November 2000

The previous two issues of Business Cycle Indicators discussed major enhancements to the methodology of the Leading, Coincident, and Lagging Indexes planned for January 2001. We intend to improve the composite indexes by imputing data components that are not available at the time of publication and use them, together with already available data, to release the composite indexes on a more timely schedule. Specifically, we plan to publish the composite indexes at least two weeks earlier than was previously possible.

In addition to these improvements, beginning with our first 2001 Leading Economic Indicators release, The Conference Board will also remove the process that equalizes the volatility of the composite indexes. The current methodology pegs the month-to-month variance of the Leading and the Lagging Indexes to the variance of the Coincident Index.

In technical terms, the current procedure calculates the sum of the component contributions to the Leading Index, and then multiplies that sum by a standardization factor that makes the average historical monthly volatility of the Leading Index equal to that of the Coincident Index. The standardization factor is calculated as the ratio of the standard deviation of the monthly percent changes of the Coincident Index to the standard deviation of the monthly percent changes of the unadjusted Leading Index. The Lagging Index is standardized similarly.

The original purpose of this standardization process was to simplify comparisons of monthly changes in the three composite indexes. In a standardized index, any monthly change that exceeds 1.0 percent is an above-average change, and thus carries some significance. Also, when they are charted together, the volatility of all three standardized indexes is identical. The standardization process, therefore, adds some consistency to the composite cyclical indexes, and facilitates analysis of them.

These properties, however, have turned out to be unimportant. Month-to-month changes in the indexes have always been too volatile to have significant analytical content, and thus no one pays much attention to one-month changes in and of themselves. Business cycle forecasters, instead, focus mainly on movements in the indexes across 3, 6, or 9 month periods (econometric models do so as well, incidentally, and for the same reason). Standardization of month-to-month changes in the indexes affects those somewhat longer movements hardly at all, and thus is not particularly helpful to analysts.

Effects of the Change

The accompanying charts, which compare the indexes with and without standardization, demonstrate this clearly. Chart 1 shows that the standardization process slightly changes the long-term trend of the Leading Index while the cyclical patterns remain the same.

Moreover, the months at which the two series reach cyclical peaks and troughs are identical. In other words, the length of leads at cyclical turning points remain unchanged. While month-to-month movements in the unadjusted composite indexes are more pronounced, the correlations between the Coincident Index and each version of the Leading Index remain virtually identical. Most telling, as you look at this chart, nowhere can you find a time across these four decades of business cycle history where your forecast would differ depending on which of these two series you relied upon.

Charts 2 and 3 make the same comparison for the Lagging Index and the ratio of the Coincident Index to the Lagging Index. In each case, the differences are even smaller than those described earlier for Chart 1. Again, the peak and trough dates are identical, as well as correlations between the Coincident Index and the ratio. Here, too, the standardization process amounts to a scale adjustment that has almost no impact analytically.

Finally, it should be emphasized that the month-to-month changes in the standardized Leading and Lagging Indexes are not comparable to changes in the unstandardized indexes. For the same reason, the month-to-month variance in the Coincident Index will no longer be comparable with the monthly variance of the unstandardized Leading and Lagging Indexes.

Summing Up

Standardizing the month-to-month changes in the cyclical indexes does not make a meaningful difference to their analytical value. It does not hurt them, but neither does it help, at least not significant enough for it to count. Accordingly, beginning January 2001, The Conference Board will publish the Leading and Lagging Indexes without standardizing the month-to-month changes.