Société Générale’s Andrew Lapthorne lays out the danger of relying on P/E ratios.

“The essential message is that although one is of course simply the inverse of the other, using P/E ratio instead of earnings yields can give dramatically different results when making historical valuation comparisons.

This disparity between average P/E and average E/P is greatest where profits have a tendency to head towards zero. This has the effect of sending the P/E ratio out towards infinity, before ultimately profits turn into loss and then P/E becomes negative and hard to work with. In 2009 for example the P/E on the S&P 500 hit 146x, and as anyone who has tried to build a market valuation model for Japan will attest, the P/E ratio is not much use.

For example, taking the same price and earnings history from Robert Shiller, the average S&P 500 P/E can be either 16.5x when calculated using the average P/E, or 14x when the average is calculated more correctly by using earnings yield. This represents a massive 17% difference – see below. In the same vein therefore a dividend yield versus a history model will always produce a more pessimistic outcome than a P/E model.”

Another way to think about **Earnings Yield **– it allows a better reflection of *companies with no earnings*, and further allows a market analyst to incorporate periods when earnings are negative.

The difference between the two: The current **P/E ratio **of 15.4 looks historically cheap –a bout 28% undervalued; However, using **Earnings Yield** , the market appears fairly valued. The current yield is 6.5%, the average 6.6% making SPX 0.1% overvalued. (P/E lower is cheaper; Earnings Yield higher is cheaper)

>

*Click to enlarge:*

*
Source:* Societe Generale Research

Quant Quickie, 21 March 2012

Category: Analysts, Data Analysis, Earnings

Please use the comments to demonstrate your own ignorance, unfamiliarity with empirical data and lack of respect for scientific knowledge. Be sure to create straw men and argue against things I have neither said nor implied. If you could repeat previously discredited memes or steer the conversation into irrelevant, off topic discussions, it would be appreciated. Lastly, kindly forgo all civility in your discourse . . . you are, after all, anonymous.

As a finance outsider, I always wondered why the main valuation indicator for stocks was upside down. It makes it difficult to compare to other yields, for example, bond yields.

I’d find a graph of CAEP more useful than CAPE. I would like to see it plotted against interest rates, bond yields, inflation expectation etc… You could also plot the spreads between these.

Please note that the area between the PE line and the average for the past 15 years of PE in the CAPE graph is enormous compared to the areas for other 15 year periods. The equity bubble fo the past decade or so has actually significantly influenced the calculation of the mean so that our current PE (or EP) numbers look more “normal” than they would have a decade or two decades ago. I think the average PE ratio for the past century has actually gone up by about 1 over the past decade even as we go thorugh a secular bear market.

[...] Why using P/E ratios can be misleading. (Big Picture) [...]

Salient point, but there’s more reasons why p/e is a dangerous metric, such as diffrences in deprecation and financial engineering that occurs within companies(e.g. taking on leverage to buy-back stock)

My beef with PE ratios is that earnings used are either past earnings, telling you nothing about the future, or are ouija-board earnings estimates that are almost always wrong. The price, is of course, immediately visible, so that PE using past earnings tells one how a stock is valued now relative to past earnings. Not terribly helpful when assessing future prospects.

But I think that free cash flow does a better job of more clearly indicating the true profit picture — although it too, suffers from not being great in the predictive department.

hmmh… I would have thought the correct way to calculate a market P/E is total earnings / total market cap. Then it is consistent with earnings yield.

ratios are misleading if they’re calculated by innumerate people.

(sorry for the snark LOL!)

esp. since I reversed and gave formula for earnings yield LOL

Anna, you wrote:

This is based on 12 months trailing earnings, isn’t it? So, SPX is much cheaper today than it was in March 2009 when SPX was allegedly much more overvalued according to this metric, although SPX has more than doubled since then?

Anna writes:

“The current P/E ratio of 15.4 looks historically cheap”

Isn’t that comparing apples and oragutans? Doesn’t Barry consistently warn about using stupid current or forward P/E’s to value the market compared to TRAILING historical P/Es? And she does it using a 1yr number when CAPE is an inflation adj 10-yr number. Or am I mistaken that BR warns about this often enough (or is it another writer I’m thinking of)? Regardless, it is a useless and stupid comparison.

You can solve the problem by using the median P/E, rather than the average. Since taking the reciprocal reverses the ordering, with the middle being the same in either way, the median(P/E) = 1/median(E/P). The median could be thought of as the most typical P/E (or E/P). 50% of the the time the P/E is higher and 50% of the time lower it is lower.

The biggest issue with earning yield for me is that it invites comparision the ten year treasury. The ten year has a duration of roughly 9 and the s&p around 50. So what’s the basis of comparing the two. I see none.

constantnormal, you wrote:

Depending on what PE ratio you are talking about. I think PE based on 12 months trailing earnings is quite useless because of the high volatility of earnings within the business cycle. On the other hand, CAPE seems to be a quite reliable statistical predictor for the longer-term (about 10 years) average stock market returns, i.e., it can be helpful to decide if you want to get in for the longer run. It’s not as useful for short-term market timing purposes, though.

The average PE is a useless number.

It might be of some value if there were a tendency for the PE to converge on the average PE.

But if you do a histogram of the historic PE you find that the chance of the PE being at any number between 10 and 20 is about the same. Actually, it is at 15 less than any other number between 10 and 20.

Look at the chart you just displayed. How much time does the actual PE equal, or is near, the average PE?

Hardly ever. Maybe for about one month out of every decade.

spencer, you wrote:

I disagree. The average (or as someone else suggested taking the median instead) is useful as a reference to determine whether stocks have high valuations or low valuations compared to historically averaged (median) valuations. No convergence toward the average (median) needed for this.

Spencer and Rootless:

I prefer to look at the upper and lower quintiles instead of the average.

Entering those quintiles for an extended time is an indicator that the bear or bull market is on its last legs.

I like PEs but prefer to look at downside risk vs. upside risk. The higher the PE value, the greater the downside risk, lower divvies etc.

Right now the market is a terribly bad place to be imo price wise. You get a ton of downside risk, very little yield (no I don’t buy divvy increases nonsense) and are waaaay late to the party. I also don’t like stocks as an inflation hedge although they certainly beat most other classes in that regard.

[...] we trotted out SocGen’s chart showing markets are not cheap, by way of Earnings Yield (Why Using P/E Ratios Can Be Misleading). GMO’s James Montier peak profit comments were widely circulated — he discussed how [...]

[...] Why Using P/E Ratios Can Be Misleading [...]

[...] Why Using P/E Ratios Can Be Misleading | Societe Generale I always use Earnings Yield instead of PE for two reasons: it accounts for negative earnings, and it’s easier to compare to interest rate/bond yields. [...]

Statistics 101: The way you calculate the average of ratios is to calculate the average P and the average E and then divide: Average P/E = Average P / Average E. This way you get the same result whether you do it via P/E’s or E/P’s.

(It’s probably futile to try to explain why to an economist. We can calculate the arithmetic mean from anything, but it only has a meaning and is useful under certain conditions: If there is a well-behaved underlying distribution of values, the mean is an estimate of “the center” of this distribution. Ratios, even in the best of worlds, follow a Pareto distribution which happens to have no center, and what ain’t can’t be estimated. But one can estimate the distribution center of prices and that of earnings. And there is no rule against calculating the ratio of these averages and call them the ratio of average prices and earnings or in short the average P/E).