Could Stocks Still Be Undervalued?

On my vacation last week, I mentioned to someone I met that I work as the equities strategist for Morningstar. Immediately he asked me, "Where's the market headed?" Before I could answer, he quickly followed up with, "I'll bet you get that question all the time."

Yep, I do. Almost as much as I'm asked about  Cisco (CSCO),  General Electric (GE),  Intel (INTC), and (in all seriousness) Sirius Satellite Radio (SIRI).

Almost invariably when people ask me where the market is headed, they want to know about the next one to six months, rather than the next one to six years. I have a well-rehearsed answer: "I don't have a clue."

There are a couple of reasons why I don't have a clue (and I don't believe anyone else does either). First, the market is completely unpredictable in the short term. Just when you think valuations have peaked, the market throws you a curve ball and goes up another 20%. And conversely, just when you think stocks are undervalued, the market falls another 20%.

Anyone who dares to make predictions about the short-term direction of the overall stock market is overconfident at best, and irresponsible at worst. As Warren Buffett has often said, a prediction about the direction of the stock market tells you nothing about where stocks are headed, but a whole lot about the person doing the predicting.

Second, the several different fundamental models one can use to value the market often flash conflicting signals. Even someone who does this stuff for a living (like me) has a hard time figuring out which model to believe. In the end, there is no one right answer.

But although it's nearly impossible to consistently predict which way the market will go, some fundamental valuation metrics can offer insight about where the market should go. In the short term the market will often diverge from fundamental valuations; it often doesn't go where it should. But in the long term, two fundamentals override everything else: earnings and interest rates. Where these go, the market will eventually follow.

The Fed Model
The most common fundamental model for valuing the aggregate stock market based on earnings and interest rates is the Fed Model.

Market strategist and portfolio manager Ed Yardeni coined the name of this valuation model, although the Fed hasn't admitted that it actually uses the model to value the stock market. In any event, the model compares the interest yield on 10-year Treasury bonds (currently 4.05%) to the forward 12-month earnings yield on the S&P 500 Index (currently 5.4% according to Wall Street stock analysts, or 4.5% according to market strategists' and economists' estimates). The forward earnings yield is simply the inverse of the forward P/E ratio (1 / forward P/E ratio).

According to the Fed Model, the S&P 500 Index should currently sell between 1,291 (using economists' earnings estimates) and 1,549 (using analysts' earnings estimates). The actual current level of the index is 1,155. Thus, according to the Fed Model, the S&P 500 should rise 12% to 34% to get to fair value.

The Fed Model seems to do a good job of explaining the relationship between interest rates and stock prices. When interest rates are very low (as they are now), stocks should carry high P/E ratios. When rates are very high (as in 1981), stocks should have low P/E ratios. Over time, the correlation between long-term interest rates and forward earnings yields has been high. For example, in 1981 the forward P/E ratio for the Dow Jones Industrial Average was around 9. Today it's 18, based on analyst estimates of 2004 earnings. The rise in P/E ratios can be explained, in large part, by the drop in interest rates and inflation.

The Model's Limitations
Unfortunately, the Fed Model does a better job of explaining stock prices than predicting them. This is because everything hinges on the 10-year bond rate. Interest rates are notoriously difficult to predict, so the direction of the stock market is, too.

Under the Fed Model, if the 10-year bond rate rose to 5%, stocks would suddenly be overvalued by 9% (using economists' estimates of forward corporate earnings). A 10-year bond rate of 5%, while higher than today's rate, is very low by historical standards. It's quite possible we could see rates rise at some point over the next year or two. Since 1962 (as far as my 10-year bond data goes back), the average 10-year Treasury bond rate has been 7.3%. More recently, the 10-year rate averaged 6.6% during the 1990s. So anyone arguing that the rate will stay below 5% for very long is essentially saying that "things are different this time."

Further, the historical record of the market's interest rate predictions is not very encouraging. As evidence, we can look at data compiled by Ed Easterling of Crestmont Research. After crunching the numbers, Easterling found this (quoted directly from the Crestmont Web site):

"In the past 35 years (with a two-month exception), there has not been a 6-month period during which interest rates did not change at least 50 basis points--interest rates are much more volatile than most investors realize. As history demonstrates, more than half of the time, interest rates change by more than 1.5% over all 6-month periods." (Click here to see all the data behind this conclusion.)

If bond investors had expected interest rates to change so dramatically over the following six months, then by definition, rates would not have been so volatile--whatever caused rates to move so much would have already been "priced in." Thus, sophisticated institutional bond investors are not very good at predicting future interest rates. And if they're not good at it, do you think stock investors are good at it?

Another problem with the Fed Model is that it can spit out some wacky numbers in times of extreme interest rate fluctuation. Last May, for example, the 10-year Treasury bond carried an interest rate of just 3.35%. That means the forward P/E of the stock market should have been 1 / 3.35% = 30. Just two months later, the rate was back up to 4.47%, for an implied forward P/E of just 22. If stocks had followed the Fed model precisely, they would have fallen 26% between May and July. Instead, they rose about 3% during that period.

Future Shock
To get a sense of what could happen to future stock prices according to the Fed model, consider this: If the 10-year rate goes back to its July 2000 level (6%), stocks would be 10% to 25% overvalued right now, depending on whose forward earnings estimates you use. So while the Fed model does give us hope that stocks are still undervalued, it's likely that interest rates will rise, and thus, that stocks could fall.

Finally, I must question why the 10-year Treasury bond yield is the standard, instead of the 30-year bond yield. Stock prices are based on all expected future cash flows (or earnings as a proxy for cash flow) discounted back to today. These expected future cash flows include much more than just the next 10 years, so we should probably use the 30-year bond instead of the 10-year bond to compare earnings yields to bond yields. Even this comparison, though better, isn't completely apples to apples because bond yields are stated in nominal terms, while earnings yields are real (after inflation) yields. Also, earnings grow over time, but the cash thrown off by a bond doesn't. Still, using the 30-year bond seems better to me.

By the way, the rate on the 30-year Treasury bond is currently 4.91%, implying a forward P/E ratio of 20 (1 / 4.91%)--almost exactly where the stock market P/E is today.

Next week, I'll dissect a market valuation method I like better than the Fed Model. Be forewarned, though, that there is no magic bullet for determining when to buy stocks. If I had one, I wouldn't tell anybody--and that vacation I just returned from would have been permanent.