Printer Friendly

The statistics corner: interpreting the unemployment statistics.

My guest columnist for this issue is Janet L. Norwood, formerly U.S. Commissioner of Labor Statistics. Janet gave this excellent discussion of how to interpret the unemployment statistics at the NABE Annual Meeting in Dallas, but the paper deserves wider distribution.

FOR THE PAST two quarters, the gross domestic product has grown, but the expansion has yet to pick up steam. Job growth has been quite weak. Even last July, when the Bureau of Labor Statistics reported an increase of nearly 200,000 jobs, almost none of the increase came from the private sector. The August data showed an even clearer picture: payroll employment in the private economy dropped by 167,000. The overall jobs number was bolstered by about 100,000 federally funded summer youth jobs, Job losses were especially large across manufacturing industries and in retail trade. In fact, factory jobs hit the April 1983 low point,

Bad news, you say? But the overall unemployment rate edged down another tenth in August. The Bush Administration, focusing on the improvement in the jobless rate since June, called the report "encouraging." The Clinton forces, focusing on the employment numbers, pointed to the weakness of the private sector.

The point is that we need to dig below the topside numbers to find out what is really happening. The jobless rate is always affected by the size of the labor force: those who are working plus those who are looking for work. Labor force growth has been unusually slow in recent years, especially for young people, whose labor force participation rates had been trending downward. The government summer youth program may have encouraged some of them to start job search once again.

We know that, when analyzing the unemployment rate, we can expect certain kinds of changes once economic growth really picks up:

1.The labor force will begin to increase. As the economy expands, many who have left the labor force during the recessionary period think that jobs are now available and so they decide to look for work. This labor force increase, of course, makes it harder to reduce the jobless rate because so many more people are in the labor force looking for a job. It has been the unusually slow labor force growth over the past few years that has exerted downward pull on the unemployment rate.

2. A part of that increased growth should come from the discouraged worker group -- those who were not looking for a job because they believed no job was available. The number of discouraged workers always rises during recession and shrinks during expansionary periods. We must remember, however, that the discouraged worker group did not increase as much as usual during the recession that began in 1990, so the increase in the labor force from this group, while important, will not be very large.

3. The proportion of the unemployed who have been jobless for six months or longer -- the long term unemployed -- will continue to increase as the business expansion sets in. This occurs because the most experienced and most valuable workers are laid off last and rehired first. Workers fired first (those who are not critical to the operation of the business) are usually the last to be rehired.

4. We can expect that the unemployment rate will probably lag at the beginning of an expansion because of the labor force growth that occurs.

In analyzing the current situation, of course, we must remember that this time around we face several unusual circumstances, all of which have an effect on employment and unemployment. For example, in recent memory, we have not had to face a recovery against the background of a banking crisis, which has resulted in more careful and slower extension of credit. In addition, the construction industry, as the result of overbuilding during the 1980s, has been extremely weak, even in a period when interest rates are quite low. Furthermore, the fact that consumer confidence remains quite low has slowed sales in the retail field and in services. As a result, the retail trade industry, which was hit hard during the recession, has been slow to expand its work force. This industry is one of the largest, employing more than 19 million workers; it is also a source of employment and job experience for many of the nation's young workers.


But what about the second issue that I mentioned at the start -- what the unemployment rate means and how it is measured.

Many people know that the data come from a statistical survey, but even they are surprised to find that no one is ever asked whether or not he or she is actually unemployed. Others are sure that there must be some connection between the jobless rate and the insured unemployed. Still others cannot understand why those who have very low income may not be counted at all, or why those working only a few hours a week are counted as employed. Questions about the meaning of unemployment rise during economic recession and recede from public interest when recovery is well under way.

In the early days of our country, unemployment was considered a private affair to be dealt with by individuals not by government. There was no labor force survey. A few questions were included in the Population Census, but the definitions used were vague, the questions asked varied across states, and the respondents often provided questionable information. The concept used in most of these early attempts was not consistent with the data as we know them today.(1) But the issue received little attention because most officials believed that the task of data collection and analysis of unemployment was not properly a function of government (de Neufville 1975; Norwood 1989).

In fact, there was so little agreement on the definition of unemployment that in 1921, when President Harding called a Conference to discuss unemployment, in what has been called, ". . . one of the lowest points in the history of social measurement. . . ," the Conference voted on the number to announce as unemployed, choosing a suitably high range of figures to attract the nation's attention." (de Neufville, p. 72, emphasis added). The country was informed that the number of unemployed was somewhere between 2.5 and 5.5 million. (Goldberg and Moye 1985 and de Neufville 1975).

Even during the Great Depression of the 1930s, estimates of the jobless ranged from about 13.5 million to 16 million. The definition of unemployment that we use today began about fifty years ago when a small group of innovative survey statisticians working at the Works Progress Administration (WPA) selected a concept and designed a scientific survey to represent the nation's labor force experience.

Although many changes have been made in the design of the survey and some elements of the definition of unemployment have been find-tuned, the basic definition of joblessness has remained essentially the same for the past half century. It has been adopted by other countries and has been established by the International Labor Organization (ILO) as a goal for all countries. The U.S. definition is based on activity, i.e., job search. Those persons in the survey who (1) have not worked during the survey week; (2) are available for work; and (3) have looked for work during the preceding four weeks are counted as unemployed. The respondent must fit this activity definition to be included in the official definition of unemployment.

Twenty-five years ago, the Gordon Committee (President's Committee 1962), which was appointed by President Kennedy to review unemployment statistics, recommended further research on unemployment measurement concepts. The Committee recognized that the many uses of the unemployment measure made it difficult to select a single definition (President's Committee 1962, p. 42). Labor force attachment covered a broad spectrum, the Committee wrote, from those working continuously at full-time jobs to those who were clearly outside the labor force. Between these two extremes, there are many who work intermittently or part time or who are not looking for work but might do so if circumstances were different (pp. 42-3). And then, there are those who are self-employed for whom unemployment as measured in the official definition is almost nonexistent. The Gordon Committee agreed that the ". . . relatively simple dichotomy between those in and out of the labor force does not, for many uses, provide a satisfactory measure of labor supply." (p. 71). But the Committee did not opt for changing the concept; instead, it recommended modifications to reduce some of the ambiguities (which were adopted in 1967) and suggested that further research be done to produce measures that would be easier to understand. Since that time, many have complained about the concept, but few have attempted development of a conceptual framework within which measurement could take place.

Nearly twenty years later, a new Presidential Commission chaired by Sar Levitan (National Commission 1979) reviewed all the criticisms (Cain, National Commission 1979), and pointed up the need for supplementary measures to shed greater light on those in economic hardship. The Commission suggested a number of changes, the most relevant of which for purposes of this paper was revision of the definition of "discouraged workers." After much discussion, however, the Commission recommended very little basic change in concept.

What are the unemployment data telling us now? This summer (June 1992), more than three-fifths (61.4 percent) of the population sixteen years of age were working. Some 10 million people who were not working, (7.8 percent of the labor force) were looking for work and, therefore, were classified as unemployed. Of that 10 million, nearly 8.5 million (8.4 million) were adult men and women and a little over 1.5 million (1.6 million) were teenagers. Nearly three-quarters of those counted as unemployed were white, although jobless rates for blacks at 14.9 percent and for Hispanics at 12.1 percent were far higher than the 6,8 percent rate for white workers. Eighty-three percent of the unemployed were full-time workers, with the remainder (1.7 million) part-timers. Some 6 million additional workers were counted as partially employed -- the group of part time for economic reasons who wanted full-time work but could only find part-time jobs.

We can also look at the unemployed from a family perspective. About half those who were jobless were husbands, wives, or women maintaining families by themselves. Some 2.3 million of the 10 million were married men living with their wives, 1.7 million were married women with husbands present, and 740,000 were women who maintain families on their own.

We can also look at the unemployed by occupation. Most of them were technical, sales, and administrative support personnel (2.4 million) and operators, fabricators, and laborers (2.1 million). Nearly three-fifths of the unemployed had worked in service-producing industries.

So, we know a great deal about the characteristics of those whom we consider officially unemployed. But we do not know very much about the concept that underlies the classification structure into the employed, the unemployed, and those out of the labor force. Much has been written about the relationship between unemployment and inflation, about the interaction of aggregate demand and the labor market, and about joblessness and business cycles. We have studies of frictional and structural unemployment, models about labor force entry and the interaction of unemployment insurance and joblessness, and work has been done on the social issues and economic hardship resulting from unemployment. In spite of the large body of research on these issues, economists have paid relatively little attention to the development of a comprehensive framework for measurement of unemployment. Instead the problem has been left to the survey designers who require a definition that can easily be applied and who must develop data for many different kinds of users.

Criticism of the data come from many directions. Unemployment is generally described as a labor supply measure, but we all know that some people have greater attachment to the labor force than others. The data include those seeking part-time work, counts as employed those working one hour or more during the survey week, does not include those too discouraged to look for work, and includes the self-employed where unemployment has a different meaning from wage and salary work. It has been alleged that the data system places too much emphasis on "marginal workers" -- students, youth, those not living in families with others to support, etc. (Cain, 1979) Indeed, some have questioned whether we are experiencing a declining commitment to work, especially among our young people. Yet, information on marginal attachment to the labor force is not available, nor are data on reservation wage collected. We know that people seem to place themselves somewhere along a continuum -- some always working or, when not employed, looking for work, others wanting to work but prepared to take some time off to do other things if finding a job becomes too difficult, and some who want only part-time work. The labor force, made up of those working and those looking for work, is considered by some to be too small because it excludes some of these groups and by others to be too large because it includes those not wanting permanent full-time jobs. Indeed, a whole host of factors can affect the size of each of these demographic groups: family status, income, education, and training. But one of the most important determinants of labor force behavior, wage rates, is not regularly available in the detail that could be helpful in determining the trade-off that workers make between their own well-being or living standard, work, and leisure. However, wage information is difficult and burdensome to collect in household surveys. We have only data on income from the March Supplement to the CPS and information on usual weekly earnings each quarter from the CPS itself.

One group of people about whom there has been much discussion is the group classified by the federal government as discouraged workers -- those who tell the survey interviewers that they want to work but believe that no job is available and that job search, therefore, would be useless. Because discouraged workers do not search for jobs, they are not included in the jobless count. Because discouragement is a state of mind, it is very hard to measure, and, as a result, the government provides the data each quarter but excludes the group from the unemployed. Several social interest groups deplore this practice as an understatement of unemployment, and some of them put out their own unemployment rates.(2)

The treatment of discouraged workers is related to what is perhaps the most prevalent criticism of the unemployment rate. It does not measure economic hardship. The problem is that many Americans who are fully employed are paid very low wages; these working poor may suffer far more economic hardship than some not employed but looking for work. A significant portion of the jobless are eligible for at least some unemployment compensation or other government income support program. Others may suffer unemployment of short duration. Still others may be teenagers or those who live in families, have no family support responsibilities, and do not have to earn their own living. The link between unemployment and economic hardship has weakened over time. Nevertheless, because most Americans equate joblessness with the lack of income, they find it hard to understand that the unemployment rate is not a measure of economic hardship.

There are other problems with unemployment as a social hardship measure. We do not yet know how to measure the homeless in this country, a group that is becoming increasingly visible especially in the nation's cities. Clearly, the homeless are not employed, and some of them may not be employable. In any case, they generally do not search for work nor do they appear to fit into the discouraged worker category. There are other situations in which the official definition of unemployment does not provide data that are particularly useful e.g., the reservations on which Native Americans live where very little work is available or the rural areas in the winter months when there is very little demand for workers on the farms.

Under these circumstances, the task of defining unemployment has become laden with value judgments, making it very difficult to bring about any significant change in definition. In an attempt to deal with these issues, about a decade or so ago, the Bureau of Labor Statistics began publishing a series of different unemployment rates: U-1 through U-7. Beginning with U-1, the most restrictive measure, unemployment for 16 weeks or more, and moving through other definitions to U-7, which includes the part time for economic reasons and the discouraged in addition to the official count of unemployment, these measures have probably not received the attention they deserve. The reason for this, I believe, is that the measures generally show the same trend; the differences show up only in level -- U-2 is higher than U-I; U-3 is at a higher level than U-2, but the trend over time seems to be almost exactly the same.

Public officials, researchers, social activists, to mention only a few groups, all have made a great deal of use of the unemployment and other labor force information that comes from the Current Population Survey. Economists have produced studies using a number of theoretical concepts to explain relationships in the labor market, but there is still no comprehensive conceptual framework for the design of the labor force survey. Instead, we have a large group of users who focus on the unemployment rate as a measure of labor supply even though they know that the potential labor force might well include many not counted either as employed or unemployed. Other users focus their attention on the unemployment data as a measure of labor market hardship. They use the unemployment data as a proxy for a welfare measure even though it is not really satisfactory for this purpose. Still others decry the fact that the current data system provides insufficient help to those wishing to study the social or psychological effects of unemployment.

A comprehensive conceptual framework is needed for the nation's labor force statistics system. Until that time, however, the government must continue to improve the Current Population Survey and help users to interpret labor market developments as they occur. Sustained public interest in this important survey -- even when the economy is functioning well -- is critical to that mission.

Janet L. Norwood is a Senior Fellow, The Urban Institute, Washington, DC.


1 The notion was that of the "gainful worker" -- a worker who usually worked but had been forced into idleness and needed a job.

2 See Americans for Democratic Action May 1992 unemployment rate, which adds 9.5 million more to the jobless count, or the Urban League, which defines discouraged workers as those "who are not in the labor force but who want to work immediately." Daily Labor Report, June 8, 1992, no. 110, p. A-9.


Bregger, John E. (1971) "Unemployment Statistics and What They Mean," Monthly Labor Review, November.

Cain, Glen (1979) "Labor Force Concepts and Definitions in View of Their Purposes," National Commission on Employment and Unemployment Statistics, Counting the Labor Force, Appendix. Volume 1.

"Counting All the Jobless: Problems with the Official Unemployment Rate." (1986) 37th Report by the Committee on Government Operations, House Report 99-661, 99th Congress, Second Session.

Counting the Labor Force (1979) National Commission on Employment and Unemployment Statistics.

Daily Labor Report. June 8, 1992.

de Neufville, Judith (1975) Social Indicators and Public Policy.

Freeman, Richard B. (1979) "Why is There a Youth Labor Market Problem?" NBER Working Report No. 365. June.

Goldberg, Joseph P. and William T. Moye (1985) The First Hundred Years of the Bureau of Labor Statistics.

Hauser, Philip M. (1974) "The Measurement of Labor Utilization," Malaysian Economic Review, April.

Keyssar, Alexander (1986) Out of Work, The First Century of Unemployment in Massachusetts.

Klein, Bruce W. and Philip L. Rones (1989) "A Profile of the Working Poor," Monthly Labor Review, October.

Maurer, Marilyn E. (1991) "Data Needs for Analyzing Labor Force Status and Transitions," paper prepared for Committee on National Statistics, March.

Measuring Employment and Unemployment (1962) President's Committee to Appraise Employment and Unemployment Statistics.

Norwood, Janet L. (1987) "The Measurement of Unemployment." Proceedings, American Economic Association, Chicago, Illinois.

Shack-Marquez, Janice (1991) "Issues in Labor Supply," Federal Reserve Bulletin, June.

Stein, Robert L. (1967) "New Definitions for Employment and Unemployment," U.S. Bureau of Labor Statistics.

Wallis, W. Allen (1991) "Conversation with W. Allen Wallis," Statistical Science, vol, 6, no. 2.
COPYRIGHT 1993 The National Association for Business Economists
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Norwood, Janet L.
Publication:Business Economics
Date:Jan 1, 1993
Previous Article:The changing fortunes of bank economists.
Next Article:Latin America's Economy.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters