Debate over the “Shiller CAPE” normalized P/E ratio, popularized by Nobel Prize winning economist Professor Robert Shiller, has been particularly pointed this year. As we’ve discussed in the past, this measure takes the average of 10-year trailing earnings for the “E” in the P/E ratio in order to smooth out and eliminate the volatility present in typical expressions of simple trailing 12-month S&P 500 earnings. Supporters of the widely followed valuation measure correctly point out that future 10-year total index returns have a statistically significant relationship to CAPE valuation levels. Detractors have leveled a number of charges varying from complaints that 10-year trailing earnings unfairly capture the massive, and in their view highly unusual earnings cliff dive that occurred in ‘08/’09 (but ignoring the fact that earnings were overly inflated ahead of the crisis), to the fact that the supposedly mean reverting ratio has remained above mean for the vast majority of the past two decades undermining the credibility of the indicator.
A particularly interesting argument against Shiller’s version of the normalized P/E came from a blog by the name of Philosophical Economics back in December. The blogger pointed out that changes to the treatment of “goodwill impairment” when calculating GAAP earnings had created real problems in terms of the consistency of Shiller’s earnings data over time. The blogger demonstrated this issue by comparing GAAP earnings since the early 2000s (when the rules changed) with Bloomberg’s S&P 500 earnings calculation of scrubbed non-GAAP earnings from continuing ops and showing that the deviations between the two series since the early 2000s had increased dramatically since the shift. We thought this was a particularly compelling argument and therefore used the alternate Bloomberg earnings data series in the same manner as the above blogger to reconstruct a historical normalized P/E ratio to create a more consistent measure.
First, we’ll take care of a few housekeeping items. Since the Bloomberg pro-forma data only goes back to 1952, and considering that deviations between pro-forma and GAAP are much, much small prior to the goodwill changes, we took a bit of “artistic license” and added Shiller’s publicly available data for the period from 1925 to 1952 so we could get a longer term view incorporating the bubble of the late 1920s and the subsequent depression years. Now that that’s out of the way, here is the chart:
Now that we’ve addressed inconsistency due to GAAP changes, what do we learn from the current normalized P/E levels using the alternate “pro forma” data? We see that the real (inflation-adjusted) P/E ratio currently sits at 22x versus the 25x level of the traditional Shiller P/E. This compares to a long-term average of approximately 17x and a long-term median of 16.5x. While not as overvalued as the traditional Shiller P/E, we see a market that’s still decently above historical levels. How does this translate into a prediction for future returns? The 10-year future annualized return forecast comes in at 3.54% for inflation-adjusted total returns (i.e. with dividends), 6.38% for non-inflation-adjusted total returns, and 4.47% for non-inflation-adjusted returns without dividends. This compares to historical averages of 5.71%, 8.85%, and 6.63% respectively. These numbers, therefore, look better going forward than those produced by Professor Shiller’s calculation. Still, we’d say at current levels, even with the GAAP adjustments, there’s a good chance that market returns over the next 10-years will be substandard relative to historical averages.
Regarding the criticism by some that the ratio has remained above median for a good part of the past 15 years, we see looking over the future returns streams that the above average P/Es resulted in sub-average returns. The massive spike associated with the late 90’s tech bubble? That spike resulted in outright negative 10-year annualized nominal and real returns. How about those elevated P/E ratios observed in the period from roughly 2003 to 2007? So far, the 10-year annualized returns that have been realized from 2003 and 2004 onward have been roughly in-line with the sub-par results P/E ratios would have predicted. Take September/October 2003, for instance, when the normalized P/E was trading at approximately 22x, the current observed level: 10-year future total real returns were 4.5% per annum, over a percent lower than historical averages and nominal total returns were 6.9%, approximately 2% lower per annum than historical average. The pattern has continued. And, don’t forget that there was one heckuva wild, dangerous ride during that 10-year period, with the S&P 500 falling over 60% peak to trough at one point during the financial crisis.
There are some other interesting historical observations worth pointing out, in our opinion. Look at the Great Depression period and the secular bear market associated with that era. Leading into the ’29 crash, valuations were extremely elevated. At the beginning of the Depression, valuations dipped into the single-digits at a velocity similar to the velocity of valuation compression we’ve observed at various points over the past 14 years. Over the five-year period from 1932 to 1937, however, markets recovered materially and valuations actually pushed back above 20x to a level almost exactly where we currently reside. Sure enough, a relapse sent markets and valuations tumbling from overstretched highs back towards cycle lows.
Of note, the next single-digit valuation low in 1942 proved, though, to be one of the better buy-points of all time. Buying at the valuation low in 1942 would have produced double-digit annualized returns over the next 25 years. Further showing how important starting valuation is, buying the market at the height of the Depression in 1932 when valuations hit a modern low produced double digit total returns over the ensuing 33 years until the next valuation peak was reached in 1965. Similar observations and outcomes arise out of the valuation lows of the 1970s and early 1980s.
What’s the point of all this? Set expectations appropriately for US returns and make sure you’re keeping your eye open for opportunities elsewhere. They do exist. Again, at current valuations, we’re hard-pressed to believe that US returns will be able to approach or exceed long-term historical averages over the next decade. On the flip side, many other markets overseas hold much more attractive valuations in the low to mid teens. In fairness, just because US valuations are elevated doesn’t mean that total market Armageddon is around the corner. There are a lot of breathless, hyperbolic bears out there right now. But, thinking that a decade of 80’s and 90’s like equity returns are in our future isn’t a very prudent way to approach US markets under current circumstances.
