Arnold Kling  

Lectures on Macroeconomics, No. 12

Another Perspective on LTCM... Techno-escapism...

This lecture speculates on some possible behavioral economics of booms and recessions. The idea is that herd behavior (buying at the top, selling at the bottom) is a form of procrastination.

As I write this, on January 1st, 2009, the U.S. economy is in recession. There is evidence that consumers have cut back on their purchases of automobiles and other durable goods. They have cut back on their purchases of jewelry and other discretionary purchases. As a result, prices on these goods have been marked down.

One might think that it would be rational for consumers to shift some of their purchases to times like this. That is, it would be rational at the margin to buy more during a recession and to buy less during a boom. Yet we know that the opposite is what happens. It appears that:

--consumers cut back especially on durable goods purchases during a recession, when such goods are cheap

--the typical investor buys more stock near the peak than near the trough

--the typical firm hires more labor near the peak of a cycle, when labor is expensive, than near the trough of the cycle, when labor is cheap

Let us call this phenomenon herding, because people behave as a herd in a way that is contrary to their interests as individuals. A rational economic agent tries to buy low and sell high. The herd does it the other way around.

There are two reasons for economists to be interested in herding. One reason is that it poses a puzzle about behavior, since herding seems inconsistent with rational decision-making. Second, herding accentuates business cycles. If instead people behaved more countercyclically, the cycles would be dampened.

As a specific example, consider the real estate boom and bust. In 2005 and 2006, fewer people should have been trying to buy real estate and more people should have been trying to sell. Today, we seem to have the opposite, with many people foregoing opportunities to buy and many people anxious to sell, even though prices are low.

As of January 1st, 2009, is it a good time to buy real estate? To most people, the answer is, "No. Wait until prices decline further." This reasoning may prove correct, in the sense that prices may very well decline further. However, I predict that the average person who reasons this way will end up buying real estate at prices that are considerably higher than they are today. Instead of buying at the bottom, they will miss the bottom and instead buy at some point well into the next upswing.

People confuse perfect market timing with feasible market timing. When prices are high, they think to themselves, "If I get out now, I may miss out on more profits. I'll keep putting money in." Implicitly, they are assuming that they will be able to time the market better by waiting. They are afraid that if they sell now, they will have failed to time the market perfectly. They do not take account of the fact that perfect market timing is not really feasible.

Similarly, when prices are low, they think to themselves, "If I get in now, prices may fall further. I'll wait until they go lower." Again, they are acting as if they can time the market perfectly.

In a famous lecture, Procrastination and Obedience, George Akerlof argued that a lot of irrational behavior could be explained by procrastination. You don't quit smoking today, because you know you can quit smoking tomorrow. What you fail to account for is the fact that tomorrow you will once again think that postponing quitting smoking by another day is a good idea.

I think that herding behavior takes place for the same reason. People realize that prices are too high, but they put off selling until tomorrow. Or, people realize that prices are too low, but they put off buying until tomorrow.

The problem is that nobody can predict when the herd will turn, and when it does it turns so quickly that the procrastinators lose. For example, a retailer's executives may worry that they have expanded too quickly, but they put off closing marginal stores until too late. Instead, the firm waits, the boom ends, and the firm finds itself selling excess inventory and stores in the midst of a recession, just as everyone else is trying to do the same thing.

At some point, the herd will once again buy U.S. stocks. When that happens, more people will be late than early.

What the foregoing assumes is a model in which asset prices fluctuate wildly relative to a "true mean." It suggests that if you as an investor have a reasonable estimate of the true mean and sufficient patience, you can do well. Invest more in the stock market when prices are below their long-run mean and invest less when prices are above their long-run mean. Ex post, this model always works. That is, if you take some model of average stock prices, estimate it on historical data, and then use that model to determine buy and sell decisions, you can go back and show that you would have done very well. The challenge is coming up with the right model of the true mean ex ante. Your model of the true mean that fits past data really well (think of Shiller's model that looks at the ratio of stock prices to the past ten years of earnings) could end up not being the right model in the future.

From a macroeconomic perspective, the key decisions that may be subject to herding are labor demand by firms and spending on durable goods by consumers. (Investment by firms could also be subject to herding.)

On the margin, it is smarter for a firm to hire during a recession, when good workers are readily available, than to hire during a boom, when the quality of available workers is lower. Similarly, during a boom, it would be smart to be reluctant to hire workers and eager to shed workers, because at the margin worker quality is likely to be low. However, what we get is herding behavior--firms tend to hire too much at the peak, and they all tend to lay workers off at the same time.

Similarly, consumers--even those who are safely employed--do not buy durable goods in a recession, when they are cheap. Instead, they procrastinate, and they make more of their purchases during a boom, when durables are expensive.

To the extent that procrastination and herding are important, there is a theoretical justification for countercyclical government policy. However, there is an even stronger justification for ordinary individuals to behave countercyclically.

Do not assume that just because government might behave countercyclically that it in fact will behave that way. It would not surprise me if it turns out that most of the spending in the soon-to-be-enacted stimulus package ends up taking place after the recession has bottomed out.

Previous lectures in this series:
11. Leijonhufvud and the corridor

10. Money or Credit?

Comments and Sharing

CATEGORIES: Macroeconomics

COMMENTS (7 to date)
MattYoung writes:

Let's not denigrate herding too much for it is the same phenomena that moves thousands to the factory each day. We use our herding instinct to get those thousands of job classifications Arnold referred to in his first lecture. Herding mammals rule the roost for a reason.

In mathematics one wants to look up the Jensen Inequality, estimation theory and Hayek. How do we estimate the general well being of society? With Hayek's minimum of transactions, so we group into sectors and use those as counting units, with the error of Jensen (equivalent to economies of scale).

If housing is going to count in national wealth, then housing must have momentum, that is herd movement, to reduce the uncertainty of the calculation. The housing industry must obey a high self-correlation.

That is until the herd breaks; then Jensen inequality no longer holds, the unions break up and reform.

So one can see that information shocks cause chaos because they provide alternate aggregations in which to count national wealth. In that case one gets a much better estimate of wealth than is provided by the current Jensen error bound.

The larger the shock, the greater the herd disorientation as Hayek's computing tree must reform and rebalance.

I would expect biologists to someday tell us that mammals have a specific error band for uncertainty about the herd security.

Alex J. writes:

There's a recession, durable goods are cheap and yet consumers buy less. Surely the simplest way to characterize this is that durable goods are especially cheap during a recession because consumers cut back especially upon them. In "Why I Am Not an Austrian Economist", Your colleague has an explanation for why this would be:

You buy durable goods to make up for depreciation and to adjust your total stock based on your prediction of future income. In a recession, your predicted income goes down.

often this means that there is no point even making up for depreciation, since natural wear-and-tear simply moves you closer to your new, lower total stock.

For companies, employees are like durable goods. First, they are costly to acquire, because of the searching and training. If you are pessimistic about the economy, you will be worried that the market clearing wage will go down in the near future. Then you will be stuck paying your recent hires too much, because layoffs and wage cuts are morale killers.

I suspect that the general error in a recession, and in bubbles, seems to be giving too much weight to (available, concrete, socially conforming) recent trends and examples and not enough to (distant, abstract, socially disdained) longer term trends or events.

On the other hand, the market can stay wrong longer than you can stay solvent.

Lee Kelly writes:
.One might think that it would be rational for consumers to shift some of their purchases to times like this. That is, it would be rational at the margin to buy more during a recession and to buy less during a boom.
People have seen their expected income decrease. They feel less wealthy and are adjusting their consumption today in anticipation of a more impoverished future. It is nonsense to suggest that they should be buying more as prices drop when wealth has dropped correspondingly.

If income, now and in the future, stays constant in real terms, and prices decrease, then we would expect an increase in demand. But if income, now and in the future, has plummeted in real terms, then an increase in demand may not occur. Moreover, there is nothing irrational about it.

People in the US have stopped saving money, in part because home prices were rising so much. Home equity was a substitute and provided a sense of security which savings used to (a safeguard against unexpected job loss, injury, etc.). Now home equity is evaporating people are scared and are demanding less consumer goods like cars and dishwashers.

It does not seem like 'herding' to me.

Grant writes:

Are we sure we're interpreting the statistics correctly? I think I'm with Lee Kelly on this.

People have seen both their expected incomes and savings drop significantly. They should rationally reduce their spending.

What happens if we look at the subset of the population who's expected incomes and savings have not dropped? Are they taking advantage of the low prices and consuming more? In my personal experience they are, but those people seem to be outnumbered by the sort who have been harmed by the recession.

The saying "be fearful when others are greedy, and greedy when others are fearful" has been around for a while. If it were that easy, I think we'd see more people doing it.

fundamentalist writes:

I don’t understand why academics are so quick to label people as irrational. Rationality differs when assumptions differ. If the only factors to consider in buying durable goods was price, and the only assumption regarding consumer behavior is getting the lowest price, then sure it’s possible to call consumers irrational. But doesn’t anyone see the artificiality involved here? Consumers have far more things to consider than just getting the lowest price on durable goods. They have to look at total wealth, uncertainty about future income changes, and debt levels. Only those consumers with cash or very high levels of confidence in the future of their jobs and good credit ratings can buy durables in a recession, but those are getting very good buys right now. Still, the others aren’t irrational because they consider variables other than just price.

As for the herd behavior in stock investing, it has rational aspects as well. It has been well documented that fund managers find career safety in the herd because if they’re wrong at the same time the majority is wrong, they get a pass from management. If they’re right when everyone else is wrong, they get little credit, are considered lucky and cause a lot of jealousy. But if they’re wrong when the crowd is right, they get their heads handed to them.

Individuals follow the crowd for similar reasons. They lack confidence in their own investment skills and so tend to follow the advice of those they respect, often the talking heads on TV. Being a contrarian also has social liabilities. Success when the crowd fails can make others think you’re just lucky and spur envy. Failure makes you look like a fool when the crowd succeeded. So there isn’t much social upside to being contrarian. It can be very lonely. All that says of people is that they don’t value making money on investments as much as they value good social relationships. How is that irrational? It would be better to say that people don’t value money as much as economists think they do.

John S. Molini (Simbaking3) writes:

I don't think I understand why this correlation has to be explained. The prices are low during a 'bust' precisely because no one is buying, and during a 'boom' prices are high precisely because many people are buying. This relationship is as clear as day in the definitions of 'boom' and 'bust.' So what about the relation isn't a 1.00 coefficient ceteras paribus.

Watching buying patterns and watching 'boom-bust' cycles means watching the same data. The larger question of what causes people to change their buying habits is of course complex, but the question of why they correlate with boom/bust cycles seems like a complete non-starter to me.

The 'herd' actors buy high because they expect the price to go higher, and end up wrong. Their decision to buy is in no process different from the 'rational' actors who buys low because they expect the price to go higher, and end up right. This isn't a question of rationality, and herd psychology, while valuable, has no place being applied to this particular problem.

Raphael writes:

Let's hope everything will be better this year..

Comments for this entry have been closed
Return to top