For

ployment goal at a lower unemployment rate than does the United States, and they have in recent year been more successful in achieving the goal. What does this contrast mean in terms of our social welfare function? one thing it means that the marginal contribution to welfare of declining unemployment, after allowing for associated costs and benefits, is viewed in Europe as remaining positive for a lower range of unemployment rates than is the case in the United States. When a single figure, or a narrow range, is given for the full employment goal-say, 2 percent unemployment in Europe or 4 percent in the United States-this implies that welfare increases as the unemployment rate falls toward the target figure. It also implies that, on net balance, welfare does not increase further or possibly declines as unemployment falls below the target figure. Welfare declines in the latter case because the increased benefit from a further decline in the unemployment rate taken by itself is more than offset by the marginal disutility of "side effects"-accelerated price and wage increases, effects on the balance of payments, retarded gains in labor productivity, and so on. The reader so inclined can express this more rigorously in terms of marginal rates of substitution and partial derivatives.

Similar considerations hold as to contrasting attitudes toward increases in the unemployment rate above the target figure. For some range of unemployment rates above the target, welfare is apparently considered to decline relatively more rapidly in most European countries than in the United States, even after allowance is made for ancillary benefits from greater price stability and improvement in the balance of payments. We have commented earlier on one corollary of this contrast. European governments would almost certainly be willing (or feel compelled) to override the balance-of-payments constraint at a significantly lower level of unemployment than would be the case in the United States."

We have already referred to the contrast in employment targets between Europe and the United States. We have also noted that sights in Europe have been gradually raised during the postwar period. This has not been the case in the United States.

28

Today, in Western Europe, 2 percent unemployment is the target most frequently mentioned." When translated into American definitions this may mean an unemployment rate from something below 2 percent to perhaps 3 percent as a maximum. Virtually all countries are very loath to announce an official quantative target." But a variety of scraps of evidence permit one to infer the approximate goal which motivates policy. From such scraps of evidence, I should put the goals in some of the leading European countries today, expressed in terms of American definitions, about as follows:

It is to be emphasized that these rough approximations are expressed in terms of American definitions. This makes a significant difference in the case of France and the United Kingdom, whose official figures have to be adjusted upward to correspond to American concepts. Hence, to Frenchmen or Englishmen my estimates of the employment goal may seem high for these countries.

30

All of this can be illustrated, in artificially simple terms, by the use of isoquants and indifference curves for the case of two target variables say, the unemployment rate and the rate of change in prices. The indifference curves can be also drawn to reflect some of the constraints previously described.

Frequently in Europe one encounters the triad of numbers: 2-2-4. That is, goals of 2 percent unemployment, increase in the price level not exceeding 2 percent, and 4 percent rate of growth in GNP.

Swedish policy and experience suggests a target closer to 1 percent.

20 Thus, "while most governments have accepted the obligation to make the achieve ment of a high level of employment one of the aims of economic policy, the majority of them do not want this aim to be expressed in any rigid formula embodying an obligation to keep unemployment below a clearly defined level." "International Labor Review," vol. 74 (July 1956), p. 2. UNESCO has regularly asked member countries for the submission of quantitative standards, but few have been offered. One was offered by the United Kingdom, which "replied in 1951 that it had adopted a standard of 3 percent although it hoped to maintain unemployment at below 2 percent. It had put the standard at 3 percent, however, in order to make allowances for unfavorable repercussions on employment that might arise from adverse developments in foreign trade.' Ibid., p. 3. 20 It should be noted that the figures in table i have already been standardized to a modest extent by OECD. The adjustment factor needed to make the original official measures comparable to the American seems to be a bit less than 100 percent for France and about 50 percent for Britain. See the source cited in footnote (b) of table 1.

In

I am sure that none of these targets would be officially admitted by the governments of any of the countries listed. And it may be that in some cases I have understated the range within which the employment goal is presumed to fall. But there is a fair amount of evidence that the welfare functions implicitly conditioning policy action in European capitals suggest employment targets approximately as we have listed them. We shall defer discussion of the targets suggested for the United States until the next section.

As overall unemployment has fallen to levels that seem very low by prewar and early postwar standards, European governments have come to place increasing emphasis on the differentially high unemployment rates that remain for particular sectors of the labor force. Throughout Europe, the labor force is much more homogeneous than in the United States. There is no color problem; teenagers are readily absorbed into the ranks of the employed, despite an earlier average school-leaving age than in the United States; and there are no particularly marked differentials between average unemployment rates for blue-collar and for whitecollar workers. Nearly every country, however, does have some variant of a depressed-area problem-more serious in some countries than in others—and in some countries, particularly the United Kingdom, unemployment rates are relatively high among the unskilled.

The development of an integrated labor market policy to deal with such differential unemployment rates-and more generally to expedite the adjustment of labor supply to the changing pattern of labor demand-began earlier and has proceeded further in Europe than in the United States. The most fully developed of such programs is that of Sweden, which has probably set its employment target higher than in any other country in the West."

THE EMPLOYMENT GOAL IN THE UNITED STATES

I should like to turn now to the problem of formulating a full employment target for the United States. There has been a good deal of debate since the Employment Act was passed in 1946 as to what the quantitative target should be in order to achieve "high," "maximum," or "full" employment. Most of the numbers bandied about fall in the range of from 3 to 5 percent of the labor force.

The early reports of the Council of Economic Advisers suggested a moderate range around 4 percent. Thus the January 1947, report referred to a then current unemployment rate of about 3.9 percent as "close to the minimum unavoidable in a free economy of great mobility such as ours.' A year later, a rate of 3.6 percent was referred to as "probably the practical minimum." Again, toward the end of the first Eisenhower administration, full employment was said to have been "practically reached" when the current monthly rate was about 3.6 percent. On the whole, it is fair to say that, between 1946 and 1962, official policy tacitly or explicitly accepted a target goal of about 4 percent, plus or minus a range of perhaps half a percent."

this connection, it is worth noting that Beveridge was certainly correct when he said that this proposed target of 3 percent was conservative rather than "unduly hopeful," "Full Employment in a Free Society," p. 128. By American definitions, his target for England was presumably an overall unemployment rate in excess of 4 percent. The following by the Director of the Labor Market Board in Sweden is worth quoting: "An active employment policy may be defined as 'measures which affect labor as a factor in production and are so varied, so individualized as-in time to fit every single person on the employment market'." The main features of such an employment policy are "geographical mobility of manpower, increased training and retraining and a policy for the location of industry that is well adapted to prevailing conditions.' B. Olsson, "Employment Policy in Sweden," "International Labor Review,'" vol. 87 (May 1963), pp. 411, 413. Economic Report of the President, January 1947, p. 9. This would have been about 4.3 percent on the basis of the definitions now in use.

33 Economic Report of the President, January 1948, p. 2. See also the report for January 1951, p. 22, where 3.6 percent was again referred to as "near a practical minimum."

34 Economic Report, January 1956, p. 28. But perhaps I should also include here Arthur Burns' statement to the Joint Economic Committee a year earlier that "although 4 percent of the labor force is nowadays widely regarded as an approximate measure of the average amount of frictional and seasonal unemployment, the Council has not favored this or any other rigid figure to serve as a trigger to governmental action or as a measure of good performance." January 1955 Economic Report of the President: Hearings, 84th Cong., 1st sess. (Washington, Government Printing Office, 1955), p. 45.

This is to say no more than that the official view assumed that, abstracting from the behavior of other associated variables, welfare would rise as the unemployment rate fell until the latter was in the neighborhood of 4 percent. But, when the other target variables were taken into account, the value judgments of successive administrations, as well as their views as to how the target variables were related to each other, might have varied so that "maximum welfare solutions" did not always yield the same unemployment rate as an immediate policy goal.

The debate of the last few years, which still goes on, as to whether "structural" unemployment has been worsening has had an effect on the way the full employment goal is viewed in some circles both in and out of Washington. Those who lean strongly toward the structuralist side say or imply that, in the short run, the full employment target has to be set at an unemployment rate higher than 4 percent-perhaps 5 or even 6 percent. In the long run, through various types of manpower policy centered on education, training, and retraining, and the like, the structural-frictional minimum can be reduced until the country can again have a short-run target of 4 percent or less. This view carries weight in the Halls of Congress and may have some influence on Federal Reserve thinking; but, as suggested in the following paragraphs, it has not represented the official position of the Kennedy or Johnson administrations.

The first Economic Report of the Kennedy administration contained an unusually full and explicit formulation of the employment goal." A temporary target of 4 percent was accepted. "In the existing economic circumstances," this was considered to be "a reasonable and prudent full employment target for stabilization policy." But this was to be only an interim goal. "If we move firmly to reduce the impact of structural unemployment, we will be able to move the unemployment target steadily from 4 percent to successively lower rates." This was still the position being taken in the January 1964, Economic Report.

37

These targets have been given a more explicit formulation in the recent report of Senator Clark's Subcommittee on Employment and Manpower, which recommended "that it be declared the public policy of the United States to maintain a level of unemployment no higher than 3 percent of the labor force. The year 1968 should be set as the target for attainment of this goal."" Here, as in the 1962 Economic Report, the full employment goal is viewed as being approached in two ways: (1) Through eliminating unemployment associated with a deficiency of aggregate demand, given the hard core of "frictional" and "structural" unemployment that presumably exists today; and (2) operating through an aggressive and integrated manpower policy to reduce this frictional-structural minimum from its presumed present level of about 4 percent to something like 3 percent.

Statements from the White House or a Senate committee do not, by themselves determine de facto policy or fully reflect the goals implicit in such policy. In terms of our welfare function, I should describe the present state of employment policy in the United States about as follows: The 4-percent target is still official policy. The fact that welfare is assumed to decline significantly as unemployment rises above, say, 4.5 percent is suggested by the recent tax cut. At the same time, the slowness in getting the tax cut, the reluctance of Congress to approve more vigorous fiscal measures, the official and public concern with the balance of payments, and the fact that public discontent with unemployment of 5.5 percent or more for so long has been no greater than it has, all suggest that the rate of decline in welfare associated with a rise in unemployment above the 4-percent target leads to marginal rates of substitution between unemployment and the other target variables such as to maximize the welfare function, under present conditions, at an unemployment rate of more than 4 percent.

In our earlier discussion of the balance-of-payments constraint, we pointed out that this constraint eventually has to give way to another one-the highest unemployment that is politically tolerable. In the current American context, I should make a wild guess that this limit is in the range of 6 to 7 percent for the unemployment rate. To put the matter very crudely, the "ex post" welfare function revealed by policy actions and dominant public opinion is such that welfare declines between the target of 4 percent and the upper constraint of, say, 7 percentbut the decline is moderate by European standards. In Europe, the decline from the target of, say, 2 percent to an upper limit of 3 or 4 percent is apparently relatively sharper. Equally significant is the constrast between other features of the European and American presumed welfare functions. The American function should probably include as additional variables the rate of change in the Federal debt and the rate of change in Federal nondefense expenditures. Both of these are inversely related to congressional and popular notions of national welfare

See particularly app. A.

Subcommittee on Employment and Manpower of the Senate Committee on Labor and Public Welfare, "Toward Full Employment: Proposals for a Comprehensive Employment and Manpower Policy in the United States," 88th Cong., 2d sess. (Washington, Government Printing Office, 1964), p. 40.

in a way that does not exist in Europe. "Marginal rates of substitution" between unemployment and each of these two variables are such as to yield "maximum welfare solutions" at unemployment rates that seem very high by European standards."

The reader is invited to rewrite the welfare function presented earlier and offer his own surmises regarding the way the function should be interpreted to apply to American conditions. I have suggested that two additional target variables-change in the Federal debt and increase in nondefense spendingshould be included within the parentheses. I might add one final comment on such a function. I am prepared to argue that congressional and popular habits of thought are such as to generate notions of a more intimate relation between these two additional variables, on the one hand, and notions of welfare, on the other, than is thought to exist between changes in the price level and welfare apart from balance-of-payments considerations."

DISAGGREGATING THE FULL EMPLOYMENT GOAL

In this last section, I should like to pick up the theme of the Clark subcommittee report previously cited and consider the following question. If we conceive of an employment policy as being concerned with differentials in unemployment rates as well as the overall rate, how can we formulate a set of targets to take account of this double approach to the problem? We might put the problem in the following manner. The conventional definition of full employment may be thought of as unidimensional-in terms of total unemployment of a homogeneous labor force. In effect, the question asked is, How low can aggregate unemployment fall before it becomes resistant to further expansion in aggregate demand?

The unemployment that remains is the hard core of frictional and structural unemployment-presumably 4 percent or so in the United States. We seek, instead, to formulate a set of employment targets for the different categories of the labor force and then to see what overall unemployment rate this implies. The labor force can be classified in a number of ways-by age, sex, color, occupation, education, industry of last employment, and so on. A full-fledged multidimensional employment goal would have targets for each segment of the labor force along each of these dimensions, and the targets would need to be mutually consistent."

Let us define "aggregative full employment" as the percentage of the total labor force employed when, given existing structural-frictional reasons for differential unemployment rates, the unemployment rate for the most favored group ceases to fall in response to an expansion of aggregate demand and the higher rates can be made to fall further only with an unacceptable rise in prices. We then ask: In what ways and at what cost can the structural-frictional differentials existing at aggregative full employment be reduced? If we know the costs involved in reducing these differentials by various amounts, application of our value judgments permits us then to say how far we are prepared to go in narrowing these differentials and thus reducing the aggregate unemployment rate that corresponds to full employment. The recent steps toward the development of a labor market policy in the United States-ARA, OMAT, etc.-suggest that we are now fumbling toward such a multidimensional approach to the problem of redefining the target of full employment.

How far can we expect to go in reducing these frictional-structural differentials? Is Senator Clark's subcommittee being realistic in urging a target of 3percent unemployment by 1968? What changes in unemployment-rate differentials are implied by this goal, and are the administration and particularly the

40 "To use deficit financing in order to drive unemployment down below 6 percent is therefore very dangerous. It will tend to do far more harm through inflation than the good it will do by absorbing some of those who are unemployed from seasonal and transitional causes." "When unemployment is between 6 and 8 percent. the governmental budget should at least balance and therefore be neutral in its effects." These statements are by a prominent Senator (and past president of the American Economic Association) who is not generally considered to stand very far to the right in the political spectrum. They are taken from Paul H. Douglas. "Economy in the National Government" (Chicago, University of Chicago Press, 1952), pp. 253-254.

41 Within the limits imposed by the upper price constraint previously described. 43 For a similar statement of a disaggregated approach to the formulation of a fullemployment goal, see Seymour L. Wolfbein, "The First Year of the Manpower Act," in A. M. Ross, ed., op. cit., p. 56. The same approach has, of course, been urged by others, also.

37-181-64

Congress likely to value the costs and benefits such as to be moved to take the steps needed to reach the goal by 1968 or later?"

Tables 2 and 3 are presented as a small exploratory step toward answering these questions. Table 2 deals with occupational differentials; Table 3 looks at unemployment in terms of age, sex, and color. Both tables are constructed in the same way. Let us look first at table 2.

The table contains four sets of three columns each. Three of the sets present actual data for 1953, 1957, and 1963. The first of these years was probably a year of overfull employment, with an overall unemployment rate of 2.9 percent by current definitions (2.5 percent by the definitions then being used); 1957 was the last year in which unemployment was close to 4 percent-1963 is included to reflect the recent and current picture. The fourth set of columns investigates the effect of one pattern of reduced differentials on the overall unemployment rate.

By weighting the unemployment rate in each occupation (col. (a) for each year) by the fraction of the labor force in the occupation (col. (b)), we obtain the contribution of each occupation to the total unemployment rate (col. (3) for each year).

In laying out our multidimensional full-employment target, we have to ask two questions. First, what does past experience suggest as to the lowest unemployment rate that is likely to prevail for the most favored groups under full-employment conditions. Second, how much (and by what means) can we hope to reduce the unemployment rates for the less favored occupations that have existed in past periods of full employment. The fourth set of columns in table 2 suggests one possible set of answers to these questions.

Let us look at the first three lines of column (a). These are the occupational groups with the lowest unemployment rates. The highest unemployment rate for any of these groups in 1953 and 1957 was 1.2 percent. For my full-employment target, I have taken the unemployment rates as they were in 1957 rather than the somewhat lower rates in 1953. For the remaining occupations I have simply made a guess as to how low we might hope to bring down each unemployment rate under a combination of an adequate level of aggregate demand and an expanded and intensive manpower program.

On the whole, my guesses are perhaps on the conservative side. This is suggested by the fact that my target rates are in most cases higher than the rates which actually prevailed in 1953, though lower than those in 1957.“ My most radical assumption is in reducing the unemployment rate for laborers from 6.1 percent in 1953 and 9.4 percent in 1957 to 5 percent.

Some perspective on the problem of structural differentials can be gained from looking at the three columns for 1957. Note that all but three groups had unemployment rates below 4 percent. The remaining three occupations, with unemployment rates between 5.1 and 9.4 percent, accounted for a bit more than a third of the labor force and for more than half of total unemployment. It is on these three groups, plus those without previous work experience, that our manpower program obviously needs to be concentrated.

A word about those unemployed without previous work experience, for whom the unemployment rate is always 100 percent, no matter how few or many there are in this group. I have simply assumed that this group would constitute about the same percentage of total unemployment as it did in 1957, despite the current upward trend in the share of the labor force made up of young people."

If we now weight our assumed unemployment rates by the share of each occupational group in the total force in 1963, we get each group's contribution to the total unemployment rate, and the latter turns out to be 3.1 percent-or very close to the goal set by the Clark subcommittee. This, then, is one possible occupational pattern by which a 3-percent overall rate could be substantially 43 When employment goals are suggested, they are almost always expressed in terms of the number of persons seeking employment, not in terms of total man-hours lost through unemployment. In 1963, the percentage of total labor time lost through unemployment was reported as 6.4, whereas in terms of persons the unemployment rate was 5.7 percent. A full-fledged employment goal clearly should take account of involuntary part-time unemployment, as was strongly urged by Thomas K. Hitch over a decade ago in "Meaning and Measurment of 'Full' or Maximum' Employment," Review of Economics and Statistics, vol. 33 (February 1951), pp. 1-11. A 3-percent goal in Conventional terms might repre sent, say a 3.5-percent goal in terms of man-hours. This is a question which will be investigated further in the larger study on which the present paper represents a progress report.

In general, for the middle range of occupations, I have taken unemployment rates which are about halfway between the 1953 and 1957 rates, after the former have been adjusted upward by about 10 percent to allow for the change in definitions that occurred in 1957.

45 The percentage of total unemployment with no previous work experience was abnormally low in 1953. See my paper, "Has Structural Unemployment Worsened?" in Industrial Relations, vol. 3 (May 1964).