Forecasting Business Trends, Part Three: Method of Analysis
AIER analyzes individual series rather than using them to construct a composite index, such as the Conference Board’s Leading Economic Index. The data for each series are plotted on a chart. Historical data are replotted when revisions are reported. If necessary, we calculate and plot a moving average for each series to smooth irregular fluctuations and isolate cyclical movements. The number of months a series is smoothed ranges from one to six.
To assess the cyclical status of each series, we compare its most recent changes with past patterns. If a series changes in the direction of an already established trend, its cyclical status remains the same. If, however, a series diverges from an identified cyclical trend, we must assess whether the divergence is temporary or is the beginning of a cyclical reversal.
A one-month reversal in the moving average of a series does not necessarily establish a new cyclical trend for that series. A trend of several months’ duration and of a significant magnitude usually is needed for that. The longer the duration of a decreasing trend, for example, the greater the probability that the downward trend reflects a cyclical contraction.
A series does not have to decrease every month, however, to establish a downward trend. It may increase for one or more months and still be trending downward during the period. As long as the series does not regain the peak from which the decrease began, the trend is considered downward.
No single series defines the business cycle. Any individual series in a given cycle may peak or trough sooner or later, in comparison with the turning points in economic activity, than would be expected from its role as a leader, coincider, or lagger. Sometimes a given series may not even exhibit any cyclical fluctuation at all throughout a recessionary episode.
This is not only why business-cycle analysts must consider many different series but also why judgment cannot be eliminated from the process. What we look for is an overall pattern to the changes in the indicators. In assessing the cyclical status of a series, our procedures are not always as “cut and dried” as the foregoing discussion may suggest.
In our monthly appraisal of the cyclical status of a series, we may classify a series as expanding, probably expanding, contracting, probably contracting, or indeterminate.
After assessing each series, we calculate the percentage of each group (leaders, coincident, and laggers) of indicators that we appraised as expanding cyclically.
The percentage of a particular group expanding during a given month is important, as are the relationships of each group to the other groups and to past percentages. We begin with the leading series. Their function is to provide warning of changes in economic activity. Although a cyclical turn in the majority of the leaders may be a false indication of a turn in general business activity, since 1947 no cyclical turn in business activity has occurred without a prior turn in such leaders.
A decrease in the leading indicators from 100 percent expanding to a lower percentage during one month, or even for several months, does not necessarily indicate an impending contraction in business activity. Only when less than 50 percent of the leaders are appraised as expanding do we suggest that a contraction in business activity is probable.
A few months (but sometimes as much as a year) after the percentage of leaders expanding has decreased to less than 50, the percentage of coincident series expanding usually will begin to decrease, supporting the expectation that a contraction of business activity is underway.
Confirmation of such would be needed from the lagging indicators, which measure structural imbalances within the business cycle. Imbalances that develop during the late stages of an expansion, such as inventory overhang and increasing labor costs, help to further dampen economic activity measured by the leaders. Thus, the laggers should not be viewed only as a result, but also a cause of changes in business activity.
Once a cyclical contraction of business activity has been identified, a similar method of evaluation is used to identify signs of recovery. In both instances, the leading indicators generally will turn first, followed by the coincident series, and then the lagging series.
