Dumb Alpha: Trailing or Forward Earnings?
Value investing is a form of creating dumb alpha. After all, in its simplest iteration, value investing boils down to selecting stocks with the lowest price-to-earnings (P/E) ratio or lowest price-to-book (P/B) ratio. What could be simpler than that?
But alas, the financial industry, in its eternal drive to make things more complicated than they need to be, has managed to confuse investors about this concept as well.
All too often, investment reports or the financial media maintain that a stock is attractively valued because its forward P/E ratio is such and such. These kinds of statements have become a pet peeve of mine. Why? Because very few people ever stop to check if forward P/E ratios are actually predictive of future outperformance.
All the studies on the value factor have been conducted with trailing P/B and trailing P/E ratios, not forward P/E ratios. Recognizing this difference is critical to becoming a successful value investor.
Picking Value Stocks Based on P/E
To demonstrate how reliable P/E ratios are as a source of dumb alpha, I took the current index constituents of four major stock market indices: the S&P 500 in the United States, the FTSE 350 in the United Kingdom, the Euro StoXX 300 in the eurozone, and the Nikkei 225 in Japan. For each, I chose the 20% of stocks with the lowest P/E ratios and compared their average returns to the 20% of stocks with the highest P/E ratios going back monthly for the past 20 years. I first used trailing 12-month P/E ratios for each stock and then switched to forward 12-month P/E ratios. For the latter, I followed common practice and used the P/E ratio calculated with Institutional Brokers’ Estimates System (I/B/E/S) consensus earnings forecasts.
The results of the exercise are instructive: In the United States, the cheapest quintile of stocks based on trailing P/E outperformed the most expensive by 1.2% per year. When using forward P/E ratios, on the other hand, the most affordable stocks underperformed the priciest by 1% annually. In other words, relying on forward P/E ratios destroyed performance!
In the United Kingdom, the picture is not quite as bad. Sorting stocks based on trailing P/E led to a 10.1% annual outperformance by the cheap stocks, while sorting based on forward P/E created a 7.4% outperformance. But here again, employing forward P/E was significantly less successful than trailing P/E.
In the eurozone, based on trailing P/E, inexpensive stocks did 4.6% better per year compared to 3.5% for forward P/E. And finally, even in Japan using trailing P/E led to a 6.6% annual outperformance by the cheapest quintile of stocks compared to just a 0.6% outperformance with forward P/E.
Forward P/E Is Useless: Discuss
These exercises demonstrate that, regardless of region, relying on forward P/E ratios as a value indicator is less effective than trailing P/E ratios and can sometimes completely destroy the value premium investors seek to harvest. The same picture emerges when risk-adjusted returns are the measure of success or if country or sector indices are employed instead of single stocks.
Why don’t forward P/E ratios work as a measure of value? The answer is rooted in analysts’ estimates of future company earnings. As I demonstrated in the second installment of this series, analysts are terrible at predicting interest rates, exchange rates, or stock market performance over the coming 12 months. And they are similarly inept at predicting company earnings. In fact, using trailing 12-month earnings is typically a better predictor than analyst-estimated forward earnings. This is why trailing P/E ratios do a better job at selecting value stocks than forward P/Es.
What explains the lack of accuracy in forward earnings estimates? Quite simply, analysts are overly optimistic. Forward earnings are, on average, about 10% higher than subsequently realized earnings. However, this excess of optimism is not stable over time or across stocks. Paul Hribar and John McInnis show that analyst overoptimism rises when investor sentiment is buoyant — particularly for growth stocks with hard-to-estimate future earnings. And recent research by Ulrike Malmendier and Devin Shanthikumar demonstrates that while analysts on average are overly optimistic, those affiliated with an underwriter of a company’s securities tend to be strategically overoptimistic and thus systematically distort their forecasts.
Of course, Ben Graham, the founder of value investing, knew this a long time ago. In The Intelligent Investor, he wrote, “While enthusiasm may be necessary for great accomplishments elsewhere, on Wall Street it almost invariably leads to disaster.”
In the spirit of dumb alpha, then, we can say that simple trailing P/E ratios are far better value indicators than forward P/E ratios. Or as I tell my colleagues at work: “Never ever use forward P/E ratios. Ever.”
For more from Joachim Klement, CFA, don’t miss Risk Profiling and Tolerance: Insights for the Private Wealth Manager, from the CFA Institute Research Foundation, and sign up for his regular commentary at Klement on Investing.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©iStockphoto.com/CSA-Archive
29 thoughts on “Dumb Alpha: Trailing or Forward Earnings?”
Key word is “forward”.
We cant predict the future and never will.
Thanks Joachim – nicely written!
There’s another serious problem with forward PE ratios. Apologists usually use a reasonable-looking forward PE to justify an obviously high trailing PE. Now suppose that the necessarily high earnings growth does actually materialize AND the trailing PE does become what the forward PE had anticipated. Then by basic algebra, the total return on the stock is zero (or, more precisely, just the dividends earned over the period, but if it was a high PE share, chances are that the dividends were negligible anyway).
My empirical work on South African equities since the 1960s is in broad agreement with your findings.
Lucas – in my experience, PEG is even worse than forward PE. It’s often used in practice as (price:forward earnings/expected earnings growth), so the probable error in earnings expectations is actually compounded.
There is not just one agreed upon PEG formula. There are several. Some use trailing PE; some use forward PE. Some use 1-year projected growth; some use 5-year projected growth. The point of PEG is that a low PE (trailing and/or forward) by itself might not indicate a value stock, but actually a failing company. Since PEG relates the PE to growth prospects, it would seem to be a more, not less, accurate estimate of value than (any) PE alone. You said, “In your experience”. How has your experience shown that PEG is worse. I mean it does not require experience to suppose a relationship where two of the three variables (P, E, G) are future guesses might be inaccurate. Could you clarify, please?
Great post. We wrote a paper in the JPM http://www.iijournals.com/doi/abs/10.3905/jpm.2012.39.1.112 that made this point as well. Sorting stocks on forward E/P actually underperforms the market, on average. And yet, nobody seems to pay attention, as most “value” based index methodologies include some form of forward looking price/fundamental. Crazy.
Interesting. Did you use GAAP/GAAP-equivalent or “ex-items” trailing EPS?
GAAP equivalent EPS
Investing in low P/E stocks (regardless trailing or forward) is not necessarily a correct strategy in the first place. A stock’s low P/E might arise from investor’s pessimistic view on the company.
This is exactly what value investing is about. If Mr. Market is pessimistic about the future he is willin to sell you a perfectly good stock at a low price.
He never talked about if it was the right strategy or not. He only talks about if trailing will outperform forward. Kindly read better.
I thought Intrinsic value of stock is calculated using discounted cash flow (dcf) and PE or forward PE can as be used as a proxy and nothing more. dcf and PE are poles apart. Please dont confuse everyone.
I would argue the opposite. DCF is so fraught with instabilities and highly sensitive to input parameters that it is effectively a garbage in, garbage out model to determine something that is called value. I generally do not rely on DCF because it is unreliable and highly sensitive to estimation errors.
Maybe I should write a dumb alpha article about that one day 🙂
The “usefulness” of forward PR depends on 1) realistic growth and 2) politics. When the realistic / actual growth is above trend or accelerating, the corporate earnings are be more likely to beat investors’ expectations which will lead to outperformance. In other words, multiple periods with different earnings growth profile requires to validate the usefulness of forward PE. Politics and business relationship have been a key driver on analysts’ recommendation and hence earnings forecast of listed companies. The recommendations on corporates listed, especially in many emerging countries, tend to have more BUY than SELL. The reliability of earnings forward have been tinted by politics and business relation, not to mention we cannot predict the future.
Then what about PEG? There seem to be two or three different formulas for PEG. Which one do you think is the most useful?
Can’t say because unfortunately, I never looked into this metric specifically.
Going into the 1987 crash the S&P 500 PE on trailing earnings was 21 –my model said the PE should be 14. But people kepi telling me that it was not a problem because EPS was forecast to be up 33% in 1988.
Guess what, they were right about 1988 EPS, but it did not prevent the 1987 crash.
This article starts off on the wrong foot: “Value investing is a form of creating dumb alpha. After all, in its simplest iteration, value investing boils down to selecting stocks with the lowest price-to-earnings (P/E) ratio or lowest price-to-book (P/B) ratio.”
Just no. Value investing is about finding underpriced securities. It is not about simply low valuation ratios. MPT junkies and product creators turned this into sifting the stock universe by these variables, but that is not one and the same.
Current constituents -> survivorship bias! Using accurate point in time data could completely change your results.
Unfortunately it does not. The point is not whether the index constituents have changed in the past. You could take any random selection of stocks, hold it constant and do the exercise all over and the results remain the same.
How would a strategy that relies on foresight — selecting stocks based on the ratio of price to next year’s earnings — do? If it does very well then it suggests that a price-to-forward-earnings strategy could do well if earnings forecasts were improved.
New to all of this – what do you compare the trailing P/E with then? Is is vs a historical range?
Thank you Joachim.
I don’t think your method contradicts using forward ratios as a time series for one stock. Comparing current forward ratio to its historical average and assuming mean reversion seems to me as a valid check for under/overpricing. What are your thoughts?
You mentioned that there hasn’t been a study on Forward PE. Have you seen this?
Apologies — pasted the wrong reference.
Reposted below, would love your take
You mentioned that there hasn’t been a study on Forward PE. Have you seen this?
Benji Graham did always defend this approach and the useless of foreasting. Backtesting and normalized earnings are powerful tools for value investors. The industry always try to make things complex. Great post.