Scott Sumner  

Endless bubbles?

PRINT
The Worst They Can Do... Socialism Was Born Bad: The Ca...

I distinctly recall that Robert Shiller did not recommend that people buy stocks in 2009. That made me wonder when Robert Shiller did say it was a good time to buy stocks. Stephen Kirchner pointed me to an Alan Reynolds column from March 2010:

On the March 9 anniversary of the stock market implosion a year ago, a front-page story in the Wall Street Journal featured one of the same bears making the same bad argument he made a year ago.

The article, "Worries Rebound on Bull's Birthday," was almost entirely devoted to trying to explain a graph by Robert Shiller of Yale, titled "Stocks Still Expensive." The New York Times ran the same graph on March 15, 2009, to warn us that the ratio of stock prices to earnings "hasn't fallen as far as the market bottoms of 1932 and 1982."

By then, reports from Barron's, Bloomberg and the Wall Street Journal had already suggested that the Dow could fall to 5000 and the S&P 500 to 500. The Journal's headline on March 9, 2009, was "Dow 5000? There's a Case for It."

. . .

Here's the bottom line: Following Bob Shiller's "over 20" rule would have kept you out of the stock market every single month from December 1992 to September 2008.


Then Stephen pointed me to a passage from an article written by Robert Shiller in 2003 (p. 38):

In Irrational Exuberance, I gave what I thought was a long list of precipitating factors for the stock market bubble that began in 1982, accelerated in the late 1990s, and peaked in early 2000.
The market bubble began in 1982? Even more puzzling, one of the factors listed was the worldwide decline in inflation. But how can a fundamental factor create a bubble? After all, declining inflation sharply lowers the real tax on capital. Capital gains tax cuts are also mentioned. Is it true that "bubbles" can occur when fundamental factors that should push up stock prices, do in fact push up stock prices?

Of course it's easy to criticize people like Shiller in hindsight. But the Alan Reynolds article is now 4 years old. How does Reynolds' analysis hold up today? Consider this, from the same article:

Looking at the figures through September of last year (the latest available), the trailing E-P ratio was down to about 1% while the bond yield was closer to 3%. That disconnect is simply because trailing earnings toward the end of a recession are deeply depressed, so that stock prices based on expected recovery look high relative to the previous year's earnings.

Earnings over the past year are a poor predictor of future earnings, and earnings over the past 10 years are even worse.

When the E-P ratio was unusually high, as in the inflationary 1978-81 period, it means the P-E ratio is very low and bond yields very high. To suggest that the E-P or P-E ratio will invariably return to some long-term norm, as Shiller does, is to suggest that bond yields will likewise return to some long-term norm. But interest rates are determined by variables -- such as inflation and real returns on invested capital -- that are not simply determined by predictable past trends.

The permabears might make a plausible argument against buying stocks if they argued that a big spike in bond yields was imminent. But that would be inconsistent with their usual forecast of stagnation and deflation. So they're still peddling the fallacy of "above average" multiples, as they were a year ago.

For the press to still be recycling Robert Shiller's stale arguments against buying stocks in March 2009, despite what happened since, is a remarkable example of the media's inclination to favor downbeat theories over any actual good news.


I suppose it wasn't hard to predict the bounce back of earnings after the recession, although it's been stronger than forecast. But Reynolds' claim that P/E ratios and long term bond yields are not necessarily trend reverting now looks quite prescient. Both real and nominal 10 year bond yields have been declining for over 30 years, with real yields falling from over 7% to near zero. Rates are likely to rise modestly in the recovery, but I'd guess they stay lower than the norm of the 20th century. Lots of famous people like Larry Summers are also making this prediction, but Reynolds saw the implications back in 2010.

I met Alan Reynolds at a conference last November and discussed the stock market. He had a smile on his face like the cat that ate the canary. I'm pretty sure he's been putting his money where his mouth is.

Robert Shiller has the Nobel Prize, and Reynolds is at the Cato Institute. I'm sure Shiller is more respected at elite Ivy League institutions. But I can't help thinking that Reynolds actually has the better understanding of stock valuations.

Shiller bleg: Can people find me the dates where Shiller recommended people buy stocks? We all know about his famous bubble warning (1996), but when did he tell people to buy into this seemingly endless "bubble?"


Comments and Sharing





COMMENTS (22 to date)
Hunter writes:

The problem is that Shiller was looking at historical data to find the mean. In 1981 the 401k came into existence meaning there were more buyers than ever and a large number of them had automatic investment contributions. This did not exist before. The underlying system changes therefore the mean is pushed up. To what I don't know.

Andrew_FL writes:

@Hunter-There's not really much indication of a sudden shift in the mean associated with that factor, but then I think it would make more sense if that gradually became a more important factor.

But judging from some indices I've looked at, you really should be asking what changed after 1995.

Robert Shiller has the Nobel Prize, and Reynolds is at the Cato Institute. I'm sure Shiller is more respected at elite Ivy League institutions.

I'm sure he is, although he wouldn't need the Swedish Bank Prize to be more respected. His views on economics are politically correct, that's really all that is necessary.

The more you stop to think about it, the more bizarre it is to give a prize simultaneously to Fama and Shiller, for literally the exact opposite claims.

Jack PQ writes:

[Comment removed for ad hominem remarks. --Econlib Ed.]

Jack PQ writes:

@Andrew_FL: The prizes to Fama and Shiller are not necessarily a contradiction. Fama says, short-run price returns are unpredictable, you cannot outguess the market, there is no easy money to be made, and passive investing is best for most investors.

Shiller says, in the *long-run* there are strange things that suggest price returns could be predictable.

They disagree on the interpretation. Shiller says this shows people are not rational. Fama says this shows people care about more than just Beta. New risk factors are needed.

Neither is entirely satisfactory. Saying investors are not rational means you're basically giving up, "Anything goes", and we have no theory.

"More risk factors" is like Ptolemy's epicycles. Gone fishin' for more factors. Hmm...

Benjamin Cole writes:

Nice blogging...Shiller always very cautious...I agree the secret sauce is the secular decline in inflation and interest rates...which may be the new normal...

James G writes:

@Hunter

To be clear, an increase in demand for a fixed amount of securities would cause expected returns to decrease. That is, according to your hypothesis, prices should have risen in 1981 and returns going forward should have been lower.

@Jack

I wouldn't say that the statement investors are not rational means we have given up or that time variation in the investment opportunity set necessitates "factor fishing."

Behavioral Finance is largely concerned with putting structure to models of systematic irrationality and there's some interesting work being done on subjective probability.

Good (in my opinion) articles that do propose new "asset pricing factors" link those factors to changes in the investment opportunity set in the spirit of the ICAPM (although many published papers do not).

Jack PQ writes:

The debate about Robert Shiller's predictions is really a debate about what constitutes a valid prediction. Statisticians talk about "forecast calibration", loosely speaking, predicting what happens and not predicting what does not happen.

Shiller has, time and again, identified weaknesses in the US economy and plausible tipping points. His research in finance and real estate is carefully done and mostly convincing.

The problem is when we talk about his success in predicting US downturns. "Permabears" (not my term) like Shiller or Nouriel Roubini have predicted doom and gloom every year since the early (?) or mid 1990s. In terms of forecast calibration, their performance is not impressive. If I toss a coin 100 times and call "heads" every time, I will be right about 50 times. But it is not an impressive feat.

David R. Henderson writes:

@Jack PQ,
If I toss a coin 100 times and call "heads" every time, I will be right about 50 times. But it is not an impressive feat.
Shiller’s performance is way worse than that. He has been right on this issue well under 50% of the time. He would have done much better by flipping a coin.

Andrew_FL writes:

@Jack PQ-Perhaps, but from where I'm standing it looks like a kind of bizarre attempt by the Bank to avoid coming down on one side or the other.

Gordon writes:

This part caught my eye:

"After all, declining inflation sharply lowers the real tax on capital. Capital gains tax cuts are also mentioned."

I've seen some economists point to the capital gains tax cuts as a significant factor that helps to explain changes to wealth inequality over the last few decades. But wouldn't the declining inflation also fit the bill? Or are there factors such that it doesn't help provide an explanation?

Marcus Nunes writes:

Scott, the only occasion I saw "buy and stocks" associated with Shiller was in a CNBC inetrview in Davos early this year:
http://thefaintofheart.wordpress.com/2014/03/24/bubble-mania/

tom writes:

@ David Henderson,

Frequency of being right is meaningless without magnitude when it coes to investing.

Andy Harless writes:
Even more puzzling, one of the factors listed was the worldwide decline in inflation. But how can a fundamental factor create a bubble? After all, declining inflation sharply lowers the real tax on capital.

There is research (eg Modigliani & Cohn) suggesting that the market undervalued stocks in the period preceding the bull market that began in 1982 because it was discounting real cash flows in nominal terms. So I don't think Shiller is necessarily wrong to suggest that the decline in inflation affected stock prices to a degree that went well beyond the fundamentals. I do think he's wrong to suggest that the earlier pricing was rational and the later pricing was a bubble.

Roger McKinney writes:

There is a lot that is wrong with both the EMH and behavioral finance. The most egregious is that both assume an objective value for stocks. Another is that risk is measured by variation in prices.

Viewing the stock market in terms of subjective value conditioned by the business cycle makes market gyrations a lot easier to understand. I do that in my book "Financial Bull Riding" published by Laissez Faire Books.

Scott Sumner writes:

I agree with most of the comments, and unfortunately don't have time for detailed responses.

I like Jack PQ remarks about the strengths and weaknesses of each approach. I am also very impressed by Shiller's work on excess volatility using historical data. But that's precisely what makes me surprised by what I see as a rather poor "advice" record by Shiller in recent decades. The last thirty years have been a great time to own stocks, and yet all I ever see from Shiller is worries about overvaluation. If stocks have been such a great historical investment, even adjusting for risk, why isn't he arguing that there was a negative bubble prior to the 1980s? And that perhaps now stocks are fairly priced. It's just a thought, but if I am right it might explain why he nailed the excess volatility, and yet was still unable to come up with particularly useful advice for investors over the past few decades.

Think about this folks. This guy just won a Nobel Prize for the claim that he's smarter than the stock market, and he knows when it's mis-priced. And he was NOT advising the public to buy stocks in 2009 when the S&P was at 675. Isn't that kind of odd? Fama doesn't claim to be able to give advice on when to buy stocks, but Shiller does. So if you are going to win a Nobel Prize based on that claim . . .

Tom P writes:

Shiller's work on excess volatility shows that most variation in stock prices come from changing discount rates, rather than changing expectations of the cash flows being discounted.

If discount rates are changing, there must be times when expected returns are relatively high and times when it they are relatively low. In that sense, it is possible to time the market.

In practice, Shiller's preferred measures like the DP and EP ratios have not been incredibly great guides to market timing.

Nevertheless, unless we think these ratios will tend to infinity or zero in the long run, there must be some force restoring them to long-run averages. That adjustment seems to have come from price levels reverting towards trend, since dividends and earnings are not very predictable from these ratios.

Scott Sumner writes:

TomP. All I can do is repeat my point. He claims his model allows us to offer good market timing advice. And yet he seems unable to do so. Why is that not a big problem for Shiller? It's his model after all.

Ebsim writes:

I think that economists like Shiller are confusing "animal spirits" for macro fluctuations.

Tom P writes:

Well, Shiller is not saying you should stay out of the market. He is currently predicting a 3%/annum real return on stocks, and he must have been predicting even higher returns when the 2009 article was written, since valuations were lower then. (http://www.cnbc.com/id/101031584)

That said, I do agree with the first commenter that Shiller's model may need to be modified to take into account the big explosion in equity ownership via mutual funds; this big inflow into equities may have raised prices permanently (so there would be no mean-reversion).

Additionally, I'm not sure what to make of Reynolds' argument that interest rates and stock valuations are highly related -- if you extend his graph back to the 1920s, the relationship seems to break down (see here, page 14: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=381480&download=yes).

Terran writes:

I also did not go long in 2009 because I agreed with Shiller.

Philip Tetlock talks about this in his book on political forcasting. As I think he would say, my options now are either to say "I was almost right/I'll be right soon", or to be a good Bayesian and update my beliefs by an appropriate amount. One of these options feels better than the other, but it probably isn't the one I should do.

Mike Rulle writes:

On 9/13/2013, on CNBC, he did say "it was still ok to invest in the stock market, but do not expect miracles". S&P was at about 1700. I have no idea what that means.

I have to admit, Professor Shiller's work has always irritated me---a lot. His critique of the EMH always had a straw man feel to it. In the long run, stock price growth has approximated growth in earnings, however, with a large amount of volatility---as should be expected.

The "correct" price even using a variety of fundamental measures is always an extraordinary wide range. Using the term "irrational exuberance" is the ultimate punt in stock price valuation commentary.

Shiller acts as if there are not plausible changes which can radically alter future values virtually on a daily basis. It is as if he thinks the long term Sharpe Ratio of the stock market of .30 is some kind of proof the world is "irrational" as demonstrated by volatility. He is right and the world is wrong.

Shiller got famous for "calling the Nasdaq bubble". I called it too----at about 50% from its high.

By metaphor, the following explains my view of his work.

For purposes of this comment, I will assume the long term distribution of the US stock market is "known". Using annual data for 100 years, a 16% volatility of annual returns, one gets an excess return of 4.8% per year. But the standard error is 1.6%, or a third of the annual excess return. Or, pretending we are political pollsters, the 95% confidence interval of the "true" long term excess return of the stock market is between 1.4% and 8%.

I use this obviously simplistic method to show just how unknowable things are.

So why in the world would anyone believe a 10 year average backward looking P/E ratio is a remotely plausible framework for "fair value"? I would bet hard cash he got a good regression----I would love to see his in sample and out of sample results.

There are extreme times. To believe 700 was a "fair" price for S&P in 2009, one must assume a lot of other things about the nature of the economy, financial system, politics, demography, war, etc., etc. To believe Cisco's 20 year forward value (in 2000) would equal the 20 year US GDP forward value--which it did under certain assumptions, one would also have to believe a lot of other things. Sometimes these "other things" actually do happen----but nothing so radical yet in the US.

EMH is best thing we have so far. Trying to curve fit fair value around a 10 year P/E because people like Kahnemann wrote interesting books is absurd. He cannot and has not made predictions.

His 9/13/2013 comment says it all to me about his views.

Brian Donohue writes:

Great post Scott. Being a professional pessimist earns you gravitas and the devotion of sandwich-board wearers everywhere.

10-year average P/Es sounds oh so prudent, but they obscure a lot of information. The important things are where the price is today and what future earnings will be. 10-year averaging adds a lot of extraneous information about the path to today's prices.

Comments for this entry have been closed
Return to top