"But then economists' models all say that at historical price patterns stocks are substantially undervalued, so they shouldn't go back to historical price patterns. They should never have followed the "historical price patterns" in the first place."

Then perhaps the problem is not the "historical price patterns" but the economists' models?

Are you telling me that economists can't come up with one stinking model which conforms with the real world prior to 1998?

If you socialised more widely with economists, you would not be so amazed that the incompatibility of standard models with observable facts gives them so little concern ...

And Tobin's Q?

Don't trust the Dow-Jones--it's a bad index, not representative of anything, that preserves its media footprint only because of the power of the Dow-Jones media conglomerate.

Categorically false.

The Dow holds symbolic sway based on the psychological inertia of its longevity...the fact that it has been the *only* index quoted regularly and publicly over the course of most people's lives. It's familiarity, period.

Dow Jones and Company has gone from fumbled chance to fumbled chance for generations, and has been in relative decline for decades. Their "power" has shrunk relative to Reuters, Bloomberg, etc.

This is an appalling inaccuracy.

That said, yes, the index stinks.

S&P 500 only covers about 80% of US equity market. Wilshire 5000 is better - it covers 100%.

3.02% earnigs yield. Note that at August 30, S&P reports p/e of 37.09 (2.7% yield) and Barra reports 31.94 (3.13% yield). The S&P 500 is off 10.5% since then. If nothing else, this shows using two decimal places is not the best method.

Many economists take 10 year average earnings to compute p/e, rather than a point estimate, to eliminate transitory movements, such as might be caused by a recession.

then economists' models all say that at historical price patterns stocks are substantially undervalued

"All" may be an overstatement. A lot of economists believe the realized equity premium has been way above the expected equity premium. See, for example, a recent paper by Fama&French. Efficient market theorists would never say the market is wrong. Shiller certainly does not believe stocks are undervalued.

In any event, if your model shows different values than the market, the odds are the problem is with your model, not the market. If memory serves, you once said that.

Re:

>>If you socialised more widely with economists, you would not be so amazed that the incompatibility of standard models with observable facts gives them so little concern ...<<

Oh no! It gives us great concern! The "equity premium puzzle" is something that I, at least, worry about every single day...

I'll say a word or two in favour of the Dow Jones Industrials as an index; I don't think it's bad at all.

First up, it's equally weighted, which I prefer to market-cap weighted on the grounds that 1) it doesn't get skewed by megacap stocks 2) it doesn't, for that reason, need to be rebalanced so often and 3) it doesn't give the impression of utterly spurious analytical rigour. The last point is the most important one; despite what anyone thinks, the S&P composite is *not* even a reasonable proxy for the CAPM's market portfolio.

Second, it's only 30 blue-chip stocks, which is all you need for a big stock index, which should then be combined with a decent smallcap index. Going from the Dow to the SP500 or Composite is (once more) spurious rigour, and allows you to confuse yourself about what the purpose of a stock market index is. It's not, because it can't be, a measure of how the average investor's wealth has changed, nor is it in any meaningful sense a representation of what happened to the wider market. What it is, in Charles Dow's original phrase, is a "barometer", and the DJIA does as well in that role as anything. If the Dow's not rising, then big investors (in general) are not buying stocks (in general).

I used to buy into the idea that the Dow doesn't matter, but I think that's an exaggeration. The Dow components represent almost 25% of the total value in all companies publicly traded in American markets, right? Plus, it includes several of the largest companies - Microsoft, GE, Exxon Mobil, Wal-Mart. These are companies that many people have invested in, either via stock or indirectly in mutual funds. Since the Dow has considerable value and its stocks affect very many investors, the Dow has relevance - as a barometer, as Daniel Davis said.

At the height of the media coverage of the Microsoft antitrust case, I remember reading something along these lines: "For God's sake. I don't know about the merits of the case, but I have stock in Microsoft. My friends have stock in Microsoft. Everyone has stock in Microsoft! And your God damn litigation is driving the price down, so knock it off and everyone will be a lot better off." Everyone is not an economist; they don't bother with modeling and equations and esoteric terms. They just want their investments to increase in value. And since the Dow components affect a large number of investors - I would imagine a disproportionately large number relative to its value - it matters more than the numbers suggest it should.

One thing to remember is that large bets are being and have been placed on the idea that historical stock valuations are wrong. There are hedge funds betting that stock and bond returns should be pretty much equal in the end. So there is a practical side to this. In fact this theory is responsible for a lot of the price support as stock prices have been sliding.

The Dow is not equally weighted. It is price weighted. If it was eqaully weighted, it would need to be constantly rebalanced.

There are 2 problems: Why should the p/e ratios be so high after 30 months of the worst bear market in more than 50 years? There is no similar data for any bear market or recession since 1950.

Why should low interest rates allow for higher p/e ratios? After all, Wall Street analysts were telling us interest rates did not matter in early 2000.

>> Don't trust the Dow-Jones -- it's a bad index, not representative ... the tech-heavy more speculative NASDAQ index is down 77% <<

An interesting fact that no popular index reflects, and which really hasn't drawn as much comment as I'd have expected, is the bifurcated nature of the stock market.

That is, as of a month ago when I last checked, tech stocks were down 75% but non-tech stocks in the S&P 500 were down only about 18% -- which is not even a "bear market" in the common parlance. The 40% fall in the S&P 500 reflects the historic hammering of tech stocks (which were 36% of the S&P by value at the market high, about 14% today) not a broad market decline of that amount.

The rest of the stock market -- the great majority of stocks -- hasn't done nearly so poorly. In fact I'd say it's held up surprisingly well considering the recession, all the financial trauma caused by the tech stock calamity, the scandals, the war uncertainties, et. al.

It's like there are two different markets here -- a smaller one that has plunged off a cliff, and a much larger one that has been pulled into a slump as a result but which hasn't done nearly so badly, especially all things considered. The broad indices only conflate them.

(Of course, all this is the mirror image of what happened during the up side of the boom, when tech stocks rocketed upward while the non-tech majority of stocks rose too but at mortal rates, lagging far behind. )

"as of a month ago when I last checked, tech stocks were down 75% but non-tech stocks in the S&P 500 were down only about 18% -- which is not even a "bear market" in the common parlance."

As far as I can tell, and the telling has been quite important, there is a deep bear market in growth and value stocks that began 30 months ago.

The S&P value index is down about 38%, the growth index about 55%. European stocks have fallen as long and even more in dollar terms.

Yes, there is a bear market in value stocks....

Indeed, the small cap value index has down over 30% from 5/03/02 to 9/24/02. There is a bear market. Knowing valuations stocks, and knowing the bond market, has been most important in protecting against the bear market.

Most stocks, even "value" stocks, were an historically poor investment in early 2000, while investment grade bonds were a terrific investment. John Bogle and Warren Buffett were telling us just this.

Terrific talking points. I think they do a good job of representing the best professional judgment about the issue.

I'll say this for the economics profession.

1) when stock prices were at historical norms (i.e., cheap by our valuation), we tended to hold more stocks in our portfolios than other people. (I remember a study of TIAA-CREF allocations that showed this.)

2) economists were more cautious than many others when the stock market was at its 2000 levels

3) economists were much more cautious than others about the Internet Bubble. I haven't talked to anyone in the profession who was a buyer of Internet stocks. I'm sure if I talked to enough economists I would find some who were duped, but it seems to be much fewer than in the investment community at large.

>>The "equity premium puzzle" is something that I, at least, worry about every single day...<<

That's good for you, but you're not the typical economist:-)

Seriously, have you considered the proposition that the "equity premium puzzle" is not a puzzle but is a genuine antinomy for the mean-variance view of investment behaviour?

I'm currently labouring through Paul Davidson's latest book, "Financial Markets, Money and the Real World". I seem to remember you edited an issue of the JEP on the Post-Keynesians -- I'm coming round to the view that they've got a lot to offer.

>>That is, as of a month ago when I last checked, tech stocks were down 75% but non-tech stocks in the S&P 500 were down only about 18% -- which is not even a "bear market" in the common parlance. .."

Exchange "tech stock" for "railroad stock" or "automobile stock" and then see if there is any difference in previous bear markets. You will see that it is typical that something like this happens in both bull and bear markets. The market is driven up by one sector and goes down based on that sectir,

I've never understood why the equity premium is a puzzle, rather than just a measure of how risk-averse people really are, especially when dealing with significant sums of their own money.

Ask people the following question:
(Emphasize that you are not testing their knowledge of probability, just their attitudes towards risk).

How much would you pay to play a game where you flip a coin 1000 times and get a dollar for every heads that comes up, and nothing for a tails?

The answers I have gotten are astonishing, often below $400.

The reason the equity risk prmium is a puzzle may simply be that those who think about it are less risk-averse, at least with respect to equities, than the population at large.

>>I've never understood why the equity premium is a puzzle, rather than just a measure of how risk-averse people really are, especially when dealing with significant sums of their own money. <<

Basically, the answer to this is that if you try to use the estimates of risk-aversion that you get from the equity risk premium in any other context, (for example, if you try to base a demand for money function on them), you get results which are either theoretically ridiculous or completely at variance with empirical experience. The original "puzzle" from the original Mehra and Prescott paper was trying to come up with a utility function which could deliver both a demand for money which was consistent with evidence, and the observed equity risk premium.

>Ask people the following question:
(Emphasize that you are not testing their knowledge of probability, just their attitudes towards risk).<

How much would you pay to play a game where you flip a coin 1000 times and get a dollar for every heads that comes up, and nothing for a tails?

The answers I have gotten are astonishing, often below $400. <

Two questions about this experiment:

1. How is the question phrased exactly? If you just ask "How much are you prepared to pay?", you make it sound like a negotiation over price. If I am negotiating with you about how much I'm going to pay to play the game, then of course I have an incentive to name numbers that are lower than the actual maximum that I'm prepared to pay.

2. If you are not testing their knowledge of probability, then maybe you should tell them the relevant probabilities involved. If you inform them that there is less than a one-in-seven-billion chance that they would get no more than $400 from the game, do they still give a maximum price less than $400? If not, then it wasn't only risk aversion but also cognitive error that led to their low answers.

At this point it must be time for Monty Hall to take the stage. I think I'll catch an early night instead.

Why don't we talk about something interesting like whether the age structure of the population has anything to tell us about all this - I don't know the answer, but I certainly thinks it's a topic that needs more research. I'm afraid its a topic that keeps me awake more at night than the equity risk premium.

And thinking of it, what about the impact of global funds on all this, and how does globalisation affect what we think we're trying to measure.

Should the relative level of interest rates also be taken into account when making judgements about relative pe levels?

Should the relative level of interest rates also be taken into account when making judgements about relative pe levels?

Thank you.

Historically, PEs and inflation/rates have a very strong inverse relationship. Discussing PEs without setting them in a cost-of-capital/discounting rate context is like discussing the weather without once mentioning the temperature.

If there is a very strong inverse relationship between p/e ratios and inflation/rates than this stock market must easily be 30% below fair value.

Of course, the market is still down about 45% in the last 30 months. Europe as well is down about 45%. So much for the relationship as a timing device.

Jeremy Siegal has a fair value for the market of about 22 p/e. I agree. Then the market may be decently valued now, and we may hope for but should not expect 30% more to soon.