Accuracy and reliability
Contact infoLabour and Income, Social Statistics
Flemming von Hadeln Løve
+45 39 17 37 27
Get as PDF
The statistics is mainly based on the Labour Market Accounts (LMA). LMA integrates and harmonizes a wide range of data sources in a statistical system. This means that LMA can illustrate the labour market better than individual statistics can. LMA is at the same time based on a total census of the population, so there is not the same uncertainty as with statistics based on sampling. The quality of the statistics has also been significantly improved by the fact that the projection period has been reduced compared to previous versions.
The margins of statistical uncertainty are related to the statistical uncertainty of the individual primary statistical data that are used.
The statistics is mainly based on the Labour Market Accounts (LMA), which is a longitudinal register based on integration and harmonization of a large number of registers in Statistics Denmark. This means that LMA can illustrate the labour market better than individual statistics can. LMA is at the same time a total census of the population, so there is not the same uncertainty as with statistics based on sampling. Data is available at a very detailed level, which makes the delimitation of the population and concepts very accurate. Against this background, quality is judged to dump other labour market statistics at home and abroad.
The quality of paid hours is not as high as other concepts in the LMA, but relatively high compared with survey-based statements. In contrast to survey-based statements, where they ask the employed themselves how many hours they work, and as a result, among other things, is subject to memory and random sampling, the registry-based working hours are based on the administrative payroll systems that are also reported to the Danish Customs and Tax Administration. Because the information is thus linked to both income and tax payments, they are considered to be relatively high quality. It also means that hours worked are limited to not including illegal activities and unpaid hours of work, which are information that, for natural reasons, is not available in administrative sources. Illegal activities and unpaid hours are information that can only be provided by surveys asking the employed themselves. Thus, this information is subject to both the memory and the willingness of the employed to respond. Also due to sampling errors the information will only be available at a relatively aggregated level.
Information on pay hours of work is from the Employment Statistics for Employees (BfL), where paid hours of work are sometimes imputed and therefore there is uncertainty related to this information. However, the proportion of imputed pay hours for employees has fallen from 2008, where the share was more than 14 per cent to less than 2,5per cent by 2022. There is greater uncertainty about the paid hours of work for self-employed and assisting spouses. This is due to the fact that hourly information for self-employed and co-spouses is imputed on the basis of paid hours for employees from BfL adjusted by how much more self-employed and assisting spouses say they work compared to employees in the labour force survey (LFS). The Working Time Accounts (WTA) converts paid hours from the LMA at job level to hours worked using factors calculated using the annual Structure of Earnings at a more aggregated level. The hourly information is considered to fit an overall level, but at the detailed level they should be interpreted with caution.
Since data includes provisional structural data from the Labour Market Accounts (LMA), the projection period in the WTA has been reduced compared to earlier versions of the WTA so that the maximum length of projection is 18 months. The 18-month projection occurs in the calculation of Q2 in September, while for example the calculation of Q# in December only will be projected for six months. This increases the quality of the statistics considerably compared to previous versions of the WTA. By combining structural and short term statistics, both high-quality data from structural statistics and statements of recent periods using faster short term statistics is obtained. In addition, it contributes to the consistency of the WTA both over time and between variables, that the main source used for projection is employment for employees (BfL), which is also the main source of employees in the LMA.
Not applicable to this statistic as data is not based on samples.
Specific conditions for publication in 2020, 2021 and 2022
The labour market, and hence the Working Time Accounts (WTA), have been strongly affected by the shutdown of society from mid-March 2020 as a result of COVID-19 and by the measures put in place to mitigate the effects of the shutdown. The exceptional circumstances in the latter part of the first quarter of 2020 and up to the start of 2022 mean that the statement of hours worked is subject to greater uncertainty than usual. Similarly, the new holiday law which came into force on 1 September 2020 has caused a change in the pattern of paid hours, particularly for newly recruited officials who, unlike in the past, are entitled to paid holiday from the start of their employment.
Employment: Employment of employees (BFL) is one of the main sources of the Working Time Accounts. The closure and the measures to mitigate its effects affect the employment account, but the impact is limited by the fact that employees temporarily absent from work continue to be included as employed. This also applies to employees who have returned home without working, including employees for whom the enterprise receives compensation of employees.
Hours worked and COVID-19 restrictions: The shutdown and mitigation measures have significantly reduced the number of hours worked. To reflect this, the usual calculation of hours worked has been corrected on the basis of new alternative sources.
Seasonal adjustment: The seasonally adjusted series are generally more uncertain than usual due to atypical labour market developments in 2020, 2021 and 2022 as a result of COVID-19.
In general, the quality of data has improved significantly with the use of the new eIncome sources, and even better with the transition to using AMR. Previously, the ATR was calculated by compiling data from many different sources. After the 2012 changeover, the ATR relies primarily on eIncome sources. Thus, the data basis became the same for most of the sources included in the ATR, ensuring a high degree of internal consistency. The transition to AMR improved accuracy once again by integrating and harmonising a very large number of data sources into one statistical system. This means that the AMR can shed much more light on the labour market than the current stand-alone statistics can, particularly in relation to the periodisation of transitions from one labour market status to another. At the same time, the AMR is a total census of the population and thus does not have the same uncertainty as sample-based statistics. Read more about [precision in AMR]https://www.dst.dk/da/Statistik/dokumentation/statistikdokumentation/arbejdsmarkedsregnskab--amr-).
Statistics Denmark follows the recommendations on organisation and management of quality given in the Code of Practice for European Statistics (CoP) and the implementation guidelines given in the Quality Assurance Framework of the European Statistical System (QAF). A Working Group on Quality and a central quality assurance function have been established to continuously carry through control of products and processes.
Statistics Denmark follows the principles in the Code of Practice for European Statistics (CoP) and uses the Quality Assurance Framework of the European Statistical System (QAF) for the implementation of the principles. This involves continuous decentralized and central control of products and processes based on documentation following international standards. The central quality assurance function reports to the Working Group on Quality. Reports include suggestions for improvement that are assessed, decided and subsequently implemented.
The margins of statistical uncertainty are related to the statistical uncertainty of the individual primary statistical data that are used. The source used absolutely fundamental to describe the level and developments, is the Labour Market Accounts (LMA). The conceptual consistency and the uniform adaptation of sources over time contribute to a reduction of the margins of statistical uncertainty in the Working Time Accounts. Especially, the juxtaposition of information from the primary sources in a joint system implies that the results will automatically be compared and thereby reveal any errors and inherent problems of consistency in the basic concepts and data. These errors and inconsistencies are reported back to the primary sources. The work on integrating statistical systems will thus be instrumental in enhancing the general data quality of the primary statistical data.
For a description of the statistical uncertainty of the primary sources, see the respective Documentation of statistics: 1. Labour Market Accounts see Documentations of statistics LMA 2. Structure of Earnings see Documentations of statistics SES 3. A-Income Statistics see Documentations of statistics AINCOME 4. Employment Statistics for Employees see Documentations of statistics BfL.
Self-employed and assisting spouses make up the group for whom the lowest quality of data is available on number of jobs (with activity over 1 weekly working hour), length of job (duration of each individual job) and number of hours worked in each individual job in the data sources. The consequence is that the information on employment, jobs and hours for self-employed and assisting spouses are subject to a relatively greater degree of uncertainty than it is for employees.
When the Labour Force Survey (LFS) is applied in the Labour Market Accounts (LMA) in adjusting how many more hours are worked by self-employed and assisting spouses compared to hours worked by employees, it is impossible for us to take into account that there is a tendency for self-employed to overstate, to a greater extent, than is the case for employees. It would be extremely subjective, if we were to introduce a factor for the extent of this overstatement made by the self-employed compared to the employees. However, we have an assumption that this overstatement is greatest in cases where the workplace for the self-employed (and the assisting spouse) is the residence of the self-employed, as it must be assumed that the relation between working time and leisure time becomes more blurred. This applies, especially, to employment in agriculture, etc. and small businesses in retailing and hotels and restaurants.
In eIncome information is reported with regard to paid hours in the jobs in the individual reference month. This information is the primary source on paid hours of work in the LMA. The quality of this information naturally reflects the quality of the data reports. Generally, we think that the information has a high quality. However, particularly related to the data on unpaid hours of absence and overtime hours there may be quality problems in the primary data of the eIncome register. Some problems have been revealed with respect to data reports of paid hours for employees paid by the month who are not paid for in periods of absence. Lack of impairment of hours paid for as a result of unpaid absence leads to an overestimation of paid hours. Lack of registration of paid overtime will result in an underestimation of hours paid. Typically, the registration problem worse related to hours of unpaid absence. There can be, especially for salaried who have not paid absences be a problem in that the hours are not always written down sufficiently during periods of absence.
There is associated uncertainty in relation to developments in hours actually worked over the year. In the Danish Working Time Accounts (WTA) the development is identified by paid hours for hourly workers, who have not paid absence. As described in the section on sources, in the calculation of hours actually worked, conversion factors are used which are calculated as the relative distribution of hours actually worked compared to paid hours over the months of the year. The development in the ratio between hours actually worked and paid hours over the year is calculated on 19 industry groups. The assumption here is that the time distributions over the months of the year for hourly workers in each of the 19 industry groups represents the hour distribution for all the employed in the same industry groups. This information is used to convert reports of paid hours from eIncome over the year to monthly information on hours actually worked over the year in the WTA. Because of compensated absences as holidays and sickness etc., the distribution on paid hours and hours actually worked is not the same throughout the year.
In the absence of structural statistics (SES and LMA for 2018), the relative distribution of hours actually worked compared to paid hours over the year's months is based on previous year's allocations. This implies an inevitable uncertainty. Easter has significant impact on the distribution of hours actually worked between the first and the second quarter.In 2018 the Easter holidays fell primarily in March, but a few fell in the month of April. The only year in the time series, where Easter fell approximately in the same way is 2013. However, there was a lockout in April 2013 in public line of industries. This is why 2016 has been used to distribute hours worked for public industry group activities (NACE groupings Public administration and defence compulsory social security, Education and Human health and social work), as Easter fell in March in 2016. The actual distribution of hours worked between the first and second quarter very much depends on how the persons in employment take their Easter break up to and after Easter. In September 2019, hours actually worked will be published in the Working Time Accounts, in which data based on Structural earnings 2018 and The Labour Market Accounts for 2018 have been incorporated.
Data revision - policy
Statistics Denmark revises published figures in accordance with the Revision Policy for Statistics Denmark. The common procedures and principles of the Revision Policy are for some statistics supplemented by a specific revision practice.
Data revision practice
The Quarterly Working Time Accounts (WTA) will be published in accordance with Statistics Denmark's specialized goals of timeliness, which with respect to the quarterly statistics implies not later than by the end of the subsequent quarter.
Data on employment for employees is available both in a provisional version as well as a final version, so the two most recent quarters will be subject to revision in conjunction with the compilation of each quarterly statistic.
When new structural data are incorporated (the Labour Market Accounts and Structural Statistics on Earnings) in connection with the compilation of the Annual Working Time Accounts, the levels as from this year will be revised, i.e. also during the entire period of projection. When final structural data are incorporated, the data in the Working Time Accounts are considered to be final.
However, data in the Working Time Accounts can be subject to revision as a result of updated values in the primary sources, in the case of methodological changes or use of new information and sources.