Stocks and Shares

Strip out Amazon Inc. and US stocks look cheap

There are many stock valuation ratios. None is perfect, and all can give false readings at times. Right now, I believe the elevated Cyclically-Adjusted P/E ratio (CAPE) – also known as the Shiller P/E or PE10 – is giving a very false reading. This is for multiple reasons. Once these are taken into account, the S&P 500 index looks cheap.

Back in August, I took a look at how the S&P 500’s trailing price-to-earnings ratio is skewed upwards by relatively few, high P/E stocks with big market capitalizations. A particular culprit is Amazon Inc. (see here).

Taking out that one stock alone brought the index P/E ratio down from 22.9 at the time (now 21.7) to 16.6. Stripping out the whole tech sector as well brought it down further, to 14.4. That was below the historical median of 14.7 since 1871.

Today, I’m going to focus on a variant of the P/E – the Cyclically-Adjusted P/E ratio (CAPE). The current reading is 29.7. That’s 79% above the (arithmetic) mean since 1881 of 16.6. It’s also 90% above the median reading of 15.7.

For reference, the all-time low was 4.8, reached in December 1920 (more recently, it bottomed at 6.6 in August 1982). The all-time high was 44.2, reached in December 1999.

At the current level, the headline CAPE is at the 90th percentile of all readings since 1971. That means it’s been lower in nine months out of ten since the dollar was unpegged from gold. On the face of it, that looks pricey.

Below is a chart that I did previously. It shows the distribution of the monthly CAPE readings between January 1971 and August 2017. The current position is just below the red dot (the green dot is the median, or mid-point of all readings). Everything to the right of the red dot relates to the late ‘90s bubble.

A normal P/E divides a price (the numerator) by earnings per share, or EPS (the denominator). CAPE does something similar, but with applying a twist to the EPS denominator.

A standard trailing P/E uses one year of earnings. This is problematic from time to time, since corporate earnings may be abnormally low (leading to an inflated P/E reading) or abnormally high (leading to a depressed ratio). CAPE seeks to solve this problem by looking at earnings over a period of 10 years. Thus, in theory, it captures a whole business cycle or two.

The way it’s done is to take each of the EPS figures for the past 10 years. These are then adjusted upwards for inflation, since the year they were reported, so that they’re expressed in today’s money. The 10 results are then averaged to give a measure of “normal” earnings power. When the technique is applied across a broad index of companies, such as the S&P 500, the “normalness” should be even greater than for a single company.

Most of the time…

However, the current US CAPE is pushed artificially upwards by a number of factors. These are:

  1. Amazon skew
  2. Recent reduction in US corporate tax rate
  3. Effect of net stock buybacks
  4. Depressed bank earnings during and after the global financial crisis (GFC)

Let me explain each of those, starting with the Amazon skew. At an index level of 2,651, the S&P 500’s market capitalization is $23.5 trillion. That’s the free float value of the shares of all the companies included. Amazon’s free float is $683 billion (84% of total company market capitalization), at a stock price of $1,663.50. That means Amazon is 2.9% of the index.

I’ve looked at Amazon’s EPS for the past 10 years and worked out a CAPE ratio (adjusting each year’s EPS for inflation and taking the average). Since Amazon made no money to speak of until very recently, the CAPE comes to a massive 458.

Backing out Amazon from the index brings the CAPE of everything else right down to 16.9, a drop of 43%. Put another way, imagine if Amazon went bust tomorrow and the stock went to zero. An S&P 500 index investor would lose less than 3% of their investment, but the CAPE would plummet by 43% overnight.

Next is the effect of the cut in US corporate taxes this year. I’ve previously estimated that this boosted post-tax profits by about 11% (see the earlier link above). That means nine years of EPS used in the CAPE calculation is understated. Applying this adjustment reduces the headline CAPE by 9%.

The third adjustment is the effect of net buybacks. Since the early 1980s, US corporations have made bigger cash distributions in the form of stock buybacks, and reduced the share paid out as dividends. Although combined cash distributions have remained around the same level.

However, offsetting the increased buybacks there’s been higher stock issuance, particularly related to executive stock and option programmes. The result is a reduction in share count of about 1% a year for the index (note: much lower than the gross buybacks of about 3% a year).

That reduced share count boosts EPS over time. More importantly, it means that the past EPS figures used in the index CAPE ratio are understated when compared with today. Making this adjustment reduces the headline CAPE by 4.2%.

Finally, bank earnings were abnormally depressed during and after the GFC. Since banks are a big sector, this has a big effect on the index EPS. Real (inflation-adjusted) EPS for the S&P 500 was $102 in 2006, $80 in 2007, just $18 in 2008, $60 in 2009, $89 in 2010 and $97 in 2011.

So I’ve made an adjustment to strip out this extreme event, albeit a crude one. I’ve increased the real EPS to $90 in each of 2008 and 2009, and $95 in 2010 – more in line with the pre- and post-GFC levels. This reduces the headline CAPE by 9.7%.

Below is a summary of the effect of each adjustment:

Also, here is a summary of the effect on the CAPE of all four factors, or excluding any one of them at a time.

Combining all the factors reduces the CAPE by a massive 55%, to just 13.3. That’s only in the 26th percentile since 1971, meaning the CAPE has only been lower a quarter of the time since the dollar supply was free to inflate.

Now, I know that this is not a perfect analysis. I also know, in an ideal world, that I’d be able to adjust the whole historical time series for these (and other) factors. But I still strongly believe it shows that the headline CAPE figure of 29.7 is wildly overstated.

Just stripping out Amazon alone, let alone a handful of other bubbly stocks, takes it right down to 16.9. That’s 14% below the median of 19.7 since 1971. Taking account of all four factors slashes the CAPE to 13.3, which is 32% below the median.

Does this mean the S&P 500 is much cheaper than it looks? Certainly. Does it mean that it’s actually pretty cheap on an absolute basis? Probably.

But I have a couple of caveats. With wages and interest rates rising, not to mention trade disputes hitting top-line sales of many companies, profits could come under pressure once more.

Despite those factors, on balance, the S&P 500 actually looks like pretty good value. Which is not the same as saying it can’t fall sharply in the short term. That’s especially true if most investors stay fixated on the artificially high, headline valuation ratios, such as the unadjusted CAPE.

Stay tuned OfWealthers,

Rob Marstrand

Previous ArticleNext Article
Rob is the founder of OfWealth, a service that aims to explain to private investors, in simple terms, how to maximise their investment success in world markets. Before that he spent 15 years working for investment bank UBS, the world’s largest wealth manager and stock trader with headquarters in Switzerland. During that time he was based in London, Zurich and Hong Kong and worked in many countries, especially throughout Asia. After that he was Chief Investment Strategist for the Bonner & Partners Family Office for four years, a project set up by Agora founder Bill Bonner that focuses on successful inter-generational wealth transfer and long term investment. Rob has lived in Buenos Aires, Argentina for the past eight years, which is the perfect place to learn about financial crises.