Investors need to be on their guard against the so-called Fed Model, which is being resurrected as a reason why U.S. stocks are likely to slide. The Fed Model is no more valid now than in the past.
The Fed Model is a market-timing model based on the difference between the 10-year Treasury yield
and the U.S. stock market’s earnings yield (which is the inverse of the P/E ratio). It’s supposedly bearish when the Treasury yield is higher than the earnings yield, or even when — like now — the spread between the two begins to narrow.
The idea that interest rates are relevant to assessing the stock market’s prospects is hardly new. The Fed Model’s attempt to exploit this idea traces to a single paragraph in the Federal Reserve’s report to the Congress in 1997 that, more or less in passing, compared the stock market’s earnings yield to the yield on the 10-year Treasury.
The model does not have the Fed’s endorsement. Many times over the past couple of decades the model has been shown to have neither theoretical nor empirical support. Yet it seems to have more lives than a cat. In recent weeks it’s even spawned an offshoot: a “Tech Index Fed Model,” which compares the earnings yield of the Nasdaq 100 Index
to the 10-year Treasury yield.
It’s this tech-index version that is telling the most bearish story right now. While the S&P 500’s
earnings yield currently is almost two percentage points above the 10-year Treasury yield, the Nasdaq 100’s earnings yield is less than a percentage point higher.
The theoretical case against the Fed Model, and by extension its subsequent offshoots, was made two decades ago by Cliff Asness, founder of AQR Capital Management, in the Journal of Portfolio Management. I can’t do a better job of making the case, so I refer you to his article, entitled “Fight the Fed Model.”
To mount my empirical case against the Fed Model, I analyzed the U.S. stock market, corporate earnings, and interest rates over the past 150 years (courtesy of data from Yale University’s Robert Shiller). The table reports a statistic known as the r-squared, reflecting the degree to which a given model predicts returns over the subsequent 12-month period. Note that the results for the Nasdaq Composite
and the Nasdaq 100 encompass far fewer years than the S&P 500; data for the Nasdaq Composite extend back to 1971 and for the Nasdaq 100 back to 1986.
|Using raw earnings yield to forecast S&P 500’s subsequent 1-year return||1.2%|
|Using Fed Model to forecast S&P 500’s subsequent 1-year return||1.3%|
|Using raw earnings yield to forecast Nasdaq Composite’s subsequent 1-year return||0.8%|
|Using Fed Model to forecast Nasdaq Composite’s subsequent 1-year return||2.2%|
|Using raw earnings yield to forecast Nasdaq 100’s subsequent 1-year return||0.2%|
|Using Fed Model to forecast Nasdaq 100’s subsequent 1-year return||0.1%|
Note carefully that none of the r-squareds listed in this table is significant at the 95% confidence level that statisticians often use when determining if a pattern is genuine. Note also that, even if they were significant, none of the r-squareds suggest more than a sliver of explanatory power. But, most importantly, notice that in no case does explanatory power increase significantly when taking interest rates into account.
The best you can say about the Fed Model, therefore, is that it doesn’t have a worse track record than simply focusing on the earnings/price ratio alone.
These results don’t automatically mean that stocks in general, or tech stocks in particular, will rise from here. The implication instead is that the recent increase in interest rates is not, in and of itself, a good reason for concluding that they should drop. So if rising interest rates are the reason you recently turned bearish, whether on the overall stock market or tech stocks in particular, you may want to reconsider.
Mark Hulbert is a regular contributor to MarketWatch. His Hulbert Ratings tracks investment newsletters that pay a flat fee to be audited. He can be reached at email@example.com