A Better Way to Value the Stock Market

Last week, I highlighted a common method for valuing the stock market called the Fed Model. The essence of this model is that it compares the interest yield on 10-year Treasury bonds to the 12-month forward earnings yield for the stock market. The forward yield on the stock market is currently between 4.5% (using economists’ estimates of 2004 S&P 500 earnings) and 5.2% (using Wall Street analyst estimates of S&P 500 earnings), and the yield on the 10-year Treasury note is 4.05%. According to the Fed Model, then, stocks would have to rise 12% to 34% from their current level to get to fair value.

However, if interest rates rise to a level anywhere near their long-term average (6% or so), stocks will suddenly appear overvalued by the Fed Model. No one can predict where interest rates will go, of course. For this reason, I came to the conclusion that the Fed Model is much better at explaining past market valuations than predicting future ones (unless you are able to predict interest rates).

Is there a model that does a better job of predicting the future? Yes, there is. Before I get there, though, I’d like to briefly discuss some other methods for determining whether the market is over- or undervalued.

Historical P/E Ratios
This is probably the most common logic you’ll hear when someone is arguing in favor of buying or selling stocks. This valuation method compares the long-term average P/E ratio of the stock market (often defined as the Dow Jones Industrial Average) with the current valuation.

Over the past century, the average P/E for the Dow has been about 16. Currently, it’s 20. (The current P/E ratio is printed every day on page C2 of The Wall Street Journal.) Thus, the Dow is about 20% overvalued based on its long-term average P/E ratio.

There are at least three problems with this valuation method. First, it uses the trailing P/E ratio on the Dow. Stocks are priced based on future expectations, not past ones. In other words, trailing P/E ratios are not necessarily predictive of the future (though to be fair, there’s some good evidence that trailing P/E ratios and future stock returns are negatively correlated. See, for example, Robert Schiller’s book Irrational Exuberance.)

Second, and more important, there are some logical reasons why current P/E ratios should be a little higher than the long-term average. For one thing, trading volumes are much higher today than they were 100 years ago. Thus, liquidity is better, making it easier to get out of a stock if something goes wrong. Also, the rise of mutual funds and the dramatic drop in brokerage commissions has made it less costly for investors to diversify. And an argument can be made the Fed’s monetary policy is more advanced today than it was at the beginning of the 20th century, so that the economy is more predictable and less prone to extreme swings. For example, at the beginning of what became the Great Depression, the recently created Fed actually raised interest rates and tightened the money supply, causing massive deflation and a banking panic. This would never happen today--we have reams of data on the money supply and its effect on the economy. Today, our banking system is healthier and not easily susceptible to panics. We have learned something from our mistakes, so to speak. Thus, an argument can be made that, all else equal, the U.S. economy is more stable today than in the past.

For these reasons, and a few others that I didn’t mention, it may well be that the trailing P/E ratio for the Dow should be 20 today, and not 16. Or maybe it should be higher than 20, or maybe a bit lower. The jury is out. As a result, I’m skeptical whenever someone tries to justify buying or selling stocks based on historical average P/E ratios. History may be set in stone, but valuations are a moving target.

The Ben Stein Method
Ben Stein (yes, that Ben Stein) and Phil DeMuth have come up with a better method for valuing stocks based on trailing data. Stein and DeMuth use the trailing 15-year moving average of the market P/E ratio and compare it with the current market P/E ratio. The authors also compare the current inflation-adjusted price of the index, as well as the price/book, price/sales, and price/cash flow ratios to their 15-year moving averages. These metrics seem to work pretty well, based on back-testing the authors present in their 2003 book Yes, You Can Time the Market!

This method seems intuitively better than the one I presented earlier comparing the current P/E to the 100-year average P/E. If it’s true that market valuations are a moving target, Stein and DeMuth’s method should capture that movement because it relies on just 15 years of trailing data, rather than 100. Of course, the data is trailing and so not necessarily predictive. However, I like this method and check regularly for updates on Stein and DeMuth’s Web site. The most recent update has the market looking significantly overvalued. You can see for yourself by going to www.YesYouCanTimeTheMarket.com.

The Q Ratio
The "q ratio" measures the aggregate stock market valuation compared with the economic replacement cost of the assets in place. It might seem similar to the price/book ratio, but it isn’t really. The q ratio measures the current inflation-adjusted replacement cost of assets. Book value is based solely on what a company paid for its assets in the past, and is often distorted by such things as inflation, arbitrary depreciation methods, one-time writeoffs, and other assorted accounting conventions.

Researchers Andrew Smithers and Steven Wright wrote a book about the q ratio called Valuing Wall Street in which they discuss the q ratio in far more detail than I have room for here. Suffice it to say that they currently estimate the adjusted q ratio to be 1.9 (1.0 is the long-run average). This means the S&P would have to fall about 49%, to 580, to get to fair value. To get regular updates on this data, go here.

The big problem with this measure is that it discounts the replacement cost of intangible assets. Thus, if the U.S. economy had $4 trillion in hard assets and $4 trillion of intangibles (patents, goodwill, etc.) on its aggregate balance sheet, the q ratio would include only the hard assets in its calculation of market value-to-asset value. Over the past 30 years, the U.S. has shifted rapidly toward an economy driven more by intellectual property than by hard assets. The authors claim that this doesn’t make a difference because, in aggregate, the value of intangible assets relative to hard assets hasn’t changed (though for any one individual company, this is not necessarily the case). I’m skeptical of this claim.

Morningstar Median Price/Fair Value Ratio
I’ve highlighted four methods for valuing the market: the Fed Model, historical P/E ratios, the Ben Stein method, and the q ratio. While all these have their strengths and weaknesses, there are two common problems they share: First, they all rely on aggregate data, rather than data on individual companies. Second, they're all market-cap-weighted (in the case of the S&P 500) or price-weighted (in the case of the Dow). Thus, these measures don’t give you a feel for the valuation of the average stock. Most investors don’t weight their portfolio based on market caps of the companies they hold--save for index funds--so it seems odd to rely on such data as a measure of valuation. What these measures really tell you is the valuation of index funds, not of the average stock in the marketplace.

Which brings me to what I think is the most logical method for valuing the aggregate stock market--the Morningstar median price/fair value ratio. Here’s how we calculate it:

First, our analysts use a discounted cash-flow model to estimate fair values for each of the 514 companies we cover. These 514 represent about 83% of the aggregate value of the Wilshire 5000 index. Most of these are large caps like  Amgen (AMGN),  Home Depot (HD), and  American Express (AXP), though we do cover a few dozen mid-cap companies as well.

Next, we compare the price of each stock to its fair value in order to get a price/fair value ratio for each stock. Then we rank all 514 of these from lowest to highest.

Finally, we take the median price/fair value ratio in this universe of 514 stocks. And that gives us the unweighted median price/fair value ratio for the stock market.

The Morningstar median price/fair value ratio essentially treats the entire U.S. stock market as if it were one large company with 514 separate divisions. We take the price/fair value ratio of each division and calculate the median. When you hear people say, "It’s a market of stocks, rather than a stock market," this is essentially what they mean. We treat each stock individually to get a bottom-up estimate of the aggregate market.

Now, this method makes sense only if you agree that stocks should be valued based on the risk-adjusted value of their future cash flows. And of course that must be the case--the price of any financial asset (bond, real estate, or whatever) is just the present value of the future cash flows you expect to receive from it.

The advantage of this method is that it’s based purely on projections of the future, rather than being a measure of some backward-looking historical average. I also like this method because it’s based on human judgment--namely, the estimates of each of our 35 analysts. These analysts make unbiased estimates of future cash flows on companies with which they are intimately familiar. If you assume that our analysts are, on average, representative of "the market," then, on average, our fair value estimates should be rational. We have no conflicts of interest, so we have no incentive to produce artificially high or low fair value estimates. We will, of course, be incorrect on many of our fair value estimates, but on average we should be about as good as the market at predicting future cash flows, and thus fair values.

We’ve created a graph that shows the median price/fair value ratio over time for our whole coverage universe. You can also manipulate the graph to see the median price/fair value for various subgroups of our coverage universe, such as low-risk stocks, wide-moat stocks, and NYSE stocks. To see the graph, click here.

A teaser: Our data shows the median price/fair value for the market is 1.08 right now, meaning the average large-cap stock is about 8% overvalued by our estimates. A year ago, the market was about 15% undervalued. Play around with the graph a little bit; I think you’ll find it fascinating.