Reference metadata describe statistical concepts and methodologies used for the collection and generation of data. They provide information on data quality and, since they are strongly content-oriented, assist users in interpreting the data. Reference metadata, unlike structural metadata, can be decoupled from the data.
The labour cost survey records the input and all costs of the production factor labour in the calendar year. Input is measured in four different measures: as an annual average of employment and full-time units, as the sum of hours paid and as the sum of hours actually worked. Costs are recorded and broken down in detail. For apprentices, part-time employees and civil servants, labour input and labour costs are recorded separately so that, in principle, they can be shown separately and included or excluded in the statement.
The labour cost survey provides information on the amount and structural composition of the costs per input unit of labour. The most important indicator is the net labour cost per hour worked in the economy as a whole or by sector and size of the enterprise. Other important indicators are the share of non-wage labour costs in labour costs - especially statutory non-wage labour costs - and the number of hours actually worked per full-time employee.
2.2. Classification system
• Nomenclature of territorial units for statistics NUTS ("Nomenclature des unités territoriales statistiques", as at 31.12.2020).
• Economic activity: Classification of economic activities, 2008 edition.
2.3. Coverage - sector
Economic sections B to S of the Classification of Economic Activities, 2008 edition.
2.4. Statistical concepts and definitions
Labour costs
Labour costs comprise the total of all expenditures incurred by employers in connection with the employment of labour. The definitions in Commission Regulation (EC) No 1737/2005 and the recommendations of the International Labour Organisation (ILO, An integrated system of wages statistics) are taken into account.
Gross labour costs include:
• gross earnings (D.11) and
• all ancillary wage costs, which include
• employers' social contributions (D.12), including the costs of vocational education and training (D.2),
• employers' statutory social security contributions,
• the expenses for the company pension scheme,
• employers' imputed social contributions (continued remuneration, severance payments and pensions and allowances for civil servants),
• Recruitment costs and professional clothing (D.3),
• and taxes and levies on employment (D.4).
Net labour costs (D) are obtained by deducting wage subsidies (D.5) from gross labour costs.
Compensation of employees (D.1), also known from national accounts, is the sum of gross earnings (D.11) and employers' social contributions (D.12).
Labour input
Full-time units (full-time equivalents) and hours worked (B.1) are primarily used as units of measurement for the input of labour. Hours worked are part of the volume of work actually performed by employees. They, therefore, do not include paid hours lost, such as for holidays or sickness. Full-time units are all full-time employees as well as part-time employees, part-time employees and apprentices converted into full-time jobs according to their hours worked.
2.5. Statistical unit
Survey units (economic sections B to N, Q to S, and economic groups P85.5 and P85.6 of the Classification of Economic Activities, 2008 edition): enterprises.
For Economic Division O84 and Economic Groups P85.1, P85.2, P85.3, and P85.4 of WZ 2008 the data were not collected but calculated.
For economic group Q86.1, the smaller part of the data was calculated (for legally dependent public hospitals), and the larger part was collected.
2.6. Statistical population
Companies with 10 or more employees or more in manufacturing and services [Sections Bto S of the Classification of Economic Activities, 2008 edition].
2.7. Reference area
• Germany,
• former federal territory (including Berlin) and new federal states in Germany,
• the 16 federal states of Germany.
2.8. Coverage - Time
Comparability with earlier surveys is only limited, as the surveys differed with regard to the sectors of the economy covered:
• 1992: Sections C, D, E, F, G (partly), J (partly) of the Classification of Economic Activities, 1993 edition
• 1996, 2000: Sections C, D, E, F, G, H, J of the Classification of Economic Activities, 1993 edition
• 2004: Sections C to O of the Classification of Economic Activities, edition 2003
• 2008, 2012, 2016, 2020: Sections B to S of the Classification of Economic Activities, edition 2008.
However, the economic sectors included have a central influence on the results. For temporal comparisons, it is indispensable to use identical economic sectors as a basis. On the basis of the respective publications, the results for the manufacturing industry can be compiled as the longest available time series from the 1966 reference year onwards
2.9. Base period
Not applicable.
3.1. Source data
Random Sample of enterprises, maximum 34,000
3.2. Frequency of data collection
The survey is carried out every four years.
3.3. Data collection
The survey was conducted decentrally by the statistical offices of the Länder. The characteristics were collected through a written survey of the selected enterprises. The owners, the companies included in the survey, and the persons responsible for their management were obliged to provide information.
Figure 1: Proportion of statistical offices of the Länder by fieldwork deadline
The questionnaire is attached at the end of the quality report (see Annex). Almost all respondents submitted the information electronically via the form-based IDEV (“Internet Datenerhebung im Verbund”) reporting.
3.4. Data validation
Correction of missing values (Item Non-Response)
Primary survey: Although a large part of the survey characteristics was taken from the company databases for payroll accounting and human resources management, there were numerous incorrect reports. The discrepancies were corrected by the statistical offices of the Länder through queries and subsequent reports in direct contact with the parties responsible for providing information.
The remaining missing data were individually supplemented on the basis of plausible comparative values.
Correction of real non-response (unit non-response)
The legal obligation to provide information ensured a very high unit response rate: 97.5 % of the
companies obliged to provide information reported this. The unit non-response rate was corrected in the extrapolation.
3.5. Data compilation
Some data are not collected but are calculated (see 6.3.5). In particular
1. Estimates of the results for local units
2. Estimates of D.1113 payments for days not worked, D.1221 guaranteed remuneration in the event of sickness, D.11111 direct remuneration, bonuses and allowances paid in each pay period
3. Estimates of results for sections O (in full), P (by and large), and Q86.1 (in part)
Estimation of results of sections O, P, and Q 86.1 is based on the data of the personnel statistics, were available at the Federal Statistical Office; data collection was not necessary. The supplementary results of the finance statistics were also available at the Federal Statistical Office. The collective bargaining parameters applicable in 2020 (special payments, working hours, holiday entitlement) were taken from collective agreements collected by the Federal Statistical Office for collective bargaining statistics. Data on sick days, social security contribution rates, and other parameters were taken from relevant publications. Aggregated data on supplementary public health care were provided by the supplementary health care institutions. The characteristics were derived from or calculated on the basis of the available characteristics of the headcount statistics. Only a few characteristics of the headcount statistics could be transferred directly. These included the contractual working time, the share of the normal working time of a full-time employee, and the gross monthly earnings in June 2020 (extrapolated to the calendar year 2020). All other characteristics were calculated. The calculations were based on data from finance statistics, collective bargaining parameters (special payments, working hours, holiday entitlement), information on sick days, social security contribution rates and supplementary public provision.
3.6. Adjustment
A bound extrapolation according to the method Generalised regression estimator (GREG) was carried out. The typical auxiliary variables for GREG were the number of full-time employees, part-time employees, and part-time employees of the company in the annual average 2020 according to the administrative data memory 2020 of the statistical offices.
4.1. Quality assurance
A Federation-Länder working group of the technical officers of the statistical offices accompanied all work steps of the survey from the legal and methodological preparation to the publication of the statistical results. The working group held regular meetings three to four times a year. The survey questionnaire took into account the harmonized definitions of Commission Regulation (EC) No. 1737/2005 (see legal bases), and standards of the International Labour Organisation (ILO) on the design and implementation of the survey were observed. The reported data were subjected to extensive checks for completeness and plausibility in the statistical offices of the Länder. In case of doubt, the parties responsible for providing the information were requested to check the report and to confirm or correct it.
4.2. Quality management - assessment
Positive:
The data collected for the most important types of costs (gross earnings, social security contributions) are relatively accurate. They mostly come from the personnel administration of the companies, which is subject to internal and external audits. This does not always apply to data on working hours and absenteeism. These were partly estimated by the enterprises. All data were subject to extensive checks by the statistical offices of the Länder, which demonstrably led to considerable improvements in accuracy. The number of employees covered was exceptionally high at 11.4 million so the unavoidable random error in the sample was comparatively small. There was also an obligation to provide information so that distortions due to selective participation or non-participation can be ruled out.
Negative:
Employees in enterprises with less than ten employees were not included due to the relevant provisions of EU law. The results thus do not cover about a quarter of Germany's employees, which could not be corrected in the extrapolation either. All results therefore refer to the population as a whole, i.e. all employees in enterprises with ten or more employees, but not all employees in Germany. This affects the informative value of the statistics, in particular of absolute figures, which are therefore not actively published by the statistical offices. However, time comparisons are hardly affected. The latter only applies if the reporting periods compared cover identical branches of the economy (see 8.2). International comparisons - the main purpose of these statistics - are also hardly affected because the exclusion of micro-enterprises applies in all EU Member States.
5.1. Relevance - User Needs
In Germany, the results of the statistics are used by the Federal Government and the regional governments, the employers' and employees' organisations, the regional representatives of trade and industry, economic and social science research, as well as by companies and management consultancies. For these users, the focus is above all on the international comparison of sector results, i.e. the level of labour costs per hour worked in Germany compared with other EU countries or worldwide. Politically, the development over time of the share of non-wage labour costs, in particular those induced by law, is also of importance. Both use serve to observe and assess Germany's competitiveness as a production location.
Also at the level of the European Union (Eurostat), comparative evaluations for the assessment of aspects of competitiveness dominate. In addition to the indicators of labour costs per hour and share of non-wage labour costs, the comparison of actual hours worked per employee also plays a role here.
5.2. Relevance - User Satisfaction
Information on the data needs of the main users, their assessment of completeness or redundancy and their satisfaction with the data offered was not collected in a targeted manner. The Federal Ministries, the Statistical Offices of the Länder, the central municipal associations as well as representatives from industry and science are represented on the Statistical Advisory Board, which advises the Federal Statistical Office on fundamental issues in accordance with § 4 BStatG.
Subject-specific questions or suggestions are brought to the "Prizes and Earnings" expert committee appointed by the Statistical Advisory Board and discussed in the "Earnings and Labour Costs" speaker meetings. In addition to the institutionalized committees, the statistics on earnings are in an ongoing dialogue with associations, companies, universities, and private users, whose wishes arising from practical work are also incorporated into the further development of the statistics.
5.3. Completeness
The survey represents all companies based in Germany with ten or more employees. Not included are companies from the agricultural and forestry sector (Section A of the Classification of Economic Activities)
5.3.1. Data completeness - rate
Survey units (economic sections B to N, Q to S, and economic groups P85.5 and P85.6 of the Classification of Economic Activities, 2008 edition): enterprises.
For Economic Division O84 and Economic Groups P85.1, P85.2, P85.3, and P85.4 of WZ 2008 the data were not collected but calculated.
For economic group Q86.1, the smaller part of the data was calculated (for legally dependent public hospitals), and the larger part was collected.
See below.
6.1. Accuracy - overall
The statistical results are comparatively accurate. On the one hand, the individual data collected are comparatively very accurate if they come directly from the personnel management of the companies, which is subject to internal and external audits. However, this is not the case for all characteristics, for example not always for working time and days lost. These were partly estimated by the enterprises. All data were subject to extensive checks by the statistical offices of the Länder, which demonstrably led to significant improvements in accuracy. Secondly, the sample size is sufficiently large so that the unavoidable random error of the sample is comparatively small, at least for the economy as a whole. There was also an obligation to provide information so that distortions due to selective participation or non-participation can be ruled out.
6.2. Sampling error
Not provided.
6.2.1. Sampling error - indicators
For Information on the relative standard errors by section in NACE, see Table 1.
6.3. Non-sampling error
See below.
6.3.1. Coverage error
The aim of the survey is to estimate ratios such as labour costs per hour worked. This makes any under-coverage of the total population more acceptable than in the case of total values such as the sum of labour costs. An uncorrected under-coverage of 10% means that the total value will always be underestimated. For ratios this is not always the case: if the ratios for the units covered and not covered are the same, there is no distortion. If the ratios differ by, for example, 20%, the distortion is 2%.
The results for sections O (in full), P (by and large) and Q86.1 (in part) of NACE revision 2 are obtained from model-based estimates rather than a survey. The results for section O are based entirely, and those for section P almost entirely, on the data for civil service employees recorded in a specific national public-sector workforce statistics, the Personalstandstatistik the basic data source. Measured by the number of employees, the labour cost survey in economic section O "Public administration, defence; social security" reached a comparable level (93.7%) to the national accounts data for 2020. However, in economic section P "Education and training" it was even 99.8% of the employees registered in the national accounts were included in the labour cost survey. Enterprises in sections O and P85.1 to P85.4 were not part of the 2020 labour cost survey’s selected population. The local units of these very largely public enterprises, which were assigned to economic sections B to N and Q to S, were therefore not recorded either. This was because local units were included in the survey only via the enterprises. The exceptions were the local units of economic group Q86.1 Hospitals, for which calculations were made on the basis of the labour force statistics in order to avoid significant under-coverage since they formed the largest group of the local units affected.
It did, however, cause under-recording; newly founded enterprises in particular were not recorded. This under-recording was corrected since the reporting year 2012 by grossing up by the generalised regression estimator method (GREG). Employee data from the administrative data were used as auxiliary characteristics for the GREG.
6.3.1.1. Over-coverage - rate
The over-recording rate of the random sample of 32 000 enterprises taken from the enterprise's register was 2.4%. Thirty-six of every 1 000 random sample enterprises, therefore, did not belong to the basic population. Of these, 13 had discontinued their activities, 6 were below the threshold, two were double entries and one was outside the scope of their economic activities.
6.3.1.2. Common units - proportion
Not provided.
6.3.2. Measurement error
Characteristic
Full-time employees(A.11)
Source of error
Enterprises sometimes classified part-time workers on grounds of age as full-time employees
Effect
Overestimates of full-time employees, underestimates of part-time employees
Correction
After consultation with the enterprises, adjustment of the figures reported for full and part-time employees
Characteristic
Total hours paid(C.1)
Source of error
No recording in enterprises that pay by assignment or piece rates
Effect
Some of the figures for the enterprise missing or estimated
Correction
Imputation of an estimate by multiplying the average annual number of employees, weekly hours worked, and the number of weeks available for work in 2020
Characteristic
Total hours paid(C.1)
Source of error
The figures for hours paid were sometimes too low because holidays and sick leave were not taken into account
Effect
Under-recording
Correction
Consultation with the company; imputation of average values
Characteristic
Wages and salaries (excluding apprentices)(D.111)
Source of error
The definitions relating to figures for persons who are working part-time on grounds of age were sometimes unclear; sometimes the figures could not be obtained from enterprises or only with difficulty
Effect
Underestimates
Correction
If possible, imputation ofestimates
Characteristic
Payments for days not worked (D.1113), Total hours actually worked (B.1)
Source of error
It was not clear which days not worked were to be included
Effect
Sometimes too many public holidays, weekend days and unpaid days of sick leave were counted; overestimates of D.1113 and underestimates of B.1
Correction
Consultation with the company; stripping out of days not worked included in error; imputation of average values
Characteristic
Payment for days not worked (D.1113), Total hours actually worked (B.1)
Source of error
The enterprises did not always keep lists, hence leave entitlement was reported rather than the days of leave taken
Effect
Over-/underestimates of D.1113 and over-/underestimates of B.1
Correction
Consultation with the company; stripping out of days not worked included in error; imputation of average values
Characteristic
Hours actually worked by part-time employees (excluding apprentices)(B.12)
Source of error
Enterprises sometimes included days not worked by apprentices in the figures for part-time employees
Effect
Underestimates of the hours worked by part-time employees (B.12) and the number of part-time employees converted into full-time units (A.121)
Correction
26 days leave and/or 8 days of sick leave were deducted per apprentice; if the number of days was still too high, the entry was reduced to 0 (largely payment on the basis of presence)
6.3.3. Non response error
See below.
6.3.3.1. Unit non-response - rate
The unit non-response rate for the labour cost survey was 1,0%:
(referring to random sample enterprises which belonged to the target population, i.e. in-scope respondents)
Nine hundred and forty-five of every 1 000 sample enterprises filed reports. Of the 55 which did not respond, 27 did not belong to the target population (see over-recording rate under 2.2.1), 25 were part of the target population and it could not be ascertained whether tree belonged to the target population.
6.3.3.2. Item non-response - rate
Not provided.
6.3.4. Processing error
See below.
6.3.4.1. Imputation - rate
Item imputations were not generally documented. It is therefore not possible to calculate the item imputation rate for the characteristic “annual labour costs” and the total imputation rate.
From the information available, the following can be said: The survey was conducted by 14 statistical offices of the Länder in Germany.
Missing or implausible data were handled differently in the various offices, by, for example, enterprises being consulted in writing or verbally, or plausible data being calculated from average values and imputed immediately. Most frequently it was the characteristics on working hours and days not worked required to calculate the number of hours actually worked (B.1) that were corrected. The rate of corrected cases for each characteristic was between 9% (other payments for days not worked for full-time employees) and 16% (paid hours for full-time employees) of all reports.
6.3.5. Model assumption error
Correction of unit non-response
The shortfall of the 1.0% unit non-response was made good by a higher grossing-up factor being attributed to respondents in the grossing-up procedure.
Estimates of results for sections O (in full), P (by and large) and Q86.1 (in part)
The estimates were essentially based on the data sets which are available at close to the individual level for the 4.5 million employees in public service taken from the public-sector workforce statistics as at 30 June 2020, with taxable gross monthly earnings in June 2020 and characteristics determining earnings, including area of employment (employer), pay and wage group and contractual weekly working hours. For each data set the gross annual earnings and the commensurate additional costs borne by the employer for social security and occupational pension schemes were estimated. Data from the financial statistics on subsidies for civil servants and on costs of occupational initial and further training and from the annual reports and accounts of public service accident insurance schemes were also processed. For civil servants, imputed employers' contributions to pension schemes were calculated in the same way as in the national accounts procedure. It was not possible to estimate benefits in kind, employers’ contributions to maternity benefits, recruitment costs or severance payments. Overtime and working time accounts could not be taken into account in estimating hours worked.
Estimates of the results for local units
Enterprises reported the detailed information on employees, labor costs and hours worked for the whole enterprise rather than for each local unit. The data on local units, which were indispensable for producing the statistics, were then restricted to a few key
characteristics: Federal Land, economic activity, employees, wage bill, hours paid. Enterprises could group all the local units of one Land and economic activity to form a combined report. The statistical offices in the Länder calculated the target characteristics for local units mechanically by proportional distribution of the enterprises’ reports on the basis of the key characteristics. The aspect of the still different labour cost structures , broken down into parts of enterprises in the former Federal territory (including Berlin) and in the new Länder was taken into account in this context by transferring the proportions of the cost types from enterprises with only enterprises in the new Länder and only enterprises in the former Federal territory (including Berlin) to those enterprises which had enterprises both in the new Länder and in the former Federal territory (including Berlin).
The procedure meant that the labour costs of an enterprise with several establishments were ultimately distributed and evened out over the economic activities and the Länder of its local units. The key characteristics selected, however, ensured that, although this did apply for the structure of labour costs, it was hardly the case for their level. Even for the structure, the equalising effect was limited for the most part: the share of an economic activity’s labour costs (D) stemming from reports over 25% of which covered local units of other economic sections was between 1.3% in economic section K and 28.7% in economic section B. The median figure was 6.1%. When the economic activities were broken down in more detail the median proportion was 7.2%. The effects of the estimate procedure are therefore not expected to be critical in virtually any of the economic sections. At most, one might suspect that some adverse effects might arise in economic divisions S91, C27, C37, with proportions of between 26% and 43%.
Estimates of D.1113 payments for days not worked, D.1221* guaranteed remuneration in the event of sickness, D.11111 direct remuneration, bonuses and allowances paid in each pay period
Only a small proportion of enterprises can report these characteristics directly and without extensive calculations. For this reason the questionnaire did not contain characteristics D.1113 and D.1221* but instead all days not worked owing to leave (SUM_U), sickness (SUM_K) and other reasons (SUM_S) totalled for all full-time employees. The total number of public holidays not worked (SUM_F) was estimated mechanically. The figures were surveyed for full-time employees only, because part-time employees could distort the result as a result of the different working-time models. For each local unit the target characteristics D.11111, D.1113 and D.1221* were approximated by dividing the regularly paid salaries and wages (D.11111+D.1113+D.1221*) on the basis of the proportions of worked or non-worked days on all working days of the year (2020: 262 for a five-day week):
D.1113_i = (D.11111+D.1113+D.1221*) * SUM_i / [262*A.11] i = U, F, S
D.1221* describes here the main component of D.1221: guaranteed remuneration in the event of sickness. The second component of D.1221 - the employers’ contribution to maternity benefit - is added later. A.11 describes the annual average number of full-time employees (excluding apprentices).
Estimates of the characteristics of hours actually worked by full-time employees (B.11), by part-time employees (B.12) and by apprentices (B.13)
The characteristics on the number of hours worked always proved to be particularly difficult to record in past surveys, because only some enterprises could extract them from their accounting systems. The reported information on the hours worked was set in the context of calculated limit values. For these, in principle, the Statistical Offices calculated the number of hours actually worked by deducting from the number of paid hours those hours which were spent on paid but not worked days for leave, sickness, public holidays and other purposes (survey SUM_i see above). The days not worked were converted into hours not worked on the basis of the reported weekly working hours. Accumulation or reduction of unpaid overtime which is not recorded by the number of paid hours reported were taken into account, by economic section, by adding estimates from the working time volume calculated by the Institut für Arbeitsmarkt- und Berufsforschung (Institute for Employment Research). In the case of apprentices, an additional 280 hours per person of vocational schooling were deducted, this value being derived from a recommendation of the Cultural Ministers Conference on the design of framework syllabuses.
6.4. Seasonal adjustment
There was no season adjustment.
6.5. Data revision - policy
No preliminary results will be published. Therefore, published data are considered final.
6.6. Data revision - practice
No preliminary results will be published. Therefore, published data are considered final.
6.6.1. Data revision - average size
No preliminary results will be published. Therefore, published data are considered final.
7.1. Timeliness
As the survey collects data for the whole reference year in retrospect, it cannot start until January of the year following the reference year. In view of the many survey characteristics and a large number of queries and corrections, the field and processing phase lasts until March of the second year following the reference year.
7.1.1. Time lag - first result
The first results were published 19 months after the end of the reportingyear (press release 4 July 2022).
7.1.2. Time lag - final result
No preliminary results will be published. Therefore, published data are considered final.
7.2. Punctuality
The fieldwork was conducted by 14 statistical offices in the Länder. The deadlines for the fieldwork phases differed from one office to another. Figure 1 shows the proportion of statistical offices subscribing to the major deadlines. The statistical offices were weighted on the basis of the number of enterprises they surveyed.
The selected enterprises received the survey documents from January 2020 Due to the special Corona situation, general deadlines for response in the individual Land Statistical Offices varied for this survey between March and May. After this, enterprises that had not responded received a series of reminders, formal notices to respond, and administrative offense proceedings. During this time the questionnaires were recorded and checked and consultations were held with enterprises. The plausibility checks and the forwarding of the individual data surveyed to the Statistisches Bundesamt were completed, depending on the office, between December 2021 and March 2022.
The Statistisches Bundesamt worked from December 2021 to May 2021 on producing results for economic sections O, P and Q86.1, for which the characteristics were not surveyed but were derived from other statistics, meeting of the delivery requirements to Eurostat and the methodological
preliminary work for publication (calculating the relative standard error, conducting statistical confidentiality procedures).
The Statistisches Bundesamt forwarded the final data to Eurostat on 30 May 2022, a correction report was made on first August 2022.
7.2.1. Punctuality - delivery and publication
A statistic is punctual if the results are published on the planned and possibly announced date. The results of these statistics were transmitted to Eurostat on time before the statutory deadline (30 June 2022) on 30 May 2022. The national publication of the results also took place on time with the release of the online database and a first press release on 4 July 2022.
8.1. Comparability - geographical
In principle, the data are geographically comparable within Germany and with regard to EU requirements. There may be some shortcomings for economic sections in the public service sector. The results for the economic sections O (in full) and P (in part) are produced by model-based estimate procedures rather than a survey. Inclusion of the economic section O was optional under Council Regulation (EC) No 530/1999 and the option was not taken up by all the Member States. The estimate procedures detract from the geographical comparability of the results for economic sections O and P and led to a certain amount of under-coverage, primarily in section P and with regard to some missing characteristics (see 6.3.1). Both may have led to unquantifiable over- or underestimates of labour costs. In these sections too, no clear distinction could be made between local units and enterprises. All the estimated data were attributed to enterprise size band 1 000.
8.1.1. Asymmetry for mirror flow statistics - coefficient
Not provided.
8.2. Comparability - over time
See below.
8.2.1. Length of comparable time series
The results of the 2020 labour costs survey are basically comparable with the results of the 2008, 2012 and 2016 surveys.
8.3. Coherence - cross domain
For the Labour Cost Survey 2020, Eurostat implemented for the first time standardised cross-domain plausibility checks to be provided by each country. In addition to the comparison of central results of the Labour Cost Survey (LCS) with the
- the Labour Cost Index (LCI)
- the National Accounts (NA),
- the Labour Force Survey (LSF) and
- the Structural Business Statistics (SBS)
also methodological, definitional and delimitation-related explanations for the differences (see Annex Cross-domain plausibility checks of the LCS 2020). There are no approaches for inconsistencies that cannot be explained methodologically and definitively.
See argumentation in Annex Cross-domain plausibility checks of the LCS 2020.
8.5. Coherence - National Accounts
See argumentation in Annex Cross-domain plausibility checks of the LCS 2020.
8.6. Coherence - internal
Internal statistical coherence is the extent to which results for different characteristics are consistent for the statistics described. With one exception, no internal incoherences are known for these statistics. In the data calculations for the majority of public services, it was not possible to determine a size of enterprise comparable to that of private enterprises. The results were uniformly assigned to the highest size class "enterprises with 1 000 or more employees". The results of the economic divisions O 84 and P 85 are comparable with other economic activities in the breakdown by size of enterprise only if this limitation is taken into account.
9.1. Dissemination format - News release
The first results were published at national level on 4 July 2022 in a press release. This constitutes an interval between the reference period and initial publication of 19 months.
9.2. Dissemination format - Publications
The main results at national level were disseminated in the following publication:
Federal Statistical Office: Earnings and labour costs, labour costs in manufacturing and services 2020, Fachserie 16 Hefte 1, 2 und 3, Wiesbaden: 13. Juli 2022.
The publication is available free of charge on the Internet in PDF format on the website of the Federal Statistical Office. (www.destatis.de/publikationen)
9.3. Dissemination format - online database
Since July 2022, Federal Statistical Office provides results of 1992 to 2020 Labour Cost Surveys in its online database Genesis-Online.
9.3.1. Data tables - consultations
Not provided.
9.4. Dissemination format - microdata access
The micro-data of the Labour Cost Surveys are not available.
9.5. Dissemination format - other
Not provided.
9.6. Documentation on methodology
A quality report prepared according to national requirements, the questionnaire and a detailed description of the publication characteristics were published together with the results:
Federal Statistical Office: Earnings and labour costs, labour costs in the manufacturing sector and in the service sector 2020, subject series 20 issues 1, 2 and 3 Wiesbaden: July 2022.
The quality report was included in the collection of all quality reports of the Federal Statistical Office on the Internet (www.destatis.de).
9.7.1. Metadata completeness - rate
Not provided.
9.7.2. Metadata - consultations
Not provided.
According to current measurements taken in 2020, a respondent needs an average of almost 7.5 hours to complete the questionnaire. All respondents taken together, this corresponds to costs of around 11 million euros (at an assumed wage cost rate of 46.03 euros per hour). Measures to relieve the burden on those responsible for providing information: The legal upper limit of 34 000 companies was not exhausted at 32 000. A rotation of the reporting companies was successfully carried out: Only 14.1 % of the companies in the sample had already reported in the previous reporting year 2016. Among small enterprises with 10 to 49 employees, only 0.8 % had already reported four years earlier.
11.1. Confidentiality - policy
The individual data collected are kept secret in accordance with § 16 of the Federal Statistics Act (BStatG). The names and addresses of the respondents will never be passed on to third parties. Pursuant to Section 16 (6) BStatG, it is possible to provide universities or other institutions with the task of independent scientific research with anonymized details for the implementation of scientific projects. The obligation to maintain confidetiality applies to all persons who are recipients of individual data.
11.2. Confidentiality - data treatment
In tables, cells are suppressed if the cell value allows conclusions to be drawn about an individual specification (primary confidentiality). In addition, further cells are suppressed so that primarily secret cell values cannot be revealed by difference formation (secondary confidentiality).
The labour cost survey records the input and all costs of the production factor labour in the calendar year. Input is measured in four different measures: as an annual average of employment and full-time units, as the sum of hours paid and as the sum of hours actually worked. Costs are recorded and broken down in detail. For apprentices, part-time employees and civil servants, labour input and labour costs are recorded separately so that, in principle, they can be shown separately and included or excluded in the statement.
The labour cost survey provides information on the amount and structural composition of the costs per input unit of labour. The most important indicator is the net labour cost per hour worked in the economy as a whole or by sector and size of the enterprise. Other important indicators are the share of non-wage labour costs in labour costs - especially statutory non-wage labour costs - and the number of hours actually worked per full-time employee.
Not Applicable
Labour costs
Labour costs comprise the total of all expenditures incurred by employers in connection with the employment of labour. The definitions in Commission Regulation (EC) No 1737/2005 and the recommendations of the International Labour Organisation (ILO, An integrated system of wages statistics) are taken into account.
Gross labour costs include:
• gross earnings (D.11) and
• all ancillary wage costs, which include
• employers' social contributions (D.12), including the costs of vocational education and training (D.2),
• employers' statutory social security contributions,
• the expenses for the company pension scheme,
• employers' imputed social contributions (continued remuneration, severance payments and pensions and allowances for civil servants),
• Recruitment costs and professional clothing (D.3),
• and taxes and levies on employment (D.4).
Net labour costs (D) are obtained by deducting wage subsidies (D.5) from gross labour costs.
Compensation of employees (D.1), also known from national accounts, is the sum of gross earnings (D.11) and employers' social contributions (D.12).
Labour input
Full-time units (full-time equivalents) and hours worked (B.1) are primarily used as units of measurement for the input of labour. Hours worked are part of the volume of work actually performed by employees. They, therefore, do not include paid hours lost, such as for holidays or sickness. Full-time units are all full-time employees as well as part-time employees, part-time employees and apprentices converted into full-time jobs according to their hours worked.
Survey units (economic sections B to N, Q to S, and economic groups P85.5 and P85.6 of the Classification of Economic Activities, 2008 edition): enterprises.
For Economic Division O84 and Economic Groups P85.1, P85.2, P85.3, and P85.4 of WZ 2008 the data were not collected but calculated.
For economic group Q86.1, the smaller part of the data was calculated (for legally dependent public hospitals), and the larger part was collected.
Companies with 10 or more employees or more in manufacturing and services [Sections Bto S of the Classification of Economic Activities, 2008 edition].
• Germany,
• former federal territory (including Berlin) and new federal states in Germany,
• the 16 federal states of Germany.
Not Applicable
The statistical results are comparatively accurate. On the one hand, the individual data collected are comparatively very accurate if they come directly from the personnel management of the companies, which is subject to internal and external audits. However, this is not the case for all characteristics, for example not always for working time and days lost. These were partly estimated by the enterprises. All data were subject to extensive checks by the statistical offices of the Länder, which demonstrably led to significant improvements in accuracy. Secondly, the sample size is sufficiently large so that the unavoidable random error of the sample is comparatively small, at least for the economy as a whole. There was also an obligation to provide information so that distortions due to selective participation or non-participation can be ruled out.
Not Applicable
Some data are not collected but are calculated (see 6.3.5). In particular
1. Estimates of the results for local units
2. Estimates of D.1113 payments for days not worked, D.1221 guaranteed remuneration in the event of sickness, D.11111 direct remuneration, bonuses and allowances paid in each pay period
3. Estimates of results for sections O (in full), P (by and large), and Q86.1 (in part)
Estimation of results of sections O, P, and Q 86.1 is based on the data of the personnel statistics, were available at the Federal Statistical Office; data collection was not necessary. The supplementary results of the finance statistics were also available at the Federal Statistical Office. The collective bargaining parameters applicable in 2020 (special payments, working hours, holiday entitlement) were taken from collective agreements collected by the Federal Statistical Office for collective bargaining statistics. Data on sick days, social security contribution rates, and other parameters were taken from relevant publications. Aggregated data on supplementary public health care were provided by the supplementary health care institutions. The characteristics were derived from or calculated on the basis of the available characteristics of the headcount statistics. Only a few characteristics of the headcount statistics could be transferred directly. These included the contractual working time, the share of the normal working time of a full-time employee, and the gross monthly earnings in June 2020 (extrapolated to the calendar year 2020). All other characteristics were calculated. The calculations were based on data from finance statistics, collective bargaining parameters (special payments, working hours, holiday entitlement), information on sick days, social security contribution rates and supplementary public provision.
Random Sample of enterprises, maximum 34,000
Not Applicable
As the survey collects data for the whole reference year in retrospect, it cannot start until January of the year following the reference year. In view of the many survey characteristics and a large number of queries and corrections, the field and processing phase lasts until March of the second year following the reference year.
In principle, the data are geographically comparable within Germany and with regard to EU requirements. There may be some shortcomings for economic sections in the public service sector. The results for the economic sections O (in full) and P (in part) are produced by model-based estimate procedures rather than a survey. Inclusion of the economic section O was optional under Council Regulation (EC) No 530/1999 and the option was not taken up by all the Member States. The estimate procedures detract from the geographical comparability of the results for economic sections O and P and led to a certain amount of under-coverage, primarily in section P and with regard to some missing characteristics (see 6.3.1). Both may have led to unquantifiable over- or underestimates of labour costs. In these sections too, no clear distinction could be made between local units and enterprises. All the estimated data were attributed to enterprise size band 1 000.
