant idea at the heart of CDOs was to divide up the investors into different “tranches,” and to have all the losses absorbed by the most “junior” tranches first.

Thus buying into a senior tranche of a CDO typically gave the investor not merely the right to collect payments from a bunch of different loans, but also the right to collect payments from the best of these loans. This is why lenders began to relax their standards. When you take one single individual with perhaps a patchy credit history and unstable employment, the risk of default on a loan is quite high. But suppose you could take one hundred such individuals, or a thousand. While some will no doubt have trouble making their payments, the chances that more than a quarter will default are actually quite low.

Thus a lot of the behaviour that on the surface seemed quite crazy did have an underlying rationale. (One can find an admirably clear explanation of the basic mechanics in John Hull’s contribution to The Finance Crisis and Rescue.) This is not to say that it was wise, but merely to say that it was not crazy.

Of course, there was an Achilles heel to all this, and it was the same Achilles heel that threatens the integrity of banks—correlation. If the chance of one person defaulting on a home loan is substantially independent of the chance of others defaulting, then the law of large numbers will insulate those at the senior levels of a CDO from all losses. But if one default starts to cause more defaults, then the whole model stops working, which is exactly what happened a year ago.

The problem—and this is a problem quite specific to the U.S. housing market—is that most mortgages in the United States are so-called non-recourse loans. This means that if you default on your mortgage, the bank can seize your house, but it cannot seize any of your other assets. As Hull puckishly observes, this means that banks, in making zero-downpayment loans to homeowners, were essentially selling them a put option on their properties. (Selling someone a put consists in agreeing to buy some asset at a predetermined price in the future—in this case, the price was the market value of the house at the time of purchase.)

When house prices began to decline in the U.S., homeowners started exercising this put option (by defaulting on their mortgages, thereby “selling” the house back to the bank at the price they paid for it). As a result, the type of insurance offered by CDOs against credit default became either sketchy or worthless. Banks and investors, not knowing how much exposure anyone had to this now toxic debt, stopped lending to each other. Catastrophe ensued.

Again, the same features that one finds in an old-fashioned bank panic can be seen at work here: a dramatic tipping-point effect that creates volatility, along with a vicious cycle of collectively self-defeating behaviour. But this does not mean that any of the individuals involved acted irrationally, or that global finance is nothing but a castle built out of thin air. It just means that the market for credit default risk failed, for a variety of prosaic and fairly well-understood reasons.


Why then did so many economists miss all the warning signs? The problem is not that they assumed people were rational. The problem stemmed from what Harvard law professor Robert C. Clark once described as “facile optimism about the optimality of existing institutions.”

“Thinking like an economist” is an ideal that harbours significant ambiguity. One way of doing so is to assume that everyone is egoistic and bloody-minded. Think of this as the Machiavellian strain. Another way is to assume that all agreements entered into by private parties are mutually beneficial, and that all outcomes produced by free markets are efficient. Think of this as the Panglossian strain.

All the rhetoric over the years about the invisible hand of the market has persuaded many people that there is no great tension between these strains. This is what led to mistakes. In the context of a properly structured capitalist economy, egoism and bloody-minded behaviour are capable of generating mutual benefit. But when that structure is less than perfect, the two have the capacity to come apart in spectacular ways.

Consider the earlier dust-up over stock options and executive compensation. A person of Machiavellian temperament looking at this would be inclined toward the view that CEOs were just ripping off shareholders, taking advantage of information asymmetries and collective action problems that resulted in weak oversight. Since this analysis happens to coincide with the verdict of common sense, many people would be surprised to discover how strongly it was resisted by academics in economics, finance and corporate law.

What the apologists for CEO pay—and they are legion—proposed instead were all sorts of baroque schemes through which these compensation arrangements could be seen as in the best interests of all. Corruption, for instance, was interpreted as merely an implicit form of executive compensation, and therefore part of the “tacit” employment contract agreed to by shareholders.

Why would anyone think this? Because everything is for the best in the best of all possible worlds.

Similarly, when more cautious souls began to sound the alarm about the run-up in housing prices in the United States, or the complexity and opacity of the derivatives that were being traded, the Panglossians responded in force. These bankers are clever people, they said, and they know what they are doing.

This turned out to be true. Most of them were clever, and they did know what they were doing. Unfortunately, there was never any reason to think that the sum of all this individual cleverness would be stable in the long term, but much less beneficial for society as a whole.

Books Mentioned:

Animal Spirits: How Human
Psychology Drives the
Economy, and Why It Matters
for Global Capitalism
George Akerlof and Robert J. Shiller

Princeton University Press

A Failure of Capitalism:
The Crisis of ‘08 and the
Descent into Depression
Richard A. Posner

Harvard University Press

The Finance Crisis and
Rescue: What Went Wrong?
Why? What Lessons Can Be
Rotman School of Management

University of Toronto Press

After the Crash: How to
Guard Your Money in These
Turbulent Times
Garth Turner

Key Porter