Why ETF Price/Earnings Ratios Lie

Be careful when making fruit-basket comparisons; you’re likely to come up with lemons.

Last week, one of my analysts, Boris Valentinov, penned a blog titled “Brace For Irrational Exuberance 2.0.” In it, he looked at the price-to-earnings ratios of various ETFs, noting in particular that the iShares Russell 2000 ETF (IWM | A-79) had a surprisingly high P/E of 83. He went on to explain that the ratios get high because some of the companies in the index are actually losing money, and that somewhat paradoxically has the effect of raising the portfolio P/E.

The email traffic we received for the article wasn’t particularly kind. Most of it fell into one category: “You’re an idiot. IWM’s P/E is 28, so your entire argument is invalid and how can I ever trust anything you guys say ever again?”

So I felt it worth going into all the reasons people have fallen in love with P/E ratios, and why they’re almost always wrong, especially when you’re comparing ETFs.

First, why do we even look at P/E ratios? Back in the day, when people focused mostly on single stocks, price-to-earnings was the go-to measurement of growth. The theory was that if you were buying a stock to receive a future dividend stream, that dividend stream would have to come from earnings.

If you could buy a lot of earnings for not so much money, you had a better chance of continuing to get that dividend stream in the future. So, low P/E stocks were seen as “safe,” while high P/E stocks were seen as speculative.

The dirty little secret of investing—well one of them—is that of course these kinds of shorthand metrics are massive oversimplifications. JPMorgan Chase and China Fruits Corp. each have a P/E of 10, but the prospects for a $5 million fruit beverage company that sells in China and a quarter of a $1 trillion global financial company are hardly related. Still, if you’re trying to separate out different baskets of stocks for further analysis, P/E is a convenient first place to start.

The biggest problem (aside from the fruit versus bank problem) is accounting. Earnings are far from a pure, or even accurate, representation of what’s really going on in a company.

Companies regularly manipulate their reported earnings numbers through all sorts of shenanigans, whether it’s the rash of manufactured goodwill and bartered income that confused investors in the ’90s, or the magical disappearance of toxic assets from bank earnings statements following the Rule 157 changes coming out of the financial crisis. Earnings itself is a polluted number.

But it gets even more complicated when you start working with portfolios of stocks, instead of individual companies.

If I put China Fruit and J.P. Morgan into my portfolio, we could probably both agree that the P/E of the portfolio should be about 10. But how we come to that 10 is important. You could just make a weighted average of the two P/Es, giving J.P. Morgan 99.9 percent of the weight and China Fruit its tiny amount.

But what if instead of China Fruit, our two-stock portfolio had a company that was losing money?

According to Bloomberg (and a lot of financial data providers), no company in the world has a negative P/E. Not even Twitter, which, despite being a $29 billion company that loses $500 million a year, earns a negative P/E. Instead, it just gets an “N/A,” even though my 10-year-old can divide $29 billion by -$525 million and get -58.

So if you use Bloomberg’s P/E numbers and add Twitter to your J.P. Morgan stock, what’s your P/E going to look like? 10. And what if you add 1,000 other money-losing companies? 10.

It is, frankly, insane, and yet that’s the convention of many very smart companies and analysts. Not everyone just ignores negative P/Es, of course. Some assign artificially high P/Es to any company losing money. The point is that everyone does it differently. For a deep dive, check out our guide to negative P/Es written by Paul Britt in April.

So is there a “right” answer here? A rebuttal to the folks who thought Boris was an idiot? Sort of. In my opinion, how we do things here is about as not-bad as it gets. We take all the money earned by all the companies in a portfolio—positive and negative—and use that as the divisor against the total market cap of the portfolio (all weighted by the positions value in a portfolio, of course).

That way, the Twitters offset the J.P. Morgans, and you create a kind of “what if this was all one company” view of the real earning power of the entire pool of stocks, relative to how the market is valuing them.

It’s fine if you don’t like that way of comparing ETF stock portfolios. But whomever you lean on for data, the most important thing is to make sure you’re comparing two identical methods of calculation.

At the time of this writing, the author held no positions in the security mentioned. Contact Dave Nadig at [email protected] or follow him on Twitter (which he doesn’t own either) @DaveNadig.