tion of the shape of things to come. The oldest have always been in the worst financial plight. It has been assumed that as the older beneficiaries die and as others enter retirement with years of high wage levels behind them, beneficiaries as a group would be much better off. The small income advantage enjoyed by the age group 65-72 compared with beneficiaries aged 73 and over raises some question concerning this assumption.

Furthermore, by the close of 1962 almost onethird of the women aged 65 and over who were drawing benefits as retired workers, and more than two-fifths of the retired women beneficiaries aged 62 and over (married plus nonmarried), had taken an actuarially reduced benefit. This action has been possible for women since late 1956. Of the women drawing benefits as dependent wives of retired workers at the end of 1962, the proportion with actuarially reduced benefits was 34 percent for those aged 65 and over, 45 percent for the entire group aged 62 and over. 13

It was not until August 1961 that men were eligible for a reduced benefit at age 62 and then on even less favorable terms than women because of the method of computing their benefit. By the end of that year, however, there were 273,000 men with actuarially reduced benefits. By the end of 1962 the number had advanced to 657,000, or onetenth of the retired men receiving OASDI benefits. Nearly one-fourth of all men aged 62-64 in the Nation were receiving OASDI benefits at that time. Although some workers may take advantage of the OASDI provision for retirement at ages 62-64 because they can also draw a private pension, it is clear that many of the men who retire before age 65 are unemployed at the time or have had a history of low earnings or intermittent employment.

In considering adequacy of benefits, thought must be given to the reduced amounts for which many beneficiaries will settle. One may well wonder whether a provision intended to ease the way for workers forced out of the labor force prematurely may not be creating a new group of poor-people who will have many years with little income but a benefit, and that a small one.

There seems little doubt that OASDI will re

13 The average benefit of retired women whose benefits were actuarially reduced was $58 a month, compared with the $66 that would have been payable as a benefit were it not for the actuarial reduction. For aged wives the corresponding averages were $37 and $45 a month.

main the major source of retirement income. The level of protection afforded by the program becomes a measure of what our society intends for its aged members.

Technical Note on Source and Reliability of the Estimates

SOURCE OF THE DATA

In 1962 the Social Security Administration of the Department of Health, Education, and Welfare undertook a nationwide survey of the socio-economic characteristics of the aged, with the Bureau of the Census responsible for collecting and tabulating the information.

SURVEY DESIGN

1. Interview unit

The basic interview unit for the Survey was an "aged unit," which was defined as a married couple living together, either member of which was aged 62 or older, or a nonmarried person (including persons whose spouse had a usual residence elsewhere) who was aged 62 or older.

2. Sample design

A representative multistage area probability sample of the universe was used as the basis for the Survey. (The universe was composed of the civilian population aged 62 and over residing in the 50 States and the District of Columbia.) Ultimate sample units consisted of a representative subsample (one-half) of the Current Population Survey (CPS) sample 14 and the full Quarterly Household Survey (QHS), to create the sample for the 1963 Survey of the Aged. The ultimate sample units in the 1963 Survey sample, therefore, were selected after the following stages of sampling :

a. The standard metropolitan statistical areas and counties of the United States were grouped into about 1,900 primary sampling units (PSU).

b. These primary sampling units were then grouped into strata of one or more primary sampling units that are relatively homogeneous according to socio-economic characteristics. (There were 357 strata for the CPS and 333 for the QHS. The 333 represent an earlier phase of the evolution of the first-stage design of the CPS.)

c. Within each of the strata a single primary sampling unit was selected to represent the stratum. The 357 area CPS design is composed of 701 counties and independent cities and the 333 area QHS design of 641 counties and independent cities-with very substantial overlap between the two sets.

d. Within each of the primary sampling units a sample of housing units with addresses from the 1960 Census

14 For a complete description of the CPS sample see Bureau of the Census, The Current Population Survey— A Report on Methodology, Technical Paper No. 7, 1963. The QHS sample design is similar to the CPS design.

Because of the amount and complexity of the information being requested, the field survey was conducted in two stages. In the first stage, begun early in January 1963, respondents were identified and the Survey was explained to them. They were asked to provide their social security account number and such identifying information (not already available from the CPS or QHS interview) as is usually obtained on an application for an account number. Respondents were then given a questionnaire to complete and hold for an interviewer to pick up at a subsequent visit. In the second stage, completed in February 1963, the interviewer reviewed the answers on the self-administered form and filled in a second questionnaire relating to additional topics. Altogether, useful questionnaires were completed for 7,515 aged units, a completion rate of about 88 percent.

Persons in institutions were included (at half the sampling ratio used for the aged units in households). Only a limited amount of information-primarily on income and medical care was requested. Where feasible, the answers were obtained directly from the respondent; in other cases, personnel of the institution and/or hospital records provided the needed detail.

4. Nature of information

Information was collected on such topics as income by source, work experience, assets and liabilities, health care costs, health insurance coverage, and living arrangements, as well as other facets of socio-economic status of persons aged 62 and over. Information in this detail will be available for the first time for a representative sample of all aged persons in the United States rather than only of OASDI beneficiaries.

The first-stage questionnaire covered health insurance, medical care costs, assets and debts, and income. The follow-up interview obtained more detail on these subjects and included additional questions on other subjects such as home tenure, living arrangements, housing and food expenses for those living alone, and on labor-force participation and work experience, as well as special questions for recent widows.

The information obtained from these two questionnaires was supplemented by information on household composition and family income from the CPS and QHS interviews as well as the Social Security Administration's record data described below.

5. Match with social security records

All cases were checked against the Social Security Administration's National Employee Index and other records to determine if the individual respondent had an account number or, by cross reference, if he had any possible claim status. All cases with a social security account number or a possible claim were then further screened to determine if a claim had been filled. Information was abstracted on type of benefit, primary insurance

amount, benefits received during the survey year, year of first benefit, and other factors. Of the 8,878 units screened against OASDI records, positive identification as beneficiaries or nonbeneficiaries was completed on all but about 10, for which there was no evidence of a claim. Benefit record data were compiled on all but five of the 5,258 units identified as beneficiaries. Anyone who had received at least one benefit by the end of 1962 on an existing claim was classified as a beneficiary.

ESTIMATION

The estimates presented in this report are therefore derived from both the field collection and the OASDI program information. The basic data for each unit were weighted as follows:

1. Adjustment for noninterview

Some of the sample units did not provide usable questionnaires. For most households, however, there was some limited information that could be utilised in the noninterview adjustment process. Interviewed units having characteristics similar to those of noninterviewed units were selected at random and given a weighting factor of 2 to adjust for units not interviewed. The characteristics used in identifying similarities between interviewed and noninterviewed units were geographic area, size of aged unit (1 or 2 persons), age and race of the head of the unit, and sex for one-person units.

2. First-stage ratio estimation

The first stage of ratio estimation takes into account differences at the time of the last Census in the distribution by race and residence of the population estimated from the sample PSU's and that of the total population in each of the four major regions of the country. This stage of estimation has the effect of reducing somewhat the contribution to sampling variability arising from the selection of sample areas in the first stage of sampling. 3. Second-stage ratio estimation

The second-stage ratio estimation used the results of the 1963 Survey of the Aged after the noninterview adjustment and the first-stage ratio estimation to provide distribution of characteristics within age and race groups. Independent estimates of the civilian population aged 62 and over by race, sex, and age groups were then multiplied by the distributions derived from the Survey to create the estimates shown in this report. The number of OASDI beneficiaries calculated in this way was found to be less than 2 percent below the Social Security Administration estimate of the number with benefits in current-payment status and within 5 percent of the number with benefits in force--that is, on the rolls, whether or not a benefit had ever been received. At the end of 1962, more than 400,000 of the 14.5 million persons aged 62 or over with benefits in force were not actually receiving payments.

RELIABILITY OF THE ESTIMATES

Since the estimates in this report are based on a sample, they may differ somewhat from the figures that would have been obtained if all aged persons in the United States had been surveyed and the same schedules, instructions, and interviewers used. Estimates of the sampling variability of the Survey results will be available in the detailed report on the 1963 Survey of the Aged.

In addition to sampling variability, as in any survey work, the results are also subject to errors of response and nonreporting. In many cases the data were based on memory rather than on records. In most income and expenditure data derived from field surveys the memory factor probably produces underestimates because of the tendency to forget minor or irregular sources of income and outlays. There are indications, however, that the tendency to underestimate income was less in this Survey. Other errors of reporting result from misrepresentation or misunderstanding as to the scope of a concept. Incomplete responses to questions were handled in a variety of ways, depending on the question. Every effort, short of mechanical imputation, was made to obtain for each schedule a total income and a total medical expense figure, each built up from a detailed series of questions. In the case of income, for example, when an asset was reported and there was no entry for income accruing from assets of that type, income at the rate of

4 percent was recorded. If, on the other hand, the respondent reported on most income items but failed to make an entry (of an amount, "None," or "Don't Know") for certain infrequent income sources, such as unemployment insurance or individual annuities, this was tabu. lated as a zero entry. In the case of medical care, if the cost of care by doctors, dentists, and care in hospitals was recorded, but there was no entry at all for "Other" (miscellaneous) medical care, this item too was tabulated as a zero.

In addition to the results available from the match against the social security records, a series of comparisons with other reports on the number receiving income from specified sources is in process. Data on size of income, amount of assets, health insurance coverage, and hospital utilization are also being compared with those yielded by other field surveys. The results of these comparisons will be published in the detailed report on the Survey.

Health Insurance

for Retired

Employees

LIKE GROUP life insurance, group health insurance now is an almost universal element of corporate employee benefit plan structures for active employees. By 1960, more than 80% of plant and office workers in private employment in the urban labor force were covered by hospital insurance; in addition, 42% of the office workers and 20% of the plant workers were covered by major medical insurance.'

Judging from the information obtained from the companies in this study, an important segment of private corporations extend health insurance coverage to employees after they retire. However, the general pattern of this postretirement health insurance protection is much different than that of group life insurance:

• The extension of health insurance is significantly less common than the continuation of group life insurance to retired employees. It is also a more recent development.

• Postretirement group life insurance generally is paid by the company. A minority requires pensioner contributions, and relatively few place the entire burden on the pensioner.

Postretirement health insurance is fully financed by only one third of the companies. Another third shares the premium cost, and the final third, pays nothing.

The basic argument for noncontributory group life insurance after retirement (page 48) also applies to health insurance: the retired worker has substantially reduced income from which to pay premiums. That a majority of the companies requires the pensioner to pay part or all of the health insurance premium implies that the program is too costly to totally finance.

"Supplementary Wage Benefits in Metropolitan Areas, 1959-60," Monthly Labor Review, April, 1961, p. 386.

• This is reinforced by the one practice in the continuation of health insurance that is similar to that under group life insurance. Practically all companies that continue group life after retirement reduce the benefit levels; so do a sizable majority of companies that continue the group health insurance coverage.

A case can be made (page 49) for reducing group life at retirement, based on equity and on the reduced need for life insurance after retirement. But neither principle holds for health insurance. As a rule, the amount of health insurance protection for a particular group of active employees is not geared to salary levels, as is the case in group life insurance; thus the argument that a reduction in salary at retirement should lead to a reduction in insurance benefits has little force for health insurance. More important, the need for health insurance unquestionably tends to be greater after retirement, not only because of reduced income, but also because of the greater incidence of illness and disability among older persons.'

The greater incidence of disability among pensioners means that the cost of health insurance will be a good deal more than for younger employees. Furthermore, the past decade, at least, has been one of general increases in the price of group health insurance.'

Unquestionably, this combination of increasing costs for health insurance and the specific higher cost of health insurance for the aged has been a key factor in the extent to which health insurance is continued after retirement, the extent to which benefit levels are reduced for retired employees, and the way in which the postretirement benefits are paid for among the companies in this study. These basic characteristics of the procedure for

1 The National Health Survey (1958-1960) provides documentation of the higher incidence of disability among older persons and of their greater utilization of medical services. Comparative medical expenditures in 1957-1958 by persons under and over 65 have been compiled by the Health Information Foundation. Pertinent statistics from both sources have been assembled by the Department of Health, Education and Welfare in "Medical Care Financing and Utilization," Public Health Service Publication No. 947, Health Economics Series No. 1, 1962.

A recent study estimates that the price of group hospital insurance increased from 7% to 8% annually between 1949 and 1959. H. M. Somers and A. R. Somers, "Doctors, Patients and Health Insurance," Washington, D. C.: The Brookings Institution, 1961, p. 287.

extending health insurance to retired employees are described in the following sections of this chapter. The analysis is limited almost exclusively to the continuation of commercial group health insurance, however; continuation of Blue Cross-Blue Shield protection after retirement is a much different matter and, in important respects, is not under the control of the company but results from the particular way in which these nonprofit arrangements operate.

As a rule, a company with a Blue Cross or Blue Shield plan automatically insures that an employee has available after retirement essentially the same coverage at group rates. For example, under Blue Cross an employee who leaves the company, at any age, usually can continue the identical protection by paying the premium directly to the Blue Cross organization; the premium he pays under this group conversion provision is only slightly larger than the premium charged a member of the company group.'

The company can prevent even this increase in premium after retirement if it is willing to assume the administrative costs of collecting the retired employee's premium-for example, by deducting it from his pension check. Furthermore, by doing this--that is, by allowing the retired employee to remain a member of the company group-the company is not likely to be faced with higher costs that are attributable directly to the addition of retired employees to the group. The Blue Cross premium usually is a community rate and does not vary according to the age distribution in each company.

Commercial Health Insurance
Coverages

With the development of major medical insurance, three different types of commercial health insurance plans are used to provide medical-expense protection for active employees: plans that cover basic medical expenses; supplementary major medical plans that cover expenses over and above those paid for by the base plan; and comprehensive major medical plans that replace the

For example, in 1959, the median monthly cost of a Blue Cross family plan was $6.35 for a group member and $7.58 under a group conversion contract. Ibid., p. 312.