Prediction by Index
by Clifford J. Sherry, Ph.D.
The Composite Index of Leading Economic Indicators is a summary measure designed to indicate
changes in the direction of aggregrate economic activity. The Index measures the average behavior of a
group of 12 economic time series that show similar timing at business cycle turns. They represent
different activities or sectors of the economy. The Index was first developed in the 1930's by the National
Bureau of Economic Research and has been published since 1961 by the Department of Commerce in
Business Conditions Digest. The Index tends to lead at business cycle turns and this look ahead quality
makes it a valuable tool for predicting changes in the economy, such as impending recession.
The procedures used to combine the individual economic series into the Index are designed to prevent the
volatile series from dominating the Index and to give more influence to the better performing series. The
economic series that are incorporated into the Index tap the following economic processes: employment
and unemployment; production income, consumption, and trade; fixed capital investment; inventories
and inventory investment; prices, costs, and profits; and money, credit, and interest rates. Selecting and
classifying these individual economic series that become part of the Index is a place where economic
theory and empirical observation closely interact.
The value of the Index by month from 1980 is shown in Chart 1. These values are often used to
determine the end of economic recovery and the beginning of a recession before it occurs. There are a
number of different methods for looking at these figures in order to determine if a recession is eminent.
Four filters are commonly used. Filter one looks for a decline of any size from the previous month
(technically not a filter at all); filter 2, a decline of any size, for two months in a row; and filter 3 for a
decline of any size for any three months in a row. The fourth filter seeks two consecutive months of
negative and decelerating growth. For example, if the changes in the Index is -1, -1, -2, this would not
constitute a signal of a recession, since the middle month, although negative, is not decelerating from the
previous month. If the value were -1, -1.5, and -2 or -0.5, -1, and -2, etc., in other words, the growth in
each of the last two months was negative and algebraically lower than the previous month, then it
satisfies the criteria of the fourth filter. The performance of these four filters as reported by Dr. Howard
Keen for each of the postwar expansions are shown in Table 2.