Why the CAPE multiple is fatally flawed

The Cyclically-Adjusted PE (CAPE) multiple – usually calculated as the market price divided by 10yr average inflation-adjusted EPS – is fashionable amongst value investors. It is particularly popular as a way to value broader market indices. Unfortunately, it has become a deeply flawed and misleading measure.

Once upon a time it made a lot of sense. That was the time when stock buybacks were a rarity and corporates returned cash to shareholders almost exclusively by way of dividend payments. However, today, stock buybacks are quite common – particularly in the US. Probably not coincidently, the US is also the market where the CAPE multiple is most frequently cited as an argument for why the broader market is overvalued. 

It is fairly easy to demonstrate why buybacks are a major problem for CAPE-based valuations.
Consider a situation where a mature company trades at $15, earns a stable, cyclically-neutral $1 a year for ten years (lets ignore inflation), and pays out 100% of its earnings in dividends. In 10yrs time, the company's cyclically-adjusted 10-year average EPS would be $1, and if the stock was still trading at $15, the stock’s CAPE multiple would compute at 15x.

Now, let us suppose the same company, with the same cyclically-neutral operating results, elected to forego paying dividends and instead use its earnings to buy back shares. If the stock traded at a constant 15x earnings throughout this period, the company would be able to retire 6.7% of its shares annually (1/15). With constant operating earnings, EPS would therefore compound at 6.7% a year from $1 in year one to $1.79 by year 10. The company's average EPS during this period would compute at $1.38, and at the end of the period, at 15x year-10 EPS, the stock would trade at $26.81.

The latter is the same multiple of the same, unchanged operating earnings as in the prior example (the market cap would also be the same). But the CAPE multiple would now compute at 19.7x, not 15x! Someone using the CAPE multiple would conclude that the stock was trading almost 5x multiple points above its long term average, but it would be an utterly erroneous conclusion.

What is happening here is that in the latter scenario, returns to shareholders are being compounded explicitly instead of implicitly. In the first example, shareholders were free to reinvest their 6.7% cash dividend yield into further stock on market. If they did so, at the end of the 10 year period, they would own more shares with a constant per-share EPS/DPS, whereas in the latter scenario, investors would own the same number of shares with higher per-share earnings and dividend capability. Either way, returns would be the same, but under the former scenario, the stock price chart would be flat and commentators would bemoan a 'lost decade' of returns, whereas in the latter scenario, the share price chart would marched steadily higher in a long bull market, and commentators would bemoan an extended bull market unwarranted given the lack of organic profit growth. The bears would point to CAPE multiples being stretched, but the apparent overvaluation is a total mirage.  

Given the popularity of buybacks these days – a rational popularity given their tax-preferred status over distributions to shareholders by way of dividends – CAPE measures need to start to be adjusted for buy-backs or they will lead to systematically flawed estimates of the market’s fair valuation, and in particular, they will tend to indicate the market is more expensive than it really is.