How to Read Those Statistics

Every winter the public is treated to a blizzard of year-end statistics purporting to measure the myriad business trends that affect the pocketbooks of everybody. Yet many Americans find that official Government indexes seem contradictory and confusing. The fault lies with the way statistics are compiled and reported.

Some, like the Consumer Price Index, are limited but fairly consistent and reliable estimates, while others, such as retail sales, inventories and quarterly productivity figures, are little better than ballpark guesses. One of the weakest is the index of leading indicators, which is supposed to foreshadow economic trends. Often the Commerce Department releases preliminary figures that give false signals and then, like Stalin rewriting history, subjects the numbers to revision after revision. Statistics can be made to dance to almost any tune, depending on how they are presented, particularly at year's end. Warns Economist Walter Heller: " Tis the season to be wary."

One of the most important distinctions in reporting growth, inflation or jobless trends is whether they are calculated "year over year" or "fourth quarter to fourth quarter." The difference can be substantial. Say that the economy rises fast during one year but is absolutely flat the next year. Even so, the averages would show a fairly strong year-over-year rise.

How can that be? Well, the production index began the first year at 100 in January and rose to 110 in December. The next year was flat, so the index also measured 110. No gain, right? Not really.

In fact, the average for the first year was about 105, and the average for the second year was 110. Thus, when measured on a year-over-year or average basis, the second year would show a gain of about five points over the first. Of course it would be more accurate to use the fourth-quarter-over-fourth-quarter measure, which would show no rise. If the forecasts of TIME'S Board of Economists are correct, next year's G.N.P. will rise about 2% year-over-year but will not increase at all in a fourth-quarter-to-fourth-quarter measure.

This file is automatically generated by a robot program, so viewer discretion is required.