Experimental Index Will Test Traditional Theories
The index of leading indicators is familiar to many business-cycle observers. Developed in 1937 at the National Bureau of Economic Research by Wesley Mitchell and Arthur F. Burns, and amended by Geoffrey Moore and others, the index averages the movement of several economic variables chosen for their ability to anticipate booms and recessions.
The index for August, the last available month, declined sharply, consistent with the emerging view among most forecasters that the economy is slowing.
Amid this growing concern about recession, a new forecasting index--the experimental index of leading indicators--has produced a startlingly strong forecast of nearly 4% growth in the economy the next several months. A companion index places the probability that the economy will be in a recession next February at a minuscule 3%. Why such a big discrepancy between these predictions and the conventional ones? And should we stop worrying about recession and plan for a boom?
The XLI has serious credentials. It is the result of sophisticated statistical research by James Stock and Mark Watson, two highly regarded econometricians working under the auspices of the National Bureau of Economic Research. Their index is constructed in much the same spirit as the official leading indicators were but presumably benefits from using the extensive database and number-crunching capacity that is available to econometricians today.
The official index of leading indicators contains 11 variables, chosen because they have shown predictive power historically. They are: hours worked in manufacturing, initial claims for unemployment insurance, manufacturers’ new orders for consumer goods and materials, vendor performance, contracts and orders for plant and equipment, building permits, change in unfilled orders, change in sensitive-materials prices, stock prices, the money supply and an index of consumer expectations.
The XLI is a weighted average of seven variables. Although the list includes building permits and unfilled orders--in common with the official index--and an index of workers employed part time, it contains several financial market variables that have not been part of the leading indicators. These are the trade-weighted U.S. exchange rate, the yield on 10-year Treasury bonds, the spread between interest rates on six-month commercial paper and six-month Treasury bills, and the spread between interest rates on 10- and one-year Treasury bonds and bills.
The last two variables are contributing the most to predictions of economic growth.
It is easy to see why these variables might help predict recessions. The spread between private and Treasury borrowing costs should reflect the risk of default on private debt. If lenders foresee a recession, they will see a greater risk of default and require a higher interest rate to compensate.
Thus, if private investors are able to forecast recessions, a wider spread between private and Treasury interest rates should signal a higher probability of recession and a smaller spread the opposite.
The yield spread between long- and short-term Treasury bonds and bills is interpreted as a forecast of the course of interest rates. Rates fall in recessions. So if investors expect a recession in the near future, they will push down long-term rates relative to today’s short-term rates.
Also, to the extent that recessions are brought about by the Federal Reserve raising short-term interest rates to slow down the economy, recessions will be preceded by a period of exceptionally high short-term rates and a narrow or even inverted spread between long- and short-term rates.
Why might these stories be wrong this time? The long-short spread is most persuasive when monetary policy is raising short-term interest rates to slow inflation and, incidentally, causing a recession. That behavior was an important factor in many of the past recessions. But a recession emerging from sources other than a tightening monetary policy would not be preceded by a strong signal from the long-short spread. The present situation may fit that description.
In addition, long-term interest rates may be exceptionally high today for special reasons. The flood of long-term borrowing for the thrift industry bailout and the apparent withdrawal of some Japanese investors from our bond market are two that come to mind.
Lack of a recession signal from the private Treasury spread is a bigger puzzle. Indeed, it may just be that the official series used in the XLI is simply not reflecting what is going on in credit markets. Recently, credit ratings of well-known firms have been downgraded. In addition, the yield spread between junk bonds and Treasuries has soared. On this interpretation, an alternative indicator of the risk premium on private debt might be telling a very different story.
The XLI is an interesting addition to the business-cycle literature. Perhaps its next reading, due at the end of this month, will tell a gloomier story than the last. If not, the index will either become a sensation for correctly seeing strength in the economy where others saw weakness, or Stock and Watson may want to go back to their data bank to make some alterations. That is the nature of research, and it is why experimental is the first name of the new index.
