Normally, yields on long-term government bonds are higher than yields on short-term ones. Investors demand a bigger return for the risk that comes with holding an investment that takes longer to repay.
The relationship has been upside-down since July, however, with yields on short-term U.S. Treasury bills exceeding those on long-term Treasury notes. Late Friday, the yield on the three-month Treasury bill stood at 5.05%, well above the 4.648% yield on the 10-year note.
This unusual state of affairs -- known as an inverted yield curve -- has gone on longer than many economists expected and has some wondering whether the bond market is signaling that the economy itself could turn upside down.
Even non-Wall Street types are starting to notice. Charlotte Observer sports columnist Rick Bonnell likens an inverted yield curve to a basketball player whose shooting percentage is lower at the free-throw line than from the field. It's uncommon and nerve-wracking.
Just as hoops star Shaquille O'Neal's ineptitude at the free-throw line gives his opponents strategic information -- foul the big man to prey on his weakness -- a yield-curve inversion may offer insights about the economy's own Achilles' heel.
Yield inversions, many analysts say, are harbingers of hard times. When bond investors see a recession coming, they tend to buy long-term Treasury securities for two reasons. First, they are safer than stocks. Second, they are appealing when inflation is low, and recessions tend to beat down inflation. The buying that comes with recession fears drives down a long-term bond's yield, sometimes below the prevailing yield on short-term Treasury securities.
The market, in effect, is betting that the Federal Reserve, which dictates short-term rates, will have to cut its overnight fund rate to boost the economy, and investors are pushing long-term rates down in anticipation.
Inverted curves often come toward the end of Fed rate-increase cycles. Fed Chairman Ben Bernanke halted the central bank's latest rate-increase campaign in August.
The Fed's overnight funds rate now stands at 5.25%. "The market is pricing in that the Fed funds rate is not going to stay at 5.25% over the next 10 years; it's pricing in that it's going to be considerably lower," said Bill Irving, who runs a long-term Treasury-bond-index fund for Fidelity Investments.
Some economists doubt the yield curve's effectiveness as a recession-forecasting tool. They think long-term rates are exceptionally low right now for other reasons, including lower long-term expectations about inflation and growing demand for U.S. government bonds from foreign investors needing somewhere to park their money.
But those who think highly of the yield curve's predictive power have history on their side. Seven times between 1965 and 2005, yields on the 10-year note have dropped below those on the three-month Treasury bill for an extended span. In six of those instances, the U.S. economy went into recession soon after.
For example, the 2001 recession was predated by a yield-curve inversion that lasted from July 2000 to January 2001. In the one time when a recession didn't follow, in the mid-1960s, there was still a sharp slowdown in growth.
"The yield curve gives you information that you should pay attention to," said David Roberts, chief economist at Dominion Bond Rating Service.
When the yield curve inverted for two weeks last February and March, the inversion between long-term and short-term rates was so narrow and lasted so briefly that many economists dismissed it.
The current inversion is harder to ignore. The disparity has grown deeper and has endured for months; the three-month/10-year inversion is entering its 26th week.
"The longer it persists, the more of a puzzle it becomes," Mr. Roberts said.
One economic-forecasting tool using Treasury yield-curve data pegs the chances of a recession at nearly one in two. The model, which was developed by Fed economist Jonathan Wright, takes into account yields on 10-year and three-month Treasury securities as well as the Fed's overnight funds rate.
Another forecasting model -- developed by Federal Reserve Bank of New York economists using only the 10-year/three-month spread -- puts the chances of a recession in 12 months at just under 40%.
Those predictions are at odds with the consensus among economic forecasters.
A recent survey of economists by The Wall Street Journal pegged the chances of a recession within the next 12 months at 27%.
Two researchers who focus on recession forecasting, Lakshman Achuthan and Anirvan Banerji of the Economic Cycle Research Institute, argue that the yield curve is overrated as a recession harbinger. They note that the yield curve failed to invert before recessions in the 1950s and early 1960s. They also point to the misleading signal sent in 1966-67, when a lengthy inversion didn't precede a recession.
Messrs. Achuthan and Banerji argue that the economy and financial markets have changed greatly in recent years.
They note that pension funds, oil-producing nations and other cash-laden institutional investors around the world have been pouring money into long-term Treasurys, which they say creates "artificial" pressure that pushes long-term bond prices higher and their yields lower.
"Once the dust settles and we look back a year or two from now, we're going to see [the inverted yield curve] as a false alarm with respect to a recession," Mr. Achuthan said.
Dominion Bond Rating's Mr. Roberts also isn't predicting a recession. But he said the yield curve's long inversion does sound a cautionary note for him -- and should do the same for anyone who is trying to figure out where the economy is going.
"I think that a prudent businessperson would say: 'Maybe I should look
more carefully at some other sources of information. Maybe I should be more
skeptical,' " he said.
Write to Michael Hudson at michael.hudson@wsj.com