The Not-So-Big Picture on Financial Models

Rarely do I disagree with fellow investor and financial blogger Barry Ritholtz, but the time has finally come. Last week Barry wrote a column for Bloomberg View called “Why ‘Peak-Earnings Models’ Are Nonsense .” Though I don’t agree with everything I read, and I hardly ever expend the energy to write a long response, this time I saw an opportunity to discuss several important topics: the Fed model (where Barry and I actually agree) and profit margins and the P/E model (where we disagree).

In his article Barry argues that the Fed model is flawed because it compares two variables — valuations implied by interest rates (the inverse of the ten-year Treasury yield) and actual valuations, market price-earnings — to tell you whether stocks are cheap or expensive. Barry writes, “The problem with the formula is that it contains not one but two variables. ... 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.”

I agree that the Fed model could be telling you that stock valuations are too low or too high in relation to interest rates, but this assumes that interest rates are at the right level and will be at this level in the future. Current interest rates could be too high or too low as well, especially in an environment in which governments globally set the levels of short- and long-term interest rates, rather than letting the market do it. Also, the Fed model confuses an intuitive relationship (that is, higher interest rates lead to lower P/Es and vice versa) with a direct relationship.

Aside from the interest income and expenses lines on corporate income statements, there is no direct relationship between interest rates and valuations. Also, today’s interest rates are low because the Fed has bought $4 trillion of our bonds, and no one is really sure (except for the Fed, which is certain!) what impact the unwinding of this debt will have on future interest rates.

So far, Barry (who has written about the flaws in the Fed model in his blog, the Big Picture) and I are in agreement: You have to forecast a ratio that has been extremely difficult to forecast to determine another ratio, and we are not even sure about the impact that one has on the other.

Then Barry goes to argue that the P/E ratio has two variables as well (price and earnings), and therefore, “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.”

In the P/E ratio there is no question what the current price is, but earnings are an assumption. You can look at what earnings were last year — that is not an assumption — but last year’s earnings are not very relevant; what the future E is going to be is what matters. You can look up analyst estimates for forward earnings for a stock or the whole stock market, but those are just estimates.

I implore you to think about earnings as sales times a profit margin. Let’s put sales aside for a second and focus on the margins. When you talk about profit margins for the whole economy, it is a bit of an equivocal exercise, but once you start thinking about profit margins on the company level, they become a bit more intuitive. If you have a company that earns very high economic profits, another company will want some of those profits; a competitor will lower prices or provide more services and generally try to compete those profits away (think Apple versus Samsung).

Or consider equipment company Caterpillar, an American icon that has benefited tremendously from the commodity supercycle: Its sales suffered a brief pause during the financial crisis and then continued to go up, as if overcapacity in construction in the U.S. and Europe had not taken place. Caterpillar has been greatly helped by operational leverage — as its sales went up, its costs grew at a slower pace, and thus its margins expanded. But even supercycles end, and even if they didn’t, competition from cheaper Chinese and Japanese companies (helped especially by the weak yen) is always lurking. Caterpillar’s profit margins will likely decline significantly.

This is a long way of saying that profit margins mean-revert — because, as GMO co-founder Jeremy Grantham (Trades, Portfolio) so tidily puts it, capitalism works. It is almost like a law of financial gravity: Profit margins always revert to the mean. (There are some exceptions on a company-by-company basis, but they are rare.) If profit margins are too high, they decline; if they are too low, they go up. Today margins are at a new modern high and will likely decline a lot going forward, driving earnings down with them.

But not so fast; let’s remember that E is profit margin times sales. So maybe sales growth will bail us out and offset the immense pressure from the imminent profit margin decline. Now let me put on the flimsy hat of an economist. I still remember this from macroeconomics class: The teacher asked us, “Imagine every worker got really serious about his job, started slurping 5-hour Energy drinks, got extremely motivated and was provided with lots of fancy computers and other tools. What do you think would happen to real GDP growth, which before was, let’s say, 3 percent?” A forest of hands shot up, and my classmates started throwing out numbers: 8 percent, 10 percent, 15 percent. I remember that the answer shocked me. He said, “It would grow 0.5 percent more, or maybe 1 percent at the most.”

Again, I am wearing a flimsy hat as an economist here, but my teacher was right. If all the wonderful, magical things happen to our economy — taxes are lowered, unemployment falls, productivity rises, and interest rates are very stable, then GDP (the sales of the whole economy) will grow just a bit more strongly — 1 percent or 2 percent more at the most. A developed economy is already a well-oiled machine; it is very difficult to make it grow much faster.

The problem is this: Even if GDP starts growing at an accelerated rate of 6 percent, this growth will not significantly offset margin mean reversion. Here is a simplified example: Let’s say sales of the economy in 2013 were $100, and profit margins were 16 percent (we are at about that level); thus earnings were $16. If in 2014 the economy grows, let’s say, 6 percent, sales will be $106. But if profit margins mean-revert to 10 percent, then earnings will drop to $10.60 — down 33 percent. (I am being very kind here; the mean is about 9 percent, and historically margins drop below average at first when they mean-revert.) So if the price of the market was $180 and you thought you were buying a cheap market for 11 times $16 in earnings, you’d be unpleasantly surprised when earnings dropped to $10.60, because suddenly you’d discover that you owned a very pricey asset at 17 times earnings.

Yes, there are multiple variables in P/E: price, sales and profit margins. But unlike the Fed model, in which one variable was not directly related to another, in the P/E equation there is a direct mathematical relationship between variables. It is an equation with which you have to perform some analysis — just as most of us do when we analyze stocks, and just as Barry did when he was a sell-side analyst.

Of course, there is a truism that shines through Barry’s article. Even if I am right about all this, I have no idea when profit margins will mean-revert or what P/E that investors will decide to put on declining earnings. The market was up more than 30 percent last year, but earnings saw a mediocre increase. The market’s irrationality may outlast my reputation as a prognosticator.