Committee on Economic Security (CES)

Social Security In America

Part I


Chapter IV


FOR an established type of insurance it is customary to use the most comparable results of past experience within the group insured and to weigh their applicability to the problem of cost estimates for the future. Well-designed, accurate record keeping is basic to this procedure. The ability to recognize changes or trends, to evaluate the credence which can be given to past experience while applying it to the future is supplemental to the simpler process of accumulation of experience.

In the analysis of a new type of insurance, since no exact data are available, accuracy of past records is less important than judgment in the selection of the data to be analyzed. In the experimental advance procedure, any accurate determination of scope of coverage and rate and duration of benefits is difficult but fundamental and is reached through the method of successive trials to determine what benefits can be given as a result of various rates of advance contribution. Especially must the immeasurable factors be recognized as such and allowed for by a careful contingency margin. The need of such a margin has been most thoroughly demonstrated in the administration of workmen's accident compensation where the failure to provide sufficiently for the unknown has created more trouble than any other factor. It has been pointed out by numerous serious students of the problem that in unemployment compensation there is less random chance and more possibility that individual businesses in time can create the contingency insured against than is common in life insurance, but sufficient margin must be left so that faulty cost calculations will not impair the protection given to the insured.

Even in that simplest form of insurance--life insurance--there are marked differences by regions, by industrial categories, and by time periods. In order to obtain uniformity of protection, it is, therefore, necessary to secure as broad a view as possible so as to determine what may be regarded as long-term over-all probabilities of unemployment, and the value of unemployment statistics will be seriously


limited if too small an area, too short a period of time, or too little industrial differentiation enters into the statistical compilations.

While unemployment insurance has been carried on for more than 20 years in Great Britain and for several years in other European countries, no foreign country has maintained a uniform plan for any long term of years without drastic modifications either in coverage or benefits; therefore, little satisfactory actuarial experience is available from other countries.

In the United States neither exact unemployment rates nor exact data as to what proportion of the apparent unemployment would have been compensated are available. The measure of protection afforded by an unemployment compensation law will depend to a large extent on the behavior of the employment curve, and experience has shown that no unemployment insurance system in existence has ever made a sufficiently adequate estimate of benefit requirements. The government actuary in Great Britain, in 1919, using trade-union records of unemployment and material collected by the Ministry of Labour, estimated the average rate of unemployment before the war at a little over 4.5 percent for the trades to be covered by the Unemployment Insurance Act of 1920. In the following decade, however, Britain's unemployment average was about 12 percent and even higher from 1930 on. The British law of 1920 provided certain definite benefits which were based on estimates of unemployment in pre-war years. The 1921 depression which immediately confronted the newly established scheme was more extreme than any other similar crisis in British industrial history. With the upturn in business conditions Great Britain was faced with chronic unemployment in certain heavy industries, of a type that the unemployment insurance scheme had not been designed to cover. The recognition that this was a permanent condition was stubbornly resisted with the result that the fund was usually in financial difficulties until 1934, when contributions and benefits were finally brought into balance. The German system, begun in 1927, was soon affected by a depression, which necessitated doubling the contribution rate and revising benefits downward in an effort to keep the scheme intact. The Swiss Federal scheme established in 1924 did not make proper allowance for depressions, and as a result a greatly increased amount of Federal aid was inevitable. A growing volume of benefit claims has also been the experience of every voluntary plan in the United States.

The inherent limitations which accompany the application of data accumulated in the past to future conditions are greatly accentuated when the past data are inadequate. It is necessary to estimate what has occurred in the past, and to be even more cautious in the application of this material to the future than when


the past is a matter of accurate record. It has been suggested that a single 10-year period is only a part of a much longer employment cycle. Since unemployment is the result of intricate forces--individual, industrial, and social-and since a complete understanding of these forces is impossible at present, not even the hardiest would venture to claim that the course of future unemployment can be foretold. The insurance technique, however, allows for this inadequacy by drawing from past experience as well as possible and by adjusting for the contingency factors involved. The smaller the unit of observation, the larger the essential loading to cover contingencies, and, in addition, the greater its fortuitous variation in one portion of the exposure from the aggregate experience on the whole.


The British system was originally intended to take care of normal unemployment in certain selected trades. The eligibility requirements were framed so as to keep from benefit those not genuinely unemployed through lack of work. The act worked well. When, however, coverage was extended to the vast majority of the British working population almost simultaneously with the beginning of the serious post-war depression, before adequate reserves had been accumulated, difficulties immediately set in. From then on, the many changes in the British law and, in particular the introduction of uncovenanted benefits, grew out of the basic, fact that the British Government did not wish to require insured persons who had either exhausted their right to, or were not qualified for, insurance benefits to seek aid from poor-relief authorities.

The British actuary, Sir Alfred Watson, has pointed out that, while he had established successive actuarial bases for the system in Great Britain, they were soon broken down by action of Parliament in changing rates without relation to experience. In periods of serious unemployment, legislators are under social pressure to liberalize the compensation terms, and they frequently do so at a time when it is difficult to increase income proportionately. The balance between contributions and benefits is therefore destroyed, the system is soon forced into insolvency, and the legislators are compelled to provide other measures of relief outside insurance which should have been established as supplementary to insurance at the outset.

Recognizing the importance of these considerations, the new law of Great Britain, passed in 1934, provided for an unemployment insurance committee of experts, of which Sir William Beveridge has been appointed chairman. The committee is to report to the


Minister of Labour early in each year on the financial condition of the fund at the end of the preceding year, and, in addition, to report whenever it is felt that the fund is likely to become inadequate to meet its liabilities. In any such case the committee is to suggest amendments to the law and estimate their effect on the fund. The Minister is required to report to Parliament on such recommendations within a specified time. The Minister may modify the proposed amendments but not to the extent of changing the anticipated effect on the fund; he must also give Parliament his reasons for altering the committee's suggestions.

Foreign experience has shown that the authorities administering unemployment insurance, especially in the early years of the systems, are confronted with many unexpected problems and questions affecting the solvency of the fund and the just treatment of the insured. It is quite impossible to anticipate all situations and to provide for them in the initial legislation. Obviously as much discretion as possible must be vested in the administrative agency so that prompt adjustment can be made to changing conditions, thus keeping the fund solvent. Yet without some limitations, flexibility may prove a definite liability. It is probable that such important changes as altering the amount or duration of benefits should be left to legislative action, even if a special session of the legislature becomes necessary in emergency situations.


In building a scheme of unemployment compensation on an actuarial basis, estimates may take two forms: (1) The rate and duration of benefits may be set, and contributions sufficient to meet the costs of such standards may be levied, or (2) contributions may be set, and benefit rates and duration may be estimated within these financial and other limitations. The first type of estimate is that commonly used in insurance schemes of all kinds; the second is based on the principle that industry can assume only a certain additional cost without suffering undue hardship, resulting, perhaps, in contraction of employment, and that consequently employers' contributions should be limited. The second type of estimate has been made at contribution rates set at 3, 4, and 5 percent of pay rolls.

In order to judge the validity of the actuarial estimates of unemployment compensation it is necessary to analyze the interreliationship of the constituent elements. The choice of' which elements are used depends upon the available data. The British and Canadian actuaries could estimate the number of claims over a period of time, the average duration of claim, and the average total benefits, and from them factors could obtain an estimate of the total


cost and rates of contribution. In this study approximate figures on the number of unemployed at any one time, the proportion of those unemployed eligible for benefits, an assumed rate of benefits for the unit of time selected, and an anticipated income based upon a specified pay-roll tax were available.

The estimates that follow are based on the United States as a whole, because of the more comprehensive statistics available on this basis. Adjustments will be necessary for individual States to meet their particular conditions and unemployment experience.

Since a tentative estimate of the probable duration of the maximum benefit period is based upon a hypothetical experience in the past, such estimates have been prepared covering the operation of several types of plans for the United States as a whole; for the period 1922 through 1933, the summary of which will be presented herein.

The period 1922 through 1933 was chosen because it included what might be called a complete business cycle, covering the initiation of recovery after the depression of 1921, the subsequent years of normal conditions, the crash of 1929, and the 4 years of major depression which may be said to have swung into recovery in 1933. The place of these years in a longer cycle cannot as yet be appreciated. Nevertheless, during the combined period a wide range of unemployment rates has been experienced and in the aggregate there is reason to believe that the period is a fair background from which to predict the future.

The dearth of data concerning all phases of employment and unemployment as well as of income necessitated the use of much indirect methodology in deriving the estimates presented. A word of caution is therefore injected to warn against too literal an application of the figures which appear, although it is felt that they are the best indications obtainable of what would have happened had an unemployment compensation system been in force in the United States in the past. It should be definitely noted that the statistics presented are representative of the United States as a whole and cannot be accepted without further research as typical of any State.


For the purposes of calculating the number of persons who would be covered by a uniform Nation-wide unemployment compensation system in the United States on the basis of the April 1930 census, it was assumed that all persons employed in establishments with 8 or more employees during at least 20 different calendar weeks of the year would comprise the compensable labor force, except that the following occupations were excluded from coverage: agricultural labor; domestic service in a private home; service performed as an


officer or member of the crew of a vessel on the navigable waters of the United States; service performed by an individual in the employ of his son, daughter, or spouse, and service performed by a child under 21 in the employ of his father or mother; service performed in the employ of the United States Government or of an instrumentality of the United States; service performed in the employ of a State, a political subdivision thereof, or an instrumentality of one or more States or political subdivisions; service performed in the employ of a nonprofit corporation, community chest, fund, or foundation, organized and operated exclusively for religious, charitable, scientific, literary, or educational purposes or for the prevention of cruelty to children or animals. These exclusions are the same as the exclusions from the Federal tax on pay rolls imposed by title IX of the Federal Social Security Act.

TABLE 14. Estimated compenaable labor force in the United States, April 1930










All gainful workers







Number of workers excluded from plan by occupation {1}

20,133 669






Size of firm {2}







Total number of workers excluded from plan







Compensable labor force







{1} Occupational exclusions eliminate all employments excluded by title IX of the Social Security Act.

{2} Size-of-firm exclusion eliminates workers employed by firms with 7 or less employees not eliminated by occupation.

The Fifteenth Census of the United States, taken in 1930, reported approximately 48,800,000 gainful workers, of whom about 54 percent would be excluded by the above provisions--some 20,134,000 because of occupation, and approximately 6,416,000 because of the "size-of-firm" exclusion--leaving about 22,280,000 as the total compensable labor force. (See table 14.) These eliminations result in the total exclusion of workers in public service, agriculture, construction and maintenance of roads, independent hand trades, and preserving and canning, and leave in the covered group only about 338,000 in domestic and personal service, of whom none would be in private homes. About 90 percent of all persons engaged in professional service are excluded. (See also table I-12 in appendix I)

The division of the gainful workers who would have been covered had an unemployment compensation system been in operation in April 1930 between those employed and those unemployed is also shown in table 14. Employment and unemployment by industries were estimated from figures in the unemployment census of that year


and from other data. Deductions of the unemployed by socio-economic classes--managerial, professional, clerical, skilled, and unskilled--yielded an estimated distribution of the employed in each industry and made possible the approximation of the number of employed and unemployed eliminated from the plan by virtue of the exclusions stated above. The fact that on this basis 72 percent of the unemployed and only about 43 percent of the employed are included in the compensable group indicates that the coverage assumed would apply to the group obviously most in need of it.


The extension of this coverage inquiry on the basis of yearly employment estimates applied to the compensable employed labor force of April 1930 made possible the computation of the insured employed labor force for each year from 1922 to 1933. {1} Estimates of the change in the number of gainful workers from year to year applied to the total coverage of April 1930 resulted in the derivation of the total compensable labor force for the period. Table 15 displays the changes in the number included in the compensable labor force for

1 For the period 1922-27 employment estimates were derived from the unemployment estimates of Leo Wolman in Committee of the President's Conference on Unemployment, Recent Economic Changes (McGraw-Hill Book Co., New York, 1929), vol. II, p, 478. For the years 1928-33 Robert R. Nathan's estimates were used. These estimates by Mr. Nathan have appeared in slightly revised form in "Estimates of Unemployment in the United States, 1929-36", International Labour Re-utew, vol. XXXIII, no. 1, January 1936, p. 49.


each of these years and the estimated employment and unemployment in the compensable labor force. This table, as well as the general experience of other countries, indicates the wide variation in unemployment rates year by year. Unemployment in the covered group appears in greater proportion than in the general population; in this discussion consideration is given to the unemployment rates believed to occur in this group.

It should be noted in this connection that the estimated unemployment rate of the covered group averages about 15 percent for the period 1922-33, from a low of 5.41 percent in 1929 to a high of 41.56 percent in 1933. Since contributions increase with employment and expenditures increase with unemployment, a ratio of contributions to benefits must be maintained over periods of prosperity and depression, and the reserves in good years must be very sizable in order to carry the fund through bad ones. Viewing the unemployment as a cumulative total throughout the whole period, a little less than two-thirds of the entire unemployment in the compensable labor force is found in the 4 years 1930-33 of the depression, averaging a little over 30 percent of unemployment per year, whereas the 8 years prior to the depression comprise only approximately one-third of the total unemployment, averaging less than 9 percent unemployment per year. This indicates a rate of unemployment almost four times as great in a stage of depressed business activity as in a period of normal business conditions in the insurable group.


In this inquiry estimates are made of contribution levies of 3, 4, or 5 percent on the total weekly wage or salary of the compensable labor force.

The estimates of total contributions were based on national income figures found for the period 1922 through 1928 in W. I. King's study, The National Income and Its Purchasing Power, and for the remaining years 1929-32 in National Income, 1929-32, Senate Document 124. The income for 1933 was estimated by the staff of the Committee on Economic Security.

Table 16 shows that the application of a 1-percent levy in 1922, a 2-percent levy in 1923, and a 3-percent levy thereafter would have resulted in a total income to the fund of approximately $8,746,000,000 for the period 1922 through 1933, ranging from a high of about $991,000,000 in 1929 to a low of about $517,000,000 in 1933. During the 6 years, 1924 through 1929, contributions would have averaged approximately $890,000,000 per year. Contributions would


have greatly diminished, however; with the depression, during 4 years of which an average of only about $664,500,000 per year would have been collected. The variation in annual income is the result not only of the variation in the number of employed contributing but also of the fluctuations in wage rates.


The amount that can be paid in unemployment compensation benefits, according to the method of approach employed, is limited by the income available. This income was assumed to be that calculated in the preceding section, based on 3 percent of pay roll. The entire


income was assumed to be available for benefits, since the Social Security Act requires that all income from contributions be used for the payment of compensation. Administrative expenses are to be defrayed through grants by the Social Security Board to approved State plans. The rate of benefits was assumed to be 50 percent of the loss of average earnings not to exceed a $15 weekly maximum. On this basis, a computation was made of the total benefits that would be paid if no other limitations on benefits payable were set. This was termed the "compensable wage loss." The total compensable wage loss was then computed by multiplying the total man-years of unemployment in each year from 1924 to 1933 by the average compensa-


ble wage loss per man-year.{2} (See table 17.) With these fixed assumptions of the amount of total income and the rate of benefits it remained to calculate the length of the period during which benefits might be paid. Unlimited payment of benefits is, of course, impossible with the limited income available, as well as undesirable from the standpoint of social policy. The time during which benefits are payable must be definitely limited, in order to keep income and outgo in balance over a period of years.

The first limitation in the time for which benefits are payable is made by the waiting period. The limitations secured through 2-, 3-, and 4-week waiting periods were considered.

For the purpose of determining the proportion of the unemployed who would have been ineligible to compensation by virtue of the


waiting period as well as of determining what benefit period could be allowed, a study of the duration of unemployment in the United States was conducted, estimating the distribution of the unemployed according to their duration of unemployment for each year, 1922 through 1933. This study utilized some 92 censuses or surveys in 46 cities and 10 different and non-consecutive calendar years, involving over 5,000,000 personal records. The 92 surveys were segregated into 5 groups according to the percentage of gainful workers unemployed at the time the survey was made. Surveys for which the percentage of unemployment ranged from 3.0 to 6.9 were included

{2} No computation was made for 1922 and 1923, since it was assumed that no benefits would be paid during the first 2 years of contributions. This is in line with the Social Security Act, which requires delay of 2 years before benefits commence, in order to increase the reserve available for the payment of benefits.


in the first group. For the second, third, fourth, and fifth groups, the range was 7.0-10.9, 11.0-19.9, 20.0-29.9, and 30.0-42.9, respectively. Ten censuses fell into the first group; 24 fell into the second; 36 in the third; 16 in the fourth; and 6 in the fifth. The percentage and cumulative percentage from these groupings are shown in table 18, together with the composite distribution for the group.

The five cumulative percent distributions shown in table 18 are presumably descriptive of the length of time that the unemployed remained idle at five different ranges of unemployment. Thus, to the years when the average unemployment for the United States was 5 percent of the total gainful workers, the distribution with limits of 3 to 7 percent would be applied. To the years when the intensity of unemployment was between 7 and 11 percent, the distribution with those limits would be used, etc. The compensable wage loss for each year from 1924 through 1933 was then distributed according to the distribution of the duration of unemployment applicable to that year. The resulting distributions were cumulated for the period 1923-33 (table 19).

This made possible a short-cut method of selecting the maximum duration of benefits possible, by adding the compensable wage loss in the waiting period to the income available for benefit and reading down the column showing the distribution of wage loss until the figures approximately coincide. {3}

This method is now applied to the estimates so far obtained which are not corrected for inadequacies. (Actuarial adjustments will follow later in the discussion.) Assuming a 3-percent contribution rate, a benefit rate of 50 percent of average earnings not to exceed a $15 weekly maximum, and a 2-week waiting period, it appears that benefits could have been paid during a 13-week period, maintaining the system in solvency to the end of 1933. On the basis of a 4-percent contribution, benefits could have been paid for 23 weeks.

At this point adjustments may be brought into the discussion to evaluate those factors in the assumed plan for which no reliable supporting data can be found and for those which are contingent upon the operation of the plan. Mr. W. R. Williamson estimated the extent to which the following assumed provisions and other factors would affect the volume of wage loss compensable

{1} Savings through the requirement that benefits will be paid only to employees for whom contributions have been paid for at least 40 weeks in the preceding 2 years;

{2} Savings through 3-week disqualification from benefits for employees who voluntarily quit their work or who are discharged for proven misconduct;

{3} This method was devised by W. R.. Williamson, actuary of the Committee on Economic Security.




{3} Savings through suspension of benefits during trade disputes or while accident compensation or other compulsory benefits are being received;

{4} Savings resulting from the requirement that the employee is able to work and available for work;

{5} Savings through compensation for partial unemployment in excess of $1 more than 50 percent of full-time wages;

{6} Savings through limitation of benefits in the ratio of 1 week of benefits to 4 weeks of contributions;

{7} Allowance of an additional maximum week of benefits for each 20 weeks of contributions without drawing benefits, up to a maximum of 10 additional weeks of benefits;

{8} Increase in costs through commutation of benefits to a lump sum;

{9} Estimated increases in costs resulting from the fact that benefits will be paid on a full-time wage basis while the contributions are made on actual pay roll, including much part time.


To these were added an adjustment upward in the estimates of man-years of unemployment resulting from inadequacy of data; and allowances for various contingencies, among them the probability of increased costs in the course of time, as is the experience in all other forms of insurance. Weighting all these factors, Mr. Williamson arrived at a loading of 30 percent above the unadjusted estimates of compensable wage loss.

Applying this increase of 30 percent to the total compensable wage loss of the unemployed in each year resulted in an increase in the cumulative wage loss from 1924 to 1933 to $31,815,000,000. (See tables 17 and 20. ) This allowance causes a wider disparity between


income available for benefits and the total compensable wage loss than was evidenced prior to the corrections. Consequently, after the wage loss is adjusted, the maximum benefit period must be shortened if the amount of income and outgo are to balance.

Using adjusted data as the unadjusted were used, 10 weeks of benefits could be allowed for the United States as a whole with a 3-percent contribution rate and a 2-week waiting period. Benefits could be increased in duration to 11 weeks, with a 3-week waiting period, and to 12 weeks with a 4-week waiting period. Other es-


timates of the duration of benefits possible with a still longer waiting period or with income from a 4- or 5-percent contribution rate can be readily computed. A summary of alternative plans possible for the United States follows, based on solvency from 1922 through 1933:

Estimated maximum weeks of benefits
Waiting period Rate of contributions
3 percent 4 percent 5 percent
2 weeks




3 weeks




4 weeks




A word of warning should be injected, however, that the reliability of these estimates decreases as longer duration periods are reached. It should also be borne in mind that the relationships shown in this summary are not applicable to the political or industrial subdivisions of this country. Each area for which a system is to be evolved must make its own estimates in order to approximate the probable financial behavior of any plan considered.


It should be again emphasized that the above estimates of the maximum possible duration of benefits with varying contribution rates and waiting periods are based on the assumption of the conservation of expenditures of accumulated reserves throughout the first 7 years of the system in order to continue paying benefits to eligible employees throughout the depression.

It is possible to estimate the maximum duration of benefits on another basis, assuming that all funds contributed during normal years and years of minor depression are expended within those years. This will mean that the emergency of a major depression with its reduced contributions from lowered pay rolls and its increased obligations for the payment of benefits to the eligible unemployed will bankrupt the unemployment compensation fund. Government subsidy or borrowing to restore the solvency of the fund or other Government provisions for the unemployed will then be necessary.


The adjusted cumulative distribution of the wage loss from 1922 through 1930 is shown in table 21. This table can be used to compute the maximum duration of benefits possible with the maintenance of solvency up to but not including a major depression. Thus from table 16 it may be ascertained that the total estimated contributions to the unemployment compensation fund would be $6,972,000,000 at the end of 1930, with a 3-percent contribution rate. This would permit the following durations of benefits with 2-, 3-, and 4-week waiting periods as indicated below.

Waiting period Maximum duration of benefits
2 weeks 17 weeks
3 weeks 19 weeks
4 weeks 22 weeks