With the U.S. economy struggling to emerge from one of the worst recessions in history, it seems reasonable to ask why the S&P 500 is up 7% for the year. Are stocks wildly overvalued?
A measure that is often used to answer this question is CAPE – the ten-year average of the S&P 500 price/earnings ratio. According to it, stocks don’t seem to be overpriced at all.
In a recent paper published by the Federal Reserve Bank of San Francisco, economist Kevin Lansing notes that CAPE was at 30.9 at the end of the third quarter of 2020, or about 50% above its 60-year average of 20.5. A naïve conclusion from this observation would be that stocks are enormously overvalued. However, CAPE was higher two years ago, at 32.6 at the end of 3Q18. Yet, stocks climbed 20% since then and CAPE is now lower. How, then, can we use CAPE to know whether stocks are expensive or cheap?
Many pundits compare CAPE to its long-term average and, if it is significantly above it, conclude that stocks are overvalued. The obvious problem with this approach is that a long-term average is a value that only exists today: comparing stocks five years ago with a ratio that includes the then-unknown values of the following five years makes little sense. Yet, some insist on using CAPE this way. A slight improvement would be to use a moving average, for example.
Lansing from the San Francisco Fed uses a different approach. He notes that some economic indicators can account for fluctuations on the CAPE ratio, and he chooses three: interest rates, the growth potential of the economy, and uncertainty.
Interest rates, as we discussed in a previous post, are important determinants of the value of stocks (Warren Buffett said that they are “the most important” determinant) because lower rates increase the value of future dividends, and therefore of equities, even if earnings do not change.
The growth rate of potential economic output (a measure computed by the Congressional Budget Office) also influences the value of equities relative to earnings because if the economy has the potential of growing faster, so do future corporate earnings.
Economic uncertainty can be measured as the forecast error of various gauges of the economy, and is in fact tabulated in an index created by academics at Columbia and NYU. The idea is to measure the forecast error of various macroeconomic indicators. If it is true that “the market hates uncertainty” as we often hear, more uncertainty should bring down equity values because it puts future earnings in doubt.
Lansing runs a simple regression with these three indicators and comes up with a CAPE estimation that fits rather well with the observed values, especially in the last 20 or 30 years. A clear takeaway is that comparing CAPE to a fixed multi-decade average value not only does not make much sense but also does not account for factors that can cause CAPE fluctuations over time.
One puzzling observation in Lansing’s paper is that the CAPE ratio predicted by his model for the second and third quarters of 2020 are much lower than the observed ratio, suggesting that the stock market could be overvalued. He attributes this to the huge spike in the economic uncertainty factor. If this variable is removed, the model prediction improves significantly. The paper concludes that “today’s investors appear to be reacting to macroeconomic uncertainty very differently than in the past.”
There may be another explanation: That investors, instead of ignoring uncertainty, are aware that seasonal adjustments applied on current indicators are inadequate and cause major forecast errors (which is what the uncertainty index measures).
On October 2, for example, 661,000 seasonally-adjusted non-farm payrolls were reported, far less than the 894,000 projected. One reason for the big miss was that seasonal adjustments adjusted down the raw data to account for the return of school personnel. Since many schools remained closed, those workers did not materialize but the downward adjustment accounted for them anyway.
In any other environment, such a large miss would have hit the stock market hard. This time, however, analysts were able to identify quickly that the issue was the way seasonal adjustments worked. Forecasting errors of this nature are likely to be more common going forward, as we discussed recently, which could explain the recent breakdown in Lansing’s model.
Another factor to keep in mind is that CAPE may be influenced strongly not just by economic variables, but also by the availability of cash. If investors are flush with money sitting in their bank accounts, for example, there is a good chance that they will end up buying stocks.
While a majority of Americans have been seriously affected by the pandemic (more than 8 million are still collecting unemployment and 800,000 file every week), a smaller number has seen their account balances skyrocket, accounting for a significant dollar amount. We added a fourth variable to Lansing’s model (“Households’ cash account balances at brokers and dealers”) and obtained a better model fit to CAPE for recent years, suggesting that despite the uncertainty, stocks may not be far from where they should be - partly because of the spike in idle cash.
Where does CAPE go from here? Two factors to watch are interest rates and household liquidity.
Higher rates bring down the price of equities because they depress the present value of future dividends. While rates have fallen to very low levels since the financial crisis, they don’t seem to have much more room to decline. They have in fact crawled higher in recent days, which could start affecting the price of stocks soon, if this continues.
High levels of liquidity among wealthy households is the direct result of poorly-delivered government help. We argued this point repeatedly, and the sudden spike in household balances at broker-dealers seems to prove the point. If more stimulus is approved, more money will flow to those balances since its delivery is unlikely to be any more efficient the second time around. Conversely, failure to provide more stimulus may well drive down the price of stocks.
CAPE is a useful measure but it is not a static value. Contrary to the conventional wisdom, we think that it shows stocks roughly in line where they should be. The flip side is that CAPE can itself change quickly. Alas, predicting where it’s headed may be just as hard as forecasting where the market will go.