LEADING ECONOMIC INDICATORS

A leading economic indicator is a statistic, such as housing starts, that is considered to signal the future direction of economic activity. Leading indicators tend to reach cyclical high and low points earlier than corresponding peaks and troughs in the overall economy, which makes them useful for predicting economic downturns or recoveries. In contrast, lagging indicators, such as the average prime interest rate charged by banks, generally trail behind changes in the business cycle. Coincident indicators, such as manufacturing and trade sales, tend to rise and fall along with overall economic activity.

The most popular forecasting tool of this type is the Composite Index of Leading Economic Indicators (CLI), which is published monthly by the U.S. Department of Commerce. The CLI was created in the late 1950s by a group of economists led by Geoffrey Moore. It appears in numerous magazines and newspapers across the country and is monitored closely by government policymakers, financial analysts, and business leaders who seek to predict and prepare for the future direction of the economy. The CLI is composed of a series of 11 indicators from different segments of the economy selected specifically for their tendency to predict business cycles in the near future. The Commerce Department periodically revises the data, definitions, and procedures used to construct the CLI to ensure its accuracy. For example, the percent change in sensitive materials prices was eliminated from the index in 1996.

The specific components of the CLI are: the average length of a work week for manufacturing employees; the average number of initial claims for unemployment insurance in state programs; the dollar value of new orders placed with manufacturers in the consumer goods and materials industries; the percentage of vendor deliveries that are slower, or require a longer lead time; the dollar value of contracts and orders for industrial plants and equipment; the number of new private housing units authorized by local building permits; the dollar change in manufacturers' unfilled orders in the durable goods industries; the average prices of 500 common stocks; the dollar change in the M2 money supply measure; and the change in consumer expectations.

In order to successfully predict cyclical changes in economic activity, leading indicators must be interpreted using a filtering rule. One popular rule of thumb for interpreting the movement of leading indicators is that three successive periods of decline signal an imminent recession, while three successive periods of increase signal the beginning of a recovery. Economists apply much more sophisticated statistical models to interpret the performance of individual leading indicators or the CLI.

In an analysis for the Journal of Forecasting, K. Lahiri and J. G. Wang found that the CLI predicted turning points in the economy successfully, with a lead time of about three months using real-time data, but also sent some false signals of impending down-turns. Evan F. Koenig and Kenneth M. Emery, in an article for Contemporary Economic Policy, found that the CLI predicted peaks in the business cycle an average of 8.2 months in advance and troughs an average of 4.2 months in advance. They still claimed, however, that the CLI "failed to provide reliable advance warning of both recessions and recoveries" for several reasons: there is usually a one-month delay in publication of the CLI; the peaks and troughs are gradual and difficult to recognize in real time; and the index is subject to frequent revisions.

Still, the researchers emphasized that their results do not necessarily mean that the CLI is useless. Instead, the index may be relied upon to a great extent by government policymakers. These officials may respond to changes in the CLI by enacting policies to counteract the cyclical trends in the economy. For example, the Federal Reserve might raise interest rates in order to slow an economic expansion that is accompanied by inflation. Any successful counter-cyclical actions would make the CLI appear less reliable as a predictor.