For years, practitioners have debated the efficacy of the so-called “Fed Model” in terms of its ability to forecast future equity market returns. For the uninitiated, the Fed Model is simply the ratio between the so-called earnings yields for the S&P 500, basically the inverse of the P/E ratio, and the yield on the 10-year Treasury bond. Theoretically, a high earnings yield vis-à-vis bond yields indicates an equity market that is undervalued, or at least poised to outperform bonds. Numerous studies have been performed on the Fed Model; many of them have proved inconclusive at best when it comes to establishing a statistical link between the model and future returns. Like other debates we’ve waded into in the past, we’re not going to attempt in any way to break any new ground on the core debate. Instead, with the massive uptick in Treasury yields and the steady progressive move higher in the S&P 500, we thought it would be interesting to take a look at the dramatic gyrations in the Fed model in recent months and years and see if we could draw any conclusions. Likewise, we thought it might be interesting to create a “Corporate Bond Model” that substitutes investment grade corporate yields for Treasury yields to see what, if any, type of insight that might provide.
First, let’s dispense with a few housekeeping items. We calculate our Fed Model monthly going back to 1962. For earnings yield, we take the inverse of the Shiller 10-year P/E at the end of each month; we feel this provides more stability and consistency than a simple trailing 12-month P/E. At each month end, we divide the earnings yield by the 10-year US Treasury yield, thus providing the Fed Model ratio. Higher values indicate equity undervalution. Lower values indicate overvaluation. For our “Corporate Model”, we use the same calculation for earnings yield, but use the “FINRA – BLP Active Investment Grade US Corporate Bond Average Yield” for the denominator. Presumably, a higher ratio indicates equities are in a better position relative to their corporate bond cousins, while a lower ratio indicates the opposite.
Let’s begin with a chart of our Fed Model:
We’ll concur with those that have worked on this in the past that there is a general weak statistical significance when it comes to the Fed Model’s ability to consistently predict future returns. A few things stand out though. First, look at the dramatic gyrations since the financial crisis in ‘08/’09 and how unusual these gyrations are compared to past history, not to mention the Fed Model ventured into territory not seen at any point over the prior 50 or so years. Obviously, this reflects the volatile downward march in 10-year yields from over 5% prior to the crisis to a low of near 1.5% last summer back to the current level near 3%. Similarly, S&P 500 multiples moved dramatically, falling from the 20s prior to the crisis, to the very low teens, back into the 20s as the recent cyclical bull market has progressed. There’s been dramatic movement in the numerator and the denominator of the ratio during the past five years. On balance, the moves have kept equities in the extreme undervalued camp. Even with earnings yields declining with valuations creeping higher, compared to historically low Treasury yields, they look fantastic (again theoretically). Second, it appears to the naked eye that the model does a decent job of foretelling the future path of equity returns when the model reaches extremes (the dark band represents the area between +1 and -1 standard deviation from the mean). The extreme undervaluation indicated in the mid 1962 time frame proved a solid point to buy equities for what turned out to be six year run to secular bull market highs. Similarly, the undervaluation extreme represented in 1974 proved to be a solid spot to buy equities long-term, though it did take several years through the late ‘70s and early ‘80s to shake off the overall malaise of the secular bear market. Conversely, we see the extreme overvaluation of the markets in 1999 captured as well. We all know the story from 2000 to 2008 in terms of frustratingly negative market performance. Then, as pointed out, the model entered extreme equity undervaluation territory with the crisis. As discussed, there have been wild gyrations and distortions over the past few years, but the model remained in the extreme undervaluation zone. Lo and behold, the S&P 500 has rallied nearly 150% (including dividends) since March 2009.
The recent downtick in the model is mostly a reflection of the fact that 10-year Treasury yields have spiked dramatically off last year’s lows. We have to keep this in perspective though. Going back to 1962, the average 10-year Treasury yield is approximately 6.5%. We’re a long way from normalcy. As such, the model remains in extreme undervaluation territory. Does this mean equity markets are poised to take off like a rocket ship from these levels. Not necessarily. Of course, as we’ve seen in recent months, the denominator can continue to move dramatically affecting the ratio. It’s probably better to frame the extreme condition of the ratio in terms of future performance relative to the Treasury market. We think it’s reasonable to believe that equity returns, whatever the tenor, have a decent shot of outpacing total Treasury returns over the next several years.
Turning now to the aforementioned “Corporate Model”, we see a similar situation. Here is the chart:
Our data on corporate yields only goes back to 2002. This chart shows less volatility than the Treasury-based Fed Model owing to the fact that this corporate index hasn’t varied as wildly. Like the traditional Fed Model above, this corporate yield-based model was showing a problematic situation for equities relative to corporate bonds in late-2007. As the crisis progressed, this reversed towards the extreme equity preference situation of late also observed in the Fed Model. Perhaps this model provides a purer look at the situation between bonds and equities since focusing on corporates dampens the accusation that the bond/equity relationship as captured in the Fed Model is meaningless because of the massive Fed intervention. Granted, Fed action has affected bonds/yields across the spectrum, but corporate bonds are more insulated from this affect in a way since the Fed isn’t directly intervening here as they are in Treasuries. Nonetheless, the picture follows the same script as written for the Fed Model.
Closing out, what’s the quick bottom line here? The Fed Model and equivalent models aren’t great for making investment decisions in many cases, but seem to provide meaningful color when they’ve reached historical extremes. Whether looking at Treasuries or corporate yields, we still see ratios that are at/near historical extremes, with models expressing a preference for equities over bonds. Going forward, we continue to believe equities, both global and domestic, will provide better returns over coming years than bonds (viewed in terms of total returns), though this does not necessarily mean that absolute total returns for equities will be stellar.
