I follow a crude measure of the equity premium that is defined as the centered-smoothed-earnings yield less the real yield offered in the 10-year Treasury note. It was helpful in keeping me generally constructive equities for about a decade after the GFC on the grounds that moderate economic growth was “good enough,” relative to what seemed to be priced. Over the past couple years, it has been less helpful because it has kept me from being enthusiastic about equities, although I don’t really do bearish or even out.
Right now, this simple metric is at its lowest level since the NASDAQ bubble in the late 1990s and early 2000s. It looks somewhat less depressed if we measure the risk-free rate excluding the (ACM) estimated term premium. However, this adjustment has little effect on the comparison with the bubble period because the estimated term premium now is about what it was then. It is just that it is no longer depressed.
I recall a colleague observing this simple metric and asking, “Why don’t you do it the right way?” By doing it the right way, he meant substituting out the idea that the earnings yields is the expected return, based on some dubious theoretical reductionism, and instead incorporate what the consensus actually expects for dividend growth and the terminal value of the market at the medium-term horizon at which expectations (and thus the index level) would be at some sort of equilibrium.
Equity valuation is unremarkable relative to 14% expected earning growth
Bond yield data are actual to yesterday’s close. Equity prices are not exactly aligned. Call them early October.
As you will know from the strategy team’s work, Aswath Damodaran, at the Stern School of NYU, does it “right” in the sense favored by my friend. He also has a somewhat unconventional measure of the cash flows generated by equities, which includes both dividends and buybacks. I hasten to add that he is very explicit that reasonable people may disagree on what the appropriate measure of the equity premium is, all the while rejecting nihilism on the point. And he seems to favor the mosaic approach in figuring out how much risk tolerance is in the market, by looking also at surveys and risk premia in other markets, such as bonds or real estate. My interest here is in pointing out what strikes me as an interesting quirk arising in his work, rather than in criticizing his work. I am not competent to do that.
Damodaran’s measure of the equity premium, which is developed in nominal space (with inflation canceling out of the algebra) remained higher for longer during the post-GFC recovery, which was clearly an advantage, one that reflected in part that expected earnings growth was higher than the EY-based approach implicitly assumed. But his measure and my simpler one were qualitatively in line. They both said a meh recovery was good enough. In the event, we got a surprisingly great recovery of profits even in a meh economy, but that is a separate discussion.
Quite recently, though, his measure of the equity premium has utterly refused to fall, despite a huge rise in bond yields and little change in earnings yields as conventionally measured. And the reason seems obvious. The consensus top-down measure of earnings growth over the next five years has exploded to a record high, although his monthly data go back less than two decades.*
Obviously, the explosive growth of earnings in recent years needed to be factored into equity strategy in real time. If earnings are soaring, it might seem a bit churlish to point out that the earnings yield looks kind of low relative to the bond yield. Totally fair point. However, it is striking that earnings growth estimates are so high with profit margins already at all time record wides. I have long been a skeptic of the notion of margin mean reversion. Graveyards, as they say, are full of fans of that. But just for fun, confirmation of the consensus earnings growth rate in a world of 5% nominal value added growth would take the record-high profit margin in the domestic nonfinancial corporate sector to 27% within five years. I am not sure what to do with that, but it does seem relevant.
* He has annual data going back to 1960, and his latest monthly observation, for October 2026, is far higher than any of those annual figures.
Consensus implicitly expects aggressive margin expansion from record high
Data are actual to 2026 Q2
