You should not read the news.

Earlier this year Goetzmann, Kim, and Shiller published “Crash Beliefs From Investor Surveys”.

For over 25 years, Shiller has been surveying people about (among other things) their belief that there will be a 1929 style crash within the next 6 months.

The chart above shows the results. Shiller surveys both individuals (the red line) and institutions (the blue line).

You can see that both lines tend to hover around the 15% mark. That means that at any given point in time, people think there is at least a one-in-eight chance of an October 1929, Great Depression style stock market crash happening in the next six months.

That’s just totally, amazingly crazy!

Goetzmann et al point out that the actual historical record shows that the probability is 1.7%. So people overestimate the chance of a catastrophe by an order of magnitude.

We find evidence that the financial press mediates investor crash beliefs asymmetrically.

A high-profile example of this is Marc Faber, who is the subject of frequent interviews & press articles. Just a few days ago he was in the press saying “S&P is set to crash 50%”. Now, before you take him seriously, his track record worthless.

So we dramatically overestimate the chances of a financial catastrophe. This is part of what drove Warren Buffet earlier this year to give his “don’t bet against America” comments in his annual letter to Berkshire Hathaway shareholders:

For 240 years it’s been a terrible mistake to bet against America, and now is no time to start. America’s golden goose of commerce and innovation will continue to lay more and larger eggs. America’s social security promises will be honored and perhaps made more generous. And, yes, America’s kids will live far better than their parents did.

But what is it that causes us to overestimate the chances of catastrophe so badly? Goetzmann et al’s real objective in their paper is to try to answer that.

They spend most of their paper analysing how the media reports market movements: how often words like “crash” or “boom” get used; whether they appear on the front page or later on; how many days after the market event the media is still talking about them; and so on. It is pretty detailed and interesting but the results are not exactly shocking: the media prefers to report on negative news and we remember negative news more than positive news.

But they add in an interesting twist: catastrophes are rare. People are bad at dealing with rare probabilities. If some other “rare event” has happened recently, does it cause people to overestimate the probability of other rare events?

The answer seems to be: Yes.

We find that recent earthquakes in the immediate vicinity of the respondent are associated with a higher probability assessment of a “catastrophic” stock market crash.

If there was an earthquake recently, then people think there is a greater chance of a stock market collapse.

Obviously the two have nothing to do with one another. Earthquakes are regional and extremely unlikely to affect national and global markets. But people appear to use them as a kind of proxy. “I thought earthquakes were rare but one just happened. So maybe other rare events could also happen sometime soon.”

It looks crazy when I write it out like that but that’s exactly how human brains work.


While there’s not much you can do about earthquakes there is something you can do about the financial press: Avoid it.

absent the financial press calling attention to the potential for a crash conditional on a market decline, investors would ignore it.
One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.