top of page

ETFs and Risky Information GA(A)Ps

ETFs are all about aggregating individual tickers to track an index or rules-based strategy. While the ETF industry gets a bum rap for producing too many small funds, it may actually be some of the largest and most successful ETFs which introduce unintended effects.

As ETFs become the language of equity investing, it’s important that investors don’t lose the ability to measure absolute and relative values. During a recent weekend’s reading, I came across at least a dozen “current” P/E (price-earning) ratios for the S&P500 and the Nasdaq-100. Articles advocating the steady course or #BTFD reported low and unremarkable levels (probably based on projected earnings), and articles predicting Armageddon reported bubble-like values (probably based on trailing earnings).  Because the P/E is casually tossed out as a simple single number like the temperature, few outlets bother with the critical data details (12mo trailing, forward-looking, CAPE, non-GAAP, etc.) and mislabeled P/Es are worse than no P/Es.[1]


Since the late 2016 market run started, commentators expecting broad regulatory, trade, and tax reform from the new administration are now rationalizing lofty market levels with strong Q1 and Q2 earnings – if that’s reasonable, let’s make sure we know the numbers and what’s behind them.

P/E Ratios.

Smart investing is all about price, and P/Es are useful for both comparing investments and assessing market cycles. If rising valuations are supported by robust (real) earnings, P/Es will remain within a comfortable range, and if valuations are rising through enthusiastic multiple expansion, P/Es signal caution – that’s the way it’s supposed to work, but only if the data is available, robust, and properly identified.

Ask a professional investor for the current Apple or Facebook P/E and you’ll probably get good guess. Ask the same professional investor for an index or ETF P/E and you’re likely to get the last number they read, and they won’t know what’s behind it.

Are P/E Ratios obsolete?

There is a debate about whether today’s P/Es are not comparable to historical values. The logic suggests that conventional metrics do not apply to today’s asset-light disruptors which have the potential to scale operations at low marginal costs. A reasonable response is that markets ultimately care about value and cash flow, and many of the most enthusiastically followed disruptors have never made a profit (to say nothing about the potential for the disruptors to be disrupted) – except for maybe Amazon, there is usually some expectation of earnings.

The historical debate around P/Es is more complex than I’ve presented, but my real fear is that ETF data presentation may lull investors into a state of indifference or complacency on market valuations and metrics.

ETF Disclosure Doesn’t Help….

The official P/E of the Nasdaq-100 is under 22 (based on 12-month trailing earnings and the appropriate index weightings), however

… if one takes the trailing 12-month (TTM) P/Es, and simply weights them by the index percentages, the result is more like 48 – more than twice the

official figure?

… and Amazon (6.78% at a P/E of 246), Broadcom (1.47% at a P/E of 190), and Netflix (1.1% at a P/E of 218) contribute almost 22 P/E points themselves and there are around 100 more tickers to add!

One. Change the P/E data to produce a lower result

Assuming one uses the individual ticker TTM P/Es, the appropriate calculation would seem to weight each P/E by its index weighting and sum the result (unfortunately without the negative P/Es).

Instead, one of the many approaches used in P/E presentation is “weighted harmonic averaging”. Weighted harmonic averaging is a technique used to lower the impact of high P/E stocks in the index (?) :

-         convert each P/E into its reciprocal. Apple’s 18 becomes 0.055, and Amazon’s 246 becomes 0.004,

-         then average the reciprocals, then take the reciprocal again…. an Apple/Amazon (assuming 50/50 weight) average would  be 132, but with the magic of harmonics it becomes a cool 33.

Assuming that the P/E isn’t just convenient window dressing, I cannot see how reciprocal averaging makes sense in equity valuations.


Two. Omit Company P/E Data


Whether one is looking at a sector or an ETF, it seems sensible to use an aggregate price and the aggregate earnings, rather than to take the individual P/Es of each company. Using individual ticker P/Es means that one has to strip out those tickers with negative P/Es – again, if one omits or massages information for the convenience of presentation it’s not beneficial to investors and traders.

As a test, I took the Nasdaq-100 2017-Q1 and 2017-Q2 earnings for each of the constituents, and computed a P/E with current price levels. This tests the market’s rationalization that today’s daily record setting valuations are fully supported by 2017 earnings. Also, adding the earnings together properly takes into account any loss making companies.

The Nasdaq-100 P/E based on 2017 trailing earnings is approximately 27 – lower than the actual TTM of 48 because more recent earnings have been valuation supportive, but higher than the official number of 22. If 27 is the correct number and the market thought they were trading to a 22 P/E, equity prices would be almost 20% too high – but obviously there are many more factors at play.

Three. Institutionalize Non-Standard Financial Reporting

As calculated above, only 3 high P/E constituents can blow-up an index’s valuation presentation – but it should be of interest to investors to know that their ETF actually has a very high P/E on a TTM basis.

Amazon wears its high P/E multiple with pride, but many public companies go to great lengths to bury their enthusiastic valuations. Companies like Salesforce, Netflix, and Tesla are in hundreds of ETFs, and despite their decades long run as public companies, they’ve each run decades long GAAP (and actual cash) losses, while being heralded by equity analysts for (really expensive) growth, and well managed non-GAAP “earnings”. Companies reliant on non-GAAP presentation aren’t bad, it’s just important to know the reality.

If we bury valuation data that’s inconvenient to an ETFs presentation, we’re making markets and investors less smart. Massaged data can create market problems if investors are buying or selling on skewed data – here is an area where the industry could contribute to inflated valuations.


The debates surrounding the effects of passive indexing and ETFs will continue for some time; classic battle lines are unintended correlations, excessive trading, and “tail wagging the dog” impacts. These debates are more about market structure issues and of only passing interest to most ETF users.

As ETFs become a condensed language of investing and position-taking, there’s a risk that ETF data techniques promote erroneous signaling – markets will be presented as cheaper or richer than they actually are. Even more important, if ETF aggregation causes P/Es to be massaged and stabilized, investors will have no reason to look under the hood.

Some believe that the ETF revolution has made expert fundamental and data analysis obsolete. Independent analysts and index providers should take note that there are real opportunities to improve valuation measurement and metrics and that the interested audience is larger than ever.




[1] “it ain’t what you don’t know that gets you in trouble, it’s what you know for sure that just ain’t so” – Mark Twain

bottom of page