On Monday we wrote a research note on the valuation of the SP500 using the Shiller CAPE P/E, which suggested that the market was overvalued by at least one standard deviation. Not surprisingly the note elicited a fair amount of feedback, some positive, some negative, but mostly constructive and all worth contemplating. Below I’ve categorized the main areas of feedback and attempted to address them appropriately.
1. On a relative basis aren’t equities still a good buy?
Clearly, money managers need to make allocation decisions between stocks, sectors, and asset classes. Therefore, one could argue, looking at U.S. equities on purely a valuation perspective and saying they are expensive only makes sense in the context of the investment alternatives, and the relative after-tax returns.
It is hard to disagree with that. As investors, we have various options of asset classes from which to consider allocating our clients’ heard-earned capital. On a very basic level, then, we should consider the relative value of bonds versus equities as an indicator of whether equities are truly expensive. (The one caveat here is that we also don’t have to be fully-invested when there are as few opportunities for true value as there are today.)
To examine this relative valuation, we’ve taken a look at the earnings yield of the CAPE P/E versus the average yield of a 10Y AAA corporate bond going back to 1954. In this analysis, a lower yield implies a more expensive asset class which is less appealing from an asset allocation perspective, in theory. Currently, bonds, based on this proxy, are a cheaper asset class as they have a 5.2% yield versus equities with a 4.23% yield.
Interestingly, on a standalone basis, while bonds are expensive versus their historical standards, not surprising given the current interest rate policy of the Federal Reserve, they are not as expensive as equities versus their historical average. In fact, 10Y AAA corporate bonds are currently only 0.74 of one standard deviation more expensive than their long run average yield of 7.2% going back to 1954 versus more than one standard deviation for equities.
2. The CAPE P/E is based on historical earnings and equities are based on future estimates, so aren’t equities still cheap?
While we would agree with this to a point, in aggregate it is important to remember that analyst estimates of future earnings are typically wildly disparate versus future actual earnings. So, while the valuation of a stock, and equities broadly, will eventually reflect their future earnings or cash flow, looking at the broad forward earnings of a market is rarely a great gauge of the valuation of the market and whether it is cheap or expensive.
Simply put, forward earnings estimates are typically inaccurate and usually too high. To highlight this, we’ve attached a chart below from McKinsey Consulting that emphasizes our point that EPS forecasts for the broad market are consistently grossly off their mark. We would submit that the primary driver of this disparity is that the bottoms up earnings projections rarely include accurate economic forecasts.
Obviously, we also have a negative view on future earnings based on corporate margins being very close to cycle highs as growth slows while inflation accelerates, so we have a hard time making the case that earnings will grow into their multiples. In addition, it is somewhat dangerous to make a “valuation” call on the collective sell-side’s earnings projections, particularly given their preponderance for inaccuracy.
3. The CPI is not a great gauge of inflation since it has changed over time, so how would CAPE P/E look with real inflation?
The calculation of CPI has changed many times over the last few decades with significant changes coming in 1983 when housing prices were replaced with owners’ equivalent rent and in 1999 when the Bureau of Labor Statistics stopped using Laspeyres indices in its calculation.
In the chart below, we have an estimate from shadowstatistics.com of what CPI would look like from 1980 onwards if the calculation of CPI hadn’t changed. The obvious take-away is that inflation would be substantially higher.
Whether we agree that CPI should be calculated the same now as it was 30+ years ago is not the point, but rather the point is that markets, commodities prices, and basic goods are indicating quite clearly that CPI understates inflation. So, the question is, if we use a more realistic, and presumably higher measure of inflation, how would the CAPE P/E for the SP500 look?
Shiller’s methodology for accounting for inflation is to use a CPI-adjusted S&P500 index and to then divide that by a 10-year average inflation-adjusted (using CPI) earnings to impute his CAPE P/E. Absent being able to actually recreate these numbers for a constant measure of inflation, which would presumably be higher than CPI, another way to look at this market valuation is on a nominal basis, which would at least normalize the valuation for changes in CPI calculations and allow us to compare apples to apples historically.
This analysis was actually done by Chris Turner (email me at if you’d like a copy of the paper with hat tip to a certain subscriber for pointing it out). He found that by using nominal numbers, for both the index and earnings, that nominally speaking (which really only makes the measure of inflation constant) the market trades at a higher multiple historically (no surpise) and is about as overvalued as it is currently without such adjustments.
We would further submit, that in a period of higher inflation the inflation-adjusted valuation of the market would be higher since the price is based on current inflation, while the earnings is based on cycically smoothed inflation. Thus, if you believe, as we do, that inflation is higher than reported today, than the market is likely more expensive. This is disconcerting because in periods of high inflation market multiples decline dramatically.
To the last point, the table below shows that as inflation accelerates, earnings multiples decline. This occurs due to a combination of margin compression due to higher input costs, which slows earnings growth and naturally leads to lower multiples; as well as, to the point highlighed earlier, the relative value trade: as inflation goes up, interest rates will as well and bonds will likely become more appealing from total-return perspectve (so “The Flows” will follow).
Daryl G. Jones