If your valuation models use forward estimates rather than twelve-month trailing data, you’re doing it wrong. Why? As we discussed in Quantitative Value, analysts are consistently too optimistic about the future, and so systematically overestimate forward earnings figures.
They are consistently, systematically, predictably ignorant of mean-reverting base rates. As we wrote in the book:
Exceptions to the long pattern of excessively optimistic forecasts are rare. Only in 1995 and 2004 to 2006, when strong economic growth generated earnings that caught up with earlier predictions, do forecasts actually hit the mark. When economic growth accelerates, the size of the forecast error declines; when economic growth slows, it increases.
This chart from JP Morgan Asset Management as of a week ago shows the chronic overestimation of operating earnings:
The chart comes via Zero Hedge, where they ask, “Is the market cheap?” My answer is not on the basis of the Shiller PE, which stands at 23.7 versus the long run arithmetic mean of 16.47 or around 40 percent overvalued. Neither is it cheap on the basis of Tobin’s q. Smither’s & Co. has it at 44 percent overvalued on the basis of q, and they note:
As at 12th March, 2013 with the S&P 500 at 1552 the overvaluation by the relevant measures was 57% for non-financials and 65% for quoted shares.
Although the overvaluation of the stock market is well short of the extremes reached at the year ends of 1929 and 1999, it has reached the other previous peaks of 1906, 1936 and 1968.
How about the single year P/E ratio as reported? The S&P 500 TTM P/E stands at 18 versus the long run mean of 15.49. But it’s cool because the “E” is growing, right? Err, no. The “E” peaked in February last year (see Standard & Poor’s current S&P 500 Earnings, go to “Download Index Data,” and select “Index Earnings”). The multiple will now have to expand just to keep the market where it is. You have to do these sort of acrobatics to get it going up:
Margins are now going to bounce free of the wreckage like those few lucky souls who remember to assume the brace position before the plane hits the ground, even though the as reported rolled over a year ago (I hope Denzel Washington is flying this plane).
So how is it cheap?
It’s at 14.5 on the basis of twelve-month forward operating earnings estimates versus a long run mean of 15.49. You gotta do what you gotta do to get the Muppets to buy.
Good luck with that.
Dear Tobias and Marvin, we finally have an updated version of our paper, available here: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2096310
BTW, earnings estimates are lagged by at least 10 days relative to subsequent stock returns.
Best
Sandro.
LikeLike
[…] that right now, there is serious concern that overall S&P 500 earnings estimates are way too high. And as Eddy Elfenbein points out at Crossing Wall Street, recent history shows that estimates have […]
LikeLike
[…] year. It would be no surprise to see analysts lower their expectations. The investment blog Greenback’d posted a very good write-up that showed that analysts are almost always overly optimistic except in recessionary environments […]
LikeLike
[…] Compare this to the Q3 2012 ratio – unfortunately the most recent point – at 100 percent. If we assume ~1 percent GNP growth in Q4 2012 and Q1 2013 (the long-run CAGR is about 1 percent per quarter), and the market has rallied around 10 percent, the ratio now stands at ~107 percent, which is around 40 percent over the long-run mean (since 1949). This level of overvaluation accords with the Shiller PE and Tobin’s q. […]
LikeLike
[…] piece on S&P 500 forward earnings estimates and the overvaluation of the market generated a number of heated emails and comments. I didn’t know that it was so controversial […]
LikeLike
[…] to my point that if your valuation models use forward estimates rather than twelve-month trailing data, you’re do…, here are the results of our Quantitative Value backtest on the use of consensus Institutional […]
LikeLike
I would argue the long run mean for forward operating earnings is much lower than 15.49 due to consitent overestimation of E.
LikeLike
Agreed. The mean for price-to-as reported earnings is 15.49. The mean for price-to-operating earnings is lower, and the mean for price-to-forward operating earnings must be lower again.
LikeLike
Only operating earnings are projected for the S&P. It just excludes things like realized gains since those are tough to project. The avg. forward PE for the S&P over the last 25 years is 14.89x. http://fingfx.thomsonreuters.com/2011/12/16/09414959ae.htm
LikeLike
Estimates for As Reported Earnings Per Share through 2014 are on the Standard and Poor’s website alongside the estimates for operating earnings.
LikeLike
I should have said people look at forward operating earnings and generally that’s it.
LikeLike
[…] Be careful: https://greenbackd.com/2013/03/21/sp-500-operating-eps-estimates-are-too-optimistic-and-the-market-is… […]
LikeLike
I don’t understand why people focus on the Shiller PE. It takes an avg. of the last ten years earnings and Graham liked doing this for companies. But what Graham would also do is throw out outlier years. The Shiller PE does not do this. 2008 should be thrown out and then look at the multiple. Tells a much different picture. The market is driven by GDP growth and inflation. A certain amount of each is discounted into the market, and to the extent that reality differs from expectations, this can also cause moves. For the market to really pull back, earnings need to drop. And for earnings to drop substantially for the majority of companies, either monetary policy or fiscal policy needs to change. So if the deficit were closed too quickly, that would be a negative (Obama has no intentions of doing this). Or if the Fed were to start tightening, this would hurt earnings since 50% of the increased margins are due to lower interest rates (Bernanke has no intentions of tightening anytime soon). I just don’t see what would take earnings down substantially. Certainly when the Fed starts to tighten it will. But that could be 2 years away. Japan will be contained when it blows up. If Europe blows up, they will print or do Eurobonds, which would be great for the U.S. China has plenty of room to ease as their interest rates are much higher. OPEC knows better at this point than to let oil go too high.
LikeLike
AQR’s Cliff Asness has dealt with many of your points in his recent piece, “An Old Friend: The Stock Market’s Shiller P/E.” It’s worthy of a post here sometime. I hear these misconceptions all the time: 1) The Shiller PE doesn’t work; 2) But-for-2008 earnings the Shiller PE would be lower; and, 3) the stock market is driven by GDP growth.
I’ve posted ad nauseam in relation to 1 on this site (see, for example, New global research on Graham / Shiller Cyclically Adjusted Price Earnings (CAPE) ratio). I’ve never seen research that found the Shiller PE is anything but predictive over long periods. Asness backtests it from 1926 to 2012, finding as follows:
Unless you have some research showing a different finding I’m going to continue to use it in estimations.
2 is another simple matter. Anyone can look at the long period of earnings on the Standard & Poor’s website and see that earnings are above trend, and the last decade has been better than usual. Asness has done it in the piece. Asness comments:
In relation to 3, there are piles of papers out there examining the relationship between stock returns and GDP growth. Researchers consistently find that there is little correlation between GDP and stock returns at the market level. The primary driver of stock returns at the market level is valuation. For example, see my piece on Dimson, Marsh and Staunton (Triumph of the Optimists: 101 Years of Global Investment Returns), Bill Bernstein’s site, or Jay Ritter’s piece.
As to everything else you say, I have no idea. Obama’s deficits, Fed tightening, Japan, China and OPEC are unknowns.
LikeLike
1.) The Shiller PE is completely insignificant. It merely shakes out, nothing more. It’s like TEV/revenue. Try to sell a company with minimal earnings off of some TEV/rev multiple and you’d see what I mean. Buyers would scoff at the idea. And people value companies off of forward earnings and cash flow. That’s just how the world works. They make their own calculations regardless of what analysts say.
2.) 2008 was an outlier. Nothing can change that. It makes zero sense to leave it in the calculation. Zero. There really isn’t any more to the argument. It’s just silly.
3.) Of course stock market gains and GDP growth are correlated. Over the last 100 years, nominal GDP growth has averaged 6%. The stock market has seen a similar average return. The 6% nominal GDP growth is made up of 3% inflation + 2% productivity gains + 1% population growth. So 3% inflation and 3% real growth. Real GDP growth can only be higher than 3% from credit expansion. So the economy is essentially driven by the credit cycle. The business cycle, which leads to recessions every 5-8 years, is driven by the credit cycle more specifically.
LikeLike
These are groundbreaking ideas. Good luck.
LikeLike
I think not. It’s common sense. The Shiller PE shakes out. Companies are valued off of forward earnings. That’s just how it is. Leaving out 2008 is a no brainer. And 3.) can be found here: http://www.bwater.com/Uploads/FileManager/research/how-the-economic-machine-works/How-the-Economic-Machine%20Works–A-Template-for-Understanding-What-is-Happening-Now-Ray-Dalio-Bridgewater.pdf
That’s pretty easily the best economic writing I have come across.
LikeLike
Thanks, Jeff. I frankly don’t understand Dalio’s argument. We happen to disagree. That’s what makes a market.
LikeLike
If you are referring to the Great Rotation, that piece has nothing to do with it. That piece discusses how the economy really works.
LikeLike
Just to clarify on why it makes sense to drop 2008. 2008 was a 1 in 75 yr event. It was the product of the end of the long term debt cycle. 2007 on the other hand is more like a 1 in 5-8 yr event. It was more in line with an ordinary business cycle as opposed to the long term debt cycle. So if you include 2002 (a lower bound of the ordinary business cycle), then leaving in 2007 is fine in my book. 2008 was the conclusion of the long term debt cycle and it laid on top of the end of a business cycle. If you read Dalio’s piece, I think it would make more sense. Again, it’s easily the best economic writing I have ever come across. Nothing else really compares.
LikeLike
Tobias, excellent article! I too find the market to be expensive and have had a bit of a tough time finding stocks that I’m comfortable buying right now. I was even planning on writing about it as well. Also like you, market behavior has been nagging my thoughts, but not the behavior of analysts. No, it is retail investors that I think about. Here are some excerpts from a recent memo from Howard Marks:
-Equity mutual funds are seeing only modest inflows, albeit the outflows have topped.
-Many institutions have allocations to equities that are well below the average of the last fifty years, and no one’s rushing to move them up. In other words, I’m comfortable saying attitudes toward equities are characterized by relative disinterest and apathy.
-A move upward can be powered by a switch from the fear of losing money to the fear of missing opportunity. When attitudes are moderate and allocations are low, it doesn’t take much.
I don’t think the “muppets” have yet taken hold of this market and it will only take one more good year, potentially this one, to trigger their return barring a Euro or China disaster. That is helped by the fact that there is no good place to put money right now. The exception might be real estate, but the majority of individuals will not and cannot rush back into that anytime soon.
Markets can remain dislocated for a fair amount of time, and I suspect that we’re only halfway to the path of absurdity. We’ll see. I’ll invest cautiously in the meantime.
LikeLike
Thanks for the comment, James. I’ve got no idea what the market is going to do. It could keep going up for all the reasons you’ve cited (although the “money-on-the-sidelines” argument is a logical impossibility). It could fall over tomorrow. I’m just making the point that the more predictive price ratios (Shiller PE, Tobin’s q etc), which are well above their long-run averages, all disagree with the PEs based on forward earnings estimates from overoptimistic analysts’ forecasts.
LikeLike
I agree. Also, to clarify, I didn’t mean to sound like speculating on the speculations of others is a sound investment idea. It’s simply a mere possibility among an infinite number of others that could prolong overvaluation. Obviously, more and more caution is required as others, in this case analysts, get more optimistic in this challenging environment as you duly noted. Great post again!
LikeLike
According to “Adding Value to Value Is there a Value premium among large stocks” by Sandro Andrade–E/P strategies based on earnings estimates outperform the market and price/book strategy. I believe this is contrary to your own research which concluded that using earning etimates was not a good strategy.
LikeLike
Hi Marvin,
Thank you for your question.
I have a major problem with their backtest method. Andrade et al use “timely available earnings forecasts and stock prices to sort stocks each month” because “using stale information to sort stocks and form portfolios is probably suboptimal.”
We lag the data six months. Why? To avoid “look-ahead” or “point-in-time” bias. From our book:
How big is this impact?
In our investment simulations, we treat the data conservatively to protect against look-ahead bias. We do so by lagging the financial statement data by six months. This helps to ensure that all financial statement reports would have been available at the time the trading decision would have been made.
I wish that Andrade et al had not muddied the results by testing both “fresh” / unlagged data and forward earnings. What are the results using “stale” / lagged data and forward earnings? You can find them in our book. They underperform. What are the results using “fresh” / unlagged data and forward earnings? Andrade et al have tested it and it outperforms. So is the outperformance due to the unlagged data or the forward earnings? No-one knows, but my guess is that there is some look-ahead bias in the results. I’d love to hear from them to explain where I’m wrong.
LikeLike
That’s a variation of a chart originally put out by McKinsey, which found out that analysts are always too bullish, except at recession lkows, where they are too bearish..
Their average estimate for earnings growth is at 12% while long term earnings have grown at 6%.
It says little about market valuation today, and lots about the analyst community’s inherent biases.
See this for more:
http://www.ritholtz.com/blog/2010/06/mckinsey-equity-analysts-are-still-too-bullish/
LikeLike
Thanks, Barry. We use in our book the original chart from Roy Batchelor’s “Bias in macro economic forecasts.” It doesn’t matter which chart we use – McKinsey’s, Batchelor’s and JPM’s charts show identical data – but JPM’s data are up-to-date as at March 14.
My point is that the market is currently way overvalued on a plethora of measures. The only one that seems to point to undervaluation is the forward operating earnings estimate, which is consistently overoptimistic, and more so at market peaks, and so I’m inclined to discount it.
LikeLike
[…] S&P 500 operating EPS estimates are overoptimistic. (Greenbackd) […]
LikeLike