Talk of bursting bubbles is getting more widespread with stocks trading at near-record valuations since last March while the Fed contemplates a retreat from the policy of keeping the Fed Funds rates set near zero and its $120 billion in monthly asset purchases.
Extreme stock market bubbles typically happen once in a generation, but it’s understandable to be concerned given the record amounts of money being printed in the US and elsewhere.
The good news is that, although there may be Fed-induced distortions in the markets, they could keep fueling returns for years to come.
So let’s deconstruct this narrative to see why this is using the discounted cash flow model (DCF).* I’m a big fan of the DCF, but the challenge with the model is that there are many variables and therefore many assumptions.
The good thing about the DCF is that it forces me to categorize all the variables that make the markets move into either the numerator (cash flows) or the denominator (the discount rate).
In the numerator are the cash flows and their expected growth rate. In the old days this would just be dividends, but these days it makes sense to also include share buybacks, even though they are not a cash flow in the literal sense. Share buybacks can be viewed as an indirect return of earnings back to shareholders, and therefore they play an important role in valuation (i.e., a higher payout = higher P/E).
In the denominator is the discount rate (or cost of capital), which consists of the risk-free rate (10-year Treasury yield) plus the equity risk premium (ERP). The ERP functions like a credit spread and represents the additional return that investors demand as compensation for the additional volatility (versus the risk-free asset).
So these are the 4 important variables that we need to project:
Then we can stress-test the DCF to see how changes in these variables might affect equity valuations. It’s no small task, but if it’s possible that the Fed has created an asset bubble, we should be able to quantify its effects using this approach.
Needless to say, fiscal and monetary policy are currently ultra-stimulative. When we adjust the Fed’s policy rate for asset purchases (which the Atlanta Fed does through its shadow rate and allows interest rates to fall below zero for modeling purposes), and we subtract the inflation rate, we see that the Fed is even more accommodative than it was during the financial crisis (green bars below).
The surge in “excess money” (money supply growth minus GDP growth) resulting from the fiscal/monetary response to the pandemic has clearly elevated the market’s valuation.
That surge in valuation has brought the ERP down to below-average levels. There are many ways to measure the ERP, but one simple method is to subtract the risk-free rate (10-year Treasury) from the earnings yield (the reciprocal of the P/E ratio).
With a forward P/E of 22x and the 10-year at 1.3%, the implied ERP is currently around 3.2%. Using a trailing P/E of 25x, the implied ERP is 2.7%. The historical average is around 4.0%–4.5%.
Another way to calculate the implied ERP (iERP) is using the DCF. This approach calculates the implied return (and by extension the iERP) based on today’s index level and the historical growth rate in earnings (6%–7%).
It’s a more sophisticated approach than just subtracting the risk-free rate (RFR) from the earnings yield, but it shows the same thing: The risk premium is currently 50-100 bps lower than the historical average.
Next let’s look at the RFR, aka the 10-year Treasury yield. It does not require a leap of faith to conclude that monetary policy is suppressing interest rates.
Based on the regression between 2 variables, nominal yields and TIPS breakevens, the current level of the 10-year is about 80 bps lower than it “should” be based on historical patterns. Again, whether today’s subdued yields are the result of policy can be debated, but it seems that way to me.
The DCF shows one thing clearly: At today’s low level of rates and reduced risk premia, the market is extremely sensitive to changes in the denominator of the DCF—the risk-free rate and the equity risk premium.
In the numerator are earnings growth and the payout ratio. With dividends being relatively stable over time and the DCF placing a lot of weight on the terminal value (long-term growth after 5 years) rather than the next few years, the main variable that I try to solve for is the pace of share buybacks.
We know that historically earnings growth has been 6% in the US, and that in recent years the payout ratio has been around 90%. Currently it’s 71%, with dividends comprising 35% and buybacks 36%.
At a 6% compound annual growth rate (CAGR) for earnings, a 10-percentage point shift in the payout ratio corresponds to a 6-point shift in the P/E ratio.
With dividends being relatively stable, share buybacks are the swing factor that drive changes in the payout ratio, which in turn affects the P/E multiple. The question, at least for this exercise, is whether buybacks are a product of easy money policies.
I don’t think they are but regardless of what drives buybacks, they impact the market in 2 ways. First, they inflate the market’s earnings per share, and therefore understate the P/E ratio. Buybacks reduce shares outstanding, which is the denominator of EPS, and that distorts the financial ratios we all look at.
The second way that buybacks impact the market is from the aforementioned change to the payout ratio. If companies in the S&P 500 only paid dividends, the current payout ratio would be 35%. That would warrant a lower valuation than the current 71% if we include buybacks.
We can see the impact in the chart below, which shows the pre-COVID earnings path compared with the current consensus estimates (from Bloomberg). Getting the earnings trajectory right is already difficult, but predicting how much of those earnings will be returned to shareholders via buybacks is another matter.
As the DCF model on the right-hand side of the chart shows, at a discount rate of 5.5% the difference between the current 71% payout and a 60% payout is roughly 600 SPX points, or 15%.
The DCF framework gives us a way to quantify the effects of today’s era of ultra-low interest rates and abundant financial engineering. Without even getting into the equity risk premium and payout ratio, we can see that today’s low rates have added 5 P/E points to the market’s valuation, and the many trillions in buybacks since 2004 have added another 5 points.
That’s 10 P/E points of additional valuation. This suggests that absent these factors the S&P 500 would trade at a 15x trailing P/E instead of 25x. Coincidently, 15x was the average P/E ratio for the market’s entire history prior to the current era.
In a hypothetical worst-case scenario in which rates revert to where they should be on the basis of inflation, the ERP goes back to average, and the buyback era ends, the DCF grid suggests that the market’s current 25x P/E could be cut in half. A 50% haircut resulting not from an earnings decline but simply a reset to “normal” conditions might indeed leave many investors thinking it was all a big bubble.
Again, I’m not convinced that we can lay all the current market attributes at the feet of the Fed, but clearly the combination of share buybacks and low interest rates have had a material impact on the stock market.
I would like to address one additional attribute of bubbles, and that’s sentiment. To see how most people are investing, we can focus on long-term flows, i.e., regular investors saving for retirement. And here we see very little evidence of exuberance (despite the surging margin debt and meme-stock speculation in other isolated areas).
Since the financial crisis, a mere $291 billion has flowed into US equities through mutual funds and ETFs. This is less than 1% of the $38 trillion US market cap. In contrast, $3.2 trillion has flowed into bonds. There may be a bubble in bonds, but I don’t see one in stocks.
In my view, interest rates are repressed by the Fed, and this will likely continue for a long time. That means that the stock market is 25% higher than it would otherwise be, but this can remain the case for a long time. Having said that, at 1.3% the 10-year yield probably has more upside than downside, and, all else being equal, that should be a drag on valuation.
In my view, only an earnings decline or regulatory action will slow this train. As long as rates are low and companies produce more free cash flow than they can put to use, they will buy back shares, inflating EPS and raising the payout ratio in the process.
Finally, while there have been pockets of exuberance and speculation in the markets, as a veteran of 36 years of market cycles I just don’t see the bell-ringing signs that the market as a whole has reached a sentiment extreme.
A bubble? No, not in the classic sense. Distortions through policy and financial engineering? You bet. Can it last? You bet. For how long? Well, my secular bull market roadmap suggests another 5–7 years of outsized returns. So I’ll go with that.