Measures of labour underutilization from the Current Population Survey

Measures of labor underutilization from the Current Population Survey

Steven E. Haugen^* 11.03.08

It is estimated that in 1933, at the depth of the Great Depression, about 13 million persons in the U.S. were unemployed, which translates into an unemployment rate of about 25 percent.¹ However, those estimates were not available at the time. Throughout the Great Depression, there was little information on the extent of unemployment in the country. More important, there was no good way to assess whether the situation was getting better or worse. The wealth of timely statistical information on the labor market that we now take for granted simply didn’t exist.

Throughout the 1930s, researchers grappled with the issue of how to measure unemployment. To begin with, there wasn’t agreement on how to conceptualize or define the condition. Simply asking those out of work if they “wanted” work or if they were “able” or willing” to work proved to be too subjective to serve as unemployment criteria. At the same time, attempts to gauge the number of jobless by looking at declines in employment or counting the registrations at public employment offices were found to be incomplete.²

Gathering data on the unemployed proved equally challenging. Efforts to enumerate all unemployed persons (i.e., a complete census) were cumbersome and costly. Moreover, the information would be out of date by the time it was collected and processed, and therefore not very useful for conducting current labor market analysis or developing national policies to combat the unemployment problem.

By the end of the 1930s, research had led to two major breakthroughs. First, the activity concept was developed as a means of determining a person’s employment status. This concept paved the way for a more objective, and hence more widely accepted, definition of unemployment. Rather than base unemployment estimates on highly subjective criteria such as whether or not a person out of work merely wanted a job, the activity concept used information on persons’ actions to determine his or her labor force status. Survey questions were designed to determine one’s labor market activity during a prescribed time period, and these questions were administered by a trained interviewer.³ If a person was working, he or she was classified as employed. If not working but actively searching for work, one would be classified as unemployed. All others were not in the labor force.

The second major development was the introduction of statistical sampling as a tool for estimating labor force characteristics. Prior to the 1940s, using a random sample of a group to estimate the size and characteristics of the entire group was not widely adopted or accepted in the social sciences. Advances in the field of statistics during the 1930s allowed researchers to devise increasingly reliable and representative sample designs and estimation methods.⁴

The combination of the activity concept and probability sampling ultimately led to the birth of a regular sample survey of the population in March 1940, called the Monthly Report of Unemployment (and shortly thereafter the Monthly Report on the Labor Force). Known by its current name, the Current Population Survey (CPS), since 1948, the survey has been the source of official unemployment statistics for the U.S. since its inception, and it became a model for labor force surveys across the world.

In the nearly seven decades since the introduction of the CPS, there have been numerous reviews of the official concept and definition of unemployment, both within and outside the Federal government. Yet these exhaustive studies resulted in only minor refinements and revisions to the official measure. In fact, the unemployment rate⁵--the conventional way of expressing the measure--has proven to be a reliable indicator of overall labor market conditions and has performed quite well as a business cycle indicator.

That does not mean, however, that everyone has been completely satisfied with the official figures. Indeed, there always have been (and likely always will be) some analysts who argue that the official measure of unemployment is too broadly or, as is more often the case, too narrowly defined. Those in the latter camp often contend that the job search requirement is too stringent; some believe that the desire for work alone should suffice in defining unemployment. Similarly, some believe that certain persons who are working should be included among the unemployed, especially if they are working fewer hours than desired--the rationale being that these persons are, in a sense, partially unemployed.

Other criticisms of the unemployment numbers arise because the figures are used to gauge more than just joblessness at a given point in time or changes in unemployment over the course of the business cycle. For many, the unemployment numbers serve as a yardstick for assessing the number of persons who experience some level of “financial hardship”--that is, the number of persons who, to varying degrees, have a lifestyle that affords them no more than life’s basic necessities. Measures of hardship (a highly subjective concept) generally are based on earnings or income. However, because most workers derive the bulk of their income from work activity (rather than from investment income, for example), joblessness or underemployment often is seen as sufficiently indicative of some level of economic hardship.

In the 1970s, Julius Shiskin, Commissioner of Labor Statistics, noted that “no single way of measuring unemployment can satisfy all analytical or ideological interests.”⁶ In response, he developed a range of unemployment indicators known as U-1 through U-7. (See table 1.) The range was first published in his Monthly Labor Review (MLR) article entitled “Employment and unemployment: the doughnut or the hole?” and subsequently issued on a regular monthly basis in the Bureau’s Employment Situation news release. All of the measures were expressed as rates, and were oriented from lowest to highest. The official unemployment rate was included in the range of estimates.

^* Steven E. Haugen is a supervisory economist in the Division of Labor Force Statistics, Office of Employment and Unemployment Statistics, U.S. Bureau of Labor Statistics.