Following useless indicators can cost you money.
Over the past few weeks, I've repeatedly heard pundits claim that stocks are now undervalued, since the Dow's dividend yield has surpassed that of the 10-year Treasury. When I asked financial economist Andrew Smithers, the author of Wall Street Revalued, for his opinion on this, he didn't mince words:
It is clearly nonsense to claim to value equities on the basis of bond yields. It assumes that equities are not correctly priced, but that bonds are. Why not point to the yield relationship and say that we have a bond bubble on our hands?
This is a version of the so-called Fed Model which argued that you could price equities by reference to the relationship between the earnings' yield on equities and bond yields [Note: The earnings yield is the inverse of the P/E ratio]. The evidence was that the relationship had been stable from 1981 to 1997. This was one of the greatest successes ever achieved by data mining, as the inverse relationship held, in a slightly stronger form, from 1950 to 1968.
Recently, my esteemed colleague Morgan Housel used the Fed model to argue that stocks are the best buy right now. He examined the historical data, making an imaginative adjustment to the dividend yield to account for the enormous change in corporate and investor attitudes toward dividends that occurred post-WWII. Indeed, as Morgan noted:
Today, some of the most profitable and fastest-growing companies -- including Apple (Nasdaq: AAPL), Google (Nasdaq: GOOG), and Cisco (Nasdaq: CSCO) -- pay no dividends at all. The slow-growers -- like Altria (NYSE: MO), Verizon (NYSE: VZ) and Consolidated Edison (NYSE: ED) -- are where you find yield. That was unheard of 60 years ago.
The adjustment was a good instinct, and it led Morgan to this discovery:
… [Once you adjust for the change in payout ratio] there would have been only two [prolonged periods in which stocks out-yielded bonds] in modern history: from 1940-1944, and 1947-1955.
And what's neat about these two periods? They were both phenomenal times to buy stocks. In the 10 years after 1944, stocks surged 161%. In the 10 years following 1955, investors were rewarded with a 145% return...
A leap too far
All that is true, but it doesn't support Morgan's final, fatal leap of logic: "Today, with the average large-cap stock out-yielding Treasuries, there's little reason to think patient investors won't be rewarded like champions 10 years from now."
Why not? Let's get back to those impressive 10-year numbers, and find out what's behind them. In the table below, I've annualized the price returns, and broken them down into two components: earnings growth and the change in the price-to-earnings multiple (P/E). The first measures corporate performance, while the second captures what the market will pay for that performance.
|
Metric |
1944 – 1954 |
1955 – 1965 |
Historical Average (1926-2000) |
|---|---|---|---|
|
Earnings Growth |
11.5% |
4.3% |
4.9% |
|
Change in P/E Multiple |
(1.3%) |
4.5% |
1.2% |
|
Return (Ann.) |
10.1% |
9% |
6.2% |
Source: Robert Shiller and Stock Market Returns in the Long Run: Participating in the Real Economy (2003), Ibbotson and Chen.
It's all in the earnings growth (and more!)
In the first period, earnings growth contributed more than 100% of the return, because the P/E multiple contracted, subtracting from the return. Note that the annualized earnings growth rate for the 10-year period, 11.5%, is more than twice the average historical growth rate (4.9). In other words, stock investors earned high returns during this period not because stocks started out cheap, but because earnings growth was unusually strong. We certainly shouldn't count on that occurring over the next 10 years, especially with consumers continuing to deleverage.
In the second period, the contributions of the two factors are almost equal, but while earnings growth is close to its historical average, the increase in the P/E multiple contributes much more heavily to stock returns than the long-term average (4.5% vs. 1.2%). This does suggest that stocks were undervalued at the start of the period.
Don't be fooled by this indicator!
The conflicting indications regarding stock valuations in the two examples show that even though stocks are out- yielding bonds, that still won't tell us whether stocks are cheap or not. If you do believe stocks are cheap now, you'd best have another reason for it.
Incidentally, James Glassman and Kevin Hassett relied on a variation of the Fed Model in coming up with Dow 36,000, which was published in October 1999. We all know how well that madcap prediction worked out.
From 1926 to 2000, dividends contributed 40% of stocks' average annual return. Over the next decade, their contribution could be even higher. Matt Koppenheffer knows where to find the best dividends.
