The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
SPECIAL REPORT ON UNEMPLOYMENT STATISTICS: MEANING AND MEASUREMENT Raymond T. Bowman and Margaret E- Martin Office of Statistical Standards Bureau of the Budget Executive Office of the President Washington 25, D. C. October 1961 i Table of Contents Introduction Basic Concepts Usefulness of Labor Force Concepts Some Popular Misconceptions The Current High Level of Unemployment A Little Background INTRODUCTION As frequently happens during periods of high unemployment, questions have "been raised recently concerning the figures on unemployment which the Federal Government publishes each month. Economic analysts, news commentators and members of the general public have asked if we are counting the right kinds of persons among the unemployed, if our concepts are too broad, or possibly too narrow, if the survey methods provide reliable results, if the figures we get today are really comparable with those for past years, if we do ourselves justice in comparison with other countries. There can be, and are, honest differences as to who should be counted among the unemployed. Most of the descriptions of what is done in compiling the U. S. series and of possible alternatives have been written by specialists and are couched in technical language which discourages the intelligent layman seeking answers to a pressing public issue. This paper has been prepared in an effort to fill the gap between the technical discussion of specialists and the series of charges and counter-charges which have appeared in the public press. The statement will discuss the meaning, purposes and problems of measuring unemployment. A brief discussion is also included on why the current rate of unemployment is causing concern, to illustrate in a limited way some of the possibilities in using our rich supply of information on employment and unemployment in interpreting the economic situation. Those of us who remember the lack of valid statistics on employment and unemployment which prevailed during the Great Depression, the estimates of unemployment made by different people which differed by millions because there was at that time no direct statistical series measuring unemployment, appreciate the major advance which has been made in providing current information. This article, by describing what is now available on the size and characteristics of the labor force, will attempt to indicate the value of this advance. We believe that the employment and unemployment statistics of the United States are probably the most objective and accurate in existence anywhere, but that does not mean they may not be improved. We have no basic quarrel with those who still wish to question concepts - although it will be asserted that the present concepts are essentially right for the major purposes f$r which the data are designed. Figures on unemployment are so important in general public discussion of the economic situation, they are looked at by so many as a "trigger11 statistic indicating when certain lines of public action may be appropriate, that it is most desirable that the purposes of the series and the concepts and methods used in compiling them be widely understood. Admittedly, the way in which unemployment is defined and measured affects the statistics. For this reason, a brief discussion of the concept of unemployment as used in the official U. S. figures is important to understanding the measurement problem. 2 BASIC CONCEPTS First, let us recognize that unenrployment is not measured by itself but as part of a system of statistics on the labor force which is related to the population in a specific way. The basic concept is relatively simple. To make measurement feasible and as objective as possible, the first step is to select that part of the population for whom the question of employment and unemployment is relevant. For this purpose, the population Ik years of age or over and not in institutions is selected. This age limit is chosen not because we believe that most youngsters of lk or 15 are likely to be working or looking for work, but because there is special interest in knowing how many actually do work, at what occupations, etc. This part of the population is then classified, as far as employment status goes, into four principal groups, which are: (l) in the armed forces, (2) employed, (3) unemployed, and (k) not in the labor force. The first three of these groups constitute the "total labor force." The total labor force less those in the armed forces comprise the civilian labor force, which in turn, is the sum of the employed and the unemployed. Our discussion will be directed to this concept of the civilian labor force. The Civilian Labor Force The concept of "civilian labor force" is a very important one for a modern free enterprise industrial economy. It is intended to include those persons who are "economically active" in the sense th&t they participate or are actively trying to participate in the organized economic activity of producing commodities or services for pay or profit. Important activities such as keeping house or attending school exclusively are not considered economic activities in this sense and persons engaged solely in such activities are not in the civilian labor force. Basically, then, the civilian labor force includes all persons who have a "job11 or who are actively "seeking" one. It is the sum of these two categories. By definition, then, the civilian labor force includes only those persons who are in the labor market. The Employed The great bulk of pjersons in the civilian labor force are employed. The concept of employment is based primarily on the activity of working for g&in or profit during the period in question. It includes persons who did not work during the survey week, but who had jobs from which they were temporarily absent during the entire week. For example, persons on vacation, paid or otherwise, are considered as employed - they have a job. Persons home from work because of illness are also considered employed - they have a job. To make this clear to users of the data, statistics on employed persons are available for two subgroups which may seem to some persons rather peculiar - (l) enployed, working; (2) employed, not working - with a job but not at work. 3 Those working may not be working full time, so separate statistics are provided on hours of work. The information on part-time employment has caused some persons to recognize that it may also be a measure of part-time unemployment. To provide information on how much part-time employment is caused by lack of demand for labor, data on those who are working less than full time because they want or could accept only part-time work are shown separately from those working part-time because of economic factors, such as cut-backs in hours because of slack business or the unavailability of full time jobs. Included among the employed are unpaid family workers who do more than incidental chores (15 or more hours a week) in a family business or on a family farm. Such unpaid family workers are shown separately from wage and salaried workers and from the self-employed. Before leaving the problem of measuring employment, it should be further noted that students who both attend school and have a job are counted among the employed, as are housewives when they have a job in addition to their housework. Only their hours of gainful employment are counted in their hours of work, of course. The Uneiqployed The concept of unemployment is based primarily on the activity of looking for work on the part of persons who did no work during the survey week. Some people have argued that those not working nor looking for work but who indicate in some fashion that they "want work" should be included as in the labor force and unemployed. We do not use this concept* Experiments have shown that if we did, many more persons would be counted as unemployed. We do not use this concept because it depends too much on a state of mind of the person involved rather than on an objective criterion indicating that the person really is in the labor market by offering his services for hire - that is, actively seeking a job. It is true that certain exceptions have been made in applying the "looking for work" concept. Most persons agree that workers who have been laid off and are waiting to be called back to work by their former employers should be counted as unemployed even if they do not actively look for other work in the interim. Persons waiting the start of a new job are similarly counted as unemployed until the new job starts. Furthermore, just as persons who would have been working except for temporary illness are counted as employed, persons whose period of job seeking was interrupted by temporary illness are counted as unemployed. Finally, if a respondent volunteers that he would have been looking except for the belief that there is no work available in his line of work or in the community, he is counted as unemployed. The survey has been criticized for including this last exception, which is designed particularly for small communities in which a large plant has closed down. On the other hand, the survey has been criticized by those holding opposite views for not explicitly asking persons who reported they did not look for work whether they would have looked had they believed work to be available. The present practice is a compromise between these two views. It is felt that the exception does not add materially to the count of the unemployed. Experiments have shown that explicitly asking the question adds persons to the count of the unemployed whose current attachment to the labor force is somewhat remote. Research has not been carried far enough to suggest an acceptable alternative to present practice. This small area of indefiniteness between those unemployed and those not in the labor force is one of the places where further research and experimentation could profitably be pursued. Those Not in the Labor Force There are in the United States approximately 52 million persons Ik years of age or over, and not in institutions, who are not, 'according to present concepts, in the labor force. They are not employed - do not have a job - and they are not looking for work. Many persons able to hold jobs or to seek them are not doing so. This is a free labor market and these persons are not in that labor market. They are neither employed nor unemployed. They may be in school, housewives, retired, disabled, or have other personal reasons for not being in the labor force. Priority Among the Categories The survey which provides employment status information is called the Current Population Survey and results are reported in a government publication, the Monthly Report on the Labor Force. The survey is taken for one week each month (the week including the 12th of the month) by asking questions about each member of a sample of households. Everyone 14 years of age and over is classified in *one of three groups for that week, as employed, unemployed or not in the labor force. The number of persons in the armed forces is then added to obtain the total labor force. Now it is possible, of course, for a person to both work and look for work during the same week. Perhaps an ambitious worker looks for a new job offering higher pay at the same time as he continues to work at the old job. Perhaps a worker is laid off during the week, so he works for two days and looks for work for three or four days during the survey week. How shall such persons be counted, if we are going to include each person once and only once in our figures, and yet have the figures refer to the week as a whole? In order to avoid double counting, or classifying people 5 as employed or unemployed depending on whether they spent more hours working or looking for work during the week, a strict system of priorities has been established to enable the civilian population to be divided into these three distinct groups. This system puts working first, looking for work second, and not in the labor force third. Thus, a person who worked one hour at the beginning of the week, was laid off and hunted assiduously for a job during the rest of the week would be counted as employed, since any amount of working take precedence over looking for work. More specifically, the priorities are: (l) working at any time during the week; (2) looking for work or on layoff from a job; (3) neither working nor looking for work but having a job at which one did not work at all for such reasons as vacation, sickness, bad weather, etc. (included among the employed in the final count); (k) not in the labor force. Thus, a person having a job who did not work at all during the week on account of sickness, and there are several hundred thousand each month, would be counted as employed, rather than unemployed, whether or not he was paid for the time off. On the other hand, a person who did not work but used his time off to look for another job would be counted as unemployed since looking for work takes precedence over having a job at which one did no work at all. Note that the priority system makes the major labor force categories exclusive as well as exhaustive. There are no omissions, no double counting. It is not sufficient to say merely that a person should not be counted as "unemployed;" in that case, he must be counted as either "employed" or "not in the labor force." USEFULNESS OF LABOR FORCE CONCEPTS How good are these concepts? Are they the right concepts? Such questions are always pertinent, but they are naturally raised most often during periods of rising unemployment. An evaluation of their usefulness can only be made in relation to the purposes for which statistics on the labor force are needed. The labor force concepts used by the United States are designed to serve a variety of purposes. First, they are designed to give a measure of the human resources available as a result of the free option of individuals to seek employment by entering the labor force. Need may induce this seeking of employment but motivation is not relevant to the count - it may vary from the direst of need to the desire for more economic goods or even for 6 the achievement of psychic pleasures associated with a job and a useful place in society. It is our objective to know to what extent these human resources are being utilized. Their use or lack of use affects both the potential output of the economy and the actual output from the production side and also from the consumption side. The growth of population and the portion of it which enters the labor force is the human potential which sets one boundary to the growth of the economy's output. But the boundaries set by numbers alone are quite flexible. Changes in the hours worked by individuals in the labor force are one obvious factor which may affect economic growth. But productivity may be affected by changes in the quality of the work force, the talents of the workers. Infoimation from labor force statistics does not tell a full story on manpower skills and utilization, but these statistics do provide much more than is generally recognized, not only on the hours per week and weeks per year contributed by employed persons, but also something about the quality of the labor force - the occupations of the workers, whether skilled or unskilled, and their educational attainments. The second use of labor force data is as a key "economic indicator.n As such, it provides a measure of the operation of the economy over time, indicates the extent of seasonal and cyclical movements in the progress of the economy, and provides more specific data on employment and unemployment as an economic and as ^ social problem. Thus, interest centers on population characteristics of the unemployed and of the employed, on shifts resulting from technological changes the growth of some and decline of other industries and occupations and the cyclical movements in the general level of economic activity. There are other uses of the labor force data, but these appear to be the major uses. It may be noted that one thing the current concepts do not provide is a complete measure of need, or loss of income on account of unemployment. It is possible for some persons to be employed and yet paid at such low rates that they and their families are in real need. On the other hand, some of the unemployed are not in serious financial straits. It also seems clear that one thing not provided by the unemployed count is a clear and unequivocal measure of those about whom "something should be done." Nevertheless, the public naturally and rightfully looks to the labor force series as one of those needed to provide the information on which policy can be formulated on such issues as proposals to extend unemployment insurance, whether or not special public works programs are desirable, th6 need for retraining courses or other possible activities affecting manpower supply and utilization and the operation of the labor market. For this reason, a large number of details are provided about the employed and the unemployed each 7 month, in addition to the total numbers. We know, for example, how many persons worked less than 35 hours during the survey week, and whether this was from personal preference or for economic reasons such as slack work (on the average, one out of six persons works short hours from choice or for various personal reasons). We know the age and sex and marital status of the unemployed, whether they are looking for their first job or are "experienced" members of the labor force with an occupational and industrial attachment - for example, whether they are unemployed steel workers, carpenters or retail sales clerks. We have information on how long the unemployed have been looking for work. We know why persons "with a job but not at work" did not work during the survey week and whether or not they were paid for the time off. Over the years, the amount of detail published each month on the characteristics of the employed and the unemployed has steadily increased, so that those who wish to use these figures to evaluate a particular program or policy, to understand the economic situation, or the conditions of employment, will have the "building blocks" which can be selected to throw light on a particular issue. Furthermore, supplementary questions are asked of the sample households from time to time, to cast further light on the conditions of the labor force. Once a year, the incomes of all members of the sample households are obtained, and analyzed for individuals and for family units. At that time the survey also shows the number of persons in the labor force for each family, how many married women work, and, of these, how many of their husbands are employed, how many are unemployed. Once a year, analysis is made of school attendance, and figures are provided showing how many of our students work during the school-term. In recent years, there has been an annual count of the number of persons who hold more than one job during the survey week. Finally, there is an annual review of the number of persons who were in the labor force at any time during the preceding year, of the amount, regularity, and nature of their work experience. It would be possible, of course, to design different concepts or definitions. Without changing the basic concepts one could say, for example, that young people who are attending school should not be counted among the unemployed, even if they actively look for work. This would reduce the number of the unemployed by about 250,000 or about 7 per cent, according to the annual supplementary survey taken in October, i960. But if students looking for work are not to be counted in the labor force, it 8 would be only fair to question whether students who find work in addition to school attendance should be included. If excluded, that would subtract about 3*8 million from the total employment count of 67.5 million, a reduction of more than 5 per cent. Should we adopt such a practice we would give up the information we now obtain on the employment and unemployment of youths by disregarding work sought for and engaged in by students. At the same time, we would be overlooking the fact that working is one of the important ways of getting a higher education these days. In October of i960, special questions directed toward college students showed that 20$ were supporting themselves solely by their own work or savings and another 30$ made some contribution from their own efforts. The same distortions of our objective criteria of measurement, only proportionally greater, occur when attempts are made to exclude other groups whose attachment to the labor force is believed by an onlooker to be more marginal than that of the chief family breadwinner. In this connection, it is widely accepted that the standard of living of a large percentage of our families would be drastically lowered if the wife or other so-called secondary worker did not contribute regular or even intermittent earnings. The fact of the matter is, our economy operates on a system in which one out of 3 in the labor force is a woman, one out of eleven is under 20 years old, nearly one in 20 is 65 or over. Twenty-five per cent of the employed women and about 10 per cent of the employed men work at part-time jobs for personal or other reasons not connected with the economic situation. It seems desirable to enumerate these various groups separately, as far as distinctions can be made, but not to try to whittle away at our basic concepts in so doing. An important analytical tool which reduces the chance of misinterpretation of the data is the regular presentation of the unemployment rate on a seasonally-adjusted basis. Seasonal adjustment corrects this key series for such recurrent and noneconomic phenomena as the influx of students into the labor market during the summer. SOME POPULAR MISCONCEPTIONS With these concepts in mind, let us specifically review some of the criticisms recently directed at the unemployment figures. In the first place, let us refute the implication that Government agencies over the past 20 years have intentionally and progressively exaggerated the amount of unemployment. This is simply not true. As the brief historical note at the end of this statement shows, the labor force survey has been under continuous surveillance and review since it was started in 19^0. If there has been any intentional exaggeration, cooperation or negligence would have been 9 required on the part of two major departments, the Bureau of the Budget, the Council of Economic Advisers and the staff and members of the Joint Economic Committee of the Congress - and this over a period of several administrations. Such a conspiracy or such negligence has not occurred. To comment on a second misconception, it is sometimes assumed that the survey enumerators determine the number of unemployed by asking each households "How many people here want a job?" The fact is that this question is intentionally not asked because we know it would lead to an overcount of the unemployed. Instead, the question is asked individually for each person 14 and over in the household who did not work in the survey week: "Was....looking for work?" Neither are enumerators free to use their own discretion in the questions they ask. Not only is the specific question wording they are to use spelled out for them on the questionnaire, but frequent training sessions are held to maintain standards of good interviewing. Furthermore, an elaborate system of reinterviews by supervisors has been developed by the Bureau of the Census as a means of controlling the quality of interviewing. Interviewers whose work does not meet specifications are retrained, or if necessary, replaced. Some critics have pointed out that the monthly survey showed more unemployment than the Population Census did in 19^0 and 1950. This is true, but the fault does not lie primarily with the current series. Careful examination on a case-by-case basis was made of the discrepancies in 1950. Both sets of enumerators were hired by, both supervisory staffs belonged to, both sets of training materials were prepared by, the Bureau of the Census. However, the enumerators for the current survey were experienced, well-trained enumerators compared with the Census enumerators who had to be trained in a few hours to cover a great many more subjects than labor force, employment and unemployment. Analysis of these comparisons has convinced impartial students of the problem that the reason for the lower count of unemployment reached in the Population Census is almost wholly due to the inexperience of the temporary Census enumerators and the tendency of untrained interviewers to skip pertinent questions. It is too soon to tell if the use of self-enumeration in the i960 Census of Population has overcome this deficiency. A recent critic has directed attention to the experience of changing the sample design in 195^ as an indication of the effect of the attitude of the enumerators on the level of unemployment. At that time, the Census Bureau was engaged in spreading the sample of 10 25>000 households from 68 areas into 230 areas. Other things "being equal, spreading the same number of households among more areas should increase the reliability of results. To see what effect this shift would have on the results, an overlap period was provided during which both samples were used. There was an unexpectedly large difference in -unemployment between the "old11 sample and the "new" sample during one month of the trial period, January, 19 • The following month the difference had declined to a reasonable level. A committee of experts from outside the Government was appointed to review the new sample and the changeover operation and determine if possible what led to the unexpected result in January. This committee, composed of well-qualified statisticians from a university, a business group and a labor organization decided that lack of attention to maintaining standards of enumeration in the old areas as the new operation was being organized was the chief cause of the discrepancy. The committee concluded in its report to the Secretary of Commerce, "•••the new sample estimates are the more accurate and should be used in preference to the old survey estimates..." for the overlap period. The full report of the committee is contained in "The Measurement of Employment and Unemployment by the Bureau of the Census in its Current Population Survey" by the Special Advisory Committee on Employment Statistics, August, 195^• THE CURRENT HIGH LEVEL OF UNEMPLOYMENT It seems desirable to comment on the real point of concern which the figures highlight - the continuing high percentage of unemployment since the recovery from the 1958 recession. This discussion will not provide the answers as to the causes of this high level of unemployment. But it is hoped that an illustration or two of the use of the labor force figures in relation to current economic conditions will, together with what has gone before, make it clear that this high level of unemployment is real and not a statistical mirage. Search for the causes is continuing, by private as well as governmental groups, and requires analysis not only of labor force, employment and unemployment figures themselves, but of all other types of economic data capable of throwing light on current economic conditions. The most direct way to clarify the major issues raised by the unemployment figures is in terms of a chart showing the seasonally adjusted percentage of the civilian labor force who are unemployed each month. Also shown on the same chart is the percent unemployed among men age 20 years and over in the labor force, seasonally adjusted. 11 SEASONALLY ADJUSTED UNEMPLOYMENT RATE P , n o Cvlo tbr Fr. . e . f iiin oo oe Jul 1948f > ° When seasonally adjusted unemployment is expressed as a percentage of seasonally adjusted labor force, the major emphasis is on the cyclical movement in unemployment. Since the chart uses percentages, unemployment is measured relative to the size of the civilian labor force, (in absolute terns, unemployment for a growing labor force would also be expected to grow, as would employment*) The chart clearly shows the major cyclical periods recognized by economic analysts, namely, the recessions of 19^9, 1951*, 1958 and 1960-61. The peaks of the chart are in these years and represent high percentages of unemployment. The concern is primarily over the fact that in the prosperity period after the 1958 recession, namely part of 1959 and early I960, the unenqaloyment rates did not revert to the lower levels reached in 19^8, 1951-53, and 1955-57. 12 The relatively low unemployment rates in 1951-53 involve certain special factors. The period 1951-53 was the period of the Korean war - the civilian labor force was reduced as the result of mobilization while simultaneously employment opportunities were expanding. But the period 1955—5T did not have these unusual circumstances and the unemployment rate averaged, seasonally adjusted, just a little over h percent. During the prosperious period prior to the 19^9 recession it averaged just under k percent. We will not here discuss whether these rates were too high or too low. It is sufficient to note that they were considerably better than those achieved in 1959-60, which averaged above 5 percent. Furthermore, although it is customary for recovery in the unemployment rate to lag somewhat after improvements in other indicators, the rate is currently sticking at a recession level longer than usual. The upturn from the 1960-61 recession is generally considered to have started about March 1961. If the experience following the 19^9> 195^ and 1958 recessions is taken as a guide, we could expect the unemployment rate to start falling about July of this year. This did not occur either in July, August or September, the seasonally adjusted rate remaining just short of seven percent of the civilian labor force. There is every reason to be concerned about this situation. Although it is quite appropriate to ask the question, we can see no reason to believe that the problem is with the figures themselves. It is a real problem, not one of changing concepts or measurement techniques. The concepts and measurement techniques make the data reasonably comparable for the entire period I9V7-I96I. It has been suggested that inclusion of certain so-called "marginal11 groups in the labor force inflates the number of unemployed so that the figures do not depict the "true" course of unemployment on a comparative basis from year to year. It is time that unemployment rates are higher among young new workers, slightly higher for men over 65, and frequently, although not always, higher for women - the groups sometimes considered to be "marginal." Since these groups have been included in the labor force since the beginning of the series, they could affect the year-to-year unemployment comparisons only if they were entering the labor force in much larger proportions now than formerly. But these groups have not all been entering the labor force in larger proportions in recent years. In fact, there have been conflicting trends. Although a greater proportion of women are in the labor force now than in 19^8, proportionately fewer boys and older men are in the labor force now. These changes in labor force participation have occurred, quite consistently, year-by-year. As a result, the civilian labor force as a whole 13 has remained a stable percentage of the population Ik years of age and over. Although concealing divergent trends among the components,the overall figures provide a rough check that the unemployment figures have not been inflated because of counting undue and growing proportions of the population in the labor force. A more complicated statistical technique termed 11 standard!zing" indicates that if the age-sex composition of our labor force had been the same in i960 as it was in 1957* the unemployment rate in i960 would have been lowered only fractionally, by less than two-tenths of a percentage point. The following table provides the percentages of the population in the civilian labor force, together with the percent of the labor force unemployed, for the last Ik years. For brevity, annual average figures are shown, avoiding problems of seasonal adjustment, but concealing the more exact timing of charges possible from monthly figures. An examination of these figures shows that the labor force has not been a higher percentage of the population during periods of relatively high unemployment. On the contrary, looking at recent years, the high rate of labor force participation in 1956 occurred during a prosperious period. In the recession year of 195*b on the other hand, the proportion in the labor force was relatively low - excess unemployment apparently was not caused by excess numbers in the labor force. The percentages of persons in the labor force in 1951^53* years of high level economic activity, are lower than most other years, not because our concepts have changed, but because the Korean mobilization took out of the civilian labor force persons who were needed in the armed forces. Year 19^7 19hd 19^9 1950 1951 1952 1953 Per cent of population* Per cent of in civilian labor force . labor force unemployed Year 55-9 56.6 56.6 56.9 56.1 55.6 55.** 3-9 3.8 5-9 5.3 3.3 3-1 2.9 195b 1955 1956 1957 1958 1959 I960 Per cent of population* Per cent of in civilian labor force labor force unemployed 55.5 56.1 56.9 56.4 56.3 56.3 56.3 5.6 k.h k.2 b.3 6.8 5.5 5.6 *Non-institutional population 14 years of age and over. These factors are not stressed to indicate that the proportion of the persons in the labor market should not change, or if they did, Ik that there would be something wrong with the figures. They are provided merely to show that overall labor force participation rates did not, in fact, change in the periods in question in a way that would add significantly to unemployment • There are a variety of factors which would make for quite appropriate changes in the size of the labor force relative to the population of working age, such as, for example, changes in the age and sex distribution of the population, changes in school attendance or changes in general social attitudes toward retirement, employment of married women, etc. Further light on the reality of the increasing unemployment rate is afforded by the unemployment insurance system. Insured unemployment is not as large as total unemployment since the State unemployment compensation programs generally exclude job seekers having no recent work experience, the self-employed, agricultural wprkers, State and local government employees, the unemployed who have already exhausted benefits, and in some States, former employees of very small firms. Nevertheless, it is possible to compute the rate of insured unemployment to insured employment, and compare this rate with the rate of total unemployment. On that basis, despite the differences in coverage of the two statistical series, they show the same general movements from year to year* The rate for 1959> for example, was approximately a third again as high as for 1956 for both insured unemployment and total unemployment. It is not the intention here to attempt a definitive explanation of the employment-unemployment situation but to show that the figures are appropriate to tell us that a problem exists and an answer should be sought. Another example relates to the question of what we know about the growth of employment relative to the civilian labor force and associated population growth over the period 19^7-1960. It is instructive first to divide the period into three parts 19V7-1953, 1953-1957 and 1957-1960. The last year of each of these periods includes the turning point at which a cyclical increase in unemployment started. For these periods we compare the average annual percentage rate of growth of (l) the non-institutional population Ik years of age and over; (2) the civilian labor force; and (3) total civilian employment. u Periods Average Annual Percentage Rate of Increase Non-inst. Civilian population labor Ik yrs. & over force Employment 191*7-53 i.i 1.0 1.2 1953-57 1.1 1.6 1.2 1957-60 1.2 1.1 .7 1/ Data for Alaska and Hawaii are excluded from the i960 figures in calculating the average annual rates of change to make them comparable with the earlier years. 15 The figures show that the population and the labor force grew at roughly equivalent rates on the average, except for the middle period when civilian labor force growth accelerated following the Korean war. Employment kept pace with this growth in the earlier period but after 1957 employment has grown more slowly. In interpreting these facts, we must remember that 1957-1960 is still a short period relatively speaking. If employment had increased at the same rate as the labor force from 1957 to i960, it would have reached 67.2 million by i960, about 0.8 million higher than the actual average for that year. The facts indicate, then, that the higher levels of unemployment in 1959-60 and currently are related to the slower rate of employment growth relative to the population and the civilian labor force. This is important information. Why these things are so and what should be done about them are important questions. The policy problems aie not easy ones and there are and will be differences of opinion as to cause and cure. Neither these statistics nor any others will give clear and simple answers as to what needs doing, but they are the tools with which the economic analyst isolates problems so as to find solutions. No statistical series is perfect and improvements are always in demand. Larger samples, more frequent results, more detailed characteristics are some of the possibilities. Reconsideration of concepts in the light of uses made of the series is always appropriate. However, the public is not well-served if it is diverted from consideration of a significant economic symptom by a denial that it exists. And unfounded suspicion concerning the reliability of a major statistical series is bound to confuse present policy and distort consideration of future possibilities. High level output and economic growth in real terms depend.on fuller use of all human and material resources available in a free market economy. Appropriate comparisons with the past are necessary but current and future efforts should not be limited to what has been achieved in the past but based on what can be achieved in terms of the current and future resource availability. The MRLF series, with related population information, is an important guide to measuring the availability and utilization of present and future manpower resources. 16 A LITTLE BACKGROUND The Monthly Report on the Labor Force was developed under Work Projects Administration auspices toward the end of the Great Depression in order to provide the administration with estimates of the probable work force with which the W.P.A. would have to contend and some measure of whether progress was being made in reducing the number of unemployed. At that time, a variety of conflicting estimates of unemployment were made, none of them based on direct enumeration. Earlier attempts to count the unemployed had been sporadic and largely unsuccessful. The series covering the decade of the thirties, now generally accepted as most consistent with the MRLF and available only on an annual basis, was reconstructed after the fact from other sources using MRLF concepts as far as possible. This series is the best available for that period, and is published to provide a continuous historical record but should be taken only to represent general magnitudes of unemployment. It is not based on a direct enumeration. In 19^0, the Office of Statistical Standards of the Bureau of the Budget began an examination of the W.P.A. survey to determine if this were just "another set of numbers11 or whether the survey, which was breaking new statistical ground, was technically adequate and should be published. When it became clear that the days of the W. P.A. were numbered, the Bureau also looked into the question of which permanent government agent might most appropriately continue the survey. After this review, it was decided that the survey should be continued and that the Bureau of the Census was the most appropriate agency for undertaking the work. At the same time, the widespread interest of other government agencies and the public were recognized by the establishment of an interagency committee, under the leadership of the Office of Statistical Standards of the Bureau of the Budget, to resolve any major issues affecting the survey in the best interests, not of a single department, but of the Government and the public as a whole. Since that time, there have been a number of changes in the sampling design and there has been one change in definition, but there has been no change in the basic concept of measurement, that of current attachment to the labor force, measured in terms of activity during a week. In July, 1959* responsibility for planning, budgeting, analyzing and publishing the Monthly Report on the Labor Force was transferred from the Bureau of the Census to the Bureau of Labor Statistics, since the series fit in better with the manpower responsibilities of the Department of Labor. However, the actual collection and compilation of the information from households in the Current Population Survey sample is still in the hands of the Bureau of the 17 Census, acting as agent for the BLS, and no change in sample, in concept, or in question wording has been made since the transfer of responsibility. The same interagency committee chaired by the Assistant Director for Statistical Standards, Bureau of the Budget, continues to watch over the policy implications of any major proposals affecting the survey, such as sample revision and seasonal adjustment procedures. In addition to the general surveillance exercised by the Bureau of the Budget through its Office of Statistical Standards with the advice of other government agencies, the Current Population Survey has had the benefit of a wide range of advice from specialists and users of the data. A Panel of Technical Consultants has advised the Census Bureau on the sample design, the Review of Concepts Subcommittee, established by the Bureau of the Budget, undertook a fundamental review of the concepts and measurement problems in the mid-fifties, and the Joint Economic Committee o f the Congress has held hearings on U. S. employment and . unemployment statistics a number of times. The Review of Concepts Subcommittee was appointed by the Bureau of the Budget in 195^ to review the labor force and related series in the light of the needs for employment and unemployment data. In addition to government agencies, every attempt was made to consult users of the data among members of the public, research organizations, business, labor, and individual experts in universities and elsewhere. The full report of this Subcommittee was published in the "Hearings on Employment and Unemployment Statistics" held before the Subcommittee on Economic Statistics of the Joint Committee on the Economic Report, November 7 and 8, 1955• With regard to the concepts used in labor force statistics, the Subcommittee recommended continuation of the basic concept of measuring labor market attachment by activity during the survey week, but it did recommend one change in the application of this concept. In consulting members of the public it found almost universal agreement that two groups, at that time counted among the employed as having a job although not at work, should be shifted to the unemployed. These two groups were those on temporary layoff with definite instructions to return to work within 30 days, and those who had found new jobs but had not yet reported to work. According to the definitions then in use, these people had jobs, even though not working and not receiving earnings. In accordance with public understanding, they were "unemployed." At the same time, there were indications that temporary layoffs were becoming more generally used as a prelude to permanent layoffs. Promised call-backs did not materialize in many cases. If persons on temporary lay-off were not included among the 18 unemployed, the usefulness of the series as an early warning system of an economic downturn might " e impaired. For this reason, the b Subcommittee recommended that these groups " e reclassified as b unemployed. The advice of the Subcommittee was accepted and made effective in January, 1957* However, it should be noted that figures were available which permitted making this change in the overall employment and unemployment series back to the beginning of 19^7> so that the series is consistent for more than 13 years, including the entire period of the three post-war recessions. Also note that the groups have been counted separately since, so that anyone who wishes may adjust them back to the former definitions at any time if he so desires. In number, the two groups have amounted to about 275,000 per year, adding less than 10 percent to the unemployed, taking about O.k percent from the employed. Minor changes in the series also occurred when the 195° census population figures were incorporated in January, 1953, and when Alaska and Hawaii were added in January, i960. These changes do not affect the overall unemployment rate in any appreciable way. Users of the unemployment rate statistics, as revised back to 19^7 to take account of the one change in concept noted above, may rely on the fact that they are using a series which uses the same basic concepts and methodology for the entire period from 19^7 to the present.