Despite the fact that variations in the Price/Earnings ratio have historically explained the vast majority of short term S&P returns, the difficulty of modeling it meant that the vast majority of market participants have adopted a mean reverting assumption for it. Unfortunately, the time it takes for PE mean reversion to occur tends to be measured in decades.
I have noted in the past that, based on my own work, various macro variables suggest that we are likely to see a secular decline in PE ratios. The list of factors includes long term factors like demographic shifts, high debt levels and low growth rates. Beyond making sense, various studies have suggested that there are statistically significant relationships between those factors and the PE ratio. (A society with more retirees than workers will dis-save, decreasing PE as fewer dollars are available for purchasing stocks. High debt and low growth both suggest lower PE as earnings growth is lower and capital leaves for higher growth assets / countries) In the US, these factors all suggest that the decline in PE ratio could persist over the next decade.
Until this year, this thesis has broadly played out. On a trailing basis, the S&P PE ratio has declined from 21 in 2002 to 17.5 in 2008 to less than 13 by 2Q this year. However, since then, the PE has steadily increased, and has made the first higher high last month in a very long time. In fact, over the past year, almost the entirety of the year’s returns has been driven by expanding PE.
As a result of this unexpected turn of events, I have been reconsidering my original thesis. I have been thinking about some possible rebuttals to the original thesis given the macro backdrop. Here are some ideas – readers are welcome to come up with their own:
1. Possible mitigating factors for the demographics effect: The very unequal wealth distribution means that equity holders are much more likely to be well off and thus do not need to sell their stocks for lower risk assets likes bonds or cash. Furthermore, the introduction of Medicare in 1965 may have offset the need for seniors to save for medical costs, which represents a large component of post-retirement expenses and mitigates dis-saving. Finally, the QE programs conducted on a global basis has pushed more people into higher risk assets. High yield bonds are trading rich to equities for the first time in many decades, which has prompted additional inflows into stocks.
2. Possible mitigating factors for high debt & low growth: QE programs have eased the real cost of debt while flooding the system with liquidity. There is simply more dollars floating around now per dollar of earnings. Meanwhile, the fact that the slow growth environment is global may have mitigated country level rotation.
The unfortunate fact is that, like the original hypothesis, these counter arguments are all also themselves hypotheses, and more frustratingly, very difficult to substantiate. As a result, even if they are accurate in the interim, it is not clear if they will persist. All we can be sure of is that the PE ratio has hit a non-recession-distorted higher high since 2003. Therefore, we must consider the possibility that this decade long trend in lower PE’s is over, even though the downward trend has not been broken just yet.
Finally, there is one more wrench that muddies the picture. A proprietary model for the PE ratio that does NOT include the factors mentioned earlier actually suggests that the PE ratio is more than a standard deviation too low. It is a regression based model with several factors and hence is at risk for a number of statistical biases. However, the in-sample results over the past ~60 years have been enticing. (EY in the chart below = Earnings Yield. So the green areas are times when the trailing S&P earnings yield is 1 standard deviation cheap to the model, while the red areas indicate when it is 1 standard deviation rich.)
It is not yet clear to me if it’s time to discard the hypothesis for a declining PE. But the evidence against it is mounting. A clear break of the trendline in the first chart (which will probably be achieved if we touch 1500 this year) could possibly qualify as conclusive evidence. In the meantime, however, with the trendline still intact, conviction behind this move higher will remain unstable.
- US ADP Employment declined to 162k in Sept vs 140k exp and 201k prev
- Services PMI in Sept:
- US: Improved to 55.1 vs 53.4 exp and 53.7 prev
- Italy: Improved to 44.5 vs 44 exp and prev
- UK: Declined to 52.2 vs 53 exp and 53.7 prev
- China: Declined to 53.7 vs 56.3 prev
There were riots in Iran over the 50% drop in the Iranian Rial this year.
Video game sales this year may be the worst since 2005 – NYT
- Thu: BoE, ECB, US Jobless Claims
- Fri: US Employment, CA Employment
- Mon: Sentix Investor Confidence, AU Business Confidence
- Tues: Japan Eco Watchers Survey, Alcoa kicks off US earnings season