Unemployment Figures

AS the March unemployment figures came out last week (see NATIONAL AFFAIRS), the big question was whether they had improved enough to show a leveling off in the recession. To some economists, both amateur and professional, they had not. The normal seasonal increase in employment between February and March is 500,000 and the normal decline in unemployment about 200,000. By contrast, the March figures showed only a 323,000 increase in employment while unemployment actually edged up another 25,000. To some thoughtful experts, an equally important question is: Should the employment figures be used for hairline appraisals of the U.S. economy? The answer is no. Employment statistics are not intended for such fine measurements.

Chief value of the statistics is to show trends and changes in the U.S. labor force. Of the overall 2,316,000 jump in unemployment to 5,198,000 since March 1957, more than 1,000,000 came in such big durable-goods producers as autos, aircraft, heavy machinery and steel. The curious fact--and the paradox of the recession--is that other industries, such as wholesale and retail, banks, services (up more than 100,000), real estate and insurance firms, are still booming ahead. In March, employment rose by 323,000 over February, bringing total employment to 62.3 million, the third highest March in history--all while unemployment also showed a 25,000 increase.

How do the statisticians account for the discrepancy? The major cause is a 778,000 increase in the total labor force over the last two months, which many people overlook in analyzing the figures. The one-month increase last February of 428,000 was abnormal for that time of year, and occurs only during recession periods; the winters of 1949 and 1954 both brought similar jumps. The reason was that most of the new "laborers" were not normally members of the labor force; they were the wives and teen-age sons and daughters of the laid-off family breadwinner, out looking for jobs to help tide the family over.

In its jobless checkup, the Census Bureau does not try to find out how many of the jobless are such new workers, how many actually lost their jobs. The census takers only ask: "Are you looking for work?" And everyone who is "looking for work," no matter how lackadaisically, is counted as a member of the labor force. Thus, as the size of the labor force increases, the number of jobless can also increase, as happened last month, even when the number of employed takes a big jump. Economists would like the Census Bureau to add more questions to separate the laid-off worker from the new job seeker. But so far the Census Bureau has said no; it fears any change will destroy the month-by-month continuity of its figures.

Nevertheless, the Census Bureau has already made changes that raise a doubt about this continuity. Many an economist has estimated that the current recession is worse than 1949-50 or 1953-54 because the unemployed percentages appear bigger. But the figures are not completely comparable, because both the definition and the sample have changed. Until last year, the census takers counted all workers laid off for 30 days or less as employed. Last year the rules were changed to count such workers as unemployed. In this fashion, the statisticians arbitrarily added another 250,000 to the unemployment totals. While the census takers have tried to adjust past figures to take into account the new rules, no one knows how accurate the adjustments are.

More important, in trying to improve the accuracy of the sampling, the Census Bureau in the past 15 years has increased the number of families interviewed from 25,000 to 35,000 and the number of areas surveyed from 68 to 330. When it expanded from 68 to 230 areas in 1954, the bureau ran two surveys, one in the smaller number of areas and one in the bigger. To its surprise, it found a huge increase of 700,000 in the computations of jobless totals for the U.S. based on the larger survey. Since then, the number of areas has been increased to 330 in the interests of greater accuracy. But on the basis of the discrepancy discovered in 1954, there is reason to suppose that the jobless totals in 1954 and 1950 were larger than officially stated--and actually larger percentagewise in the case of the 1950 recession than in the current recession.

Despite all improvements, even today's sampling methods are far from 100% accurate. The Census Bureau itself admits that the figures contain a standard error of plus or minus 120,000 at the current level. Thus no one knows within six figures precisely what the total number of unemployed workers in the U.S. really is. Therefore, such figures as a 25,000 increase or decrease in unemployment are meaningless. And so are the tortured, hairline analyses made from them.

This file is automatically generated by a robot program, so reader's discretion is required.