Second, we have always been receptive to, and have often ourselves initiated, cooperative arrangements with other agencies for the purpose of minimizing the burden upon respondents. Over 40 years ago, in the early period of the development of employment reports from employers, the Bureau entered into cooperative agreements with State labor departments which had initiated such reporting themselves. When the employment security program was established in the 1930's, this system was extended to all States through a cooperative arrangement with the Bureau of Employment Security. The way the system operates today is that the State agency collects the information from the employer, which they want to do anyway for their own purposes. They take the information off the forms for their own use and then transmit the information to the BLS in Washington for national use. Through this system only one report by an employer supplies monthly employment, hours, and earnings data for several hundred industries for the Nation as a whole, for 50 States, and for over a hundred metropolitan areas. I do not believe there is any statistical system in the world which yields such a wealth of data from a single simple report. This is also true for our labor turnover statistics program, and to a large extent for our industrial hazards statistics.

I could cite many other examples, but I shall limit myself to one more. Many Federal agencies, such as the Army, Navy, and Air Force in the Defense Department, the Bureau of Reclamation in the Interior Department, and a number of others, determine the wage scales of manual workers in their Government establishments through a system of wage boards. These boards are required by an act of Congress to base the wage scales for various occupations upon the wage rates prevailing in comparable occupations in the local community. These wage boards made their own local surveys, resulting in duplicate requests to employers for the same wage information. About 15 years ago, the Bureau of the Budget initiated a system whereby the BLS, in every community in which we operate a wage statistics program, became the collection agency for the wage boards. In this way one report by the employer to the BLS served the needs of all the cooperating agencies. As this system has become better known, many State and local agencies have joined in it. Just as one example, in the annual community wage and salary survey which we make in San Francisco, approximately a dozen Federal, State, county, and municipal agencies all accept the BLS wage reports.

Third, in many of its reporting systems the Bureau collects the information through its own agents. In the case of the Consumer Price Index, approximately 95 percent of our price collection is done by agents who visit the stores and other establishments. In grocery stores, where we obtain food prices, our agents visit the stores on Tuesdays, Wednesdays, and Thursdays, so as to avoid the weekend rush of customers. The agent records the prices as marked on the goods on the shelves. In department and appliance stores more consultation with the manager is required, but our experienced agent is able to keep the manager's time down to a minimum.

Likewise, in our major industry wage surveys, the work of copying out the data is performed by our agent, when the employer has agreed to the survey. Sometimes the larger employers prefer to do this work themselves, usually through their machine tabulation facilities. However, such wage surveys for a particular industry are generally taken only at 2- or 5-year intervals Fourth, the Bureau relies very heavily on sampling procedure in its surveys. It is economical to work with as small a sample as is statistically sound. means that many single-establishment employers would be sampled only occasionally in any large-scale survey. Of course, the multiestablishment large employers will always be represented in a sample, because at least some of their establishments would be required in every instance.

This

Over and beyond the above general principles of operation, the Bureau operates through a set of procedures which are designed to insure that any new survey will receive the most serious consideration before it is initiated. For the past 17 years, the Bureau has had the consultation and advice of a Business Research Advisory Council (and a Labor Research Advisory Council, also). These Councils have a series of committees, each specializing in one of the Bureau's major programs-wages, prices, employment, productivity, etc. In a broad and general way all Bureau programs are discussed with the Advisory Councils or with one of their committees. These consultations furnish us with some guidance as to possibility and practicability of a survey or a statistical report, and also as to the likely availability of the data in employers' recordkeeping systems.

When a decision is made to undertake a new survey we normally set up a pilot project, through which we visit a small sample of respondents, test a series of proposed questions and obtain information as to what information the respondents could readily supply from their records. On the basis of the findings, the Bureau then works up a questionnaire which must be submitted to the Office of Statistical Standards in the Budget Bureau. That Office has the authority to convene a panel of experts from the industry to determine whether the survey questionnaire is suitable. Until that Office approves the questionnaire the Bureau cannot proceed with the survey.

Finally, as I indicated previously, if it turns out that an employer cannot readily supply the information he can refuse to report, since his response is voluntary. In mail surveys, we usually make a followup request in an effort to persuade him to participate in the study, and sometimes we may offer to do the work ourselves, if that would be acceptable to him. However, in the final analysis, it is up to him to decide whether or not he wants to make a report. There is no compulsion or coercion on our part.

In conclusion, I want to emphasize that I do not think that the reporting burden on employers derives primarily from the activities of a general purpose statistical agency like the Bureau of Labor Statistics. We find on the whole that the cooperation of employers in our surveys is exceptionally good, that the response is very high, and that the time needed to the submission of a report very short. (Some of our major statistical reports are published within 30 days of the reference period.) As an indication of this generally favorable employer attitude, I can tell you that, according to my recollection, it has been nearly 15 years since any significant survey undertaken by the BLS has partially failed through a lack of cooperation by respondents. In some cases, representative groups of employers have requested that studies be made by the Bureau; and many employers are heavy users of our data for their own purposes. It is my judgment that the great majority of employers (and others) who respond to our inquiries recognize that our research and statistics activities constitute a service to them as well as to the general public.

STATEMENT PREPARED BY THE U.S. DEPARTMENT OF LABOR,
BUREAU OF LABOR STATISTICS

This statement has been prepared in response to a request from the Subcommittee on Census and Government Statistics. It presents information on each of the seven points enumerated in the chairman's letter of April 28, 1964, inviting the Bureau of Labor Statistics to appear at its hearings. A brieg background statement precedes the discussion of the seven specified items.

BACKGROUND STATEMENT

The Federal statistical system of the United States is probably the most comprehensive, most current national statistical system in the world. Data pertaining to almost all major aspects of our population and our economy are available periodically on a timely basis. The preparation of the statistical information in this national system is geared to the general functions of many Federal agencies and is not centralized in a single statistical unit.

A large volume of the needed statistics is integrated with administrative functions of certain Federal units. A relatively few agencies are not involved in governmental regulatory or administrative functions but serve as generalpurpose statistical agencies. Their primary function is the preparation of statistical information for general use by Government, by business and labor, and by research and analystical groups.

The Bureau of Labor Statistics is such a general-purpose research agency. It is responsible for preparing a variety of statistical data dealing with unemployment and employment, hours, earnings, wage rates, labor turnover, prices, industrial hazards, productivity, etc. In carrying out its mission, it is constantly assessing potential sources for needed information, and the cost and impact on respondents when data are not available for secondary sources. The Bureau has long been concerned with reducing the burden of reporting, and has had considerable success in consolidating into single questionnaires various requests for data by Federal, State, and local agencies. Such cooperative arrangements have been instituted wherever possible.

The items in this statement have been organized to answer the questions appearing in the subcommittee invitation.

1. Volume of reports, fiscal 1955 and fiscal 1963

Since the Bureau of Labor Statistics is engaged in the preparation of a wide variety of economic data, each of its program areas is dealt with separately, together with explanatory remarks. With but one exception, all of the collections of data discussed are secured on a voluntary basis. The single exception to voluntary reporting relates to employers subject to the safety and health provisions under the 1958 Amendments to the Longshoremen's and Harbor Workers' Compensation Act. Performing the function of a collection agent, the Bureau of Labor Statistics secures the information required by law and transmits it to the Labor Standards Bureau, Department of Labor. Some 1,700 employers subject to the act are sent appropriate forms which spell out what information is required by the act, and what additional information is requested on a voluntary basis for use in the Bureau's national accident summaries. Three schedules are used in this collection, and these are the only mandatory forms used by the Bureau.

One general point should be made regarding the 36 program areas described and the total number of schedule forms in use. A schedule form is designed as an instrument for the collection or recording of information. It is made as specific as possible in order to facilitate the recording of data. Since the Bureau programs as a rule are broad and comprehensive in nature, different schedule forms may be designed for particular parts of a program. For example, in our Consumer Price Index program, more than 50 different schedule forms have been devised for recording prices of the various commodity groupings. To secure the information in all of the Bureau's programs, more than 125 different schedule forms are used.

Some minor surveys are not dealt with in the following because their total impact on the Bureau's workload and that of the Bureau's respondents is very slight. Omitted, for example, are the Bureau's reports from about 35 shipyards covering the number of their production workers, their man-hours, and their straight-time payroll. The Bureau collects these reports under contract with the Department of the Navy which uses the results in escalating shipbuilding costs. Also omitted are reports secured from fewer than 10 manufacturers engaged in making hydraulic turbines. These reports are collected at the request of the Department of Defense and the Department of the Interior. They are used to prepare a straight-time earnings series for use with hydraulic turbine procurement contracts.

1.1. Current Employment Statistics Program

In 1955 approximately 110,400 industrial units submitted reports in this program. In 1963 there were some 129,000. The reports are received by mail each month. The schedule form is of the "shuttle" type and is routinely sent to and completed and returned by the respondents. Only one line of data, comprising items usually readily available from payroll records, is collected each month. Firms are requested by a cooperating State to supply a current set of data. The report is sent back to the State office where it is controlled, recorded, edited, and then forwarded to Washington. By the end of the month, approximately 2 weeks after the reference period of the data, approximately 50 percent of the entire panel has been processed by the State and is available in Washington for use in preliminary estimates. Approximately 90 percent of the panel is available by the 20th of the following month. A reporting plan based on a flow pattern from State office to respondent to State office to Washington, cannot involve a great reporting burden if the system is to keep with a 30-day reporting cycle. The speed with which reports are made available on a regular basis indicates that the system appears not to place an undue burden on our respondents.

Many States have their own current employment statistics program which requires similar data. In order to standardize the resulting statistics, avoid duplicate requests to respondents, and increase efficiency of operations, the Bureau has entered into cooperative arrangements with all States except Colorado. These arrangements provide for the use of a standard set of schedules, and a standards set of procedures for developing and presenting the resulting statistics. Reports received from respondents by the States are made available for the national program in the form of reproduced punched cards from the largest States, standard listings from punched cards from smaller States, and the shuttl ing of the actual schedules from the remaining States. Each respondent is asked only once (by the State) for his monthly report.

The survey is based on a sample of establishments rather than on all establishments in the universe. The design of the sample includes all very large

establishments and smaller proportions of the small establishments. This system minimizes the workload on small business. The 129,000 reports are less than 5 percent of the total number of business and industrial units in the United States. This program provides comprehensive industrial detail on employment, hourly earnings, weekly hours, premium-pay hours, and the proportion of women workers. The content of the program has expanded slightly since 1955 due to the addtion of the following two items: In 1956 an entry was added for premiumpay hours in manufacturing industries; in 1964 a column for the payroll associated with such premium-pay hours was added. The relatively small expansion in the number of reporters since 1955 resulted from an effort to round out the resulting data-expanding the list of industries for which hours and earnings are prepared, and securing needed respondents in certain geographic areas. These actions arose from the need for more data on a local basis and are in accordance with recommendations made by the President's Committee To Appraise Employment and Unemployment Statistics. Since 1955 the Bureau has completed arrangements so that a single standard set of schedules is now used by all cooperating States.

In 1955 approximately 11,300 industrial units submitted reports on hirings, separations, and other turnover actions during the month. In 1963 this had grown to 39,500 reports, which are received by mail each month. The schedule used here is also of the "shuttle" type and is routinely sent to and completed and returned by the respondents. As in the employment program listed above, the schedule is a one-page form and only one line of data is requested each month. About 75 percent of the reports are received in Washington by the 17th of the month following the reference period; nearly all reports are received within 40 days after the reference period.

The Bureau has cooperative arrangements with all States for the joint collection of labor turnover reports, which are used in State and area and national programs. As in the case of the Current Employment Statistics program, in order to standardize the resulting statistics, avoid duplicate requests to respondents, and increase efficiency of operations, a "standard" schedule is used by all States and the reports requested by and submitted to the States are used in the national program. Reports secured by the States are made available in Washington in the form of punched cards or standardized listings made from punched cards, or by shuttling the actual schedule to and from Washington.

In 1955 the Bureau was in the process of developing a pilot cooperative arrangement with only one State. The success of this venture led to the continuation of more cooperative arrangements with the States until at present, all States have joined in the program.

The program produces turnover data for States, areas, and the United States for manufacturing and mining industries. These data are important economic indicators as recognized by the National Bureau of Economic Research. They are widely used by employers to analyze personal transactions and the cost associated with labor turnover. Local and State officials use the data in the operation of the Employment Service. The expansion since 1955 results from the greater use made of the data at the State and local level. The information now collected by mail was formerly collected by personal interview with employers.

1.3 Monthly report on the labor force

Basic data collection in this area is carried out by the Bureau of the Census, under contract with the Bureau of Labor Statistics. The survey is part of the broader Current Population Survey of the Bureau of the Census. In June 1955 reports were secured covering 21,000 households; in 1963 some 33,500 households were included. This program secures data each month by personal interview with members of households. It is the primary source of labor force data identifying the employment-unemployment status of the U.S. population by age, by sex, by duration of unemployment, etc. Special studies provide periodic data on topics such as family income, marital status, consumer buying plans, school enrollment and dropouts from school, work experience of the population, etc. This household survey is the only source for information on labor force attachment, the self-employed, and unpaid family workers. The households enumerated are rotated so that a respondent is interviewed in four consecutive monthly surveys, skipped in the next eight surveys, resurveyed four more times and then dropped from the sample permanently.

The data resulting from the survey provide a miniature replica of the United States and are used in evaluating the economic health of the country and in describing economic and social characteristics of our population. Since 1955 the increase in the sample from 21,000 households (in 230 areas) to some 33,500 households (in 357 areas) has resulted in increased accuracy in the various items of information resulting from the survey.

The Bureau of Labor Statistics was assigned the responsibility for planning, analysis, and publication of the labor force data in 1959. Previously, responsibility rested with the Bureau of the Census. The change coordinated all major labor force functions within the Department of Labor.

1.4 Scientific and technical personnel

In 1955 some 11,600 companies submitted reports on their research and development costs and personnel. In 1963 reports were received covering 12,000 establishments (not companies). Between 1955 and 1963 the scope and character of the survey were changed. The 1963 reports include data on scientists and technical personnel by various occupational categories, functions, and types of employing industry. The expenditure data are no longer collected by the Bureau of Labor Statistics. The reports in this survey are secured once a year by mail, supplemented by personal interview for selected cases. The sample design is such that large establishments employing scientists and engineers are included with certainty; small establishments and those having no scientific or technical personnel have a smaller chance of being selected for reporting. The data resulting from the survey form the basis for estimating scientific and technical manpower resources, and for assessing the current and future demand for scientists, engineers, and technicians.

This survey is supplemented by a similar survey of scientific and technical employment in State governments every 2 years. In 1962 questionnaires were sent to 1,670 agencies. The 1964 survey will include some 1,750 reports from State agencies, the increase in coverage resulting from the inclusion of several health occupations requested by the National Institutes of Health.

A similar survey of scientific and technical personnel in local government units has been started recently. It is a mail survey covering a sample of some 8,000 local government units. It is the only comprehensive survey on scientific and technical personnel employed by local government agencies and forms a part of the basic knowledge of the scientific and technical manpower field.

1.5 National housing volume (dwelling units started)

In 1955 reports were received by mail from 7,000 units issuing building permits in the United States each month. In adition, there was a field enumeration of 53 nonpermit areas. In 1959 this survey was transfered to the Department of Commerce.

1.6 Occupational wages-Nationwide survey of professional, administrative, technical, and clerical pay

This survey did not exist in 1955. The Federal Salary Reform Act of 1962 provides for the implementation, through an annual salary survey conducted by the Bureau of Labor Statictics, of the principle that Federal salary rates shall be comparable with private enterprise salary rates for the same levels of work. The 1963 survey involved collection of salary rates by personal interview from 1,771 establishments. The response rate was over 90 percent. The survey provides nationwide averages and distributions of salary rates for 75 categories of professional, administrative, technical, and clerical jobs-each translatable to a pay grade applicable to Classifications Act employees. Although designed to meet needs related to Federal pay, the survey reports are widely used in salary administration in industry, in occupational counseling, and in the review, by contracting agencies, of compensation practices of firms with research and development contracts.1

1.7 Occupational wages-Industry studies

In 1955, the reports were obtained mainly by personal interview from 2,150 establishments; the 1963 number was 4,440. The growth was prompted by continuing demands from labor and management. The response rate is over 90 percent. Surveys are conducted periodically, mostly on a 5-year cycle, in 50 manufacturing and 20 nonmanufacturing industries. About 15 industries are studied each year. These surveys provide averages and distributions of wage

1 In fiscal year 1964, $328,800 was expended on this program.