Why 'Peak-Earnings Models' Are Nonsense
Lately, there has been a spate of research, analysis and commentary telling us that earnings are at a cyclical high and must revert. Stock valuations, therefore, are elevated and earnings will soon begin to fall, bringing stocks down with them.
This is neither a credible analysis nor a method of valuing equities. Rather, it is an interesting narrative containing two predictions, and generally fails to acknowledge multiple unknowns and variables. (Let’s hold the question of whether anything as complex as the markets can be predicted with a single variable for another time).
Allow me a few more words to explain. Many of the traditional valuation methods rely on two or more variables. For example, price-to-earnings ratio uses both stock prices and earnings to determine if a company or market is cheap or expensive. This raises the obvious problem -- obvious to anyone who is not innumerate -- of forecasting one unknown by using a second unknown.
Why is that? Simply because we do not know what earnings are going to be next year any more than we know what stock prices are going to be. You may inform me that we can use the consensus of analysts as a starting point, but I would retort that no group of professionals has been as consistently wrong for over as long a period as they have.
Some years ago, I used this same basic approach to debunk the so-called Fed Model, a/k/a IBES Valuation Model. It starts by assuming the 10-year U.S. Treasury bond yield should be similar to the S&P 500 earnings yield (forward earnings divided by the S&P price). This could in theory tell you when stocks are over- and undervalued (we will ignore the problem of the initial assumption for now).
The problem with the formula is that it contains not one but two variables: While it is commonly described as a way to evaluate when stocks are over- or undervalued, the other variable in the formula is the forward S&P 500 earnings consensus. S&P 500 prices and the 10-year yield are the knowns, while BOTH valuation and forward-earnings estimates are the unknowns. Hence, the Fed model tells you one of two things: Either equities are over/undervalued, or consensus earning estimates are either too high or too low.
The same is true for many of the arguments claiming that earnings are too high or that stocks are overvalued. Unless you know what one of the variables is going to be, you do not know what the outcome of the valuation formula will be. Hence, it is more of an “assume that this variable does this, then here is what happens to that variable” approach. If earnings mean revert to historical medians then stocks are overvalued. If profits do not increase, then stocks are pricey. But the declarations that conclude one specific outcome (valuation) based on two variables (price and earnings) are inherently flawed.
For the record: I have no idea where stock prices will go, and I do not know what is going to occur with earnings. But I can war-game a variety of scenarios with these two variables. The various outcomes can be either that stocks are cheap and will go higher, or that stocks are expensive and will go lower (there are lots of other potential outcomes, but let’s stay with these two simple variants for now).
Fidelity recently discussed four ways that earnings could improve. If the economy continues to slowly improve, if unemployment keeps falling, if rates normalize without much disruption, etc., then the market can grow into its earnings. There are untold ways profits could fall. Unless you have some insight into what the future holds, you may be doing little more than guessing when forecasting profits.