Shiller P/E Comparisons Are Distorted By Buybacks

by: Sovereign Investment Insight
Summary

Buybacks make current CAPE readings non-comparable to CAPE from older years.

Changes in CPI might also inflate current CAPE readings relative to history.

Diminishing equity risk premium and low interest rates could justify higher valuations.

Background

Many of those who are bearish on the U.S. equity markets cite that the current reading of the Shiller P/E is 30, which is well above the long-term average of 17 going back to 1871.

The Shiller P/E or CAPE (cyclical adjusted P/E ratio), aims to determine a P/E ratio that is normalized for the business cycle. The price divided by earnings ratio is heavily affected by the earnings used in denominator, and since earnings swing greatly across business cycle, it appears to have tremendous value to perform an operation that would stabilize earnings in the formula.

The CAPE takes the trailing 10 years of earnings and then adjusts each year's earnings to the present dollar level for inflation per the CPI index, and then takes an average of the 10 years' earnings data in current dollars. The average is then the "E" that goes into the cyclically adjusted P/E ratio.

Distortion

The distortion occurs because the current Shiller P/E is often compared with that of the entire history going back to 1871. However, before 1982, buybacks essentially didn't occur because that year is when the SEC issued a safe harbor to allow for share repurchases.

The CAPE adjusts 10 years' worth of earnings to the present accounting for inflation. However buybacks allow earnings to permanently grow faster than inflation since any share repurchased is permanently removed.

For example, assume 0% annual inflation indefinitely and a company with $1 Billion in net income in the year 2007 and 1 billion shares outstanding in the same year for an EPS of $1.00. The company never grows its net income and never pays a dividend, and so it has $1 billion in net income in every year from 2007 to 2015 for an EPS of $1.00 for each year, so it accumulates a large sum of cash by 2016. It decides to do a massive buyback in early 2016, where it repurchases 50% of all shares outstanding or 500 million shares. In 2016, the stock still produces the same $1 billion in net income but only has 500 million shares outstanding in 2016 for an EPS of $2.00.

Source: Created by author.

Under the Shiller P/E the adjusted earnings per share used for each of the years 2007-2015 would be $1.00. This is distortive because the same earnings power of $1 Billion in net income from 2007-2015, if it occurs in the present, would produce $2.00 of EPS because there are now only 500 million shares outstanding. So CAPE would average 9 years worth of EPS of $1.00 from 2007-15, and then the $2.00 EPS for 2016 to get a total CAPE earnings figure of $1.10. So, if the stock is trading at $22 per share in 2017, the CAPE would be 20. However, if the same earnings power than occurred every year from 2007-16, occurred in 2017, the stock would do $2.00 EPS, yet only $1.10 in EPS is produced by CAPE procedure. Using the $2.00 EPS figure, the P/E only is 11 which is much different from 20.

So the presence of buybacks understates the earnings compared to the prospective earnings power since older years' earnings aren't adjusted for buyback related share count reductions. This isn't a problem in itself, except that CAPE earnings readings before 1982 don't have this effect, so all else equal they will be lower than post-1982 readings. As a result, it becomes problematic to compare CAPE from 1982-present to those before 1982, yet many often try to perform the comparison because 1982 to present only provides a mere 35 years of data rather than 140 years if the entire history of Shiller's data is used, even if the older data is not comparable.

Implications

Obviously the above example is an exaggeration, but the point remains that under a state of buybacks CAPE will be more inflated than it otherwise would be. If the buyback run rate as percent of market cap is assumed to be 3%, as it recently has been, then due to there being a 10-year average and the midpoint is five years, the approximated understatement of earnings can be roughly assumed to be 1.03^5. That means earnings need to be adjusted up by about 16%. As a result of a missing 1.16 factor, we would conclude CAPE readings would be overstated by 14% (1/1.16 -1) in the presence of 3% buyback run rate, all else being equal.

Other Factors That Inhibit CAPE Effectiveness

There are other items that can distort CAPE readings and make them less relevant to be compared to historical averages.

One of them is the CPI, which traditionally would take a constant basket of goods and services and track price increases as proxy for inflation. However, in recent decades it has performed substitution effects. This has the impact of lower the inflation adjustments. There is strong evidence older CPI's overstated inflation by not including substitution, but regardless of whether the changes in CPI method are appropriate, the impact remains. It is generally believed to have reduced inflation readings because of things like substitution effect. If inflation adjustments in recent decades are lower than those from before say 1990s, then current CAPE readings will be higher than those of entire historical average, all else being equal.

Share Issuance

The most commonly critiqued element of the notion that buybacks distort CAPE comparisons is the assumption that share counts are permanently reduced from buybacks because companies issue lots of stocks to management which causes dilution. Yes, this is true, but one would have to argue that this has been occurring at a greater rate than in the past. That's because as long as dilution has always occurred since 1871, it wouldn't distort comparisons to long-term averages.

Reduction of Firm Specific Risk

There may be a thesis for permanently higher P/E ratios. The stock market, like any risky asset, deserves a risk premium compared to risk less securities. Financial theory states the riskier a security, the more of a premium, in terms of expected return, it deserves since investors are assumed to be risk averse. However, indexing has become much more popular in the last couple decades than ever before, since indexes were only first available for investors in 1974.

Back in the day before index funds, investors would have to buy individual stocks, and it was very expensive to diversify since commissions were high. So, investors were exposed to both systematic risk and firm-specific risk since diversification across many stocks was difficult. As more money is invested via index funds, the indexes essentially reduced firm-specific risk so many of these investors only face systematic risk. Since overall risk is reduced a bit, a lower equity risk premium may be required via a higher P/E.

Also, interest rates have been very low by historical standards, which also justifies as higher P/E ratio than historical averages.

From Shiller PE Data http://www.econ.yale.edu/~shiller/data/ie_data.xls

Conclusion

Current CAPE readings are well above the historical average. Some of the impact may be due to the buyback effect, and some of it could be from changes in the CPI. The residual elevation of the current CAPE compared to long-term averages might be attributable to low interest rates and a diminishing equity market risk premium associated with the growth of indexing.

Disclaimer: The information contained in this article is an opinion and neither constitutes actionable investment advice nor is a recommendation to trade any security.

Disclosure: I am/we are long SPY.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.