Forecasting Business Trends, Part Four: The Cyclical Score

Part Three of our four-part web series about our forecasting methods outlined this process. In this final installment, we describe our alternative measure of leading indicators, the Cyclical Score, which is also published in the first Research Report of every month.

Although the procedure for calculating the percentage of leaders expanding is straightforward, it does not allow for any shades of gray. Each series is given a specific cyclical status each month and a series reaching a new high for the cycle has the same “weight” as one that has decreased for several months and is on the verge of an indeterminate status.

The Cyclical Score was devised to address these limitations.

As with the leading indicators, the Cyclical Score theoretically can fluctuate between 0 and 100, with a score of less than 50 percent indicating that a contraction in business activity is probable. But it differs from the percent expanding series in several respects. To begin with, the Cyclical Score is a purely arithmetical calculation that does not reflect the judgments of AIER’s staff in any way.

It also is based on the current list of leaders each month. This means that the data for, say, January 2001, reflect all historical revisions and may include series that were not on the list of leaders then. Consequently, the historical record of the Cyclical Score may itself be revised whenever a series is revised or one series is dropped and another substituted. (The percentage expanding series is, in contrast, a record of the monthly findings of AIER’s staff based on the leading series then in use and then available. The percentage expanding series is never revised.)

To calculate the Cyclical Score, series that are at a new high with an established uptrend are given a score of 100, and those at a new low in an established downtrend are given a score of zero. The score for other series will depend on the magnitude and duration of the reversals of their most recent trend.

The Cyclical Score is simply the average of the scores of the 12 individual leading series. This allows for something other than an all-or-nothing contribution of a given series to the final result. Moreover, in an effort to reduce the incidence of false signals, the calculation of the overall score from the individual series provides for a heavier weighting of increases than decreases.

We rely on the cyclical score primarily to supplement the percentage expanding series. For example, if the percentage of leaders appraised as expanding indicates that a recession is probable, but the cyclical score of the leaders does not, we would be hesitant to assert that a recession is imminent. On the other hand, if both series were to decrease to less than 50, we would be more confident in offering such an appraisal.

In addition, if many indicators were indeterminate, the percentage of leaders expanding—which ignores such series entirely—could render a misleading outlook. Thus, the cyclical score, which takes into account all 12 series regardless of their appraisal, would play an important role in our assessment of conditions.

We stress again that our use of the statistical indicators of business-cycle changes is but one tool available for helping to forecast the near-future cyclical trend of business activity.