Germany (2)

Title of the survey

Survey of Labour Costs

Organization responsible

Federal Statistical Office (Statistisches Bundesamt), in cooperation with the statistical offices of the Länder.

Periodicity of the survey

Four-yearly (1988, 1992).

Objectives of the survey

To obtain data on labour costs per employed person and per hour worked and to enable the individual enterprise to compare its own labour costs with the average value for the branch or other branches of economic activity in its country and the countries of the European Communities. The results are also used by the social partners in the process of collective bargaining.

Main labour topics covered by the survey

Employment, hours of work and labour cost.

Reference period

The calendar year (e.g. 1992) or the business year ending in March of the following year (e.g. within the period 1 April 1992 to 31 March 1993).

Coverage of the survey

Geographical

The whole country (prior to 3.10.1990, the survey covered only the territory of Federal Republic of Germany; since 1992, it also covers the new Länder and Berlin (East).

Industrial

Industry, commerce (wholesale and retail trade), financing, banking and insurance, and business services (computer and related activities, research and development and other business activities - in the former territory of the Federal Republic only).

Establishments

Enterprises with ten or more persons engaged.

Persons

Employees, i.e. wage-earners and salaried employees. Apprentices are separately identified and not categorized under the employees to whom labour costs relate. Excluded are the following persons who are not regarded as employees: those exempted from social insurance contributions due to short-term employment; statutory representatives of legal entities; persons working on a commission basis; homeworkers; workers sub-contracted from other companies or firms and workers from temporary work agencies; officials who are subject to an employment contract under public law: persons participating in job-creation measures; as well as persons working short time "zero".

Occupations

Data are not collected by occupation.

Concepts and definitions

Employment

Data refer to full-time and part-time wage-earners, salaried employees and trainees, separately. Wage-earners and salaried employees are employees who are subject to contributions to pension insurance fund for wage-earners or salaried employees, respectively. Trainees are employees whose work mainly serves training purposes (e.g. trainees and unpaid trainees, employees undergoing occupational retraining as well as university graduates participating in a trainee programme). Full-time employees are persons who have a contract of employment for the full collectively agreed working hours or the full working hours generally worked in the enterprise. Part-time employees are persons with a contract of employment for less than the collectively agreed working hours or less than the working hours generally worked in the enterprise. Employment data refer to the number of persons who were employed in the surveyed enterprises on the last day of each month of the reference year. Data on the number of women employees refer to the month of June.

Labour cost

This corresponds to the sum of wages and salaries, including special payments and remuneration for days other than working days, employers' social security expenditure, expenditure for support, education, and staff facilities, and other costs related to personnel. Data are collected separately on some 70 components, as follows: Labour cost data are collected separately for wage earners and salaried employees.

Hours of work

Data are collected on total hours paid for and hours actually worked, separately for full-time and part-time wage earners and salaried employees. If the number of hours paid for of salaried employees is not recorded in the enterprise, an estimate may be provided, based on the following formula: number of salaried employees x agreed weekly working hours x 52.14 + overtime. The hours actually worked are obtained by subtracting the hours paid for but not worked (for paid vacation, sickness or public holidays, meal breaks, time spent on travel from home to work and vice versa, etc.) from the total number of hours paid for.

International recommendations

The definition and components of labour cost comply with the guidelines of the European Communities and with the international recommendations on labour cost. The concept of hours actually worked conforms to the international guidelines on hours of work and the measurement of hourly labour cost.

Classifications

Components of labour cost / compensation of employees

Data on labour costs are classified by main components. This classification can be linked to groups I to IX of the International Standard Classification of Labour Cost (ISCLC - 1966).

Industrial

Data are classified according to the Statistical Classification of Economic Activities of the European Communities (NACE), Rev.1, which itself is based on the International Standard Industrial Classification of all economic activities (ISIC), Rev.3, 1990.

Occupational

Not relevant.

Others

The data are classified by:

Sample size and design

Statistical unit

The sampling unit is the enterprise. If the enterprise is a single-unit enterprise, it is also the reporting unit. If the enterprise consists of several local units, a separate report has to be completed for each local unit. If an enterprise has employees in both the old and the new Länder, separate reports must be completed.

Survey universe / sample frame

For the 1992 survey, the Register of the production industries and the 1987 Census of non-agricultural local units served as a sample frame in the former territory of the Federal Republic. The sample frame used for the production industries in the new Länder was also the Register of the production industries. In wholesale and retail trade, the 1990 Survey of economically active persons and other sample frames were used. With regard to the banking sector, material from the Deutsche Bundesbank was utilized. Determining the number of employees required preliminary inquiries in some areas. In 1988, the universe included a total of 148,000 enterprises, of which 98,000 enterprises were in the production industries (mining, energy and water management, manufacturing and building industry), and 50,000 were in the services sector (wholesale and retail trade, banking and insurance). In 1992, the universe included a total of 189,000 enterprises in the former territory of the Federal Republic, of which 116,000 were in the production industries and 73,000 in the services sector; and a total of 13,500 enterprises in the new Länder and Berlin-East, of which 9,900 were in the production industries and 3,600 in the services sector.

Sample design

A one-stage stratified sampling design is used. The sample of enterprises is first stratified by federal Land (17 Länder). Within each Land, the enterprises are stratified according to 79 economic classes (53 for the production industries and 26 for the services sector) and within the economic classes, by size classes; there are seven classes according to the number of persons engaged: 10-49, 50-99, 100-199, 200-499, 500-999, 1000-2999, 3000 and more. No separate results are calculated for classes 6 and 7 (1000 to 2999 and 3000 and more), but large enterprises with 3000 or more employees are included with certainty, while sampling is used in class 6. The drawing of the random sample in the production industries and in the wholesale and retail trades is carried out mechanically, using the Standard Stratified Sample Programme (STIA). The drawing of the sample in banking, insurance and publishing is carried out by hand in the Federal Statistical Office. In 1992, the sample size was fixed at 28,000 enterprises in the former territory of the Federal Republic, of which 15,000 were from the production industries and 13,000 in the services sector. In the new Länder and Berlin-East, the total was 9,800 enterprises, of which 7,100 were in the production industries and 2,700 in the services sector. In order to recognise possible differences with regard to labour costs and incidental labour expenses between the Länder, different selection procedures are applied, in the sense that small Länder are included with a sampling rate which is on average higher, and large Länder are included with a rate which is on average lower. The sampling fraction ranges from 11 to 82 percent, with an overall sampling fraction of 19 per cent. Rotation sampling is used from the preceding survey: all enterprises which were covered in the previous survey are, as far as possible for reasons of sampling methodology, replaced by new ones.

Field work

Data collection

Data are collected by mailed questionnaire by the staff of the statistical offices of the Länder, in accordance with uniform guidelines issued by the Federal Statistical Office. In the case of non-response, the postal survey is supplemented by telephone calls and by reminders in writing.

Survey questionnaire

It consists of seven parts, which concern, respectively: Information on the legal basis, purpose and type of the survey, the obligation to provide information, confidentiality and auxiliary variables, as well as relevant detailed explanations, are part of the questionnaire.

Substitution of sampling units

Not relevant. The survey is compulsory. The obligation to provide information is laid down in the EC Regulations in connection with Art. 15 BStatG. The employers of the enterprises included in the sample survey are obliged to provide the required information. In rare cases of total non-response, adjustments are made to the weighting procedure (see below) by application of a replacement factor.

Data processing and editing

Data are checked manually before being processed by computer. They undergo a machine credibility test. Those which are not plausible are checked by contacting the reporting unit by telephone or letter.

Types of estimates

Estimates are made in the form of totals, averages of labour cost for wage earners and salaried employees, and distribution of labour cost by components in percentages. The time unit to which they refer is the hour actually worked. Monthly averages of labour cost are also computed. In monthly estimates of labour cost, part-time workers are converted to full-time workers by multiplying them by 0.6. The hourly estimates are based on the actual or estimated number of hours actually worked and paid for. Apprentices are excluded from the denominator of the labour costs per employee. Similarly, labour cost estimates per hour worked do not include the hours worked by apprentices. However, their remuneration and the employers' compulsory contributions to social insurance for apprentices are covered as training costs in the labour costs relating to wage earners and salaried employees.

Construction of indices

Index numbers are not constructed.

Weighting of sample results

Because of the different sampling rates in the individual strata, conversion to a uniform level must be carried out. The results for each enterprise are multiplied by the extrapolation factor (i.e. the reciprocal of the sampling fraction). Aggregate data are obtained by multiplying the sample results per stratum by the raising factor (i.e. the ratio of the number of establishments in the universe to the number within the sample), corrected by the replacement factor.

Adjustments

Non-response

A distinction is made between "false" and "genuine" non-response. False non-response indicates a firm which either no longer existed at the time the survey was conducted, was no longer involved in industry or commerce, or no longer constituted an independent local unit. In such cases, the units concerned are crossed off the list and not replaced. Genuine non-response indicates a firm which simply ignored the survey. In this case, a replacement factor is applied, which corresponds to the quotient of the number of firms contacted less the number of non-existent firms, to the number of firms which replied plus the number of non-participating firms. The number of firms contacted is the "corrected" number of firms in the sample, i.e. after crossing off the "false" non-response firms.

Other bias

No adjustment are made for any other bias.

Use of benchmark data

Not relevant.

Use of other surveys

Not relevant.

Indicators of reliability of the estimates

Coverage of the sampling frame

The Register of the production industries and the Census of non-agricultural local units are in principle complete. In 1992, the major part of the material used for the new Länder and Berlin-East was obsolete. This resulted in non-coverage which, however, cannot be quantified in figures.

Sampling error / sampling variance

For the most important data, the simple relative standard error is computed. The sampling rates within the individual strata are chosen so as to produce the minimum sampling error: the simple relative standard error for the most important data (remuneration for work done, incidental labour expenses (total), legal incidental labour expenses (total), special payments and corporate pension plans) should not be greater than 1 per cent.

Non-response rate

Not relevant. Employers of enterprises included in the sample survey are obliged to provide the required information.

Non-sampling errors

The number of hours worked cannot be covered with absolute accuracy.

Conformity with other sources

Not relevant.

Estimates for non-survey years

For the years between the survey, the labour costs are updated by the Federal Statistical Office on the basis of the development of earnings and employers' statutory and non-statutory contributions.

Available series

Regular published tables include, separately, for production industries and wholesale and retail trade and financing and insurance industries:

History of the survey

The first Survey of labour costs was conducted in Germany by the German Employers' Association in 1949. In 1955, pilot surveys were carried out in various European countries in cooperation with the ILO. The first and, at the same time, only official Survey of labour costs at the national level was conducted in connection with a Structure of earnings survey in 1957. Between 1959 and 1964, the European Community member states conducted Community surveys in selected branches of economic activity. Since 1966, Community surveys have been carried out in the production industries. Up to 1984, these surveys were carried out every three years. Since 1984, they have been conducted at four-year intervals. The most recent survey was conducted in 1993, with reference to 1992. In the services sector, a pilot survey was conducted in retail trade and parts of the banking and insurance sectors in 1970. Another separate survey of the services sector took place in wholesale and retail trade and the banking and insurance sectors in 1974. Since 1978, the survey conducted in those areas has been integrated with the Survey of labour costs in production industries.

Documentation

Statistisches Bundesamt: Löhne und Gehälter, Arbeitkostenerhebungen, Fachserie 16, Heft 1 (Arbeitskosten in Produzierenden Gewerbe) (Wiesbaden, 1988). idem: Löhne und Gehälter, Arbeitkostenerhebungen, Fachserie 16, Heft 2 (Arbeitskosten im Gross- und Einzelhandel sowie im Bank- und Versicherungsgewerbe) (ibid., 1988). Preliminary and final results of the EC Labour Cost Survey are also published by: Statistical Office of the European Communities (EUROSTAT): Labour Costs 1988 - initial results (Luxembourg, 1991). idem: Labour Costs 1988 - volume 1: principal results, and volume 2: results by size classes and by regions (ibid., 1992). idem: Labour Costs: Updating 1989-1991 (ibid., 1992).

Confidentiality / Reliability criteria

The individual data reported for the survey are used for statistical purposes only. They must not be used for taxation or passed on to third parties.

Other information

Data supplied to the ILO for publication

Statistics of average labour cost per hour actually worked in manufacturing are published in Tables 22A and 22B of the Yearbook of Labour Statistics.