3

As the individual works under covered employment, his employer uses the social security number as the numerical identifier for reporting quarterly wages. For agricultural workers, data are reported annually. These earnings reports are the primary source of data for accumulating information on taxable wages in the record of the specified employee. In a similar fashion, information on earnings for self-employed persons becomes available through information filed by the individual with his Federal tax return as part of his Schedule C or Schedule F.

Earnings data, summarized for each calendar year, are part of the machine-readable record for each account number. Quarterly data, together with codes for county of employment and industry, are also maintained for a 1-percent statistical probability sample, the Continuous Work History Sample (CWHS). It is selected on the basis of the social security number with constant selection pattern since the start. This sample design provides a basis for within-theyear as well as over-the-years linkage of the more detailed information that is maintained for statistical purposes.

Economic Data for Business Establishments

Program administration provides a variety of economic data for business establishments. Some of these data are basic classifiers; others are data from current program operations.

Every employer with a worker whose earnings are covered under the social security program is required to obtain an employer identification number (EI number). The application for employer identification number," provides basic classification information on the location and nature of the business. A copy of this application is used by the Social Security Administration as part of its administrative and statistical systems.

2 Social Security Bulletin, 1965 Annual Statistical Supplement, page 111.

3 For nonagricultural work, IRS Form 941-Employer's Quarterly Federal Tax Return; for agricultural employment, IRS Form 943-Employer's Annual Tax Return for Agricultural Employees; for domestic employment, IRS Form 942-Employer's Quarterly Tax Return for Household Employees.

4 IRS Form 1040, Schedule SE, U.S. Report of Selfemployment Income.

5 IRS Form SS-4-Employer's Application for Identification Number.

Additional information on multi-unit companies is developed in connection with the statistical activity.

The earnings reports from the employer's quarterly Federal tax return provide current data for the employer entity. Information on the number of employees during the reporting period and on aggregate taxable payrolls for each unit can then be combined with the information on nature of business and geographic location. Agricultural, State and local government, and household employer entities also report earnings. A basis for classification by specific industry and county of employment for multi-unit companies is provided in the establishment reporting program (ERP) of the Social Security Administration that permits identification of employees by unit. A self-employed person reports on his tax return the portion of his self-employment earnings that will bring his total earnings recorded in his social security records up to the taxable limit. Some information permitting classification on nature of business becomes available for these business earnings as part of the total set of economic data for establishments available at the Social Security Administration.

6

Health Service Utilization Data for Individuals

Administration of the health insurance program for the aged provides data on the utilization of health services for persons aged 65 and over. Records for the hospital insurance part of the program show the number of days of care received by each aged person in a hospital or extended-care facility and the number of home health visits received during specified periods that relate to the program concepts of "deductible" amounts and "spell of illness." 8

Each episode of hospitalization provides data on length of stay, discharge status, charge and

6 Social Security Administration, The Establishment Reporting Plan, August 1965.

7 Howard West, "Health Insurance for the Aged: The Statistical Program," Social Security Bulletin, January 1967; Aaron Krute, Newton Dikoff, and Jack Scharff, Medicare, the Statistical Elements, 1967.

8 For a full description of the provisions of the health insurance program, see Wilbur Cohen and Robert M. Ball, "Social Security Amendments of 1965: Summary and Legislative History," Social Security Bulletin, September 1965.

payment data, primary diagnosis, surgical procedures, and preoperative and postoperative lengths of stay. Diagnostic and surgical information is coded by the Social Security Administration for a 20-percent sample of the hospital insurance beneficiaries. The diagnostic code is for the primary discharge diagnosis reported by the hospital, and the surgical procedure code relates to the primary discharge diagnosis. For outpatients, coding is for a 40-percent sample; all home health bills and extended-care facilities bills are coded for diagnostic information.

Data available on utilization of physician and related medical services cover time and place of each service, the exact procedure or service provided, the condition treated, the physician's or supplier's charge for the specific service, and the charge allowed for the same procedure or service. For nonsurgical medical services, there are some descriptive data available on the type of services provided by the physician at each visit. For surgical cases, data on the surgical procedure, the diagnosis, and the overall charge for the entire procedure are available. The data become available through the administrative requirements for submittal of bills (at least $50 of services during the year is required). These data on services under the medical insurance program are centrally recorded for a 5-percent sample of all enrolled persons (a subsample of the 20percent sample for the hospital insurance diagnostic coding).

Characteristics of Hospitals and
Other Providers of Health Services

The hospital insurance program covers reimbursement for services of providers such as hospitals, home health agencies, extended-care facilities, and independent laboratories. Each provider, in applying for participation, has initially supplied (and will update periodically) a wide variety of detailed information, including data on the number of beds, type of control, and the major type of services provided. Also available are data on staff characteristics, such as the number of physicians, registered nurses, qualified special therapists, licensed practical nurses, home health aides, and other skilled medical care personnel. Information on the

volume of the provider's services is also supplied-for example, the annual total of adult admissions and discharges, the number of patient days, the number of persons served, and the current reimbursement rate.

SOCIAL SECURITY DATA AND
SOCIAL SCIENCE RESEARCH

The data available from the operations of both the cash and service benefits programs of the Social Security Administration probably represent one of the richest single aggregations of current demographic and economic information in terms of the number of people involved. The records are an important current source of information about business establishments." County Business Patterns, first published in 1946, provides statistical data on first-quarter employment and taxable payrolls by county and industry of reporting unit. Some possible uses, general and specific, of social security data are discussed below to illustrate their variety.

The data have primary importance as a research resource, not only because they provide information on covered workers, but also because the information is longitudinally linkable. Some analytic studies have taken advantage of this longitudinal characteristic of the recorded information.10 The cumulation of information on episodes of hospitalization and their characteristics and on the utilization of medical services provides an important resource for studying health conditions of the aged.

The interrelationship of variables available from other sources linked with social security administrative record information (under appropriate conditions to insure confidentiality) has become the basis for additional analysis (beyond that based solely on data directly available to the Social Security Administration). Followup of economic characteristics of worker cohorts 11 or of health characteristics of aged cohorts are additional important research potentials.

9 Bureau of the Census, County Business Patterns (various years).

10 See Lowell E. Gallaway, Interindustry Labor Mobility in the United States, 1957 to 1960, Office of Research and Statistics, Research Report No. 18, 1967.

11 Social Security Administration records are being used to follow up a cohort of workers laid off in the shutdown of operations of a large industrial corporation.

The general area of economic analysis of the labor force involves analytic consideration of the dynamics of labor supply. Social security data are an important resource for many types of analysis in this field. Studies of labor mobility, whether between industries or between different geographic locations within a given industry, are a part of such analysis.12 Differences in earnings patterns among groups are an important analytic interest. Varying patterns of lifetime earnings of various cohorts, defined in terms of available demographic variables such as age, mobility patterns, and industry may help in understanding the relationship between the processes and effects of labor mobility and economic policy questions. They also have a bearing on the concept of earnings replacement through social insurance. The nature of the processes of labor mobility may be considered a significant element for the study of the efficiency level of labormarket mechanisms in allocating human resources in the economy.

Cohorts defined by membership in a specific program are important study populations. A parallel control cohort may be studied, in addition to those of the particular program, if the experimental design elements have been identified in advance. Social security data may play a role. in examining the effects of Government training and education programs to ameliorate existing disadvantages for a specified population group. The effect of the Job Corps or of other programs of the "war on poverty" may be studied in a variety of ways. Longitudinal records of earnings and other demographic patterns of the various subgroups exposed to a program may be studied prospectively through the use of records.

Cohorts defined as a result of major economic incidents, because of attachment to specified types of industrial activity, or because of specific characteristics such as those of Selective Service rejectees can be followed prospectively. Studies of the effects of programs for attracting industries to disadvantaged areas on the retention of population in those areas or analysis of the effects of programs for retraining disadvantaged people within economically disadvantaged areas are further illustrations of some uses of social security data. The impact of technological changes

12 Lowell E. Gallaway, op. cit.

in agriculture on acceleration of migration from rural to urban areas may also be studied. The relationship of patterns and time sequences of migration into and out of depressed areas with time sequences of programs for dealing with such areas and with depressed industries may provide useful data for analysis in this area.

Data collected for a given administrative purpose often exclude many variables of significant research interest that may have been omitted because of policy or public interest considerations. Data on age or race are often in this category. In many programs, either of these (or other) variables may be deemed to have a possible detrimental effect if the data are immediately available in conjunction with other data being reviewed for a program decision. Race is not recorded on any claims forms for social security benefits, for example. Linkage with other data from Social Security Administration records may facilitate statistical analysis of subgroups otherwise not identifiable. The differential effects of various types of changes in income-tax policies on various age cohorts of taxpayer units can be studied through linkage of samples to social security data.

To the extent that analyses of differences between a worker's industry of major job in a given period and other industries of employment. are of interest, both within a year and longitudinally over time, such data are available for overall cross-sectional analysis as well as for specific cohorts. Since such data are the product of the ongoing administration of the Social Security Act, certain important variables are not available, however, except when special studies are made. Thus, occupational mobility and its interaction with other aspects of labor mobility or education are not ordinarily subjects that can be studied. From social security data, some limited information on the number of shifts between employers within a given industry in a year or longer can be considered as an added variable for its potential in helping to define groups with differing labor-force mobility. The possible effects of private pension plans and the extent of vesting, if any, in relation to labor mobility and retirement patterns may be studied if data on private pension plans are available to the researcher.

When greater detail is available from another

source, more intensive analysis becomes possible. If, for example, data on farm operators by detailed type of farm operation are available, the phenomenon of changes in patterns of the use of the agricultural work force may be studied more continuously.

new

The wealth of data becoming available from the health insurance programs represents resource for social science research-limited of course to data for persons aged 65 and over. The longitudinal patterns of hospital utilization and medical care utilization can be studied. Crosssectional analyses as well as cohorts defined in terms of available variables within the program (geographic location, age, sex, race) may be studied. Additional program variables such as diagnosis or procedures and patterns of earnings or labor mobility may be considered. As noted, cohorts defined on the basis of information from other programs can be studied.

Information may be explored on groups defined in terms of their specific previous medical care practices where data on individuals exist (Blue Cross compared with non-Blue Cross members, for example). Comparison of patterns of health care use for specified cohorts for whom data are available before and since the advent of Medicare also becomes possible. Answers to policy questions concerning the nature of the upgrading or shifts in use of health services can be developed. Rates of change among groups over a period of time can now be studied without the problems of retrospective recollection or continuous survey processes, subject only to the limitations of the program requirements and the data collection processes of the records systems.

Many suggestions for research can be developed. A few are sufficient to illustrate some of the possibilities. The data available solely from the Medicare program may provide some basis for answers to social researchers' interests in areas such as the following: (1) the patterns of utilization of medical services by people who migrate during the year; (2) the patterns of differences in use of services between areas of residence and of service; (3) the measurement of expected numbers of preoperative and postoperative visits for specific surgical procedures; and (4) changing relationships between use of extended-care facilities and hospital utilization.

When data are also available from records outside the health insurance program, the impact of the program on specific groups such as veterans, welfare recipients, the poor (identified by general geographic location as well as levels of earnings) can be measured over time. Data about individuals with or without other forms of health insurance can be studied in regard to utilization and financing of medical services.

A number of useful studies of methodological interest are possible with the wide variety of statistical data from other data sources. Reported data on age, race, or earnings in the records system data can be compared with those in other studies for the same individuals and measures of differences made available. Differentials between nonrespondents and respondents in a survey can be studied with the available variables from the social security data.

CONFIDENTIALITY OF SOCIAL SECURITY
DATA AND DISCLOSURE

Steadfast determination to keep a pledge of confidentiality for records on individuals and a conscientious desire to make information available for planning and research in the public interest may appear, at first glance, to be strange bedfellows. In the administration of the Social Security Act, the pledge of confidentiality was made at the beginning and has been stoutly maintained ever since. This fact is clearly visible at every stage in the 30-year history of the program. There is nonetheless adequate evidence that the program's principal executives understood and expected that in time the information being collected and stored for proper administration would begin to serve other useful purposes beyond the specific program purpose.

During most of its existence the social security program's administrators not only focused on the essentials of program planning and execution for effective and efficient operations but also gave thought to future needs for growth and improvement. Steps were taken to serve the immediate public interest beyond the narrow purpose of program administration. A philosophy of intention and interest to contribute to such aims was clearly formulated.

This section of the article examines the nature

of the Social Security Administration's confidentiality policy and the principles that govern the disclosures that may or may not be made from information in its possession. Such an examination seems appropriate at this time since it is now possible to see what the agency's collection of records is and what it is likely to be. It is possible to see and appreciate both present and future problems of accessibility. It is possible to explore the Administration's continued linking of a firm confidentiality policy and a positive interest in contributing through the agency's facilities to public interest social science research.

Statute and regulations protect the confidentiality of records and information in the possession of the Social Security Administration. The reasons for the statute and the regulations relate directly to "the public interest and efficient administration.” 13 These reasons have held consistently throughout the three decades of existence of the Federal social security program. Interpretation of these broad reasons has been dynamic and their meaning has broadened with program growth, but there has been no deviation from the fundamental confidentiality position.

The confidentiality policy is concerned with prohibition against the disclosure of information of a personal nature about individuals and their activities in the possession of the Social Security Administration. Both the statute 1 and the regulations 15 protect the rights of privacy of those who give information, either voluntarily or pursuant to law, with the understanding that such information is to be used only for social security and related purposes. There is no question or issue of Government secrecy about the operations of the Federal social security program. From the beginning, the Social Security Administration has published statistical summaries of all its program activities. There is no question or issue, either, of holding back information that affects the rights of individuals under the program. The Public Information Act of 1966, as amended in 1967, explicitly continues the guarantee of protection of such information.

13 Regulation No. 1 of the Social Security Board, filed with the Division of the Federal Register, June 16, 1937. 14 Section 1106 of the Social Security Act, as amended. 15 Regulation No. 1 (as amended), Disclosure of Official Records and Information (part 401, chapter III, title 20, Code of Federal Regulations), Social Security Administration, 1964.

The significance of a strong position on confidentiality of records was clearly perceived by the Social Security Board 16 well before operations began under the original Social Security Act on January 1, 1937. Since the Board was also responsible for administering a program of grants to the States for public assistance, it was deeply concerned with the possibility that access to public assistance records might be sought for political purposes by some State and local elected or appointed officeholders. In addition, before the start of the old-age insurance program, there were many rumors and news stories predicting the use of the records by employers to obtain information about the age of employees or for blacklisting purposes about a worker's previous work history. There were fears of infringement of civil liberties and the creation of a "police state."

It had been clear from the start that operation of the social insurance program would call for the collection and maintenance of considerable personal and private information from both employees and employers. To obtain the needed cooperation of both the working force and the employers, the Board publicized the types of information that would be needed for operation of the program. To allay public fears, a public pledge was made that the records would be used only for social security purposes and otherwise treated as restricted. A press release, issued on November 23, 23, 1936-the 1936 the day before the availibility of applications for social security numbers-informed the public that:

The Social Security Board announces that . . . information required of every worker on this form will be regarded as confidential within Government sources. Only the worker himself, or his immediate family, or Government employees having official responsibility in connection with the social security files will have access to this information.17

16 Independent Federal agency, placed under the Federal Security Agency (now the Department of Health, Education, and Welfare) under Reorganization Plan No. 1 approved by the Congress to become effective July 1, 1939. The Board was abolished on July 16, 1946, by an Executive Order that created in its place the Office of Commissioner of Social Security.

17 In the face of continuing sharp reaction, a second press release was issued on December 10, 1936, reiterating the Board's pledge of confidentiality: "The Board will, at all times, regard the information received from an employee as confidential. The files will be open only to those who have a legitimate interest in the administration of the Social Security Act."