Mismeasuring the market: Valuation indicators to ignore

It’s tough to know if the U.S. stock market is overvalued. This task is made tougher by the abundance of useless, redundant, and downright idiotic market indicators being peddled in the marketplace of ideas.

Here I review four commonly used measures of market overvaluation that you can safely ignore. All four measures say the U.S. stock market is overvalued today. The market might well be overvalued today, but these measures are not good reasons to say so.

The reality is that the U.S. today has very high prices relative to trailing earnings or estimated future earnings. But there’s no need to embellish this reality with useless indicators.

Household equity share

The household equity share is defined as households’ total holdings of equities as a percent of total assets. This ratio comes from the Fed’s Flow of Funds report and is currently at an all-time high of around 43%, indicating exuberant retail investors who are piling into equities and, thus, an overvalued stock market.[1]

Methodological error: Inferring trades from mark-to-market accounting.

The truth: When stock prices rise, all other things equal, the household equity share rises.

Correct approach: Look at flows (purchases and sales) from households.

It’s just wrong to think that a rising household equity share reflects buying by households. The share mechanically reflects fluctuations in equity prices as well as the changing nature of the U.S. retirement system; it’s not entirely driven by household decisions. It’s especially pernicious to interpret the share causally, as in “stock prices are high today BECAUSE households have bought equities, driving up prices.” Totally false. It could be true that households are buying, but that’s not measured by the household equity share.

Let’s take a hypothetical example. Suppose households own $30 in equity and $70 in other assets. The household equity share is 30%. Then stock prices double for some reason. If households did nothing, their equity share would mechanically rise to 46%. Suppose households are in fact contrarians and they partly rebalance, selling $4 of their equity. In this scenario, we end up with the equity share rising from 30% to 43%; but that is in spite of selling by households, not because of buying by households.

And in fact, these numbers are ballpark figures for what actually happened between 2020 and 2024;  stock prices doubled and the household equity share went from 30% to 43%. Now, obviously my example does not accurately describe all aspects of household finances over this period, but the point is that the household equity share rises when stock prices rise, ceteris paribus.

Should we view inertia – failure to rebalance to maintain a constant equity share – as an active decision by households that drives up prices? I don’t think so. It’s true that complete household inertia is consistent with rising stock prices, but it’s also consistent with falling stock prices. Inertia, like passive investing, is about not trading, and thus inertia can never be the driving factor.

As an empirical matter, the dynamics of household equity allocation have changed substantially in recent decades with the rise of target date funds. Households have become a more stabilizing force for the stock market, acting as “macrocontrarian” investors according to Parker, Schoar, and Sun (2023).

Today’s high household equity share is mostly just a reflection of today’s high stock prices. The stock market doubled in the past five years. Did you expect the household equity share to go down? Of course not.

I’m all in favor of studying retail flows, and I’m especially interested in situations where households buy equity from firms. Retail inflows accompanied by equity issuance are what I call the Third and Fourth Horsemen of the Bubble Apocalypse.  But that’s not what’s happened lately; firms have been buying equity (repurchases), not selling equity (issuance). The household equity share is not a good way to measure this activity.

The laws of math apply not just to households, but to all market participants including insurance companies and pension funds. If stock prices go up, and no other asset prices change and nobody trades, then of course insurance companies and pension funds will instantaneously have higher equity share. Somebody, somewhere is holding the stock market and when stock prices go up, the share of wealth allocated to stocks mechanically rises. When you see equity prices rising together with the equity share of investor class X, your reaction should be “ho, hum” and not “AHA!”

Here’s a poem I wrote to help you remember:

When the stock market rises on an upward path,

The household equity share does too. It’s called math.

The Buffett indicator

The Buffett indicator is the ratio of the total market cap of U.S. equities divided by GDP, described as "probably the best single measure of where valuations stand at any given moment" by Warren Buffett in 2001.[2] The ratio is currently near all-time highs, indicating an overvalued U.S. market.

Methodological error: Normalizing by the wrong thing.

The truth: The stock market is not the economy (TSMINTE).

Correct approach: Normalize by trailing earnings or expected future earnings of the companies in the stock market.

Like the household equity share, the Buffett indicator measures something, just not something very relevant to overvaluation. Using this ratio to forecast cross-country returns is especially misguided.

I will now give you seven magic words that are more powerful than any incantation uttered by Harry Potter. Seven words that will free you from the miasma of confusion that emanates from the bowels of financial journalism and sell-side research. Here they are: The Stock Market Is Not The Economy. When you are confused, just remember TSMINTE.

The Buffett indicator makes no sense because TSMINTE:

• Not all firms are publicly traded.
• GDP includes non-profits and government.
• Corporate profits are not the same as GDP.
• The stock market has firms that earn profits from non-U.S. activities.

If you really need a ratio bearing Warren Buffett’s name, may I suggest the Saylor-Buffett ratio? Like the Buffett indicator, it is also high today.

Another helpful poem:

The stock market is not the economy,

Don’t normalize it by GDP.

U.S. equity share

The U.S. is currently around 70% of the market cap of world equity markets, an all-time high. According to Ruchir Sharma of The Financial Times, since the U.S. is only 27% of the world economy, the disparity indicates that “America is over-owned, overvalued and overhyped to a degree never seen before,” and we are witnessing “the mother of all bubbles.”[3]

Methodological error: Normalizing by the wrong thing.

The truth: The stock market is not the economy.

Correct approach: Compare valuation ratios for U.S. and non-U.S. firms.

Maybe America is overhyped and we are indeed in the mother of all bubbles, but the fact that the U.S. is 70% of the world equity market tells us little about those assertions. If the world’s most profitable companies list in the U.S., then of course the U.S. will have a disproportionately large share of market cap. Who cares if the U.S. is only 27% of world GDP? TSMINTE.

Suppose hypothetically that 70% of U.S. market cap  was incorporated in Delaware, yet Delaware had less than 1% of U.S. GDP. Would that be alarming? Absurd. What we’d want to do is look at P/E ratios or other valuation metrics for Delaware vs. non-Delaware firms.

I think the alarm about the U.S. share of the world stock market reflects three claims. First, the U.S. has higher valuation multiples than non-U.S., and we’d expect these multiples to converge. Second, U.S. companies currently have higher profit margins than non-U.S., and we’d expect margins also to converge. Third, in the long run, we’d expect all countries to have stock markets that are roughly proportional to their GDP.

The first two claims are plausible. The third – that all countries should have the same “Buffett indicator” – is just silly. Standard growth theory predicts convergence in GDP per capita across countries, but there’s no reason to expect convergence in market value per GDP or market value per capita. Just as I’m not alarmed if Delaware has more market cap than California, I’m also not alarmed if the U.S. has more market cap than Europe. Again, TSMINTE.

Let’s consider a scenario in which all European firms decide to relist in the U.S. (this scenario becomes less far-fetched every day). In this scenario, the U.S. share would rise, but U.S. stocks as a whole would get cheaper since the new arrivals have lower valuation ratios. Thus, the U.S. equity share is not a good way to think about valuation.

Yet another poem:

Ain’t no reason for you to care

‘bout the U.S. equity share.

The Fed Model

The Fed Model is the earnings yield of stocks (the inverse of P/E) minus the nominal 10-year Treasury yield, which today gets you around -1%, indicating an overvalued stock market.

Methodological error: Confusing nominal and real variables.

The truth: First, high P/E predicts low future returns; the Fed Model gets that right. Second, high inflation today predicts high future returns; here, the Fed Model has the wrong sign on inflation.

Correct approach: Compare E/P to the real rate of interest as calculated from TIPS, trailing realized inflation, or forecasts of future inflation.

Comparing earnings yields with nominal interest rates is a distressingly common practice; sometimes it is called “the Fed Model” and uses 10-year Treasuries, and sometimes it has a different name and involves corporate bonds. No matter what you call it, it’s a mistake. Because future earnings will rise with inflation, the ratio E/P is in units of real yield, and it is simply incorrect to compare it to nominal yields.

Let’s take an elementary scenario. We have an economy where all assets have the same required real return of r=10% a year. We have a firm that generates $5 in earnings, with a real growth rate of g=5% a year. If inflation is zero, plug in the Gordon equation P = E/(r-g), and you get a stock price of $100. In this world the earnings yield is 5% and the Treasury yield is 10%. The “Fed Model” gives -5% for this firm. That’s the benchmark with no inflation.

Now suppose inflation will be 20% a year forever. We expect earnings to rise with inflation (both costs and revenues rise with inflation), so now the nominal growth rate of earnings becomes g=25%. The real required rate of return stays at 10% but the nominal required rate is now r = 30%. Since inflation is in both the discount rate and the growth rate, it cancels out, and the stock price remains at $100. The “Fed Model” now says -25%, wrongly indicating a hugely overvalued stock price. Lesson: nothing good happens when you confuse real (earnings yield) and nominal (Treasury yield) variables.

Here I’m playing fast and loose, skipping many details including the issue of trailing vs. forward earnings, nominal corporate liabilities, and inflation’s interaction with cost-based accounting and taxation systems. For a more complete discussion of inflation and valuation, see Ritter and Warr (2002), Asness (2003), and Cohen, Polk, and Vuolteenaho (2005).

Everything about the Fed Model is stupid, including its name. Like “Holy Roman Empire” or “Modern Monetary Theory,” “the Fed Model” contains multiple untruths. It’s not a model, and it’s not used by the Fed. Indeed, when the Fed values the stock market, it appropriately uses the real interest rate. In its latest Financial Stability Report, for example, the Fed uses inflation forecasts to adjust expected returns on equities (concluding that today valuations are “elevated” with expected returns “well below” historical norms).[4]

The widespread use of the Fed Model shows that many market participants are confused by inflation. This confusion has been studied by economists at least since 1928 and is called “nominal illusion” or “money illusion.” In the specific case of stocks, Modigliani and Cohn (1979) discuss how high inflation produces an undervalued market. If many investors are subject to money illusion, the implication is that periods of high inflation cause the market to be undervalued. Thus inflation has the wrong sign in the Fed Model; higher inflation makes the Fed Model more bearish just when it should be more bullish.

Every time you use the Fed Model, an angel weeps. It’s financial malpractice. Don’t do it. My last poem:

When you mix up real and nominal

Your malpractice is phenomenal.