Historical Data Disproves “Trough P/E Multiple on Trough Earnings” Myth

Doug Kass, a hedge fund manager dedicated to short selling and frequent guest on CNBC, made a call on the air Monday that the S&P 500 could make its lows for the year this week. A bold call indeed, given that Doug is a short seller and has been correctly bearish on the economy’s prospects for a long time. His reasoning is mostly based on extreme pessimism (not unlike in November when we made a short-term low) and low valuations.

Other commentators debate the valuation point. CNBC’s own Bob Pisani made the case that assigning a 7 or 8 P/E ratio (a typical number at bear market bottoms) to this year’s depressed earnings level forecast (currently around $50 for the S&P 500) is reasonable. Pisani concluded that unless you think that earnings in 2009 will be substantially above $50 (which is very unlikely), the market is not cheap because 7 or 8 times $50 is 350-400 on the S&P 500 index, versus today’s sub-700 level.

When Kass was on the air on Monday he rightly suggested that putting a trough P/E on trough earnings is not appropriate, but market commentators continue to insist that is where the market needs to go before a cyclical bottom can be put in.

I have argued against this logic on this blog before (sorry to keep harping on it), but I decided to dig up some evidence on this topic so perhaps we hear less of it in the future. Below you will find the earnings of the S&P 500 relative to the level of the index from 1970 through 1985, a time period that encompasses both the early 1970’s recession and the early 1980’s recession, both if which are similar in depth to what most believe will be our fate this time around.

From this data you can clearly see why everyone is using a trough P/E ratio of between 7 and 8 times earnings (the bear markets bottomed at a 7 P/E in 1974 and at 8 in 1981). The year of both market bottoms is in boldface to show these levels.

The key here is to look at the level of S&P 500 earnings during both 1974 and 1981. Although the stock market traded at the trough P/E ratios during those years, earnings were at record highs both times! The 1974 level of earnings ($9.35) had never been reached before. The same goes for 1981 earnings ($15.18). Therefore, the idea that we take trough earnings and apply trough P/E multiples is simply unfounded if we look at the very data people have supposedly been using.

Not surprisingly, I am far from the first person to point this out. John Hussman, former professor of economics and international finance at the University of Michigan, actually has created a more relevant P/E ratio called “price to peak earnings” which suggests that trough P/E ratios on previous peak levels of earnings are far more reliable bear market valuation tools.

Where would this type of P/E ratio peg the bottom of the current bear market? Well, S&P 500 earnings peaked at $87.72 back in 2006. Multiply that figure by 7.3 and 8.1 and you get a range for the bear market trough of between 640 and 710 on the S&P 500 index. Interestingly, especially given comments from Doug Kass predicting a possible yearly low this week, the index is in the 680’s currently, which is right in the middle of that projected range.

Hopefully actual data is enough to debunk seemingly popular myths about bear market low valuations for the stock market. While this evidence does not make it impossible for the S&P 500 to dip to 400-500, it would make such a move unprecedented in terms of the last four decades of market history, during which we have seen two recessions that are proving to be very similar to this one.

Enjoy this post? Subscribe and never miss another one: RSS | Email | Twitter

12 Thoughts on “Historical Data Disproves “Trough P/E Multiple on Trough Earnings” Myth

  1. Eric on March 6, 2009 at 3:54 PM said:

    I certainly agree that putting trough PE on trough earnings does not make sense. However, do you have any thoughts on that 2006 earnings number? In hindsight, it seems to me it is not a real number, ie it contains a large chunk of financial institution earnings that were clearly bogus. Not sure I have an alternative though.

  2. Rob A. on March 6, 2009 at 11:29 PM said:

    Excellent work and post. This is a pet peeve of mine, too.

  3. Trying to simultaneously predict both the level of earnings and selecting the appropriate multiple for the market in a year like 2009 sounds pretty tough to me. Especially when changes to the index will probably have a HUGE impact on earnings this year – lots of financial companies with negative earnings or no/low growth will be kicked out. At the same time, profitable mid-cap companies will be brought in. Kind of screws things up a bit.

  4. Chad Brand on March 11, 2009 at 3:29 PM said:


    To me, peak earnings do a good job of muting out the excesses of the prior expansion. I say this because each subsequent expansion produces higher earnings than the prior one. Using prior peak earnings assumes that another earnings record is not achieved. Since there are excesses at each peak (maybe not as much as this last one) I think this model makes sense.

    If we used a new record earnings level (which is a forgone conclusion at some point), the case that it includes excesses would be a lot stronger. But that said, using $75 or $80 instead of $88 as a normalized earnings level is conservative and there is nothing wrong with that at all.

  5. Pingback: Does this rally signal the bottom?

  6. 10 Point Must on March 15, 2009 at 8:10 PM said:

    We are not in a recession. It’s a depression.

  7. Pingback: Bear Market Valuation Indicators?

  8. Pingback: Bear Market Valuation Indicators? | Income Trust | Personal Finance | Real Estate SEO

  9. Nathan on April 1, 2009 at 3:40 PM said:


    I’d suggest that for two reasons your trough P/E analysis is not quite “apples to apples,” given that the data represent Dec. 31 closing price and LTM reported earnings.

    First, generally speaking equities trade on future expectations rather than events in the past. To accurately represent S&P500 valuation at any year-end we need to use forward earnings estimates, not reported numbers. This data would not necessarily yield the same conclusion. I don’t have access to these historical projections, but let’s consider the implications.

    For instance, take the ’73/’74 crash, which began Jan. 11, 1973. This Time article was published Jan. 8: http://www.time.com/time/magazine/article/0,9171,910532,00.html. Needless to say forward expectations were optimistic until days before the crash. If you were calculating the forward multiple at year-end 1972 it would be considerably lower than 19.1x due to grossly inflated expected earnings. Similarly, the forward multiple at year-end 1974 would in all likelihood be well above 7.3x, as mid-crash expectations for 1975 would probably be south of the actual post-crash report of $7.71. The upshot is that when using the appropriate valuation approach, your conclusion may not hold.

    Second, you compare a relative multiple (“trough P/E”) to a nominal price (“trough earnings”), and conclude that they are not simultaneous. This is not an apt comparison because P/E is independent of inflation, while nominal earnings are not. Furthermore, from a valuation standpoint the market will pay less for high reported and projected earnings if it fears high inflation. In the two “trough P/E” years, 1974 and 1981, inflation was 11% and 10% respectively – near peak for each timeframe.

    – Nathan

  10. Chad Brand on April 6, 2009 at 1:20 PM said:

    I don’t know if historical earnings estimates for that long ago are available. I suspect not, but would be happy to be proved wrong. Without them though, we can’t really test your theory.

    As for inflation’s effect on P/E’s, I think interest rates are more relevent there. Whether investors choose stocks or bonds has a lot to do with expected return, so if interest rates are low, stocks will be more attractive than bonds even with high inflation.

    It’s not a perfect model for sure, but I think the data does debunk the idea that we should think fair value on the S&P 500 today is 350-400 ($50 times 7-8).

  11. umm, looks like this fund manager guy was right. time for a prospectus!

  12. Pingback: Typical Stock Market Valuations at Secular Bear Market Bottoms – P/E Ratio and Earnings — The Contrary Investing Report

Post Navigation