The Shiller P/E: Alas, A Useless Friend

The high level of the Shiller P/E is one of the bears’ main arguments against equities. Bulls claim that the 10-year average earnings it uses to measure P/Es is unduly depressing earnings because it includes not just one but two of the worst profit recessions in history, the bust and the financial crisis. They also generally prefer to use “operating” earnings as opposed to the “as reported” profits that the Shiller P/E approach favors. Three articles on that debate:

Shiller P/E advocates strongly dispute the more bullish arguments. Professor Shiller admits that “corporate earnings have been unusually volatile in the past decade” but he argues that this is more reason — not less — to use normalized earnings in calculating the market’s P/E.

Others simply dismiss the bulls’ arguments with more dogmatic, almost religiously radical statements such as:

  • What about the fact that the past 10 years include two major earnings anomalies that skew the market’s CAPE? “I’m grateful that there are people who believe that, who can be on the other side of my trades,” Mr. Arnott says.
  • In short, if you don’t like what Shiller is telling you, it is because you are a bull who thinks “this time is different”.

And the ultimate:

“What alternatives do people have?”

I am entering the debate because:

  • “Operating” earnings are a better gauge of index profits;
  • Assessing current indices against the last 10-year earnings is flawed;
  • Looking at past evidence, the Shiller P/E is simply useless as a market valuation tool;
  • There is at least one alternative.

When I began as a financial analyst, back in 1975, our clients were essentially conservative pension funds and wealthy businessmen. When companies reported year-end results, one of the first questions clients asked was “What are the operating earnings?”.

In those and previous years, accounting rules and regulations were so loose that accountants had much leeway in presenting results. Investors were shown but one set of results, often highly manipulated, which may or may not reflect previous years methods and/or competitors’ standards. You had to wait 4-8 weeks after the initial release for the annual report to be mailed. Only then could you delve into the details and the notes in order to decipher what were the “operating results”, pruning out of the “as reported” all the accounting noise that, more often than not, contributed positively to the reported results.

This annual exercise helped us better appreciate and adjust the quarterlies which often came on a small folded carton with no details other than revenues, certain cost items, if any, depreciation charges and taxes. The quarterly cash flow statements were “very useful” in providing total non cash charges that may, or may not, have arisen from normal operations.

One could then approximate how companies were performing on their basic operations in order to better evaluate their stock.

“Those were the days, my friend”.

Nowadays, many investors and “observers”, often in the name of conservatism, totally dismiss “operating earnings”, advocating the exclusive use of “as reported” earnings when valuing equities.

“The times, they are a-changin’”!

Accounting “principles” changed frequently throughout the 20th century. A case in point, the treatment of unusual or extraordinary items has been fraught with difficulty. Generally,  companies had preferred to place extraordinary bad news in the earned surplus statement, and extraordinary good news in the income statement. APB Opinion 9 in 1967 endorsed the SEC’s preferred all-inclusive income statement, although it said that extraordinary items should be reported separately. Under APB Opinion 9, companies simply began rationalizing good news as ordinary and bad news as extraordinary. In 1973, APB Opinion 30 established a “Discontinued Operations” section of the income statement and defined extraordinary so narrowly that the classification no longer existed as a practical matter. In 1974, FASB’s SFAS 4 designated gains and losses on the premature extinguishment of debt as extraordinary. In 2002, SFAS 145 rescinded SFAS 4.

Other contentious issues involved inventory costing (FIFO, LIFO, etc.), goodwill recognition and amortization, foreign exchange translation, pension costs, consolidation, just to mention a few.

Here’s how the American Institute of Accountants saw its role in 1933.

Within quite wide limits, it is relatively unimportant to the investor what precise rules or conventions are adopted by a corporation in reporting its earnings if he knows what method is being followed and is assured that it is followed consistently from year to year.

When advocates of “as reported” earnings claim that “operating earnings” are manipulated and that historical stats are purer, they take little account of the fact that historical earnings were themselves positively manipulated, often much more so, than modern “operating earnings”. At least, the latters are now much better defined and consistent and are more openly displayed.

It is only in 1973 that an independent FASB became the first full-time accounting standards-setting body in the world and began to gradually improve and standardize accounting rules.

Standard & Poors has data segregating “operating” and “as reported” earnings going back to 1988.

As Reportedincome, sometimes called Generally Accepted Accounting Principal (GAAP) earnings, is income from continuing operations. It excludes both discontinued and extraordinary income. Both these terms are defined by Financial Standards Accounting Board (FASB) under GAAP.

Operating income then excludes ‘unusual’ items from that value. Operating income is not defined under GAAP by FASB. This permits individual companies to interpret what is and what is not ‘unusual’. The result is a varied interpretation of items and charges, where same specific type of charge may be included in Operating earnings for one company and omitted from another. S&P reviews all earnings to insure compatibility.

Adjusted operating earnings may exclude non-recurring expenses such as restructuring charges, asset sales gains, major litigation charges, goodwill write-downs and other write-offs.

Having an independent body to ensure compatibility and continuity adds credibility to the “operating” earnings time series. S&P effectively prunes out, when it deems appropriate and consistent, non-recurring expenses such as restructuring charges, asset sales gains or losses, major litigation charges, goodwill right downs and other write-offs that are unlikely to recur in the future and are thus regarded as one-off items that distract from the recurring earnings stream and true operating results. S&P does not only adjust earnings upwards. For example, Q3 2012 “operating” earnings were reduced below “as reported” for 50 of the S&P 500 companies.

While it is advisable to take account of “recurring non-recurrings” when analyzing individual companies in order to assess management, the notion becomes preposterous when used for an index such as the S&P 500.

The chart below shows how trailing 12-month “operating” earnings diverged from “as reported” since divergence began in 1982.


Between 1982 and 2000, the average spread was 7.4% with a 7.2% median within a stable 0-17% range. Since 2001 however, the average has shot up to 20.4% with an 11.8% median. The internet bubble and, particularly, the financial crisis have both resulted in huge “unusual” charges, substantially larger than what we had seen in the previous two decades. At its current 11.5%, the spread is back within its past normal range.

In both these truly exceptional periods, many companies, large and small, reported losses, some significant, others simply humongous. To appreciate the uniqueness of the 2008 debacle, consider that for Q4 2008, 140 S&P 500 companies, nearly 1 in 3, reportedlosses, 97 of which also recorded “operating” losses per S&P.

As a result of the carnage, trailing 12-month earnings for the S&P 500 Index collapsed to a trough of $6.86 in March 2009, down 92%from their June 2007 peak of $84.02. The last time trailing earnings were below $7.00 was in April 1973, 35 years before! “Operating” earnings, meanwhile, troughed in Q3 2009, declining 57% from their $91.47 peak of June 2007 to $39.61.

Many investors, strategists and economists have missed the March 2009 market trough because they blindly used the “as reported” data. Had they done a little more thinking and research they would have realized that:

  • While reported earnings had cratered 92%, the value of corporate America could not have shrunk anywhere near that.
  • Importantly, a very large part of the losses were in financial companies due to the  collapsing housing market and the Lehman failure. Many recorded humongous losses while their stock price sank as bankruptcy loomed. This extraordinarily unique combination of sky-high losses and stock prices diving towards zero created a very unique situation for stock indices: companies with then almost negligible market weights were recording humongous losses.

Here’s what Wharton professor Jeremy Siegel wrote in a Feb. 25, 2009 WSJ article:

(…) As the fourth-quarter earnings season draws to a close, there are an estimated 80 companies in the S&P 500 with 2008 losses totaling about $240 billion. Under S&P’s methodology, these firms are subtracting more than $27 per share from index earnings although they represent only 6.4% of weight in the index. (…)

I added these details in my March 3, 2009 post S&P 500 P/E Ratio at Troughs: A Detailed Analysis of the Past 80 Years:

At the extreme, and we are admittedly in an extreme period, a large company with a tiny market capitalization could incur losses so large as to wipe out most of the S&P 500 earnings (AIG lost over $60 billion last quarter alone). As a result, the Index PE would skyrocket even though the other 499 stocks’ valuation would actually not change at all. In effect, a casual or superficial observer looking at the Index would conclude that equities are expensive or overvalued when, in fact, 499 stocks would be cheap or undervalued.

Incidentally, the S&P Financial sub-index recorded “operating” losses in each and every quarter of 2008 for a yearly total of $21.24 ($37.96 “as reported”)

  • Another important and overlooked consideration is that many of the companies that recorded huge losses in 2008-09 either went bankrupt or were substantially restructured or acquired. A conceptually valid valuation method like the Shiller PE, measuring 10-year average earnings against a current index, is thus including in its denominator, during 10 years, the huge losses recorded by companies that are no longer included in its numerator, these companies having in fact been replaced by other, profitable, companies.

Humongous or very large losses were recorded in 2008 by companies such as AIG, GM, Merrill Lynch, Marshall & Ilsley, MBIA, Wachovia, all companies then part of the S&P 500 Index but no longer. As to the bubble, everybody remembers the infamous Enron and Worldcom, just to name a couple. Their losses still impact the 10 year average earnings even though they have no contribution to the actual index value.

The losers are long gone but their losses remain!

This is like assessing a baseball team’s current batting line-up using 10-year data that includes the dismal stats of now deceased players. How useful is that?

Cliff Asness, a Ph.D. turned hedge fund manager (AQR Capital) recently wrote a long note “An Old Friend: The Shiller P/E” which was extensively disseminated by the media and, particularly, the bear population . The notorious quant statistically looked at most arguments against the Shiller P/E and categorically ended the debate:

Those who say the Shiller P/E is currently “broken” have been knocked out.

Unfortunately, Asness did not stumble of the truth this time since nowhere does he mention that his current roster has little to do with that which generated the 10 year record.

While the internet is filled with Shiller supporters, I failed to find a good analysis of the actual track record of this valuation method. It may be because the record leaves a lot to be desired.

The long term average of the Shiller PE is 16.5 and the median value is 15.8 which most people seem to use as the dividing line between cheap and expensive. I will let you judge by yourself based on your own personal needs and risk aversion, only to point out the following (click on charts to enlarge):

  • Post WWII investors using the Shiller P/E would have had very few buying windows, to say the least.
  • The 20-year period between 1955 and 1975 was a very long one to stay on the sidelines.
  • One could have bought the 1974 low, only to be back on the fence in early 1976, buying again during the next 10 years but sell out in 1986, only to watch equities appreciate 2.5 times to 1996 (I am excluding the internet bubble years).


Cliff Asness computed 10-year forward average returns from different starting Shiller P/Es since 1926:image

(…) while not near its prior peaks, today’s Shiller P/E is high versus history. In fact, it’s higher than it has been 80% of the time since 1926.

The media and the bears loved that last sentence and the supporting table! Asness continued:

(…) Ten-year forward average returns fall nearly monotonically as starting Shiller P/E’s increase. Also, as starting Shiller P/E’s go up, worst cases get worse and best cases get weaker (best cases remain OK from any decile, so there is generally hope even if it should not triumph over experience!).

In particular, in the ninth bucket (where we are today at 22.2) the average real stock market return over the next decade does not break 1%. The worst case is a horrendous -4.4% real return per annum (those who think the disappointing post-2000 decade-long results can only happen from super high P/E’s are mistaken), and the best case is very good, though less wonderful than the much better best cases from lower starting Shiller P/E’s.

Asness then goes on listing some caveats but never mentions that readings below 15.8 occurred mainly before 1950. He does, however, confess that

(…) I would, if trading on a tactical outlook, give the Shiller P/E some small weight, particularly when it’s above 30 or below 10.

There you go! Cliff Asness shared the Truth with us. Since 1927, that is over some 1,030 months, the Shiller P/E registered below 10 thirty-seven times (3.6%), all but two months being pre-1942, and was over 30 eighty-nine times (8.6%), all but 2 being between 1996 and 2007.

The Shiller P/E may be an “old friend” to Cliff Asness (although he was born in 1966), let’s hope it is not his best friend.

“What alternatives do people have?” This authoritative question is from Professor Shiller himself.

Here’s an alternative: my personal old friend the Rule of 20 which says that fair P/E is 20 minus inflation. When the actual P/E on trailing EPS plus inflation is below 20, equities are undervalued. Above 20, risk increases as overvaluation rises.


Admittedly not perfect but, in my book, substantially more sound and useful than the Shiller P/E. I must admit that I was actually hoping to find another friend in the Shiller P/E that would complement the Rule of 20. Alas, this is not my kind of friend.

The Rule of 20 is a risk management tool, enabling investors to measure the downside against the upside in order to decide whether the risk/reward profile fits their own personal needs.

Most strategists tend to analyze the economy and the market fundamentals before assessing if the market P/E fits their scenario. I prefer the opposite approach. First, I objectively measure the risk/reward ratio, paying particular attention to trends in the 2 components of the Rule of 20: earnings and inflation.

When equities are expensive, I work on my golf game or go salmon fishing. When  equities are cheap, I then assess the economy and the fundamental trends to see if a trigger is near that might unlock values. Here’s the Rule of 20 barometer since 1980.


I have been blogging since January 2009, providing my readers with balanced, objective and detailed views, caring more about preserving than increasing capital (the return of capital concept vs return on capital). I have generally been positive on U.S. equities since March 2009 with three interim periods where I advised caution. The Rule of 20 has been very useful helping me objectively measure risk vs reward during these very volatile years. However, a disciplined following and objective analysis of the economic and financial environment is always needed to supplement the mathematical risk/reward equation (detailed track record).


U.S. equities are very cheap currently. While Q3 earnings were down somewhat (-4.2% Q/Q, -3.6% Y/Y), they are not in free fall like in 2007 when they dropped 13.2% in Q3 2007 from their Q2 peak level. Also, inflation remains contained in the 2% range and oil prices, a big driver of inflation, are behaving reasonably well currently. Inflation was rising rapidly in the U.S. between August 2007 and July 2008 which, combined with declining earnings, caused a sharp 18% decline in the Rule of 20 fair value (yellow line in the Barometer chart above) between August and November 2007.

The U.S. economy has been showing encouraging signs lately. The Fed is printing money like there is no tomorrow (literally) and is actively keeping interest rates to the floor, the ECB is taking care of the Eurozone fat tail risk and China seems to have stabilized its economy and could be about to re-stimulate more aggressively. Normally, such an environment is quite enough to unlock a cheap equity market and justify a green light on equities.

Yet, I am keeping a yellow light for now, essentially because of the looming fiscal cliff which, with odds no better than 50-50, would cause a U.S. recession as early as Q1 2013. This would most likely axe 2013 earnings by 10-20%, eliminating most if not all of the current 25% undervaluation. Betting one’s own money on politicians is generally not without peril. The current undervaluation is so large that I would rather wait to see if politicians deliver or not. This has been a wise approach in Europe.


7 thoughts on “The Shiller P/E: Alas, A Useless Friend

  1. Thanks for the fascinating history on Accounting treatments–

    “The long term average of the Shiller PE is 16.5 and the median value is 15.8 which most people seem to use as the dividing line between cheap and expensive.”

    That is how journalists write about it, not how practitioners use it. See Ed Easterling’s work at for some insight into how practitioners use normalized earnings. And even then, most of us make additional adjustments for profit margins, currency debasement (which is not the same thing as inflation necessarily), and other factors.

    The benefit of normalized earnings for long-term return forecasts (10-20 years) is that you don’t have to forecast when the next 20-50% earnings decline is going to come. Using Rule of 20 requires you to do so, if you wish to place risk management at the forefront. You might be good at providing that overlay, but most are not, including myself. Which makes Rule of 20 not so much a rule, but a very general guidepost– sort of like most everything else.

    I appreciate your work, nonetheless.


  2. How does the rule of 20 account for current earnings being approximately 70% above historical averages? Obviously earnings will return to trend in the near?? future, which would have a very big impact on the rule of 20.


    • Thanks John. The Rule of 20 makes no forecast as it uses trailing earnings. It has been helpful for many earnings cycle. That said, I carefully monitor earnings and trends in order to further assess the risk (momentum). As you point out, margins are historically high but I cannot forecast when (if) they will start declining and at what rate.

  3. Why limit oneself to US equities? Broadening the investment scope to any global equity market, one can usually find at least a few markets with low CAPEs. A buy low global CAPE strategy would have provided much better and lower risk returns than focussing only on the US, and allowed you to be in the market most of the last 30 years, with the possible exception of 2007. At the same time that US equities were cappign 45x CAPE in 2000, emerging markets were hitting lows below 10x. They then went on to shoot the lights out. US markets are overly studied, data mined, and worshipped.

Leave a Reply to jcg Cancel reply

Your email address will not be published. Required fields are marked *