A Case for Undervalued Equities: Earnings & Interest Rates

A Case for Undervalued Equities: Earnings & Interest Rates

Published June 28th, 2010 in Economy

Tags: bonds, dividends, earnings, earnings yield, guest post, interest rates, P/E, Price earnings ratio, secular bear market, valuation, Wayne Whaley.
The following is a guest post by Charles H. Dow Award winner, Wayne Whaley (CTA) of Witter & Lester. If you would like to be privy to his daily market comments and model ratings via daily email, free of charge, email him at wayne@witterlester.com with the subject title “ADD ME TO DAILY EMAIL”.

Valuation models, particularly those that rely solely on trailing earnings, are suspect as market timing tools, as the market has historically shown the ability to stay over or undervalued before returning to historic norms. With that said, there are some relevant observations we can glean from the relationship between the ‘earning’ of bonds and equities.

Converting P/E’s to Earnings Yields
The earnings yield is a theoretical construct if we imagine that all earnings are paid to shareholders in the form of dividends. It is a valuable tool because it has measurable correlation to the prevailing interest rate at any point in time. The earning yield is calculated by dividing the earnings by the S&P or if you already have the P/E ratio, you can simply invert the P/E to get E/P.

Comparing Earnings Yield to Interest Rates
I have seen the earnings yield compared to T-bills, T-notes and T-bonds. Since the spread on these three securities has varied tremendously throughout history, I will avoid getting into yield curve analysis for this study and simply compare it to what I call an average interest rate (AIR), which is the average yield for T-bills, T-notes and T-bonds. A large positive difference would intuitively suggest equities are cheap and a negative gap the reverse:

Click to see larger chart in a new tab:


Observations
The current spread is 2.74. It will go to some number between 3.0 and 3.25 as soon as we plug in a new number for second quarter earnings. That will be the largest spread since mid-1980.

The highest spread between earnings and AIR occurred on September 1974 with a spread of 6.95 and the S&P 500 index at 63.54. One year later the S&P was 50% higher (83.87). The lowest spread (-3.63) was at the end of the third quarter in 1987. We then proceeded to have the infamous 1987 crash with the S&P 500 index lower by 23% over the next quarter.

The current price-earnings ratio is 17.63 and headed to about 16.3 in a couple of weeks when we get a good handle on second quarter earnings. Some proponents of the secular bear market case argue that based on historical precedent, the bear market will end when the ratio is at single digits. If that is the case, it will have to be several years from now when rates are much higher than today.

This research has relevance not so much because it argues that stocks are cheap, but because it reestablishes the fact there is a relationship between the general level of interest rates and earnings.

Final Thoughts
Some would say that the above research is based on trailing earnings and the current earnings will never hold water down the road. As well, if you go back far enough in time (the 1950’s and the 1920’s), the interest rate to earnings relationship falls apart.

These points are well taken and I put very little weighting on valuation in my trading, because as I mentioned, the market can be overvalued for a long time. Valuation was negative for much of the 1990’s, yet the market tripled in that decade. In 1974, this valuation relationship peaked at 6.95 before having a 50% rally, however, the indicator was over 1 for the three previous quarters prior to that while the market dropped an additional 34%.

The valuation model that I use in my personal work takes a similar approach to the above but smooths the data a bit more and then attempts to extrapolate the past earnings into the next four quarters taking into consideration some economic variables. Valuation can give you a sign that your pitcher is in the late innings, but you should wait until the market starts having trouble finding the strike zone before you yank him. Conversely, the current market’s relative attractiveness doesn’t mean we have to go up this month, but it suggests that eventually the odds favor a resolution to the upside.


=============================
Rod
Jun 29th, 2010 at 8:47 am
Sorry Waine, I tend to agree with you but this time I cannot disagree more. What you are proposing is not too different from a metric known as the “Fed Model”. It is a monumental failure. It signaled stocks were expensive in 1981 (right at the start of a 20 year bull market) and cheap in the middle of 2007 (6 months before a -57% bear market). This has been documented elsewhere, both by bloggers and academic research.

The valid point you make is that, in general, valuation is not a good timing device. But I believe there are some studies that can help to identify a generational buy point. I think there is value in statistics like “Tobin’s Q” and Shiller CAPE (Cyclically Adjusted Price-Earnings).

Rod
Jun 29th, 2010 at 8:53 am
I forgot to mention that according to Shiller CAPE the market is definitely NOT undervalued. We are moving towards undervaluation, but it will still take a few years. Same conclusion looking at the Dow priced in Gold.

Wayne
Jun 29th, 2010 at 9:23 am
Rod,
Yes, The work in this post is a very close proximity to the Fed Model, it makes for interesting conversation but is not practical as trading model. In fact, you would do well to flip it upside down and trade it. Much as Shiller does, In my actual model, I attempt to smooth the earnings, albeit not over 10 years. The observations that can come out of the study are that.

1) there is a relationship between interest rates and equities. It is not 100% correlated, but the relationship would make it difficult to justify single digit P/Es with interest rates at current levels and

2) The spread between rates and earnings is largest since 1980. It that bullish or bearish is debatable

A really astute Quant would also point out that a lot of the research is based on ratios and when either the denominator or numerator approaches zero, strange things can happen to the relationship, such as is the case here since interest rates are bounded by zero and earnings can go negative.

As I said, the post is hopefully simply some research to converse on.

Mike C
Jun 29th, 2010 at 7:43 pm
Firstly, I do APPRECIATE the work. That said:

Valuation models, particularly those that rely solely on trailing earnings, are suspect as market timing tools, as the market has historically shown the ability to stay over or undervalued before returning to historic norms.

I ABSOLUTELY AGREE especially when by “market timing” you are referring to periods of say 6-12 months. Which means the analysis above is more or less academic with not much practical utility.

I think valuation is helpful for the investor, not trader, who isn’t looking to “time the market” but instead put money to work with the objective of seeing a decent return over say 7-10 years which is about the time frame it takes for valuation to mean revert. One could have made good money buying the market in 1999 and selling it in 2000 when the trend broke down, but a buy and holder for 10-years hasn’t made diddly squat. They need to buy stocks at 1979 valuation levels, not 1999 valuation levels.

Here is a different take on valuation that I think is more relevant for a buy and hold time frame

If you want to buy and hold at a good valuation and expect a decent 10-year return wait for 800-850.

I think one has to match the tool to the timeframe. I don’t see a point in forcing the valuation issue to try and make 1-year ahead decisions.

GreedsGood
Jun 29th, 2010 at 11:21 pm
I use a similar model to gauge the comparative valuation of major asset classes versus equities. The most important piece of this puzzle is what the fair “spread” or equity risk premium (ERP) should be.

For the base yield, I use a hybrid 10-year treasury, 10-year AA and AAA/BAA spread to arrive at this number.

I adjust the historical ERP (~4%) by the projected growth rate and earnings consistency as well as the current market volatility level. Based on current metrics, I could make a case for a significantly higher ERP and thus a lower P/E multiple (or conversely equity earnings yield)

Praveen
Jul 1st, 2010 at 7:32 pm
Here is an interesting take on the market valuation based spreads between 10 year treasury bonds and earnings yields.

Mike C
Jul 1st, 2010 at 8:30 pm
Interesting link.

The model says stocks were the MOST OVERVALUED in 60-years from 1980-1982. How did stocks do in the subsequent 5-10 years from those starting dates? What does that say about that model?