Explaining Multiple Expansion - A Provisional Thought
I'm not sold on this.
This will be an esoteric follow up to my recent post on examining the S&P 500 valuation metric CAPE. That analysis was more concrete that this one. This is more a thought experiment—one that I approach with a lot of skepticism.
In the prior post I said,
Still, I do believe and I think many agree that the average CAPE that prevailed before what I might call the modern era of equity investing is somewhat inappropriately low for what has prevailed and should be considered appropriate in this modern era. In other words, a good estimate of an appropriate CAPE is higher today than it was in the early part of the 20th century. When that transition should be placed is probably a fool’s errand as is getting too precise about how much it has increased.
Here I’d like to attempt some reasoning as to why the CAPE level that has prevailed recently might be higher than the historic level.
Backing up briefly, remember that the CAPE is a measure of how much we are paying as investors for the earnings that companies are generating. The fact that we would assume some growth in earnings in the future explains why we would pay more than simply a sum of current earnings projected into the future discounted for the time value of money.
If a company has been earning $1 per share per year, we don’t just add up the discounted value of $1 per year every year into the future (i.e., at a 5% discount rate summing $.952, .907, .864, .823, …). That would add up to a value of the stock being $20 (1/.05) if we assumed that $1 would continue forever.
Instead we would assume those earnings would grow at some rate. Usually one would assume a recent, reasonable growth rate for the near future and then some lower terminal growth rate. Therefore, the present value of the stream of cash flows would be greater than the $20 in our example. Perhaps it would be ~$67 because we assume earnings are going to grow at 10% per year for the next decade and then 2% thereafter (trust me on the math).
In these examples we would say the “multiple” we are paying (multiple of current earnings) is 20x or 67x, respectively.
The multiple on the CAPE historically was about 14.6 until 1987 (a date chosen not at random as this is about when it appears something changed structurally in the data). After 1985 the CAPE has been about 26.4. Like I said in the prior post, you can choose other start/stop dates, but the difference is basically the same.
Zooming in on what I’ll call the modern era for the CAPE (since about 1987), we can isolate two other distinct periods. From 1987-2016 the CAPE averaged 24.4. Since 2017 it has averaged 32.6. It is currently (June 2026) about 41.
Here is where my thought experiment begins. The 2017-June 2026 period is a ramping up to where we sit today. We went from a CAPE of 24.4 to 41, a roughly 68% increase (precision here is misleading—think of the changes in more rounded terms).
My question/hypothesis is that recent accelerations in earnings growth might account for this increase. I’m not claiming it is fully reasonable; rather, I am just saying that might be why the multiple has expanded as it has. Keep in mind, I think the current CAPE is too damn high (i.e., the S&P 500 is too expensive).
Back to the hypothesis, what if we forecasted a sustained bump in future earnings growth and compared the implied multiples?
As a simplified example of this, consider a simple bond analysis. If interest rates in the market are 5%, a 40-year bond with a maturity value of $100 that paid a coupon of $5 per year (based on the bond’s stated interest rate of 5%) would be valued at $100. In comparison a 40-year bond paying a $9 coupon annually would be valued at $168.64—the extra $4 per year means it is worth a lot (68%) more.1
Analogously (here is the hypothesis), let’s look at the change in the growth rate of real earnings for the S&P 500 between 1987-2016 and 2016-today. In the first period the growth was 5.59%, which is pretty close to the long-run average looking back much further in time. In the recent period from 2016 it has been 11.12%—about a 67% increase in the growth rate. Keep in mind these are real earnings so inflation is already adjusted out.
So the expansion in multiple over the past ten years (about 68%) happens to? roughly match the 10-year difference in real earnings growth rates. Why would ten years be important? Maybe it isn’t. But the formula for CAPE is a 10-year lookback at earnings. Perhaps a better analysis would add some complexity for changes in expected future growth rates allowing a smaller increase in the next decade and then a lower than recent but higher than historical terminal growth rate. We could certainly play with the numbers to get it to work out, but that wouldn’t prove much other than our ability to solve for our assumed conclusion—circular reasoning for sure.
My thought is that investors may have near-term hindsight bias. They see the bump up in real earnings growth and project it forward. If so, is that sustainable? I would argue strongly that it is not. It is beyond the scope of this post, but suffice it to say trees don’t grow to the sky. Stein’s Law has to kick in somewhere.
The CAPE is too damn high!
Assume no difference in credit risk, obviously.