continued to be greater than the reductions thereby continuing the increases in the labor supply of these secondary workers." He also concludes that the number of married women who enter the labor force during recessions and are unemployed is greater than the number who lose their jobs and withdraw from the labor force.

At the same meeting, Stanley Lebergott (of Wesleyan University) presented the results of work he had done for the Committee on Economic Stability of the Social Science Research Council to develop a quarterly model of the U.S. economy for use in short-term projections. In this connection, he fed vast amounts of quarterly data into computers, and came up with 12 multiple regression equations relating to the labor force participation rates of women aged 20 and over. He found that the equation that gave the most satisfactory account of short-term variations in the participation rate of adult women was one that combined measures reflecting job opportunities for women on the one hand, and male unemployment, on the other. Changes in very long-term unemployment, as measured by the percentage of the male labor force unemployed 27 weeks or more, were much more significant than changes in the overall unemployment rate, suggesting that women are not driven to enter the labor force until their husbands have been out of work for a long time. Changes in Armed Forces personnel (as measured by Department of Defense personnel procurement for service as a percentage of the male population 14 years and over) also were positively associated with changing labor force rates for women-they tended to rise and fall together. This association may be explained by the propensity for young women to go to work as their young men or husbands enter military service.

The most unexpected ingredient of this equation is the number of females employed as unpaid family workers in agriculture. Lebergott believes that short-term variations in the size of this small group (currently it is running around 550,000 and subject to very erratic fluctuations) reflect variations in off-farm job opportunities. Unpaid family workers are not counted in the labor force unless they work 15 or more hours a week. As a farm operator has more difficulty competing with non farm employers for scarce workers, his wife will work longer hours, and so become classified as in the labor force, as she passes the 15-hour mark. (It is difficult not to suspect that shortterm variations in this number are reflecting weather and soil conditions rather than off-farm migration by other family members to good jobs in the towns.)

The variable that in Lebergott's final equation has the strongest correlation with changes in the labor force participation of adult women is the volume of consumer credit extensions (deflated by the GNP consumer durable implicit price index) in the prior quarter. This apparently means that the more families go into debt the more likely the wife of the head will be in the labor force, either because with her added income the family has purchased more consumer goods or because, having gone into debt, the family needed an additional income to carry it.

The last factor in the regression equation is the average number of paid hours per week in manufacturing, an indicator of cyclical change. This varies negatively with the participation rates of adult women, indicating according to Lebergott, that in recession phases of the cycle, women are more likely to find work than men, in part because they typically accept lower wages. At peak periods, employers institute a longer week, with overtime, for their male workers. It is more advantageous to do this than to break in new women workers, if indeed they would perform the same jobs.

In evaluating this ingenious structure, one must remember that the author is not claiming that any of the factors in his equation is in itself a cause of variation in labor force activity, but that they represent independent forces that in combination are associated with shortterm changes in labor force propensities.

Others are working along these lines, using the techniques that are now so readily available by means of computers. There is reason to hope that future experimentation will lead to a real breakthrough in knowledge of what is responsible for long- and short-term changes, and how to measure and predict these causes and their effects.

Perhaps all that can be said now with certainty is that at times of extremely high demand as in World War II and in the 1955-56 boom, all segments of the population seem to react in the same direction, and labor force participation increases all along the line. Whether a future boom would have the same impact on the rates for middle-aged women, for example, when so many more are in the labor force, needs to be demonstrated.

[American Economic Association, papers and proceedings of the 65th Annual Meeting, May 1953]

IMPACT OF EFFECTIVE DEMAND ON THE LABOR

SUPPLY

(By Clarence D. Long,1 Johns Hopkins University)

INTRODUCTION

No keen insight is required to characterize the thirties as a decade of preoccupation with demand and no sharp foresight perhaps to suggest that the fifties will be one of concern over supply. There lies ahead an era of planning, of input-output tabulation, of national income accounting and projection-calculations for which the size, the quality, the hours, the mobility, and the effort of the labor force are paramount. This would be so even in peace. But the chances are that war-in some degree of temperature will be with us for the next generation and that, except for occasional lulls, the atmosphere and the competitive obsolescence of armament it fosters will mean high

I am deeply grateful to my assistants, Margaret Chen and Susan Dischka, to my colleague. Prof. Acheson J. Duncan, and to Marc Nerlove and James Terrell, of the department of political economy, for their help in developing the statistical computations underlying this paper.

employment and money income, farflung military forces, and great pressure on the labor supply.

This not entirely agreeable prospect gives new point to an old question: Will high effective demand cause more or less labor to be forthcoming and by how much?

The various attempts to deal with this problem have yielded several hypotheses. One tradition, well established in economic learning, is that homo laborans is a sloth who works mainly from hunger. Give him a wage above the bare necessities of life and, on the principle of diminishing marginal utility, he will be sure to squander some of it on surcease from honest toil. Marshall, it is true, held a candle into this gloom with the dynamic view that—

an increase of wages *** almost always increases the strength, physical, mental, and even moral of the coming generation; and, other things being equal, an increase in the earnings *** by labor increases its rate of growth; or, in other words, a rise in its demand price increases the supply of it ****

But the belief that the labor supply was necessarily backward sloping tended to prevail until Lionel Robbins showed it was based on failure to take account of the elasticity of demand for income in terms of effort. (See "On the Elasticity of Demand for Income in Terms of Effort," Economica, 1930, pp. 123-129.)

Until the thirties the effects of wages or incomes on labor supply were considered without regard to job opportunities. The depression drew some economists, notably Woytinsky, to the additional worker theory: that unemployment of husbands or fathers would force many dependents into the market as distress job seekers. Not everybody agreed, however, as to what conditions brought in additional workers. Another school held with J. H. G. Pierson that

if society were committed to providing job opportunity for all those able and wanting to work, however numerous, certain fresh supplies of labor not apparent at present would shortly be uncovered * *("Full Employment," pp. 18-19, note 22).

*

One writer has even found it plausible to expect a rise in the labor force in both depression and prosperity.

In the meanwhile, Keynes had entered the depression discussion with his now well-known postulate that below full employment

a situation where labor stipulates (within limits) for a money wage rather than a real wage, so far from being a mere possibility is the normal case ("General Theory," p. 9) —

thus suggesting that labor reacts very differently as between full and less-than-full employment conditions and that under the latter circum

2 "Principles of Economics" (fourth edition, p. 603). Frank Knight, in "Risk, Uncertainty and Profit" (p. 117), explained that one of the commodities labor purchases with increased earnings is leisure; and W. Stanley Jevons, in "Theory of Political Economy” (fourth edition, pp. 180-181), that "English laborers enjoying little more than the necessaries of life, will work harder the less they produce; or will work less hard as the

produce increases. The richer a man becomes, the less does he devote himself to business." These remarks refer, strictly speaking, to hours or effort, but they can be extended to labor force.

8 W. S. Woytinsky, "Additional Workers and the Volume of Unemployment in the Depression" (Social Science Research Council, pamphlet series 1, 1940), pp. 1, 17, 26; D. D. Humphrey, "Alleged 'Additional Works' in the Measurement of Unemployment." Journal of Political Economy, June 1940, pp. 412-419; Woytinsky, "A Reply to Mr. Humphrey." ibid., October 1940, pp. 735-740; C. D. Long, "The Concept of Unemployment," Quarterly Journal of Economics, November 1942, pp. 9-10; "The Labor Force and Economic Change," in "Insights into Labor Issues," edited by R. A. Lester and J. Shister.

stance it varies directly with money wages rather than inversely with real income.

These problems cannot, of course, be dealt with adequately in a short paper. The present one summarizes some rather detailed investigations into the size of labor force under short-term changes in wages, incomes, unemployment, and military strength. It relies mainly on the experience of the United States, Canada, and Great Britain (since these three nations began to make available annual, quarterly, and monthly statistics); but it benefits also from researches with decennial census data of a half century or more in five countries, including New Zealand and Germany. (See "Labor Force, Income, and Employment," National Bureau of Economic Research, 1950, mimeographed, now under extensive revision; also "The Labor Force and Economic Change," op. cit., pp. 329–355.)

THE LABOR FORCE

The concept of labor force varies in some degree among the countries, without, however, invalidating analysis of short-run change. (The revised manuscript mentioned above contains a detailed critique of variations in the labor force concept, both over time and between five countries.) In general, it includes the unemployed-those seeking jobs and able and willing to work (presumably for going wages)-and the employed-wage and salary earners, employers and own-account workers, military and civilian personnel of government. It excludes students and housewives (as such) on the ground that though they undoubtedly toil and spin, they get no pay and therefore have no economic standing.

My past inquiries have led to the following conclusions:

1. At any time, the labor force seems to vary inversely with real wages and salary earnings of adult male workers, even when account is taken of differences in color, nativity, marriage, child responsibilities, education, employment opportunity, size of city, density of rural population, or length of workweek.

2. Over long periods, between high-employment census dates, the labor force has manifested no such association. It has, in fact, held to a proportion of the overall population (standardized for composition) that has been remarkably stable, especially in view of the great changes in real or money hourly earnings, in disposable incomes, and in the internal composition of the labor force itself.

3. There was no net influx of desperation work seekers in the great depression and, therefore, no flight from gainful labor in the subsequent recovery. Actual decreases of 2 percent of population obtained for Great Britain in depressed 1921 and 1931, for Germany in

A criterion of stability may be had in the fact that the longrun changes in the proportion of population in the U.S. labor force have been smaller than the variation within any year between winter to summer, and at the same dates; e.g., April 1950, between the labor force reported by the Census Bureau's regular decennial enumeration and that estiated by its monthly survey based on a sample of 25,000 households. See my "Statistical Standards and the Census: Discussion." American Statistician, February 1952. See also la that issue the remarks of Morris Hansen, Assistant Director of the Census, pp. 10-14.

1933, and for the United States and its individual States and cities at various dates during 1934-40. With the peacetime prosperity, the labor force went back up to the proportions prevailing in previous years of high employment.

4. These findings refer to peacetime. My "Labor Force in War and Transition: Four Countries," has concluded that, except in Germany where no additions of natives were realized, the very large World War II labor force inflows were dominated by the military draft and, aside from normal growth, turned out to be temporary.s We now trace the behavior of the labor force in the setting of approximately full employment that has prevailed in most of recent years.

By the end of 1946, the labor forces of the United States, Great Britain, and Canada had lost their wartime additions and were close to normal; that is, to their proportions of working-age population in previous episodes of peacetime abundance of jobs.

During the subsequent 4 years, until the Korean outbreak, the overall proportion remained highly stable in each of the three nations, its maximum range of variation having been 1.5 percent of population 14 and older-half to a third of what normally occurs in this country and Canada from one season to another within the same year.

In Canada each sex also held to stable participation rates. But in the United States and Britain this overall stability was the result of offsetting movements in internal composition as males continued their long-term tendencies to drift out of the labor market with earlier retirement of men above 45, or prolonged schooling of boys and young men, and women above 25, chiefly wives, held to their longrun propensities to swap unappreciated housework for remunerated office work. However, these changes corresponded in no consistent way to the variations in Armed Forces, unemployment, hourly earnings, or disposable incomes-money or real.

5 Occasional Paper 36 (National Bureau of Economic Research, 1952), p. 61. For World War I and the early stages of World War II, see "The Labor Force in Wartime America," Occasional Paper 14 (NBER, 1944). This close tie between U.S. labor force and the Armed Forces movements is supported among the quarterly data during 1940-46 by correlations which were V12 +0.96 for simple correlation and 12.34+0.86 for partial correlation (holding income and unemployment constant). According to the partial regressions, a rise or fall in the Armed Forces of 100 persons unaccompanied by any change of income or unemployment would be associated with rise or fall in labor force of 59 persons. This is reasonably close to the overage of 70 for the war gotten from annual data and without holding the other factors constant.

Personal disposable incomes include wages and salaries, interest and dividends, rents (including those implicit on owner-occupied dwellings), and pay and allowances to the military; they exclude business savings and payments to government, both personal income tax and nontax. They are data of the Department of Commerce (Survey of Current Business, February 1952, National Income Supplment, 1951, p. 209) adjusted by that agency for living-cost changes and by me for seasonal variation. The adult male equivalent employed were converted from employment data by broad age-sex groups on the basis of the ratio of factory earnings of young people and women to those of adult males. The purpose of this latter adjustments is to prevent changes in the age-sex composition of labor force from producing a spurious association with income.

The hourly earnings data used here cover the bulk of employed workers, including those in manufacturing, mining, construction, trade, amusement, agriculture, and government. They differ from incomes in excluding dividends, interest, and other nonwage incomes and in not deducting income tax and social security contributions. Since they are payments only for time worked, they are impervious (statistically) to variations in full- and part-time idleness.

In Canada, earnings covered fewer industries, information being not forthcoming for government or agriculture, but their course was very close to that in the United States. In neither country, incidentally, did hourly earnings behave differently when weighted for given year, for years other than 1940, classified according to industrial composition of employment.