Blogiversary Themes: Analyzing Earnings

Since I started regular posting in December of 2005, I am going to declare this to be my five-year blogiversary.  The posts were a little sporadic in that first month, but I was trying hard to explain my purpose.  Here is a quotation from an early article:

It isn't difficult to convince investors that stock prices are volatile. Something less known, though, is that corporate profits fluctuate even more. Using a ratio of these two volatile measures is sure to give you misleading signals now and again.

The misleading signal often comes near the end of recession, when earnings plummet. During the 1991 economic recession, for example, the composite earnings of the companies in the S&P 500 fell by 27 percent to about $16 a share. Their stock prices actually rose during that period, as investors looked ahead to better days. The P/E multiple quickly soared to 26 from 15, as earnings fell and prices rose. Since a P/E of 26 is high compared to long-term averages, investors focused on this ratio would have likely missed the beginning of a nearly decade-long rally.

The P/E multiple is much lower now, even though we are at the end of a recession, so the advice should carry extra weight.  Regular readers know that I have taken a consistent and disciplined approach to fundamental market valuation, so I suspect that few will be surprised at this quotation.

Here is the surprise:  The quotation is from John Hussman!

My own take in 2005 was a bit different.  I questioned his failure to incorporate changes in interest rates as part of his valuation model.  Here was my position:

Unfortunately, his method ignores interest rates.  In a couple of the low PE examples he gives, like 1974 and 1982, stocks should have a low PE since interest rates were in double digits.  Just ask yourself: Would take an 8% return from stocks if bonds had a rate of 12%?  What if the bonds were 4%?

Experts may disagree about how to incorporate interest rates into analysis, but there should be no debate about whether to do so.

 The Hussman approach to fundamentals still does not recognize the importance of interest rates as an alternative asset class.

Five Years Later

During the ensuing five years I have written many articles about stock valuation, often emphasizing the differing time frames.  My position, recorded weekly, has changed with the circumstances.  Quite frankly, I expected John Hussman to be a big winner at the end of the recession.  His "peak earnings" concept was going to come into its own.  To my surprise, instead of cashing in on his theory he did something that I abhor.  He arbitrarily discarded some of his data, the time period that did not fit his multi-year market opinion.

Let's be completely clear about this.  Anyone who reaches far back into the past has many time periods of data that are dissimilar in many ways.  If you take some recent period that you do not like and throw it out, meanwhile not giving the same examination to the rest of history, it is cherry-picking.

We can all make post-facto decisions to throw out data that we do not like, but that is not part of the researcher's creed.  I now see an article where he has used multiple variables to find something that he calls a "Who's Who of Awful Times to Invest."  Big problem!  He has changed the variables from his original 2007 article, which you can check out here.

It is easy to look at past data and a large number of variables and back-fit a result, especially if you keep changing the independent variables.  If you have a big staff it is even easier.  The real test is whether the methods that were specified in advance -- that would be the methods written about in 2007 -- actually worked.  It is perfectly acceptable to change methods if circumstances warrant, but you should not pretend that it is simply an "update" of a past approach.

An Alternative Viewpoint

The Hussman viewpoint is one of the most popular reads for individual investors and for investment advisors.  By email I received an update from Georg Vrba, P.E.  Georg has written some fine articles drawing upon a number of variables to improve on the buy-and-hold approach.  Here is an example.  Georg kindly mentioned my article demonstrating why an increase in the forward P/E multiple is merely a return to normal.

I hope that Georg will publish his results soon.  To summarize, his methods confirm the first nine Hussman examples but disagree violently with the current case.  How interesing!

A Warning

I know that my readers include many Hussman fans.  I also enjoy reading his analysis.  This is really about methods and changes over time.  I am stating in advance that I am too busy with year end analysis and forecasts to do a detailed debate in the comments.  I will read and note arguments.  Next year I'll take another look, but I am not going to engage in instant rebuttal.

 

Comments

[DE]

Hussman notes that his recent update includes "equivalent" criteria. I think he has scraped the price/peak earnings, haven't heard much about that in while. Also the 4 year high in the S&P 500 index is obviously not in effect currently. I agree that he is data mining, but at the end of the day he is trying to justify his overbought/overly bullish/overvalued/rising yields syndrome concerns.

[oldprof]

Mike C -- Here are two suggestions if you are going to defend Hussman:

Explain his change in methods and throwing out data that he does not like.

Explain the scenario that would lead to "another brutal" decline. Do you predict earnings of $50 on the S&P? A P/E multiple dropping to 7? Interest rates going to 14? Oh -- I forgot. That is not relevant in the Hussman/Shiller approach. Rates don't matter.

I look forward to your answers to these questions, but I am NOT going to respond to a lot of obscure arguments that adduce even more variables to explain why we should ignore the method Hussman advocated in 2005.

Aren't you even interested in the question I originally posed? Why not respond to that?

Just a thought.

Jeff

[Mike C]

I hope that Georg will publish his results soon. To summarize, his methods confirm the first nine Hussman examples but disagree violently with the current case. How interesing!

Thanks for the Georg link. I downloaded the pdf and am looking forward to studying it.

Perhaps the situation above points that just one or a few variables are what are really relevant to the outcome?

I've been thinking a lot about this as to me one of the highest value adds one can bring to the table is effective market timing along with individual stock alpha.

Here is an interesting thing I think. The bear markets in the 80s and 90s were fairly mild compared to the 50% drop in 00-02 and 60% drop in 07-09. Perhaps Hussman's valuation metric and Shiller's P/E are ONLY RELEVANT in telling you how far a decline might go once an actual bear decline starts, but that they are useless for gauging the duration and magnitude of the upcycle.

In other words, maybe other metrics right now would indicate a run to 1500-1600 is likely (just hypothesizing), and that the Shiller P/E simply indicates once a decline starts it is likely to be yet another brutal one?