Stock valuations getting harder to explain

It's getting harder for bulls to justify the lofty valuations in U.S. stocks, and one of the last arguments left, that equities are actually cheap when compared against bonds, is increasingly being called into question.

While companies may be priced high relative to items such as sales and earnings, the values placed on fixed-income assets are much loftier. That's important because the two markets used to move in lockstep, from the early 1980s through the internet bubble. Since then, the two have diverged, with one bullish conclusion being that equities have room to soar about 40 percent before catching up to the historical relationship.

The problem for investors is that values have diverged for so long now that it's possible something has fundamentally changed - and that may keep stocks from rallying as much as they would otherwise. While companies have benefited from low interest rates and slowing inflation the past three decades, the worry now is whether it's become too much of a good thing.

"We have very low inflation or deflation and that means continued investment in fixed income, but not great prospects for earnings growth or stocks," said Jason Brady, a portfolio manager at Thornburg Investment Management, which oversees about $60 billion. While the comparison worked in the past, it breaks down when "it's clearly part of the design of central banks to push rates down below what would be a natural level," he said.

Comparing stock and bond valuations is sometimes known as the Fed model, derived by U.S. economist Edward Yardeni from a July 1997 report by the central bank. Companies may be priced high versus earnings, the theory goes, but as long as bond valuations are even higher, equities have room to soar.

Proponents of the Fed model frame valuations in terms of yields, the opposite of price-earnings ratios. Viewed this way, the S&P 500 offers an earnings return of about 5 percent, meaning for every dollar invested, a nickel comes back as profit. That compares with an interest payment of 1.4 percent on 10-year Treasuries, or 1.4 cents for each dollar spent.

People fashioning an investment case focus on the difference between the two payouts and how it compares with history. At 5 percent, stocks sport an earnings yield that is 3.6 percentage points higher than Treasury rates, greater than any time between 1980 and the financial crisis. In its classical interpretation, that's too much of an advantage for investors to resist, spurring a buying spree in the equity market that eventually brings yields back into line.

"What matters is whether or not you think they're going to mean revert," Dan Suzuki, an equity strategist at Bank of America, said by phone. "You could've made the argument for the last 10 years and it would be wrong. They haven't reverted, so what you're basically saying is something that for 10 years didn't happen is going to happen."

Of course, other things could happen to reduce the gap. Bond yields could rise, or earnings could collapse. An analysis by Suzuki's group found little statistical evidence to assume an out-of-kilter Fed model usually rights itself via gains in equities. Analysts at the firm found that readings in the Fed model have explained just 7 percent of future returns in shares.

"Convergence to the long-term average could take place from either direction," Goldman Sachs analysts led by David Kostin wrote in a July 15 note. "The most probable path involves inflationary pressures pushing bond yields higher while negative earnings revisions are among the risks that will drive share prices lower near term."

The Fed model fares poorly compared with simpler methods of judging when stocks are attractive. A study by Bank of America found that as a tool for predicting 10-year returns, P/E ratios were 10 times more accurate as an indicator of what stocks would do, while metrics like price to book value did even better. The S&P 500 trades for 20 times annual earnings, the highest P/E since 2009, and 2.9 times book, or assets minus liabilities.

In many respects, the case against the Fed model is the same as the case against any investment strategy predicated on valuation: they're fraught with false signals. Data compiled by Bloomberg show that even the most conservative applications of price-earnings ratios as a buy or sell indicator are effectively useless for predicting how stocks will perform in the coming year.

Consider the bull market of the mid-2000s and its aftermath. While the valuation disparity between stocks and bonds widened as shares rallied from 2002 to 2007, nothing in the Fed model changed heading into 2008, a year the S&P 500 plunged 38 percent. In other words, stock and bond yields gave off no signal to predict the worst equity sell-off in seven decades.

One way of thinking about how the Fed model became popular is that it was framed as signaling a bull case over a period when stocks happened to rise. And to be sure, stocks could keep rising regardless of whether the gains are being fueled by a valuation disparity with the bond market.

"It's almost ridiculous to talk about it," said Yousef Abbasi, a global market strategist at JonesTrading Institutional Services. "For a more compelling case, it would be nice to see earnings improvement."

The S&P 500 is up almost 7 percent in 2016 and last week closed at a record for the 14th time this year. Stocks have advanced in five of the last six weeks, pushed up by better-than-forecast earnings among technology companies and the two strongest months for U.S. payroll growth of 2016.

At the same time, earnings have shown few signs of reversing a string of five quarterly earnings declines. That's fodder for bears who say that when the valuation gap between stocks and fixed income narrows, it will be because declining profits widen the price-earnings ratio of the S&P 500.

"Two out of three of the ways the equity risk premium can revert are generally negative for stocks," said Bank of America's Suzuki. "Even if you want to use this model, it's not a clear-cut model for stocks."