Labor Force

views updated

Labor Force

I. Definitions and MeasurementA. J. Jaffe


II. ParticipationJacob Mincer


III. Markets and MobilityHerbert S. Parnes


IV. Hours of WorkE. H. Phelps Brown and M. H. Browne



Labor force” is a term used specifically to refer to data collection procedures developed in the United States during the late 1930s and currently being used in that and a few other countries. A more general term, and one more suited to the scope of this article, would be “working force” this is more or less equivalent to what United Nations labor statisticians call the “economically active population.” But whatever term we finally choose, it should be emphasized from the start that there are many methods of collecting statistics about workers and that any way of referring to all the workers in one country necessarily reflects one or more of these methods.

Definition of terms

In its most generalized form the working force is conceived as that portion of the population which is economically active. The definition of “economically active” in any particular society must in turn be regarded in terms of the organization of work characteristic of the culture. In primitive societies the working force is hardly distinguishable from the total population. The low level of technology requires the participation of virtually the entire population in the common tasks of producing goods and services necessary for subsistence. With the growth of a market economy in the course of economic development, the working force becomes clearly distinguishable from the total population. The working force may thus be said to be a function of the socioeconomic structure of society.

In every culture, most persons are engaged, a good part of their lives, in activities that may be considered as work. But such activities may or may not qualify them for inclusion in what may be regarded technically as part of the working force. For example, in the United States the services performed by housewives, although highly desirable from a societal point of view, are not regarded as economic. Housewives are therefore excluded from what is measured as the working force because such work is outside the characteristic system of work organization or production. Moreover, their inclusion in the working force, for purposes of economic analysis, would not help policy makers to solve the significant economic problems of American society. Thus, the concept of working force underlying statistical measurement is an artifact, created to serve purposes of social analysis, .and can be defined only in terms of economic and social structure.

Manpower is a broader term than working force; it not only includes the latter but generally also includes the potential or maximum available working force which might come into being in accordance with economic, social, or political considerations. Gainful workers (or gainfully occupied), as well as labor force, refers to specific ways of collecting data. Labor supply, as it is used in economics, is conceived in the sense of a hypothetical schedule showing how much labor would be offered at each level of wages.

Emergence of a modern working force

The previous observation that the working force in a modern industrialized country is only a portion of the total population can be seen most clearly by comparing its working force conditions with those that would obtain in a country or tribe in the complete absence of industrialization or industrializing influences (see Table 1).

Table 1Working force conditions in modern industrialized and in nonindustrialized countries
 Modern industrializedNonindustrialized
Technological developmentVery highVery low
Division of laborHighly developedAlmost none
Exchange economy (based on cash)Highly developedNone
Free contract systemHighly developedNone

The main point of contrast here is that industrialization frees many workers from the production of food and other goods essential for physical survival. Nothing, however, is implied about the form of government under which production takes place; industrial technology seems to require a centralized form of control regardless of whether the means of production are state-owned or in private hands. By contrast, in a completely nonindustrialized economy each worker tends to control and even own his individual means of production, even if his productive capacity goes little beyond subsistence level. The crucial variable would appear to be the presence or absence of money or its equivalent as a standardized medium of exchange. There is yet another important difference between these two types of society. Nonindustrial societies, as Sir Henry Maine pointed out long ago, lack not only actual contracts but even the very notion of “free contract” the individual’s labor activities are predetermined by the culture. Exactly the opposite is true of industrial societies, in which the worker enters into a free contractual relationship with his employer (there may, of course, be other constraints on him, but they are not the absolute constraints of culture).

We can visualize the process by which the modern working force emerges by starting with a non-industrialized society which meets two conditions: it is large enough to be able to use a division of labor, and its population has unfilled wants for goods and services. Although the latter condition can probably be found among all peoples, it is clear that isolated tribes of a few hundred people cannot fulfill the first condition.

If this process does get under way (and there seems to be no reason why it must begin), then the combination of division of labor and technological advancement results in the establishment of technologically complex economic institutions in which control rests in the hands of a few managers rather than with the individual workers. A highly developed exchange economy based on cash must now exist, and the worker must be free to offer his services for cash hire as he sees fit. Thus, he receives cash payment for producing a highly specialized item or service and then can use this cash for purchasing at will items or services produced by other specialized workers.

As the end result of this process we have the labor market, or that specialized portion of the market place in which the person is free to offer his services for cash hire. The farmer or own-account worker represents the special case of the individual who may offer his goods or labor services simultaneously to several buyers. Those persons, then, who voluntarily offer their services for hire in the labor market and who thereby participate (or attempt to participate) in the production of the gross national product, form the working force. Accordingly, those persons who, for whatever reason, do not offer their services for hire in the labor market, thereby automatically exclude themselves from the working force.

Slaves or serfs also are not part of the working force, according to the above definition. Amasa Walker recognized this when he wrote:

What is labor? The voluntary efforts of human beings to produce objects of desire.... Involuntary or uncompensated efforts are not to be classed as labor. They are merely the result of the use of a given amount of capital. Slaves are owned, like horses or oxen; and what value they confer is from their employment as so much capital. This distinction is not unimportant, because we shall see that capital is controlled by other laws than those which govern labor. (Walker [1866] 1872, pp. 18–19; compare Jaffe & Stewart 1951)

Social policy and working force analysis

Of overriding importance in the emergence and development of working force analysis was its connection with social policy determination. At times and in places where governments were unconcerned about unemployment, set no minimum wages or maximum working hours, or otherwise saw no problems to be solved which involved the people who produced the goods and services, there was little interest in working force analysis. In particular, where people were born into preordained social positions and occupations, there were no working force problems to solve. The son of a serf in medieval Europe was born to be a serf and work in a manner in accordance with customs and procedures known to everyone. The government might have been concerned with whether it had enough serfs or whether the serfs had enough to eat, but never with whether the serf was employed or unemployed.

In the United States the need for governmental action with regard to the working force became particularly acute during the great depression of the 1930s. If the government were to adopt any policies in an effort to cut down the volume of unemployment, information about the numbers and characteristics of the unemployed and the employed had to be available. This social policy need explains the direction in which the field has since progressed.

Data collection in the United States

Three sets of procedures for collecting statistics about the American working force are now employed; the first two furnish the statistics most often used. They can be described briefly and in general terms as follows:

Labor force procedures

Members of households are interviewed by the U.S. Bureau of the Census enumerators and queried about work experiences during a specified week. These questions are asked either at the time of the monthly sample survey, when some 35,000 households are covered, or at the time of the decennial census.

Each person living in the household who is aged 14 and over is classified as employed, unemployed, or not in the labor force by means of the following procedure, (a) First of all, he or she is asked: What was your main activity last week—working for pay or profit, looking for work, or something else? (b) Those who replied “looking for work” or “something else” are then asked whether, in addition, they also worked for pay or profit during the past week. If the answer is yes, then they are also included among the employed, (c) If the answer to the question of whether a person had worked for pay or profit is no but the reply to question (a) had been that he or she was “looking for work,” the person is included among the unemployed, (d) If the person answers no to the question about pay or profit but had originally replied “something else” to question (a), then it is asked whether he or she looked for work last week. If the answer is yes, the person is included among the unemployed, (e) If the answer to question (d) is no, the person is asked whether there was a job or business from which he or she was temporarily absent last week, (f) If the answer to question (e) is yes, the person is included among the employed, (g) If the answer to question(e) is no, the person is classified as not in the labor force.

The “labor force” thus consists of persons employed for pay or profit during the specified week, plus persons who sought work during that week, the unemployed. All others aged 14 and over are categorized as not in the labor force (U.S. President’s Committee ... 1962).

Certain additional distinctions are made. The decennial censuses include everyone aged 14 and over; the members of the armed forces, however, are shown separately from the civilian population. The monthly sample surveys cover only the civilian population, omitting those who are residents of institutions (jails, hospitals, old people’s homes, etc.). Both series distinguish, among the unemployed, between those who have worked previously and those who were seeking their first job; those unemployed who had worked previously, plus the employed, are the experienced labor force.

Persons in the experienced labor force are then asked questions about various other matters, including their occupation, industry, type of remuneration (self-employed, wage or salary worker, unpaid family worker), and hours worked per week. The unemployed are asked how many weeks they have spent looking for work. Additional questions are asked from time to time covering other aspects, such as earnings last year, multiple job holding, and so forth.

Establishment employment statistics

Payroll information is obtained each month from a sample of firms in all industries except agriculture. Items included are number of persons on the payroll and number of “production,” or nonsupervisory, workers, together with hours worked and pay received by the production workers. More information is provided for production workers in manufacturing than for workers of any type in other nonagricultural industries. These monthly establishment data approximate the corresponding figures obtained by the labor force procedures described in the preceding section.

The two sets of statistics, labor force and establishment, have different uses. The former provide information about the personal and job characteristics of workers in the total United States; only a minimum of information is available for the country’s regions or for the various industries. The establishment data, on the other hand, provide almost no information about the characteristics of the workers but do provide data on employment for state and local areas and for detailed industries.

Other by-product data

Certain other series of data covering one or another part of the American working force are available as by-products of administrative procedures. One such series comes from the reports which employers file on payroll pension deductions for workers under the Federal Old Age and Survivors Insurance Program; the series naturally covers only the employed. A series of unemployment data is available as a by-product of the operations of the unemployment insurance program. A count is obtainable weekly of the numbers of persons drawing unemployment insurance benefits. The figures on unemployment so derived range between one-third and one-half of the figures as reported by labor force procedures. Counts are also available of the numbers of persons for whom unemployment insurance has been paid, thus providing information about employment.

Many other administrative programs are available which, as a by-product of their primary operations, furnish information about one part or another of the American working force. Thus additional statistics are available on employment in coal mining, agriculture, government, schools and colleges, and a number of other segments of the economy.

In general, these by-product statistics tend to be rather specialized and therefore most useful for analysis of particular aspects of the working force rather than for over-all working force analysis. For example, if one is particularly concerned with college teachers he would turn to the data provided by the Office of Education, supplemented by the decennial census data, rather than to the monthly labor force statistics.

Data collection in other countries

Several other countries have adopted the United States labor force procedures in whole or in part. These are: Japan, Italy, Canada, Federal Republic of Germany, France, Philippines, Sweden, and Panama. Since both the length of time that these surveys have existed and their detailed mode of operation vary considerably from one country to another, we cannot discuss each country separately in this article. Some of them conduct monthly surveys, some quarterly, and some less often or irregularly. Information about employment and unemployment is generally asked as of one week, sometimes for a period of one month. The minimum age of those questioned varies from country to country in accordance with customary beginning working ages. In all countries samples of households are interviewed. In addition, there is some tendency to incorporate labor force procedures into the population censuses, which are generally taken every ten years.

Other countries, such as Great Britain, have highly developed public employment offices and all-inclusive establishment reporting; they therefore rely on these two administrative programs to furnish the available working force statistics. Thus, the public employment offices provide counts of the numbers of persons seeking jobs (the “unemployed”), while the establishment reporting, including the reporting of payroll deductions for the various social welfare programs, produces statistics on employment.

A variety of other statistical collection programs is in use; many of them are somewhat haphazard. Most countries that take occasional population censuses obtain some information about their working forces, though the information varies in usefulness. Some countries collect information only for selected segments of their working force, such as employees in factories hiring twenty or more workers; other countries have other procedures.

The safest general statement to make is that most countries, with the exception of some of those which adopted the labor force procedures or have highly developed public employment offices and establishment reporting, do not have working force data of any real use for public policy information. In summary, for the majority of the world’s people, no useful working force statistics are available.

Measurement of underemployment

One innovation which some countries are trying to introduce through modifications of the labor force procedures is the measurement of underemployment. In effect, these countries are trying to subdivide the information on the employed into two groups: those fully employed and those underemployed. Obviously these two terms are relative; the dividing line can be placed at any point in between total unemployment and working busily for 120 hours a week and 52 weeks in the year at top pay.

The procedures which have been tried in an effort to measure underemployment fall into three types: (a) arbitrary efforts to allocate workers in accordance with the number of hours worked per week (or some other time period); (b ) use of amount of time which the person had wanted to work, so that the underemployed are defined as those who worked less time than they had wanted; (c) some assessment of the worker’s desire to change his job because of the poor quality of present employment, including low earnings, little opportunity to use skills, and so on (United Nations 1958).

On the basis of the experiences to date (1964) we can only conclude that there is no uniquely correct way of measuring underemployment: alternative measures provide alternative statistics. The most important factor is the use of identical procedures over time within any given country, so that changes can be measured. In fact, only changes over time can be measured, never absolute levels.

If there is any more nearly “correct” measure of underemployment for a particular country, it is that definition which best fits the social, economic, and political context—if such a definition can be determined. This leads to the conclusion that it is impossible to have so-called “internationally comparable” measures of underemployment unless several countries happen to be identical in all manpower respects.

International comparability

Similar collection procedures need not produce similar and comparable working force statistics.

This is so, as we have pointed out previously, because the working force—its concept and measurement—is an artifact created to serve a specific purpose within a specific social, economic, and political context. Where societies are organized differently, have different values and goals, and government or private policy makers are concerned with different problems, then different concepts and procedures may be needed. If the same procedures are employed, the net result may be the provision of useless and meaningless statistics for some of the countries.

This can be illustrated by applying the labor force procedures to a semisubsistence population living at very low levels. When subjected to the barrage of questions as previously described, these semisubsistence workers will all turn out to be “employed farmers”; very few will report being unemployed even though they and their families may be living at the edge of starvation. For example, to say that there is virtually no unemployment among Panamanian farmers in 1960 (about 1 per cent) and therefore that they are better off than American workers, among whom unemployment in 1960 was about 5.5 per cent, is meaningless. The working force problems which Panamanian policy makers must consider are quite different from the problems with which United States policy makers must deal, although there is some overlap.

Even when ostensibly similar countries are being compared in terms of employment and unemployment, working force statistics, although obtained by identical procedures, may be noncomparable. For example, in the United States workers tend to work a full week of 40 hours and try to maximize their pay. There seems to be comparatively less interest among American workers, as contrasted with European ones, to spread the work so that more people may have jobs, although at lower pay. American workers, on the contrary, seem anxious to maximize their earnings even if, as a result, some workers become unemployed. Now unemployment can always be eliminated if there is enough spread-the-work; note the labor force procedures for ascertaining unemployment. Therefore, if in one country there is somewhat more of a tendency to spread the work than in another country, the resulting unemployment statistics will not be comparable in the sense of being equally useful for policy determination.

Even the size of the working force may vary from one country to another because of social or economic conditions which could tend to make the statistics noncomparable. Application of the labor force procedures in a country in which the population knows that it is extremely difficult to find jobs will result in many persons replying that they neither worked nor sought work; therefore, they are “not in the labor force.” In another country, however, if the people are more optimistic, or more determined, more may seek work and thereby increase the number of unemployed and the size of the labor force. Yet the unemployment problems with which the policy makers in the two countries may have to contend may be quite similar.

In summary, working force statistics are most useful for dealing with the problems of a specific country and for determining the changes that have occurred there over time. They are much less useful, in their present state, for comparing the working force situation in two or more countries; and this difficulty, because it is inherent in the nature of the problem, will not be solved by mere technical improvement in data gathering procedures.

A. J. Jaffe

[Directly related are the entriesEconomic data; Employment and unemployment. Other relevant material may be found inAgriculture, article onLabor; Economy and society; Industrialization.]


For detailed statements about methods used by the U.S. Bureau of the Census, see U.S. Bureau of Labor Statistics 1964; U.S. President’s Committee to Appraise Employment and Unemployment Statistics 1962. An earlier report which also provides much information on measurement procedures in the United States is Ducoff & Hagood 1947. For discussion of the measurement of the working force in other countries, including internationally recommended measurement procedures, see United Nations, Statistical Office 1958; International Labour Office 1959. Concepts of the working force are explored in considerable depth in Jaffe & Stewart 1951; Lederer 1932. The measurement of underemployment is included in Jaffe 1959. Attempts at comparison of the working force in the United States and other countries are given in Myers & Chandler 1962a, 1962b. Walker 1866 is an interesting mid-nineteenth-century treatise.

Ducoff, Louis J.; and Hagood, Margaret J. 1947 Labor Force Definition and Measurement. New York: Social Science Research Council.

International Labor Office 1959 The International Standardisation of Labour Statistics. Studies and Reports, New Series, No. 53. Geneva: The Office.

Jaffe, A. J. 1959 People, Jobs and Economic Development. Glencoe, III.: Free Press.

Jaffe, A. J.; and Stewart, Charles D. 1951 Manpower Resources and Utilization: Principles of Working Force Analysis. New York: Wiley.

Lederer, Emil 1932 Labor. Volume 8, pages 615–620 in Encyclopaedia of the Social Sciences. New York: Macmillan.

Myers, Robert J.; and Chandler, John H. 1962a International Comparisons of Unemployment. U.S. Bureau of Labor Statistics, Monthly Labor Review 85: 857–864.

Myers, Robert J.; and Chandler, John H. 1962& Toward Explaining International Unemployment. Rates U.S. Bureau of Labor Statistics, Monthly Labor Review 85:969–974.

United Nations, Statistical Office 1958 Handbook of Population Census Methods. Volume 2: Economic Characteristics of the Population. New York: United Nations.

U.S. Bureau of Labor Statistics 1964 Concepts and Methods Used in Household Statistics on Employment and Unemployment From the Current Population Survey. Washington: Government Printing Office.

U.S. President’s Committee to Appraise Employment and Unemployment Statistics 1962 Measuring Employment and Unemployment. Washington: Government Printing Office.

Walker, Amasa (1866) 1872 The Science of Wealth. 7th ed. Philadelphia: Lippincott.


The labor force participation rate of a population is defined as the proportion of the population that is either employed or in search of employment in the labor market during a given period of time. Labor force statistics of many countries contain information about participation rates of the total population and of various population groups. The data reveal large differences in labor force rates among population groups in the same economy. Available time series also reveal sizable and often contrasting changes in labor force participation rates in the various population groups.

Some of the differences in labor force participation among age, sex, and marital-status group prevailing at a given time in a given country, are easily understood. It is much more difficult to interpret the diverse patterns of long-run and shortrun changes in labor force participation. In this article an attempt is made to interpret labor force trends in the light of economic theory and of the growing empirical research in this area.

Table 1 illustrates the patterns of labor force participation of population groups in the United States, classified by age and sex, in the years 1900, 1930, and 1963. The table contains some rathe sharp and divergent trends among the several agesex classifications. Roughly similar trends, varying in magnitude, have been observed in many other countries.

Table 1 — Participation in the labor force by age and sex, United States, 1900, 1930, and 1963 (percent of each group in labor force)
Sources: For 1890 data, Durand 1948, Table A-7, pp. 216–217. For 1960 data, U.S. Bureau of Labor Statistics 1961, Table B, p. A-8.
14 and over85.782.177.720.023.636.5
65 and over63.153.928.

Economic analysis can provide insights into only some of the factors influencing labor force participation, such as incomes and relative prices. Since noneconomic considerations bulk large in labor force behavior, as in other spheres of human behavior, only a partial understanding can be achieved by this analysis. It is fair to add that even within these limitations such an analysis can never be definitive, nor does it represent a full consensus of specialists in this field of study. The present discussion contains a set of hypotheses for which the degree of empirical verification varies widely. It is, therefore, to be viewed not only as a summary of current research, but in part also as an agenda for future research.

At the outset, it is important to recognize the distinction between the theoretical concept of a quantity of labor supplied by a given population and the empirical measurement of labor force par ticipation. The notion of a quantity of labor supplied by a given population is broad, and hence ambiguous, unless the purpose for which this labor is supplied is specified and the units in which the quantities are to be measured are defined. Labor force surveys, however, are quite specific: they purport to measure the quantity of labor supplied for gainful employment or self-employment in the labor market. They measure this quantity by a simple count of people who are so employed or in search of such employment during a given short period. By this definition, the surveys focus on labor markets, on short periods of observation, and on the number of participants—without regard to the extent of participation in terms of hours of work or in terms of periods longer than the survey week. All these restrictions must be kept in mind in applying the broad language of economic analysis to the narrowly defined empirical measurements of labor force participation.

The starting point for this examination, as for most economic analysis, is the problem of individual choice in allocating scarce resources among alternative uses. In this case, it is the problem of allocating time among alternative uses. The simplest set of such alternatives is a dichotomy: time used in production (work) and in consumption (leisure). [SeeLeisure; Time Budgets.] Using this dichotomy, the effect of the wage rate on the quantity of labor supplied, measured in time units, can be analyzed in terms of the demand for leisure time, which is viewed as a consumption good. Standard demand analysis predicts the effects of changes in relative prices and in income on consumption : a rise in price of a good relative to prices of other goods leads to a decrease in its consumption—that is, a “substitution effect” in favor of other goods; a rise in income normally (excepting “inferior” goods) leads to an increase in consumption—the “income effect.” Since the price of leisure is the forgone wage, a rise in the wage rate makes leisure more expensive, inducing the worker to work more; in effect, to “purchase” less leisure. At the same time, however, an increase in the wage rate increases income, which leads to increased “purchases” of leisure, that is, to decreased hours of work. Which of the two effects triumphs cannot be determined a priori. It depends on the relative preferences of individuals between utilities afforded by the purchasing power of wages and those resulting from the availability of leisure time (Robbins 1930).

Adult males

According to a widely accepted hypothesis, the income effect of the demand for leisure is stronger, on the average, than the substitution effect, so that an increase in the wage rate normally results in a decreased amount of work (Douglas 1934). This hypothesis, known as the “backward-bending” supply curve of labor, can be used to explain the secular decline in the work week in many countries by the secular growth in real incomes per capita (Lewis 1956), provided some oversimplifications are deemed empirically inconsequential. The major one is the use of a dichotomy: “work” is work in the labor market; all other activities are defined as “leisure.” This over-simplification is perhaps not too severe in the case of adult males in industrial societies. Since the analysis does not specify the period over which the allocation of time is made, it leads to the following generalization: Looking at the lifetime behavior of individual males, and abstracting for the moment from learning processes, aging, and family life-cycle developments, we would expect the fraction of the lifetime devoted to work to decrease as real income and wealth per capita rise.

This fraction, however, can shrink in more than one way: hours worked per week or per year may decline with no change in participation; the converse may be true; or both may decrease. Thus, granted that growth in income affects the length of the work week, does it follow that it also affects labor force participation? For the relevant population group, adult males (say, aged 25 to 60 in the United States), there are economic considerations that lead to a negative answer. According to these, income changes affect hours of work, but not labor force participation, of adult males. Table 1 shows that this is in fact the case: in contrast with the sharp secular declines in participation of young and old males and with an increase in the participation of adult women, the attachment of adult males to the labor force has remained very strong and about the same in 1963 as in 1900. These high and unchanging levels of labor force participation are characteristic of adult males in many other parts of the world.

With labor income being the main source of income, it is not surprising that adult males, most of whom are the main earners in the family, are continuously attached to the labor force. The preference to absorb the gains in leisure in the form of shorter work weeks or paid vacations, rather than in the form of temporary withdrawals from the labor force, reflects, in part, a preference for steady patterns of consumption over time. By and large, such preferences hold for total consumption as well as for the consumption of leisure. The latter directly implies a preference for steady employment. A steady rate of total consumption is more easily achieved with steady employment, giving steady incomes, than with fluctuating incomes. Moreover, interruptions of employment impose reemployment costs on workers and on employers. These costs can be quite sizable, particularly where some degree of skill is involved, since interruptions of employment adversely affect the maintenance and development of skill and of a work career.

In order to extend the analysis to younger and older males, the economic effects of age must be brought into the picture. To isolate the effect of age, abstract for the moment from training processes. The important phenomenon is then the relation between age and worker productivity. At early ages productivity rises with physical and mental growth. At the other extreme, physiological effects of aging eventually bring about a decline in productivity. With wage rates reflecting productivity, it appears that leisure is less costly at younger and older ages than in the middle years. Differences in the cost of leisure, therefore, call for a greater concentration of work in the middle years and a reduction of it at young and old ages. However, assuming the same hours, currently obtainable earnings vary proportionately with the wage rate. Hence, if labor force behavior responded to current income in the manner specified by the backward-bending supply curve, the cost of leisure effects would be more than offset by income effects. This would predict that the young and old would work more than those in the middle years. Such a conclusion is, of course, at variance with observed facts.

But, to the extent that assets and credit are available to the individual, it is clear that current income is not an effective constraint on consumption decisions. The relevant variable to which consumption standards are related is the long-run income position. This is a widely accepted explanation of consumer behavior. The case for a similar income concept as a determinant of labor supply is equally compelling (Lewis 1956; Mincer 1962a). With this income concept in mind, it becomes advantageous, in view of variations in the cost of leisure, to distribute the consumption of leisure disproportionately toward young and old ages. This tendency is reflected in shorter hours, intermittent participation, or total absence from the labor force at the two extremes of working life.

An important implication of this analysis, still assuming the validity of the backward-bending supply curve, is that secular growth in real income tends to raise the age of entry into the labor force, lengthen the span of retirement, and/or increase the frequency of part-time and part-period work at the two extremes of age. As Table 1 shows, the age distribution of the male labor force in the United States and the secular changes in the various age groups are consistent with this analysis. Such patterns, including the progressive elimination of child labor and earlier retirements, are well known in countries that have experienced long periods of growth in real per capita income.

This does not mean that the explanations here advanced are incontrovertible or that they are sufficient. Other factors, some of them economic, are involved and must be considered.

Older males

A part of the sharp secular decline in participation rates of older males (65 and over) in countries like the United States might be explained by increasing longevity, one of the benefits of economic growth. But this factor is quantitatively unimportant: While the 65-and-older group grew in relation to total population, there was little change in the age distribution within the older group in the United States (Long 1958).

The growth in size of the older group relative to the younger population may have had some adverse effects on their earnings and employment opportunities, particularly in view of the lower average educational background of older people. Very little evidence is as yet available to shed light on this matter. Some decline in participation rates is undoubtedly associated with the relative decline of employment in agriculture. But participation rates have declined, and quite sharply, among older males in both urban and rural areas (Long 1958). There is little evidence to suggest that the reduction in participation rates is due to a deterioration in health at given ages. However, rising health standards, which are particularly responsive to growth in income, may mean that a worker on the border line of physical fitness is less inclined to continue working today than in the past. Contributing to this effect has been the decline in occupations permitting a gradual tapering off of activity, such as self-employment.

A financial factor of some importance is the more recent growth of the benefit provisions of public old-age security programs, public assistance benefits, and private retirement plans. This factor constitutes a reinforcement of the previously emphasized income effect of rising real earnings. This is true to the extent that the redistribution of consumption from younger to older ages is strengthened by these programs—that they increase the net volume of savings available for retirement. There are reasons to believe that this is the case, in view of the income redistribution effects and tax incentives built into these programs. Moreover, the provisions often make eligibility for benefits conditional upon giving up gainful work, or make the benefit a declining function of the income earned during the benefit years. Such provisions reduce the cost of leisure in old age and encourage the withdrawal of old people from the labor force. Indeed, in the United States as elsewhere, the declining trend in labor force participation of older people has accelerated since the 1930s, when such programs were introduced, extended, and liberalized.

The discussion thus far implies that the low and declining participation rates of older people are largely voluntary, proper account being taken of health levels and standards. It is held by many that this view is inadequate, that many old people retire involuntarily, and that declining employer demand for older people is an important force in the secular reduction of their labor force participation.

Unquestionably, many individuals in good health are retired either as a consequence of being laid off or of compulsory retirement upon reaching a certain age. However, neither layoff nor compulsory retirement age need result in withdrawal from the labor force, if the worker seeks and manages to find another job. But evidence abounds to the effect that this is particularly difficult for the older worker. Unemployment rates are higher and the duration of unemployment longer for older people. The available data indicate that, in the short run, reductions of older men in the labor force are closely associated with unemployment for the economy as a whole, as well as among industries and occupations (Jaffe & Stewart 1951).

Over the long run, however, aggregate unemployment rates have not been increasing. To what extent, if at all, they have been increasing for older males cannot be ascertained from the available record. The difficulties encountered by older workers in hiring and such practices as compulsory retirement are often described as discriminatory. Little empirical evidence exists on whether such discrimination has been increasing. Nor is it clear why discrimination, if viewed as a cultural or psychological factor, should have been growing over time.

However, a certain amount of plausibility for the discrimination hypothesis is suggested by economic factors, the most important of which is the difficulty of downward adjustment of wages to individual changes in productivity. Wage adjustments to the decline in productivity of older workers are difficult because of wide individual differences in the time at which productivity starts declining, and because of the morale-disrupting effects that continuous downward adjustments would bring about. Firms therefore find it most economical to estimate an average age at which productivity declines are likely to require significant reductions in wages and to provide for compulsory retirement at that age. The larger the work force of a firm, the more economical this procedure from a management point of view.

Moreover, to the extent that training on the job is important and is partly financed by the firm, hiring of older people frequently becomes disadvantageous. Firms engaging in such training are also likely to be large and to have seniority schemes, pension plans, and policies of hiring younger people for low-level jobs and promoting them from within. Thus, the growth in average size of firms, and the secular upgrading of skills in the labor force resulting from growth in education and in associated training on the job, may contribute to a spread of compulsory retirement and of discriminatory hiring practices.

All of the forces enumerated above may be assumed to have been at work in some degree: rising incomes, public and private security programs, and health standards on the supply side; as well as declining demand, which is related to the same technological and educational trends that raised incomes and reduced supply. The relative importance of these factors and the validity of their interpretation must await further empirical exploration.

Young males

In most industrial countries, child labor is a thing of the past. Legislation requiring school attendance is widespread, and the compulsory age for such attendance has been progressively raised. But even above those ages we find that labor force participation of young people is much lower than in the adult age groups. There is little doubt that the participation rate of teenage males, as well as those in their early twenties, is lower in more industrially advanced countries than in others, and that it has been shrinking historically.

It is clear that school enrollment and its historical growth are the basic facts in labor force behavior of young people. Indeed, using recent U.S. data, if school enrollment is added to the labor force of young males (ages 14–19) who are not at school, a total of over 90 per cent of the population of the group is obtained, almost as high as the labor force participation rate of adult males. And, historically, the declines in the labor force of young males have almost precisely matched their inflows into classrooms (Long 1958).

These are facts, not explanations. Can these facts be explained in terms of income and price effects on the demand for leisure? The answer is “Yes,” provided education is considered a specific component of the leisure category, a time-intensive form of consumption. It must be recognized, however, that neither the variables entering the demand for education nor the magnitudes of responses (elasticities) to changes in income and in prices are precisely the same as in the demand for the general category of leisure time. To use an analogy, the demand variables and the price and income elasticities are not the same in the consumption of meat as in consumption of all food.

The price, or cost, of education consists of two parts: the forgone wage and the actual direct costs, such as tuition. At a moment of time and at a given age and education level, the cost of education (of the same quality) is roughly the same for all students; hence, one would expect to find a positive relation between (parental or community) income and the fraction of young people enrolled in school. Empirical estimates based on cross-section data indicate a very strong effect of family income on education of children. This suggests that historical income growth is a powerful force in the growth of education. Over time, however, the growth in income is associated with a growing cost of education. Opportunity costs, that is, forgone wages of young people, may have risen as fast as average family incomes; but direct costs of educational outlays probably increased at a slower rate. The net result of the price and income movements —growth in education—can be interpreted in the same manner as the decline in the work week: the strong income effect dominates the price effects working in the opposite direction.

The validity of this simple analysis hinges on the proviso that education is viewed as a consumer good. True, education is often valued for its own sake, and in a literal sense the shift from conditions of child labor to school constitutes a gain in leisure for the young. But this view disregards the effects of education on market productivity. The effects of education on earning power and on upward social mobility are well known. Unquestionably, they play a part in motivating the demand for education. With this focus, education can be viewed as an investment activity, a process of investing in an income-producing asset [seeCapital, human].

To the extent that education is considered an investment rather than a consumption good, the analysis of the determinants of its growth, which underlies the changes in labor force participation of the young, must be revised. Investment demand is not influenced solely by costs of the investment goods, nor is income of the investor very relevant in a properly functioning capital market. Although a cost attaches to an increment of education, there is also a return to be expected, in the form of an augmented future income. From this point of view, what matters in the decision between school enrollment and labor force participation is not the cost of education, but the rate of return on it. If education and its growth were to be interpreted entirely as a consumption phenomenon, money rates of return could be high, low, or even negative. And, whatever the initial level of rates, a secular spread of education generated by forces unconcerned with monetary investment yields would have tended to depress these rates. Empirical research in the United States indicates that during the last several decades, rates of return to education have been neither low nor declining, despite massive educational growth (Becker 1960; Mincer 1962b). The inference which these data seem to suggest is that of a steady movement of (young) manpower resources into educational investment in response to a continuously growing demand for an educated, skilled labor force.

With rates of return on educational investment higher than in alternative pursuits, the influx of young people into classrooms is limited only by such barriers as insufficient ability, income, and information. Income is a limiting factor, since finance for educational investments is not readily obtainable in capital markets. Hence, the secular growth in income has increasingly facilitated the growth of education. Public subsidies to education have reinforced this effect. At high school and higher levels there has been a tendency to transfer some of the traditional training functions of firms (apprenticeships, engineering and business preparation) to schools. Labor force participation is thereby decreased without any real change in the activity of the young worker.

Why are educational alternatives important in the labor force behavior of young people, and unimportant in other age groups? The answer is provided by the investment aspect of education. A given increment of education benefits younger people more than older ones because of the longer remaining life over which these (monetary and nonmonetary) gains accrue. One application of this analysis, of some relevance to underdeveloped countries, is that improvements in health and increased longevity, even without an initial growth in income, are likely to bring about increases in educational investments, hence decreased labor force participation of young people.


The two-way allocation of time between work in the labor market on the one hand, and leisure or education on the other, is a severe but tolerable oversimplification in the analysis of labor force behavior of males. The dichotomy is an oversimplification of a more general condition in which the allocation of time is a three-way choice between leisure, paid work in the labor market, and production outside the market. The last is the most important activity of all population groups in primitive societies where product and labor markets are not yet developed. In advanced societies, where work at home is a shrinking sector of economic activity, the biological and cultural specialization of functions in the family continues the assignment to women of the major responsibility for the care of home and household. Hence the much smaller labor force participation of women than of men in most societies, except in those agricultural countries where census practices count all women living on the farm as employed. Since the demand for home care varies over the life cycle of the family, reaching its peak at the childbearing and child-rearing stages, it is not surprising that single women past school age have the highest rates of participation and mothers of young children have the lowest rates.

The family context of labor force behavior that is brought to the fore in the discussion of women leads to the following generalization in the economic analysis of labor supply:

The amount of work supplied by a family member to the market depends not only on his (or her) market wage rate, but also on the total income position of the family, on the individual’s productivity in household work and other activities outside the market, on prices of market substitutes for household goods and services, and on family tastes. Since income has a positive effect on the demand for leisure, a higher family income implies a smaller total amount of work (at home and in the market) supplied by family members. In the case of adult men, the effect of a rising real income has been to decrease their hours of work in the market, this being the major sector of their work activity. There is little reason to believe that the gains in the amount of leisure resulting from growth in real (family) income have been different for women than for men (Mincer 1962a).

But while the total amount of work of women shrank as much as that of men, the distribution of the smaller total shifted, historically, from home to market. In many countries this shift has been so strong that the amount of work in the market has been growing absolutely, sometimes at a spectacular rate. In the United States, between 1890 and 1960, labor force rates of all females 14 years and over rose from about 18 per cent to 35 per cent. In the same period, rates of married women rose from 5 per cent to 30 per cent. (See Table 2.) Real income per family tripled in this period.

These shifts are explainable by the historical changes in the other variables. Increases in female wage rates in industry, which were at least as rapid as those of males in the United States, and declines in relative prices of market substitutes for home goods and services (such as food preparation and labor-saving appliances) meant that increasingly larger quantities of goods and services needed by the home could be obtained by an hour’s work in the market than by an hour’s work at home. Indeed, the decline in family size, which is often considered an independent explanatory factor facilitating the shift of women’s activities to the labor market, may have been in part induced by the same growing market wage, the growing opportunity cost of child care.

The reality and importance of these factors have been substantiated by empirical research (Long 1958; Mincer 1962a). The evidence indicates that for women with similar earning power, labor force participation rates respond negatively to their husbands’

Table 2 — Participation in the labor force of women by marital status and age, United States, 1890 and 1960 (per cent of each group in labor force)
* Widowed and divorced.
Sources: For 1890 data, Durand 1948, Table A-7, pp. 216–217. For 1960 data, U.S. Bureau of Labor Statistics 1961, Table B, p. A-8.
14 and over18.236.94.528.634.844.130.540.0
14–19    25.525.337.3
65 and over7.616.

incomes: the more husbands earn, the less wives work. But for families with a similar income position of husbands, the more the wife is capable of earning in the market, the more she works. Indeed, the positive response of the woman’s labor force participation to her market wage rate is substantially stronger than her negative response to her husband’s income. This dominance of the substitution effect over the income effect is the basic behavioral characteristic capable of explaining much of the secular growth in the female labor force (Mincer 1962a).

The neglect of this factor can create confusion in the interpretation of observed moment-of-time and over-time relationships. The negative relation between labor force participation of women and family income in the cross section appears to be inconsistent with the historical growth of the female labor force. Only when the other factors, such as the sharp rise in women’s market wage rate, are taken into account is the inconsistency resolved (Mincer 1962 a).

The smaller increase in labor force participation of single women compared with the growth of participation of married women in the historical record of the United States is explainable by the same analysis. Roughly speaking, with responsibilities of single women in the household sector being intermediate between those of married women and mothers on the one hand and males in the family on the other, the result is intermediate.

It is interesting to note that in the United States the secular changes in participation rates of older women (65 and over) and of young girls (14–19) contrast with the changes in the labor force of other adult women and with the changes observed in corresponding age groups of males. The participation rate of elderly women is very small and has remained roughly constant over time. The growing incentives to participate in the labor market play a part in counteracting the declines observed in elderly males. These incentives may include the effect of the market wage rate, but there are additional ones: with growth in the proportion of households limited to nuclear families, and with increasing longevity, the demand for work at home by women whose children have established their own homes shrinks rapidly. These developments not only counteracted the factors that would otherwise have led to a decline in participation of older women but also contributed to the spectacular growth in labor force participation of women aged 45 and over.

Young girls (aged 14–19) had much lower participation rates than did boys in 1900. Over time these rates declined but little, in contrast with the steep decline in labor force participation of teenage boys. The difference cannot be ascribed to a lesser growth in school enrollment of girls. Whereas 36 per cent of the girls in this age group were enrolled in school in 1900, almost twice as large a percentage (60 per cent) were in school in 1963. Both figures are almost precisely the same as for young boys in this age group. However, whereas most of the boys not in school (62 per cent) were in the labor force in 1900, only 27 per cent of the girls not in school were in the labor force in 1900; the remainder, 37 per cent, presumably were helping out at home. In 1960, 25 per cent of the girls were in the labor force, a small decline compared to 1900. The largest shift was from home to school, with the potential shift from home to market more than offset by the shift from market work to school. In terms of economic analysis, these relative shifts in supply are intelligible as responses to the highest rates of return beckoning in education, the lowest in the household, and intermediate ones in the labor market. This ranking is consistent with the previously analyzed behavior of young males and of adult women. But such interpretations, particularly those regarding young girls and elderly women, remain in the realm of hypotheses as long as systematic studies designed to explore the magnitudes of suggested behavioral parameters have not been carried out.

International comparisons

A few remarks regarding international comparisons are in order. Differences in national statistical practices and in concepts regarding labor force membership create great difficulties in interpreting such comparisons. Despite these distortions, surveys of a variety of national censuses (see “Women in the Labour Force,” 1958; “The World’s Working Population/” 1956) reveal a rough conformity with the labor force developments that were analyzed here on the basis of U.S. data.

Thus, in all parts of the world over 90 per cent of men between the ages of 20 and 64 are economically active, and the proportion has remained virtually unchanged over the years. In industrial countries the percentage of young persons in the labor force has tended to decline. Indeed, gainful employment of children under 14 has almost disappeared in many of these countries. But children and young people form a considerably larger proxp ortion of the labor force in nonindustrialized countries. The proportion of old people (65 and over) in the labor force is generally lower in industrially advanced countries than in others. It has been declining in most of the advanced countries.

Differences in statistical practices are particularly obstructive in comparisons of labor force participation of women. In most countries their attachment to the labor force is weak, and the difficulties of distinguishing between their “economic” and “noneconomic” activities, particularly in non-industrial countries, result in diverse and arbitrary statistical classification. Still, when the agricultural sector is separated out, it appears that the upward trend in the female labor force has been shared by a number of industrial countries, with the United States and Canada experiencing the sharpest rise.

The general consistency of these data with the analytical economic structure sketched in this article, however, should not conceal the wide diversity in levels and rates of change in labor force participation of the specific population groups in the various countries. Intensive analyses of local economic and social conditions will be required to clarify these phenomena.

Jacob Mincer


Bancroft, Gertrude 1958 The American Labor Force: Its Growth and Changing Composition. New York: Wiley. → Published for the Social Science Research Council and the U.S. Bureau of the Census.

Becker, gary S. 1960 Underinvestment in College Education? American Economic Review 50, no. 2:346–354.

Douglas, paul H. 1934 The Theory of Wages. New York: Macmillan.

Durand, john D. 1948 The Labor Force in the United States, 1890–1960. New York: Social Science Research Council.

Jaffe, abram J.; and Stewart, C. D. 1951 Manpower Resources and Utilization: Principles of Working Force Analysis. New York: Wiley.

Lewis, H. G. 1956 Hours of Work and Hours of Leisure. Pages 196–206 in Industrial Relations Research Association, Proceedings of the Ninth Annual Meeting. Chicago: The Association.

Long, Clarence D. 1958 The Labor Force Under Changing Income and Employment. National Bureau of Economic Research, General Series, No. 65. Princeton Univ. Press.

Mincer, Jacob 1962a Labor Force Participation of Married Women. Pages 63–97 in Universities-National Bureau Committee for Economic Research, Conference, Princeton, N.J., 1960, Aspects of Labor Economics. National Bureau of Economic Research, Special Conference Series, No. 14. Princeton Univ. Press.

Mincer, Jacob 1962b On-the-job Training: Costs, Returns, and Some Implications. Journal of Political Economy 70 (Supplement): 50–79.

Robbins, lionel 1930 On the Elasticity of Demand for Income in Terms of Effort. Economica New Series 10:123–129.

U.S. Bureau Of Labor Statistics 1961 Marital and Family Characteristics of Workers, March 1960. Special Labor Force Report, No. 13. Washington: Government Printing Office. → First published in the Monthly Labor Review, April 1961.

U.S. Bureau Of Labor Statistics 1963 Statistical Tables: Section A. Labor Force, Employment, and Unemployment. U.S. Bureau of Labor Statistics, Employment and Earnings 9, no. 4.

Women in the Labour Force. 1958 International Labour Review 77:254–272.

The World’s Working Population: Some Demographic Aspects. 1956 International Labour Review 73:152–176.


In its broadest sense, the labor market embraces all those institutions and processes relating to the purchase, sale, and pricing of labor services. Its principal actors are: the individual worker, with his more or less unique pattern of abilities, skills, and preferences for various kinds of work and various combinations of rewards; workers’ organizations, which have as their purpose the improvement of labor’s economic and social position through collective bargaining, mutual aid, and political action; the individual employer, with his more or less distinct pattern of manpower needs, hiring preferences, and personnel policies; associations of employers, with the (not necessarily exclusive) purpose of promoting employer interests in labor relations; and government. In addition to its role as employer, the last-named plays a part in the labor market as regulator of employment conditions (for example, through minimum-wage legislation), organizer (by such means as the public employment service), formulator of the law of union-management relations, and general economic planner.

The relative influence of these actors as well as the specific functions of each vary considerably from country to country and even from sector to sector within each country. Their interaction, within the framework of the pattern of output and the character of technology in the economy, determines both the allocation of labor among alternative productive uses and the conditions of employment (including the pattern of rewards for work).

Of central importance in the labor marketing process is the mobility of labor, that is, the shifting of human resources into and out of the labor force and among firms, occupations, industries, and geographic localities. These kinds of movement of workers embrace all the processes whereby labor supply adjusts to changes in the level and composition of the demand for labor. Analysis of labor mobility, therefore, becomes a means of studying and evaluating the total operation of the labor market. In addition to its relevance to the process of labor allocation, mobility can also be studied with reference to its contribution to individual goal fulfillment, as well as in the context of the class structure of society. Since occupation is an important criterion of social class, the ease or difficulty of “climbing the occupational ladder” is indicative of the relative “openness” or fluidity of class structure and has implications for the assessment, interpretation, and prediction of sociopolitical change [seeSocial mobility].

Concepts of the labor market

The labor market, like many other useful abstractions, is easier to talk about than to define. Conceptually, it is the area within which a given set of supply-and-demand schedules operates to determine wages and other terms of employment. Consider, for a moment, an island with neither immigration nor emigration, small enough so that workers are completely indifferent to the location of their workplace. Assume further that the following conditions are fulfilled: neither workers nor employers combine for purposes of influencing the terms of employment; all workers have the same skill and are equally efficient; all employers are indifferent as to the particular employees they hire, being interested solely in minimizing labor costs; employees continuously decide for whom to work on the basis of wages alone, being solely interested in maximizing their income; and workers have complete knowledge of wage rates being paid by all employers. Under these circumstances, not only would the island constitute a well-defined labor market, but the market would be the perfectly competitive one of traditional economic theory. Workers would be in continuous competition for available jobs; employers would be in continuous competition for labor; and the result would be a tendency toward wage uniformity throughout the island. Wage differentials could not persist because workers in lower-paying firms would make themselves available to higher-paying firms, thus causing wages to rise in the former and to fall in the latter until all wages were equal.

In such an oversimplified model the market is defined by the area within which the uniform wage prevails or—what amounts to the same thing—by the area within which workers move among jobs. Indeed, traditional economic theory treats wage determination and labor mobility or allocation as continuously interrelated processes. Wage differentials, which are presumed to measure differences in contribution to the social product, induce the movement of workers in the direction of higher-paying jobs, and this movement continues until the wage differentials either disappear (for identical work) or (in the case of different occupations) are just large enough to compensate for the relative disutilities of different kinds of work. Thus, in this idealized version of the “market,” the end result is a “perfect” allocation of labor for, given the pattern of consumer preferences for final goods and services, the pattern of worker preferences for alternative types of work, and the existing technology, consumer satisfactions could not be enhanced by moving a single worker from where he is to any other job [seeWages, article onTheory].

The difficulty of defining a labor market operationally stems largely from differences of opinion as to how strongly the competitive forces of the foregoing model actually operate. On the one hand, at least since Cairnes (1874, pp. 65–68), the concept of “noncompeting” occupational groups has been recognized. This suggests the necessity for thinking in terms of a system of submarkets for labor. Also, evidence for the United States has revealed labor market behavior on the part of both employers and workers that differs significantly from that postulated by the traditional theory; as a consequence, substantial wage differentials for comparable work persist not only among local areas but within communities. Thus Reynolds (1951, p. 83) has stated that most employed workers are not “in the market,” in the sense of being aware of or interested in alternative jobs, and that as a consequence, each firm constitutes “virtually a separate labor market.”

On the other hand, irrespective of the strength of competitive forces, there is no question of their existence. Workers can and do move among firms within a community, among communities, regions, and, indeed, nations; they even move from one “noncompeting group” to another. From this standpoint, it is meaningful to talk about “the labor market,” even if one recognizes the existence of related “submarkets,” within each of which the conventional forces of supply and demand operate with somewhat greater (although by no means perfect) vigor.

Perhaps the best summary of all this is Kerr’s observation that labor markets have “vague and varying contours but no ultimate limits short of those for American society itself.... Most labor markets are ... indefinite in their specification of the sellers and the buyers. Such a labor market is merely an area, with indistinct geographical and occupational limits within which certain workers customarily seek to offer their services and certain employers to purchase them” (1954, pp. 92℃93).

Dimensions of labor mobility

As has been indicated, the movement of workers among jobs and labor market areas is at the core of the operation of the labor market mechanism. The short-run supply of labor adjusts to changes in the volume and pattern of labor requirements by movements of workers into and out of the labor force, between employment and unemployment, and among firms, occupations, industries, or localities. For analytical purposes, therefore, corresponding categories of mobility have been used to classify the types of labor market changes made by workers: labor force mobility, employment mobility, interfirm mobility, occupational mobility, industrial mobility, and geographic mobility. Although these types of movement exhaust all possible sources of change in the number, location, and function of human resources utilized in the productive process, they clearly do not represent mutually exclusive categories of labor market transactions, since a single job change frequently represents a combination of two or more of the types of movement described above.

How a particular job change is classified in the foregoing schema obviously depends upon the way in which occupations, industries, and geographic areas are classified. The broader and less detailed the occupational classification system, the fewer will be the occupational shifts registered. Some studies have treated as occupational and industrial moves only shifts between major occupation groups, such as professional, managerial, and clerical, and between major industry divisions; while others have used more detailed occupational and industrial categories. In the case of geographic movement, the concept has generally referred to a change in job status that would normally necessitate a change of residence. Sometimes this is defined in terms of a minimum distance between the two jobs; but sometimes it is taken to be any job change across county, state, or regional lines.

Data and measurement

Most mobility studies have focused upon the employment experience of a sample of workers over some period of time. Work histories have been obtained either by questionnaire or interview from the workers themselves, from employers’ personnel files, or from social insurance records. If motivational factors in job decisions are to be analyzed, interviews with workers are virtually indispensable.

Mobility may be defined as the actual movement of workers or as their willingness or propensity to move, given the opportunity and incentive to do so. The latter definition is more closely related to the basic assumption of conventional theory that workers are responsive to differentials in “net economic advantage.” However, it poses serious difficulties of measurement, since “propensity to move” has no operational meaning except in terms of specific circumstances and incentives. Generally, therefore, conclusions with respect to workers’ propensities to move have been inferred from the voluntary job changes they actually make. Use has also been made of hypothetical questions such as: “If you were offered a job with another company doing the same work you are now doing for 10 per cent more pay, would you take it?”

Measurements of mobility have generally been made in terms of number of jobs held or number of job shifts of various kinds made during the period under consideration. Results may be presented in terms of average number of jobs per worker or in terms of frequency distributions of workers according to the number of job shifts they have made. When such measures are used and if comparison is to be made among different groups of workers, account must be taken of differences in time spent in the labor force. Otherwise differences in mobility of various groups may reflect only differences in labor market exposure rather than differences in employment stability or work attachment. Another frequently used measure of mobility involves comparing the worker’s job status at the beginning and end of a specified period, ignoring any changes that might have occurred in the interim. Finally, length of service in the current job has also been used as an inverse measure of mobility. In this case, also, there must be some method of standardizing length of labor force exposure.

Amount and character of labor mobility

Systematic investigation of the amount and character of labor mobility and of the functioning of labor markets has been carried on much more extensively in the United States than in most other countries. The relative paucity of comparative data for other parts of the world makes confident generalizations about international differences difficult, although it appears that the amount of job movement in the United States is considerably greater than in most European countries and that this is a product not only of a larger volume of voluntary movement in the United States than in Europe but also of a greater incidence of layoffs (Palmer 1960, pp. 520–524). Apart from this, European studies that have investigated the same questions have produced findings that are generally consonant with those for the United States (Organization for Economic Cooperation and Development 1965, passim).

Studies of the extent of job shifting among American workers have revealed a substantial amount of flexibility, as well as considerable stability, in work attachments. Scattered evidence relating to the 1940s and 1950s suggests that the proportion of job changers each year might have been as high as one-fourth or even one-third of the total number of employees (Parnes 1960, p. 17). The only comprehensive data covering the entire labor force, however, are more recent sample surveys conducted by the Bureau of the Census, which show that both in 1955 and 1961 about one-tenth of all workers changed employers at least once (Wolfbein 1964, pp. 267–276). How much of this difference reflects a real decline in the mobility of the labor force and how much results simply from differences in coverage and methods of measurement of the studies is not certain. Whatever the size of its mobile segment, however, the labor force also has substantial proportions of workers with strong job attachments. Thus, over one-third of all those employed at the beginning of 1963 had been continuously associated with the same employer (or self-owned business) for at least 10 years.

Not all job movement, of course, is voluntary; the proportion of the total that results from layoff is very sensitive to the level of aggregate demand for labor. For example, during the prosperous 1920s about three-fourths of all separations in United States manufacturing industries were voluntary, but during the depressed 1930s voluntary quits had shrunk, on the average, to one-fourth of all separations and in certain years approached the vanishing point. In 1961, one-third of the job shifts made by a national sample of the United States labor force represented a voluntary attempt to improve status as compared with over two-fifths in 1955 when the unemployment rate was considerably lower.

The degree of flexibility in labor supply depends upon thekinds of shifts made by workers as well as upon their number. The evidence is fairly conclusive that most job shifts made by American workers are “complex”—that is, involve a simultaneous change of employer, occupation, and industry. Even when the very broadest of categories are used, a substantial proportion of job changes are across occupational and industrial boundaries. Only one-third of the job changes made by the United States labor force in 1961 involved no change in major occupation group or major industry division, while over one-third involved a simultaneous change in both.

As might be expected, geographic movement is considerably less common than job changes across industrial or occupational lines. Even unemployed workers are frequently reluctant to leave their home communities because of such factors as home ownership, family and community ties, the expense involved in moving, and ignorance of labor market conditions elsewhere. Nevertheless, although less frequent than industrial or occupational moves, geographic movement is by no means inconsequential in the United States. In the 1950s and early 1960s the annual number of males in the labor force who changed their residence from one county to another was relatively stable at between 6 and 7 per cent of the male labor force, and about half of these moves were across state lines.

The correlates of mobility

The incidence of mobility is by no means evenly distributed among the labor force. There is a pronounced inverse relationship between age and all types of mobility. Not only do older workers change jobs less frequently than younger workers, but their job shifts are less likely to involve changes in occupation and industry. In part, this relationship between age and mobility reflects the job shopping that is characteristic of the early labor market experience of young workers; in part it reflects the discrimination in hiring against older workers, which makes them more reluctant to quit their current jobs. But an additional important explanation is the fact that older workers are more likely than younger ones to have accumulated long tenure in their jobs, which not only provides greater protection against layoffs but also acts as a powerful restraint on voluntary separation. There is abundant evidence that workers with long service do not lightly give up the job security provided by their seniority. Other perquisites associated with seniority (vacations, choice of shifts, pensions) have some influence also, as does the psychological comfort of a familiar routine and a familiar circle of associates.

Evidence on the relative mobilities of men and women is not entirely conclusive. Most studies have shown greater mobility among men, but the interpretation is clouded by the fact that women’s attachment to the labor force is generally less continuous than men’s, so that they have less potential for job changes during a given period of time.

Mobility rates vary among major occupational groups, tending to decline as one moves up the socioeconomic hierarchy. Thus, professional workers and managers generally make considerably fewer job changes than do laborers. Part, though not all, of the differences result from differentials among these groups in the incidence of layoffs. Occupational differences also exist in the pattern of job changes. For example, professional workers are more likely than other occupational groups to make geographic moves, reflecting the broader scope of the market for professional services. On the other hand, because of their substantial investment in training, they show a much higher than average attachment to occupation.

The combined influence of all the factors whose effect on mobility has been investigated seems to account for only a small proportion of the total variation in mobility among individuals. It appears, therefore, that variation in mobility is attributable in large measure to personality traits or circumstances that cannot be ascertained as readily as those that have already been investigated. This is an area of research that has hardly been touched.

Labor market decisions of manual workers

Studies designed to test empirically the traditional economic theory of wage determination and labor allocation generally have focused on the extent to which the attitudes and actual labor market behavior of workers are consistent with the assumptions of rationality on which the theory is based. Most of this research has related primarily to manual workers; there is need for additional study of white-collar groups to identify whatever differences may exist between white-collar and blue-collar labor markets.

A wide variety of factors in addition to wages has been found to be important in conditioning workers’ job decisions. The physical characteristics of the workplace, the nature of the relation between the employee and his supervisors and fellow workers, the security and steadiness of his job, “fairness” of treatment, and the degree of interest of the employee in his job have all been identified as being important. Thus, the influence on workers’ job choices of relative wage rates, or even of differentials in “net economic advantage,” appears to be more diluted than the conventional theory suggests.

Numerous studies have investigated the reasons for voluntary quits, which are a fairly direct indicator of the extent to which workers change jobs in response to perceptions of “net economic advantage.” To begin with, there is mounting evidence that most voluntary separations by manual workers are made before the worker has obtained a new job, which means that only a minority of voluntary separations occur as the result of the worker’s being “attracted into” a better job. Second, only a minority of voluntary job changes, it appears, can be explained mainly in terms of dissatisfaction with wages; far more important is dissatisfaction with such intrinsic characteristics of the job as its physical working conditions and the degree of interest the worker has in it.

There is conclusive evidence that manual workers rely chiefly on informal and more or less haphazard means of finding jobs (random gate application, advice from friends and relatives) as contrasted with more formal and systematic methods (public and private employment exchanges). Also, the extent of workers’ knowledge about alternative job opportunities in the local labor market area has generally been found to be quite meager. It may be concluded that most manual workers who are employed and reasonably satisfied with their jobs—which includes the vast majority—are not “in the labor market” in the sense of knowing about (or even being interested in) the existence and characteristics of other jobs in the locality. Even unemployed workers rarely “shop” for jobs in the sense of making careful comparisons of the characteristics of alternatives; rather, they tend to take the first satisfactory job that comes along, which frequently is the first one offered.

To summarize, empirical studies have tended to confirm the view that workers, far from being concerned exclusively or even primarily with “net economic advantage,” have multiple and complex goals, that their job “choices” are bounded by considerable degrees of ignorance of alternatives, and that the typical worker is a “satisficing” rather than a “maximizing” man—all contrary to the postulates of economic theory.

Employer policies and mobility

The recruiting methods, hiring practices, and personnel policies of employers tend to explain and to reinforce the pattern of labor market behavior that prevails among workers. For example, most employers appear to use informal recruitment methods, relying on gate applications and recruitment through their existing work forces rather than on public or private employment agencies or newspaper advertisements. This makes the labor market quite different from the competitive ideal in which all workers could choose among all job vacancies and all employers could select from among all candidates. Other common employer hiring practices also tend to insulate the firm from the external market. One of these is the formal or informal hiring specifications that most firms appear to have, which prevent certain groups of workers from effectively competing for certain jobs. Another is the policy of promoting from within, for to the extent that a firm follows this policy rigidly there is contact between the internal and external labor markets only at the base of the occupational ladder.

Wage differentials. The net effect of the labor market behavior of workers and employers is to permit interfirm wage differentials in local labor markets that are both larger and more persistent than competitive theory would suggest. It is nevertheless important to inquire to what extent labor supply considerations affect employer wage policies and tend to reduce differentials. For example, do low-wage firms have difficulty in recruiting and maintaining an adequate labor supply? Is there a tendency for low-wage firms or areas to lose workers to the higher paying establishments or areas? Most empirical investigations of local labor markets, particularly those by Reynolds (1951) and Lester (1954), have answered these questions largely in the negative, although all studies show some traces of the traditional economic forces. Moreover, whatever may be true within local labor market areas, the pattern of geographic mobility is consistent with the prediction of conventional theory that movement will occur from areas of lesser to areas of greater economic opportunity. However, how strong this tendency is and whether the chief explanatory construct is wage differentials or simply differences in job availability are still unsettled questions.

Some policy implications

Despite evidence of wide differences between actual labor market behavior and that postulated by conventional economic theory, there is a difference of opinion among economists with respect to how much “damage” such evidence does to the theory. Rottenberg (1956) contended, for example, that the true measure of a theory is not the validity of its assumptions but whether it yields valid predictions, and he argues that conventional theory does describe tendencies that are actually observable in the labor market. On the other hand, the empirical evidence, while not ruling out completely the existence of the kinds of market forces described by conventional theory, indicates that they are considerably attenuated.

So long as the issue is put in these terms it is not likely to be resolved merely by additional research. Yet the issue is of profound importance because of its implications for policy. For instance, if the unrestricted operation of “market forces” is conceived to produce a precisely determinate wage structure and an ideal allocation of labor, then the intrusion of “artificial” influences such as trade unionism or government wage regulation means, by assumption, a misallocation of labor resources. But if the market, even in the absence of such “interferences,” would produce only a rather broad range of wage rates for a given type of work, there is no a priori reason for supposing that the “manipulation” of wages, at least within limits, makes matters any worse (or better). The weight of the empirical evidence would seem to make the latter interpretation far more tenable than the former.

Herbert S. Parnes

[See alsoIgration; Occupations and careers; Wages.]


Cairnes, John Elliot 1874 Some Leading Principles of Political Economy Newly Expounded. London: Macmillan.

Eldridge, Paul; and Wolkstein, Irwin 1956 Incidence of Employer Change. Industrial and Labor Relations Review 10:101–107.

Fisher, Lloyd H. 1951 The Harvest Labor Market in California. Quarterly Journal of Economics 65:463–491.

Jaffe, Abram J.; and Carleton, R. O. 1954 Occupational Mobility in the United States: 1930–1960. New York: King’s Crown Press.

Jefferys, Margot 1954 Mobility in the Labour Market: Employment Changes in Battersea and Dagenham.London: Routledge.

Kerr, Clark 1954 The Balkanization of Labor Markets. Pages 92–110 in E. Wight Bakke et al., Labor Mobility and Economic Opportunity. New York: Wiley.

Lester, Richard A. 1954 Hiring Practices and Labor Competition. Research Report Series, No. 88. Princeton Univ., Industrial Relations Section.

Myers, Charles A.; and Shultz, George P. 1951 The Dynamics of a Labor Market: A Study of the Impact of Employment Changes on Labor Mobility, Job Satisfactions, and Company and Union Policies. Englewood Cliffs, N.J.: Prentice-Hall.

Organization For Economic Cooperation and Development 1965 Wages and Labour Mobility: A Report by a Group of Independent Experts on the Relation Between Changes in Wage Differentials and the Pattern of Employment. With a foreword on the “Implications of the Study for Incomes Policy,” by Pieter de Wolff. Paris: The Organization.

Organization For Economic Cooperation and Development 1966 Wages and Labor Mobility. Supplement No. 1: Abstracts of Selected Articles. Paris: The Organization.

Palmer, Gladys L. 1954 Labor Mobility in Six Cities: A Report on the Survey of Patterns and Factors in Labor Mobility, 1940–1950. New York: Social Science Research Council.

Palmer, Gladys L. 1960 Contrasts in Labor Market Behavior in Northern Europe and the United States. Industrial and Labor Relations Review 13:519–532. Palmer, Gladys L. et al. 1962 The Reluctant Job Changer: Studies in Work Attachments and Aspirations. Philadelphia: Univ. of Pennsylvania Press.

Parnes, Herbert S. 1954 Research on Labor Mobility: An Appraisal of Research Findings in the United States. Bulletin No. 65. New York: Social Science Research Council.

Parnes, Herbert S. 1960 The Labor Force and Labor Markets. Pages 1–42 in Industrial Relations Research Association, Employment Relations Research: A Summary and Appraisal. Publication No. 23. New York: Harper.

Raimon, Robert L. 1962 Interstate Migration and Wage Theory. Review of Economics and Statistics 44:428–438.

Reynolds, Lloyd G. 1951 The Structure of Labor Markets: Wages and Labor Mobility in Theory and Practice. New York: Harper.

Rottenberg, SimoN 1956 On Choice in Labor Markets. Industrial and Labor Relations Review 9:183–199.

Rundblad, Bengt G. 1964 Arbetskraftens rorlighet: En studie av en lokal arbetsmarknad. Stockholm: Almqvist ’ Wiksell.

Silcock, H. 1954 The Phenomenon of Labour Turnover. Journal of the Royal Statistical Society Series A 117:429–440.

Wilcock, Richard C.; and Sobel, Irvin 1958 SmallCity Job Markets: The Labor Market Behavior of Firms and Workers. Urbana: Univ. of Illinois, Institute of Labor and Industrial Relations.

Wolfbein, Seymour L. 1964 Employment and Unemployment in the United States: A Study of the American Labor Force. Chicago: Science Research Associates.


Hours of work have usually been considered analytically only insofar as they are performed by the “economically active” population—the work done for pay or profit is distinguished from that done, notably by housewives, for the direct benefit of the household. The hours of work are also distinguished from those of leisure, in that work is considered to be done for the sake of its product rather than for any satisfactions it yields directly, although in practice much activity during hours of work and leisure alike yields returns of both sorts.

The observed changes in hours of work, so delimited, depend on whether we measure them by the day, the week, or the year. Divergences between the day and the year come about through changes in weekly holidays and half holidays, especially the weekend; those between the week and the year, through changes in public holidays, vacations, absenteeism, sickness, and stoppages. A given total of hours will have different significance, according to the pattern in which those hours are arranged within the day, week, and year and the variability of that pattern. The total hours of work performed by a given household or community depend in part on what proportion of its members engage in the various kinds of work measured.

The trend of prevailing hours

Insofar as men work for themselves and not under a contract of employment, they let their hours be fixed by the task rather than by the clock. If hours are fixed by the job to be done, they are liable to vary widely in the course of the year, especially in arable farming; but they may be kept up in the slack seasons by the taking on of other kinds of work, either at home or in places of temporary migration. The worker who fixes his own hours has tended to alternate between idleness and overtime—as did, typically, the domestic weaver. In the newly emergent economies today, workers are helped to adapt to the unfamiliar ways of industry by being required to complete a fixed task each day, rather than work for fixed hours.

Those who work for employers have long had their hours fixed. In Europe in the Middle Ages, guilds, municipalities, or statutes often specified hours, commonly from sunrise to sunset, with a limited break for food and rest at noon and perhaps one other break. The shorter working day in winter carried a lower wage. Over the year, the average was probably around 10½ hours a day, with a six-day week. But the yearly hours were reduced by saints’ days and other holidays: the number observed in practice varied from place to place and time to time, but often as many as 90 to 100 days in all were not worked. The yearly total would then be around 2,750 to 3,000 hours, an average of about 54 hours a week.

With the Reformation, the observance of saints’ days decreased. The rise in the numbers competing for jobs at that time would in any case have given employers an opportunity to reduce the number of holidays, some of which had been holidays with pay. But it was with the advent of modern industrialism, whether in Britain and the United States in the late eighteenth century or in Germany and Japan in the late nineteenth, that work hours were indeed extended. The motive was to use as fully as possible the greater capital per worker that the new methods required, and again employers were able to enforce longer hours because of the growth in the job-seeking population. The means chosen were reductions of the difference between summer and winter hours and of the number of holidays: the 12-hour day became common throughout the year, and there were few holidays besides Sunday, so that the yearly hours of work may be put at 3,500 to 3,750. But the hours the average worker actually completed were reduced by sickness and by absenteeism—especially after payday.

Since the 1830s the hours of work in the industrialized economies have been progressively reduced. Although the changes have been only roughly synchronous, even in economies at the same stage of development, and although there have been wide divergences within any one economy, some marked traits appear in common. One of the most striking is the extent of the reduction—from, say, 3,500 hours a year to 2,200 or less. The reduction has been effected by limiting normal daily hours to eight; shortening the week, first through a half holiday and more recently by an approach to the five-day week; and the instituting of paid vacations, which now extend to two, three, and four weeks. The chief immediate cause of these changes has been the pressure of organized labor: the differential effects of trade unionism may well be seen more clearly in hours than in pay. This pressure has also made itself felt in legislatures, where it has met with support, as well as resistance, from other parts of society: some notable reductions of hours have been brought about by statute. Reductions have also owed not a little to the initiative and example of particular employers.

In recent years there has been a rise in the proportion of part-time workers’to as much as one-tenth of the American labor force outside agriculture, for example; and whether part-time workers are included or not now has a marked effect on reported average hours of work. Increasing urbanization has lengthened the journey to work, so that total time claimed by work has not been reduced so much as time on the job. Recent years have also been marked in some countries by a persistent excess of hours worked over standard hours, due in great part to the persistence under full employment of levels of demand that before 1940 used to bring overtime only in the more active phases of the trade cycle.

The timing of reductions of hours. Hours have been changed much less often and less gradually than pay: they have commonly remained unchanged for long periods and then been reduced substantially in a movement that runs through many industries in the course of only a few years. The infrequency of change is understandable because small reductions in hours often call for extensive reorganization; and even where small reductions can be effected readily, they are not as likely to attract the worker as the rise in earnings that the employer would be equally ready to concede. When hours are regulated by collective bargaining, union members must adopt a common line, but their preferences are likely to be various; Moses (1962) has shown that it is then easier to reach agreement to claim higher earnings rather than shorter hours. On these grounds, we should expect reductions to come about only when workers— preferences for them had had some time to build up and the scope for them had become fairly wide.

Given such a period of gestation, reductions seem to have been induced by two very different states of the economy—by exceptional prosperity, but also by a check to expansion. In the first situation, the recent rise in earnings will have intensified any accumulated preference for shorter hours; a substantial reduction in hours can now be afforded without loss of earnings; and the bargaining power of labor is great enough to force employers to grapple with the difficulties of reorganization. In the second situation, the immediate impulse comes from the wish to “spread the work”—to have each man do less work, rather than have some men thrown out of work altogether—both as the best palliative for a recession of employment and as the fair way to realize the fruits of technical progress. Probably it is by a mounting threat of unemployment that most reductions of hours have been occasioned in the past, just as the prospects of automation have prompted proposals for reductions recently; although, however they were occasioned, reductions must also be seen as a way of drawing on the rise in productivity achieved in the years preceding them. There are other reasons why, at the time of a check to expansion, a reduction in hours accompanied by such a rise in hourly rates as will maintain weekly earnings should be conceded by employers more readily than a rise in hourly rates with unchanged hours: when production has recently dropped, employers can very likely reduce hours without a further reduction in output, and they do not have to finance a bigger wage bill until production rises again.

Hours prevailing at the present time. The general level of hours and the differences between economies in the mid-1950s are illustrated in Table 1. By 1963, in the United States the average actual weekly hours, outside agriculture, were below 40 (although this included part-time jobs), and paid vacations were mainly from two to four weeks, depending on length of service. In western Europe normal weekly hours remained higher, the average in Scandinavia, for instance, being generally 45, but there was a pronounced movement, especially in larger organizations, toward the 40-hour week. Paid vacations were generally from two to two and a half weeks but were tending to lengthen, notably in France, where four weeks had been adopted widely. In the Soviet Union standard weekly hours and paid vacations were much the same as those of the United States, although hours worked per household were greater; the aim had

Table 1 — Estimated annual normal hours, about 1956*
* Excluding overtime and hours paid for but not worked.
Source: International Labour Conference 1958.
West Germany2,296–2,320
Soviet Union2,248
United Kingdom2,152–2,192
United States1,984
New Zealand1,928

been set of reaching a 35-hour week in 1968. In the emergent economies weekly hours were generally longer: the standard of 48 hours was representative, but many exceptions were recognized, and enforcement was uneven. Nonetheless, these economies had escaped the earlier connection between industrialization and inhumanly long hours. Rather, the connection that does suggest itself in a survey of these economies is between shorter hours and higher participation rates. Shorter hours —including more part-time jobs—make it easier for more members of a household to take jobs; and reciprocally, a given standard of living can be maintained with shorter hours per worker when the participation rate is higher.

Industrial and occupational differences

Differences between industries are now seldom large or persistent, save that service industries tend to have rather longer hours than manufacturing, and in agriculture hours are generally even longer. There is some tendency to offset differences in standard hours by differences in overtime. The most salient occupational difference has long been that between white-collar workers and manual workers, weekly hours being shorter and vacations longer for the former. But some of the longest hours are worked in managerial, administrative, and professional occupations, by those who fix their own hours or often work beyond regular office hours. So far as work requires mental exertion and initiative and the worker identifies himself with it, the line between work and leisure becomes hard to draw and the adverse effects of sustained long hours on productivity tend to be avoided by periods of vacation rather than by limitation of the working day and week.

Shorter hours will go with lower wage rates to the extent that there is a tendency for the market to equalize net advantages of different jobs and that more leisure is seen as a net advantage. Shorter hours will go with higher earnings to the extent that one group of workers is able to secure better terms than another and realizes the advantage in the form of both leisure and earnings. Thus, the generally higher pay and shorter hours of white-collar workers as compared with manual workers may be ascribed to the relative scarcity in past periods of workers with any secondary education, which enabled the white-collar workers to get better terms of both kinds. With an alleviation of that scarcity, as a result of the social changes of the last half century, both pay and hour differentials have been reduced.

In comparing the hours in different jobs, we must regard not only the total but also the pattern in which given hours are arranged in the day, week, and year. Where processes or output have to be continuous, shifts are worked—there may be three shifts of eight hours, for example, or four of six hours. These arrangements involve night work— which is widely prohibited for women and juveniles. Fuller utilization of manufacturing equipment is provided by the double day shift, with one shift, say, from 6:00 A.M. to 2:00 P.M., the other from 2:00 P.M. to 10:00 P.M. Automation tends to bring more shift work. Workers may remain on one shift indefinitely or (more often) change shifts according to a rota. The roster of duties is often a distinctive and onerous feature of the service industries, where staffs may have to work evenings and weekends, meet morning and evening peaks of demand by working split shifts, and start and finish at times that vary widely from one day or week to another; there may also be periods when, although not at work, they must remain on call. Long-distance transport keeps men away from home for some periods when they are not actually working. In some jobs the work load varies seasonally, and short time alternates with overtime. Although shift working proves congenial to some workers because it provides more free time in daylight or other amenities, any departure from the usual working day and week raises difficulties for the worker and his family in their domestic and social activities. In negotiations it is usual to recognize that compensation is due for such departures: it may take the form not only of a higher hourly rate of pay but also of a smaller total of weekly hours or a longer vacation.

The individual worker may have opportunities to vary his own hours by working part time, doing voluntary overtime, or taking a second job. The growth of part-time jobs in recent years has been associated especially with an increase in the number of married women who work outside the home. When unemployment was rife, working overtime was regarded as taking work away from other men, and the higher rates that had to be paid for overtime were meant to be a deterrent to employers; under full employment, this objection, although still encountered, has weakened, and opportunities for overtime are now widely regarded as an advantage in a job. The practice of taking a second job (sometimes loosely referred to as “moonlighting”), within the usual weekday hours, in the evenings, or on weekends, has increased with the coming of shorter daily hours and the five-day week.

Choice of hours

Any one worker has some choice between shorter hours and higher earnings, insofar as he can choose between jobs and vary the hours he works in a given job; and a collective choice of this kind is implicit in most negotiations about terms of employment. The considerations that make up the theory of such choice were well set out by Marshall (1890). These considerations do not enable us to predict the kind of effect that a rise in the rate of return to labor will have upon the number of hours the worker prefers to work. Such a rise increases the worker’s purchasing power, including his power to purchase leisure, but it raises the opportunity cost of the unit of leisure. It increases the quantity of commodities he can obtain by working a further hour, but at the same time it increases the quantity of commodities obtainable by working a given number of hours, and so lowers his valuation of a unit increment of commodities. If the first factor predominates, he will prefer longer hours; if the second, shorter. Which effect will in fact predominate depends, among many other things, on the strain put on him by his present hours; on his stage of life—whether he is young or old, or whether he is raising a family; on the attractions of the alternative activities open to him—including work in his home and, especially in emergent economies, on his own holding; and on whether he and his fellows are content with a customary standard of living or are consciously raising their standards and have such goals before them. Where standards are stationary, a rise in the rate of pay will result in a reduction of hours worked. This was thought to be a general feature of the emergent economies, but wider experience has shown it to be only transitory in them or to be confined to certain groups. But the longrun supply curve of labor has certainly shown itself to be backward-sloping in the developed economies, in that their workers have chosen to draw on rises in productivity for both some increment of commodity income and some shortening of hours. Long (1958) gives the following estimates of percentage changes in the hours of the standard work week associated with a 1 per cent rise in real disposable income per hour:

United States1890—1950—0.27
Great Britain1911—1951—0.39

Hours and productivity

The existence of a choice between higher earnings and shorter hours implies that output varies directly with hours. But just how it does so is hard to determine even for manual workers, Whose product can be measured. Observations in comparatively controlled conditions have indicated that a change in hours may take many months to accomplish its effects on productivity, and the majority of changes have taken place in conditions far from controlled. Most extensions of hours in the present century have coincided with the exceptional incentives and difficulties of wartime; or reductions have occurred when there has been slack to take up or technical progress to draw upon; and reductions in themselves have often enforced a change of methods. The record suggests strongly, however, that at least where the rate of output depends on the effort and attention of the worker, hours in excess of 48 a week do not add to output in the long run—the worker adjusts his rate of working, so as to spread a given fund of energy over the longer hours, and the rates of spoilage, accident, sickness, and absenteeism are all higher. Except in work requiring special exertion and endurance, however, it seems probable that reductions of hours below 48 a week will be accompanied, other things being equal, by nearly proportionate reductions in output. But this holds good only at one place and time; there is reason to believe that as time goes on, a labor force that works shorter hours will develop its capabilities as it broadens its interests and education. Long hours were seen to wear men out by the age of 40; shorter hours, even where they decrease output in a given year, may raise it over a man’s life. The hours set free, moreover, although they do not add to the measured national product, are used in practice to add something to the unmeasured amenities of households and to the goodness of life.

E. H. Phelps Brown and M. H. Browne

[See alsoEmployment and unemployment; Labor force, article onparticipation; Leisure; Time budgets; Wages, article onwage and hour legislation.]


Dankert, Clyde E. 1962 Shorter Hours in Theory and Practice. Industrial and Labor Relations Review 15: 307–322.

Dankert, Clyde E.; Mann, Floyd C.; and Northrup, Herbert R. (editors) 1965 Hours of Work. Industrial Relations Research Association, Publication No. 32. New York: Harper.

Florence, P. Sargant 1924 Economics of Fatigue and Unrest and the Efficiency of Labour in English and American Industry. London: Allen & Unwin; New York: Holt.

FourastiÉ, Jean 1965 Les 40,000 heures. Paris: Laffont. Hours of Work in the United States and Abroad. 1963 U.S. Bureau of Labor Statistics, Monthly Labor Review 86: 925–934.

International Labor Office, GenevaYear-book of Labour Statistics. → Published since 1935/1936.

International Labour Conference, Forty-Second Session 1958 Report No. 8: Hours of Work. Geneva: International Labor Office.

Long, Clarence D. 1958 The Labor Force Under Changing Income and Employment. National Bureau of Economic Research, General Series, No. 65. Princeton Univ. Press.

McCormick, Brian 1959 Hours of Work in British Industry.Industrial and Labor Relations Review 12: 423–433.

Mann, Floyd C.; and Hoffman, L. Richard 1960 Automation and the Worker: A Study of Social Change in Power Plants. New York: Holt.

Marris, Robin 1964 The Economics of Capital Utilisation: A Report on Multiple-shift Work. Cambridge Univ. Press.

Marshall, Alfred (1890) 1961 Principles of Economics. 9th ed. 2 vols. New York and London: Macmillan. → See especially Book 1, Chapter 1, paragraph 2; Book 6, Chapter 2, paragraphs 2 and 3, Chapter 12, paragraph 8, Chapter 13, paragraphs 3–6 mathematical appendixes 10 and 12.

Millis, Harry A.; and Montgomery, Royal E. 1938 Labor’s Progress and Some Basic Labor Problems.New York: McGraw-Hill.

Moses, Leon 1962 Income, Leisure and Wage Pressure. Economic Journal 72: 320–334.

Northrup, Herbert R.; and Brinberg, Herbert R. 1950 Economics of the Work Week. Studies in Business Economics, No. 24. New York: National Industrial Conference Board.

Rae, John 1894 Eight Hours for Work. London: Macmillan.

Solovyov, L. 1962 The Reduction of Employees’ Working Hours in the Soviet Union. International Labour Review 86: 31–41.

Zeisel, J. 1958 The Work Week in American Industry: 1850–1956. U.S. Bureau of Labor Statistics, Monthly Labor Review 81: 23–29.