Some Wall Street analysts are playing games with the S&P 500’s price/earnings ratio.
These analysts are calculating the S&P 500’s SPX, -0.09% P/E ratio on the basis of what they think corporate earnings will amount to in the coming calendar year. Nowadays, that means their P/Es are based on 2019 earnings — a number that won’t actually be known for another 20 months or so.
Such calculations are profoundly speculative, needless to say. Think about all the things that could sabotage analysts’ estimates. The still-nascent global trade war could escalate and bring about a worldwide recession, for example. Geopolitical tensions could break out on the Korean peninsula, in Europe, or Russia — to name three current hotspots. How confident can you be in estimating the S&P 500’s earnings-per-share in 2019?
This focus on 2019 earnings is conjecture enough, but these same analysts compare forward-looking P/E ratios with historical P/Es that are calculated on the basis of actual, as-reported (i.e. trailing) earnings. Since forward-looking P/Es are invariably lower than ratios calculated based on trailing earnings, their comparisons make the market look less overvalued than it really is.
To be sure, forward-looking P/Es are always speculative. But at some times of the year the extent of analysts’ forward-looking focus is the next 12 months. In the summer, focusing on the next calendar year expands considerably the extent of speculation.
Related: U.S. economy seen busting out with 4% GDP, but it won’t mean squat without an encore
Contrast how analysts calculated forward-looking P/Es in January of this year with how they do so today. At the beginning of this year, most forward-looking P/Es were based on forecasted earnings for calendar 2018 — the subsequent 12 months, in other words. Today, in contrast, many analysts have shifted to calendar 2019 when calculating forward-looking P/Es — a period that won’t even begin for another six months, won’t end for another 18 months, and not be known until earnings season ends during the first quarter of 2020.
How big a difference does it make to extend one’s speculative horizon so far into the future? To estimate I went through two steps.
Step 1: Past versus future
The first step was to measure the historical difference between P/Es based on trailing earnings, next-12-month earnings, and earnings measured over the 12-month period that begins two calendar quarters hence. The table below shows the medians of these three different P/Es since 1965.
Median since 1965
Based on trailing 12-months
17.5
Based on next 12 months
15.8
Based on 12-month period beginning two quarters hence
14.7
In other words, by simply shifting one’s focus from the trailing 12 months to the next 12 months, the median P/E falls by 10% (17.5 to 15.8). And by shifting to the 12-month period that begins two quarters hence, the median P/E falls by an additional 7% (15.8 to 14.7). Putting both of these shifts together: An analyst who today focuses on 2019 earnings should be expected to come up with a P/E that is 16% lower than another analyst who focuses on trailing 12-month earnings.
No wonder the bulls like to focus on forward-looking P/Es.
Step 2: Hindsight bias
Significant as these changes in median P/Es are, they understate the extent of downward bias that is occurring due to analysts’ shift to focusing on calendar 2019 earnings. That’s because the medians reported in the table above were calculated on the implicit assumption that analysts would have perfect foresight into what earnings would be in the future. Of course, they don’t possess such perfect foresight. And far more often than not analysts are too optimistic, which means that their forward-looking P/Es — biased downwards anyway — are almost always still too low.
How rosy are analysts’ forecasts? For an accurate historical estimate, I would need to know what the consensus earnings forecast was at each point in the past, and I don’t have the requisite data. But we can estimate.
Consider, for example, what analysts were estimating the S&P 500’s as-reported EPS would be for calendar 2017. Their consensus estimate at the beginning of 2017 turned out to be nearly 10% higher than what transpired. As a direct consequence of their rose-colored glasses, their forward-looking P/E ratio was 15% higher than it should have been. I have no reason to believe that this number is out of line with what it’s been historically.
The bottom line? Combining the numbers I reached in steps 1 and 2, it’s possible to estimate the magnitude of the bias of an analyst who focused on forecasted earnings over the 12-month period beginning two quarters hence: We’d expect this analyst to come up with a P/E that is about 30% lower than one who focused on trailing-12-month as-reported earnings.
Sure enough, that’s close to what we’re seeing now (at least in terms of order of magnitude). The S&P 500’s current P/E, based on trailing 12-month as-reported earnings, is 24.3, according to S&P data through the end of 2018’s first quarter. In contrast, the ratio is just 17.3 when calculated on the basis of estimated 2019 earnings. The first is 40% higher than the second.
To be sure, nothing in and of itself is wrong with focusing on forecasted forward earnings rather than on trailing earnings. But it’s crucial to compare whatever P/E you arrive at with historical levels that were calculated in the same way: Forward-looking speculations should be compared to previous forward-looking speculations — not trailing 12-month facts.
For more information, including descriptions of the Hulbert Sentiment Indices, go to The Hulbert Financial Digest or email mark@hulbertratings.com . Create an email alert for Mark Hulbert’s MarketWatch columns here (requires sign-in).
