Confused Figures Lead to Confused Decision

HOW MANY U.S. JOBLESS?

HOW high is U.S. unemployment and how fast is it rising? The answer is of vital importance to the U.S. because a continued rise in unemployment would call for a radical change in Administration plans for taxes and spending. Yet the answers from the Administration's statisticians are more confusing than enlightening; their last estimates of unemployment (for January) ranged from 2,359,000 to 3,087,000, a difference of 728,000. Not one knows which figure to believe, and last week Commerce Secretary Sinclair Weeks was so disturbed that he had a panel of experts checking over the accuracy of the whole statistical system.

The trouble is that the Government uses three basic methods to chart the ebb and flow of U.S. unemployment, and all three need improvement. All are limited surveys and wide open to errors of interpretation. Of the three, the most important--and most controversial--is the Census Bureau's total count of the U.S. labor force (currently about 62 million over the age of 14). The bureau first checks a tiny, carefully chosen sample of the U.S., only 25,000 households in 68 key areas. Then it mathematically projects the figures on the size of the labor force--and the jobless--to cover the 45 million households spread out across the nation. To the layman, the sample seems woefully small. Even the slightest mistake is multiplied 1,800 times. Yet statisticians claim they can get accurate results, and point to the 1950 census, which showed a difference of only a few thousand in unemployment totals based on a 3 1/3% sample, a 20% sample and a complete count.

That a sampling system is open to big error is apparent from the huge, 728,000-worker difference between the Census Bureau's two figures for January. That month, the Census Bureau tried a new system by increasing its sampling from 68 to 230 communities. It hoped for greater geographical representation, and it wanted to include new industries, e.g., electronics, which have grown up since the war. Since one of the two sampling systems must be wrong, most experts vote for the new system and the 3,087,000 figure. They argue that accuracy should increase through a greater geographical spread.

But the census figures, even if accurate, would not necessarily tell the whole story on unemployment. The very definition of "unemployment" is a problem. Census takers ask a series of carefully phrased questions to find out if a man worked at all during the week, was temporarily laid off, was either looking or not looking for a job. Only those out of work and actually looking for jobs are counted as unemployed. Thus a housewife who is laid off from her factory job but does not look for a new job is not counted as unemployed even though she may be drawing unemployment insurance. Furthermore, the Census Bureau's system gives no weight to partial employment.

The Government's two other methods of watching unemployment are much more limited in scope and only partial answers at best. The Bureau of Labor Statistics' monthly check on factory employment is a full rundown of 155,000 companies and 23 million workers, but it is a specific count of factory employment, not total unemployment. The number of jobless is simply an estimate based on the rise and fall of payrolls. Another drawback is that it reflects no unemployment among U.S. farmers, Government workers, servants, or the self-employed.

Even less accurate as an unemployment guidepost are the much-quoted weekly figures on unemployment compensation. Jobless insurance covers a wide range of U.S. labor--some 60% of all workers--but the figures are only a rough approximation of unemployment. The reports all come from the states, each of which has different laws about the eligibility of industries and length of time a worker must wait before he can collect insurance. The figures also become increasingly misleading as unemployment climbs. Claims tend to drop sharply as fewer people have jobs to lose and as the jobless period lengthens. State unemployment insurance lasts but 26 weeks at the maximum. Thus a worker out of a job for a longer period automatically goes off the insurance rolls.

Government economists are agreed that the country needs a clearer picture of the job situation. One way the Government could help would be to get all the unemployment statistics together in an overall monthly report by some expert body such as the President's Council of Economic Advisers. This might give each figure its proper weight, and thus end much of the confusion. But even more important is the need to overhaul the statistical methods themselves, find out why there could be a difference of 728,000 between two estimates. Over the years. U.S. business has learned to plan for the future by arming itself beforehand with as accurate statistics as it can get. The businessman's Administration has not yet begun to profit by this lesson.

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