UPDATE: As Noah Smith politely points out, I did a no-no in being so lazy as to take averages of monthly returns to be "close enough" to cumulative compounded returns. Fixing that requires some edits, which I have made:
...is hovering at a worrisome level.... Above 25, a level that has been surpassed... in only... the years clustered around 1929, 1999 and 2007. Major market drops followed those peaks.... We should recognize that we are in an unusual period, and that it’s time to ask some serious questions about it...
The first question I think we should ask is: how damaging in the long run to investor portfolios were the major market drops that followed the 1929, 1999, and 2007 CAPE peaks? The CAPE is the current price of the S&P index divided by a ten-year trailing moving average of its earnings: the CAPE looks back ten years to try to get an estimate of what normal earnings are and how stock prices deviate from them. Let's look ahead and calculate ten-year forward earnings to get a sense of what signals the CAPE sends for those of us interested in stocks for the long run.
When we do that, we find that we cannot calculate a ten-year return for the 2007 CAPE peak of 27.54--we still have three years to go. But over the past seven years the S&P has produced an average annual real return of 5.2%/year: not too shabby. The ten-year average real return from the 1929 peak of 32.56 was 3.3%/year: you were in a real world of hurt if you panicked and sold or had to sell in 1931-1934, but not if you hung on. Only the 1999 peak was followed by long-run return disaster: a ten-year average real return of -2.1%/year because you would have been selling at the bottom in 2009--but even then if you had hung on until today your average 14.5 year real return would be +2.7%/year.
If you are not an investor in the stock market for the long term, you can easily get into a world of hurt with a position in the S&P composite (and an even bigger world of hurt with an undiversified portfolio). Look at the one-month and one-year return distributions:
You can lose a fifth of your money in a month: people have. You can lose more than half your money in a year. And you can do those things as well with a CAPE of 15 as with a CAPE of 40, as the next two figures show:
If you are not in stocks for the long term, your stock portfolio should not consist of money that you cannot afford to lose.
But if you are in stocks for the long term? How big are the risks, really, and how do they correlate with the Campbell-Shiller CAPE? I like to start thinking about this by plotting the full distribution for every month from 1881 to mid-2004, the last month for which we can calculate a full ten-year real return:
The left axis shows the cumulative ten-year
log compounded annual return: a value of 0.069--6.9% per year--means that you have doubled your real money over ten years; a value of 0.14--14% per year--that you have quadrupled your real money; and if there were a value of 0.21 there you would have octupled your real money. The S&P has never done this. But it came very close in the ten years starting in 1920.
An extremely naive take would be to assume the efficient market hypothesis: that the marginal trader and the marginal firm know what they are doing, and that the margin the earnings of the companies in the index are equally valuable if paid out or if reinvested. In that case we would expect the real return--the dividend yield plus capital gains--to be simply equal to the earnings yield. The warranted ten-year return would then be simply 10 times the permanent earning power. And if we take the moving average of earnings underlying the CAPE to be our estimate of the permanent earning power, the warranted return is simply 100 divided by the CAPE, like so:
Given the naiveté of the framework, that turns out to be
not at all a bad a remarkably good guide to the central tendency of the distribution of future ten-year returns conditional on the CAPE. If you are going to project an expected value for an investment in the S&P over the next ten years, simply inverting the CAPE and multiplying by 10 is a very good place to start.
But even at the ten-year return horizon the variability is enormous: earnings 10 years hence may be well above their current value compounded forward at the current earnings yield; but they may be well below as well; the earnings valuation multiple 10 years hence may have jumped up; or it may have crashed. The only places in the distribution where the naïve warranted return does not seem to capture the central tendency is where the CAPE was very low or very high. But we know that when the CAPE was very low there were no subsequent large downward reductions in the valuation multiple: that's what it means when we look back and say "oh, then the CAPE was very low". And we know that when the CAPE was very high there were no subsequent large upward leaps in the valuation multiple: that's what it means when we look back and say "oh, then the CAPE was very high".
The entire spread of the time series has something to tell us about expected returns and the variation around them. What does it say about the risk of loss--the possibility that we sock our money in the stock market and reinvest it for 10 years, but when we come to take it out it is not there or it is not all there? What I think are the ten most important points:
Risk relative to what? Trying to move purchasing power from the present into the future is a hazardous activity. War and rumors of war; inflation and rumors of inflation; depression and rumors of depression; Bernie Madoffs and rumors of Bernie Madoffs; coups, confiscations, and heavy capital taxation--risks cannot be avoided but only managed and balanced off against one another.
That said, almost always ten-year returns on the S&P have been a winner.
And on average, at a ten-year horizon, for any CAPE ratio below 35 the S&P has delivered average real asset returns pretty much outclassing all other major asset classes.
There are only
threefour historical periods during which a ten-year investment in the S&P has not at least held its real value: ten years before the post-World War I deflation and the post-WWI depression of the start of the 1920s; (barely) in the Great Depression and the WWII inflation; 10 years before the stagflation of the 1970s and the subsequent Volcker depression; and 10 years before the recent financial unpleasantness for those dates where the ten-year return window includes both the dot-com and the housing-bubble crashes:
The outbreak of World War I and its consequences was the original Black Swan. Those who were trying to shed stock market risk around 1910 by purchasing bonds of any sort found themselves much more exposed to and damaged by the inflationary component of the World War I shock. And if you could hang on to your equity portfolio after 10 years and go double or nothing for the next decade you captured in the near-octupling which was the greatest bull market in history. And, of course, the CAPE was not notably high but rather below average as this first episode commenced.
At the peak before the Great Crash of 1929 and at the peak before the secondary recession of the late 1930s that followed Roosevelt's and the Federal Reserve's turn to "austerity", real stock-price recovery and dividends over the next decade almost but did not quite bring stock-index investments back into the black.
From the early 1970s through the early 1980s the economy was hit not by a black swan but by a bunch of ugly ducklings: politicians (cough, Richard Nixon) with an excessive focus on their own reelection and Federal Reserve chairs (cough, Arthur Burns) who appear to have forgotten that they were and were supposed to be independent of both the president who appointed them and the congress that oversaw them; oil shocks; productivity slowdowns; and, ultimately, the failure to figure out any way to end stagflation and re-anchor inflation expectations at a low level other than repeatedly hitting the economy on the head with a high interest-rate brick until it collapsed. But, once again, those who were trying to shed stock market risk by purchasing bonds of any sort found themselves much more exposed to and damaged by the inflationary component of the 1970s shocks. And, of course, the CAPE was high but not notably and anomalously high as this second episode commenced.
As noted above, the first time the CAPE crossed 25 heading upward was during the last stages of the Roaring Twenties bull market that preceded the Great Depression. And--provided you held on for the full ten years--you did OK: cumulative returns were not that far from what you would have expected if you had simply inverted the CAPE and multiplied it by 10. And the same for the second time the CAPE crossed 25 heading upwards, in the years after Greenspan's "irrational exuberance" speech.
Only those who invested at the peak of the dot-com bubble which just happened to come ten years before the crash that inaugurated the Great Recession fared badly in the sense of losing any substantial component of their real wealth. Here, moreover, investing in not risky but safe bonds would have protected you: there was no inflation component to the shocks that caused the Great Recession.
Thus you can see why I am relatively unsatisfied with Shiller's writing:
In the last century, the CAPE has fluctuated greatly, yet it has consistently reverted to its historical mean--sometimes taking a while to do so. Periods of high valuation have tended to be followed eventually by stock-price declines. Still, the ratio has been a very imprecise timing indicator.... The ratio is saying the stock market has been relatively expensive for years. And that raises a question: Are there legitimate factors behind high stock prices that might keep them elevated for decades more? Such a question has been addressed before. In 1930, just after the 1929 crash, Prof. Irving Fisher of Yale published “The Stock Market Crash — and After.” The book explained why there were “sound reasons” for the high valuations of 1929. He couldn’t have been more wrong...
Shiller's rhetoric leads us to focus on graphs like this one of the Campbell-Shiller CAPE, and to think that what goes up must--someday--come down:
But is there any reason to think that the central tendency of the CAPE today is the same as what it was in the 1880s or the 1950s? There is no unchanging machine buried in the earth for the past 120 years throwing dice to determine the CAPE. It would be much better to say that extreme values of the CAPE are followed by reversion not to but toward the previous historical mean. And dividends and earnings shift too. A much better graph than the CAPE graph is the cumulative reinvested return graph:
And on a log scale:
It goes way up, and way down, but the dominant feature is not mean reversion but rather exponential growth. Think: 3.5 tenfold upward leaps in four generations--but with many periods of half a generation in there in which things at best plateau, and if you are forced to sell at the wrong time you will lose a good part of your shirt.
Given the large number of investors and institutions in our economy with very long time horizons that thought to be in the stock market for the long-term--insurance companies, pension funds, rich individuals with grandchildren--for me the anomaly does not seem to be a CAPE of 25 (or, given historical real returns on other asset classes and very low current yields on investments naked to inflation risk, 33) but rather the CAPEs of 14-20 that we saw in the 1980s, 1960s, 1950s, 1900s, 1890s, and 1880s that Robert Shiller appears to think of as "normal" and to which today's CAPE should someday return. Those are associated with warranted real average returns of between 5 and 7%/year. Why ever were stocks so unpopular as to limit demand so much as to offer such returns? Is there no long-term money? And so we have arrived at the biggest mystery of macro finance: the premium return on equities.
Yet Shiller thinks 25 is an anomaly that requires some non-fundamental psychological explanation:
I have conducted questionnaire surveys of individual and institutional investors.... When the CAPE ratio reached its record high of 44 in 2000, the confidence index hit a record low.... This year, the index took a dive for individual investors, and now is at the lowest level since 2000. The confidence of professional investors has fallen, too, but not as sharply. In short, people are beginning to worry.... It’s possible that bond prices account for today’s stock market valuations. But that raises another question: Why are bond prices so high? There are short-term explanations: the role of central banks, for example. But is there a compelling reason for prices of stocks and bonds (and maybe houses, too) to remain high indefinitely?... Perhaps today’s prices have something to do with anxiety about the future.... Such anxiety might push them to try to make up for... potential shortfalls by investing in stocks and bonds--even if they worry that these assets are overvalued.... I suspect that the real answers lie largely in the realm of sociology and social psychology--in phenomena like irrational exuberance, which, eventually, has always faded before...
That is a perspective very different from mine, which regards the failure of the CAPE to spend most of its time north of 25 as a mystery.
But given that it does not, it would be very rash for anybody who is not certain that they can wait out the market to invest more than they can afford to lose. And past performance is not only not a guarantee it may not be an indicator of future results. We have had one real Black Swan--World War I--in the past 130 years.