Beware of Seemingly Reasonable P/E’s on Growth Tech Companies

Pop Quiz:

Do all technology companies expense stock-based compensation in their financial statements? Perhaps more importantly, do sell-side analysts include such expenses in their quarterly earnings estimates, on which every quarterly report is judged by Wall Street?

Given that stock-based comp has been a hot button accounting issue for a couple of decades, and the chief accounting rule board (FASB) required GAAP financial statements to include such expenses way back in 2004, I suspect that most investors are not really paying attention to the issue anymore.

Since I am a value-oriented investor, most of my investments are outside of the high growth tech sector, where most of the stock-based compensation resides. Nonetheless, a few months ago I wanted to dig a little digging because I did not understand why market commentators in the financial media were seeming to understate the P/E ratio of the S&P 500. I have been closely watching S&P 500 index earnings for most of my career, so it struck me as puzzling when people on CNBC would claim something like “The S&P 500 trades at 17 times earnings, which is only modestly above historical averages.” In fact, the numbers I saw on the actual Standard and Poor’s web site showed the P/E to be more like 19 or 20x. Given that the historical average is around 15x, there is a big difference between 17x and 20x. So what the heck is going on?

It turns out that there is a large financial data aggregation company called FactSet, which supplies many investors with earnings data on the S&P 500. You can find their earnings data directly on their web site. After reading through it I realized that FactSet was showing higher earnings levels for the S&P 500 (which equates to lower P/E ratios by definition), and that is where the market commentators were getting their valuation information. For instance, the current FactSet report shows that calendar year 2016 earnings for the S&P 500 are projected at $119, which gives the index a trailing P/E of 19.3x. However, the S&P web site shows a figure of $109, which equates to a trailing P/E of 21.1x. Investing is hard enough, but now we can’t even agree on what earnings are? Maybe I’m making a big deal out of nothing, but this is frustrating.

The logical question I needed to answer was what accounted for the gap in earnings tallies. If earnings really were 9% above the level I thought, my view of the S&P 500’s valuation would undoubtedly change. I was shocked when I learned the answer.

It turns out that FactSet’s earnings data does not represent the actual earnings reported by the companies comprising the S&P 500, which is what the figures on the S&P web site show. Instead, FactSet uses the reported earnings that match up most closely with the Wall Street’s analysts’ quarterly forecasts. Put another way, if the analyst community excludes certain items from their earnings estimates, FactSet will adjust a company’s actual reported earnings to reflect those adjustments (for an apples to apples comparison to the Wall Street estimate) and those earnings figure are used when they tell the investment community what the earnings for the index actually are. If this sounds bizarre to you, it should.

Having followed the market for my entire adult life (and all my teenage years too), I immediately knew what accounted for much of the gap between these earnings estimates. Most technology companies still to this day report non-GAAP earnings results right along side GAAP figures in their earnings reports. For reasons I don’t understand (since the issue of whether stock compensation is an actual expense was resolved years ago), the analyst community excludes these expenses in their numbers, so when a tech company reports earnings, the non-GAAP number is comparable to the analyst estimate. As such, the non-GAAP number is incorporated into FactSet’s data. So whenever a stock market commentator quotes the FactSet’s version of the index’s P/E ratio, they are inherently ignoring billions of dollars of employee compensation that is being paid out in shares instead of cash.

To illustrate this point, Consider Google/Alphabet’s fourth quarter earnings report from last night. The analyst estimate was $9.44 per share and Google reported $9.36 per share. So today’s media headlines say that the company “missed estimates.” If you read the financial statements carefully you will see that Google’s GAAP earnings were actually $7.56 per share. The non-GAAP earnings figure, which is the ones that is reported on because that is how the analysts do their projections, was a stunning 24% higher than the actual earnings under GAAP accounting rules.

You can probably guess why there was such a large gap. During the fourth quarter alone, Google incurred stock-based compensation expense of… $1.846 billion! Multiply that by four and Google’s run-rate for stock compensation is $7.4 billion per year! That is $7.4 billion of actual expenses that are being excluded from FactSet’s earnings tally, and that is just from one company (albeit a big one) in the S&P 500 index.

So how does this impact investment decisions? Well, there are many people that look at Google and see an $850 stock price and $34.40 earnings for 2016 and conclude that the stock is quite reasonably priced given the company’s growth rate, at less than 25 times trailing earnings (850/34.40=24.7). Of course, the actual P/E is 30.5x because when you add back stock-based comp Google’s earnings per share decline from $34.40 to $27.85.

The valuation differentials get even larger when you consider younger, smaller technology companies because these firms seem to be addicted to stock-based compensation. Google pays out a lot in stock, but even that $7.4 billion figure is only 7% of the company’s revenue. Paying out 7% of sales as stock compensation is indeed a very large figure, but other tech companies dole out far more.

I looked at some other fairly large ($5-50 billion market values) tech firms and the numbers are staggering. During the first three quarters of calendar 2016, (CRM) paid out 9% of revenue in SBC, but that seemed quite low compared with some others. Zillow (ZG): 13%. ServiceNow (NOW): 23%. Workday (WDAY): 24%. Twitter (TWTR): 26%. Can you believe that some tech companies pay a quarter of revenue in stock-based compensation? Not total compensation, just the stock portion!

Importantly for investors, these companies are getting very large valuations on Wall Street. In fact, those five tech companies have current equity market values that cumulatively exceed $100 billion. I wonder if investors might view them a little less favorably if they realized they might be less profitable than the appear on the surface.

For me, the takeaway from all of this is that all investors should dig deeper into valuations in general. Don’t just take figure you hear on CNBC or read in press releases as gospel. Just because a web site says a company has earnings of X or a P/E ratio of Y does not mean there isn’t more to the story.

6 thoughts on “Beware of Seemingly Reasonable P/E’s on Growth Tech Companies”

    1. No, I don’t know of any database, screener, or anything like that. I just go through the financial statements/quarterly press releases every quarter.

  1. I struggle with this issue. If the companies did not pay SBC, they would have to pay more cash. If you paid compensation strictly in cash, there would be no issue but with SBC, not only does GAAP accounting require deducing SBC but you also have to factor in diluted shares to get EPS. So it seems to me that there is double-counting when it comes to SBC. Therefore, I do not deduct SBC in the income statement but I do use fully diluted earnings. SBC is a non-cash charge but on an economic basis, I account for it in the diluted share count. I would be interested in others’ ideas and thoughts.

    1. Here is the problem with that, as I see it:
      For Q4 2016 Google had GAAP profit of $5,333 million and non-GAAP profit of $6,593 million, for a delta of $1,260 million due to stock comp. The fully diluted non-GAAP EPS is $9.36 (vs $9.56 basic). The GAAP diluted EPS is $7.56 (vs $7.73 basic). Regardless of which share count you use, excluding stock-based comp is increasing EPS by at least $1.63 per share. To correct for that, your method says “I know that $1.63 is a real expense, and I will adjust my EPS number by $0.20 (9.36 instead of 9.56) to account for that. So, if I am understanding you right, you are essentially counting 12% of stock comp as an expense and ignoring the other 88%. I’m not sure that makes for a substantial enough adjustment.

  2. That’s a very fair point. Perhaps the difference lies in the fact that the diluted share count affects earnings each subsequent year and the diluted share count will undoubtedly go up as well every year.
    The way I look at it is that the ultimate “cost” to the shareholder stems from the share count dilution as no actual cash changes hands.
    Admittedly, I am not confident that my approach is without legitimate criticisms. I almost feel like I am trying to calculate the value of an option before Black-Scholes. I don’t feel like I have a comprehensive understanding/solution and I haven’t seen anything that clearly explains it. But that is my best attempt and thanks for your comment.

  3. For example, one might try to use the fully diluted share count and then say deduct one-half SBC, but that is not a satisfying approach and there is not a good theoretical basis for it. The problem with fully deducting SBC and dividing the market cap by the fully diluted shares is that it appears earnings are punished too much and consequently, an investor relying on that may not be properly valuing the opportunity set or making optimal decisions. Perhaps it is better to err on the side of prudence but I struggle with this issue.

Comments are closed.