EUROPEAN COMMISSION

Brussels, 28.1.2019

COM(2019) 14 final

REPORT FROM THE COMMISSION TO THE EUROPEAN PARLIAMENT AND THE COUNCIL

on the implementation of Council Regulation (EC) No 577/98

1.Introduction

Council Regulation (EC) No 577/98 on the organisation of the labour force sample survey in the Community 1 sets minimum standards for the harmonised production of labour force surveys in the European Union (EU) and its Member States.

Article 7 of Regulation (EC) No 577/98 states that:

A report on the implementation of this Regulation shall be submitted by the Commission to the Parliament and the Council every three years, beginning in the year 2000. This report shall evaluate in particular the quality of the statistical methods envisaged by the Member States to improve the results or to lighten the survey procedures.

This report is the Commission’s seventh report to Parliament and the Council; it covers the period 2015-2017 and is structured as follows.

·Section 2 gives an overview of the European Union Labour Force Survey and its role in supporting European Union policies.

·Section 3 takes stock of how Regulation (EC) No 577/98 has been implemented in the Member States, candidate countries, and European Free Trade Association (EFTA) countries.

·Section 4 focuses on ongoing initiatives launched by the European Statistical System 2 (ESS) to improve the quality of the Labour Force Survey.

2.Overview of the labour force survey

2.1.General description

The European Union Labour Force Survey (hereinafter referred to as EU-LFS) is a large sample survey of residents in private households. It provides both quarterly and annual labour market statistics on employment and unemployment, as well as on people outside the labour force. It also collects multi-annual information from ad hoc modules, and provides input for model-based monthly estimates of unemployment and unemployment rates.

The National Statistical Institutes (NSIs) of the respective countries are responsible for: designing national questionnaires, drawing the sample, conducting interviews, and sending results to the Commission (Eurostat) in accordance with a common coding scheme set by Commission Regulation (EC) No 377/2008 3 .

Eurostat is in charge of: monitoring the implementation of Regulation (EC) No 577/98; providing assistance to NSIs; promoting harmonised concepts and methods; and disseminating comparable national and European labour market statistics.

In 2017, the EU-LFS sample size was around 1.3 million people (of which around 1.2 million were in the EU-28) aged 15-74 years. The samples were taken every quarter and covered 34 participating countries 4 . This makes the EU-LFS the largest household survey in Europe.

2.2.Relevance

The EU-LFS is the largest source of official statistics on the labour markets in the EU. In total, it collects more than 100 variables on labour status; employment characteristics; working time; educational background; and training. The EU-LFS is also complemented by thematic, yearly ad hoc modules 5 .

Based on international standards and definitions 6 , the EU-LFS has relevance beyond the EU, allowing the European labour market to be compared with labour markets in other areas or countries.

Some key EU policy initiatives rely on EU-LFS data. For instance, the EU-LFS is one of the main data sources for monitoring Member States’ progress under the employment guidelines in the context of: (i) Article 148 of the Treaty on the Functioning of the European Union 7  (TFEU), and (ii) the EU’s ‘Europe 2020’ growth strategy. Three EU-LFS indicators (the employment rate among those aged 20-64; the share of early leavers from education and training; and the number of 30-34 year olds that have completed tertiary education) are used to monitor 2 of the 5 Europe 2020 headline targets 8 . Many other LFS-based indicators are used under the Europe 2020 Joint Assessment Framework.

Several indicators provided by the EU-LFS are also used to monitor Member States’ performance under two dimensions of the European Pillar of Social Rights. These dimensions are Equal opportunities and access to the labour market (which covers education, skills and lifelong learning; gender equality in the labour market; and youth) and Dynamic labour markets and fair working conditions (which covers labour force structure and labour market dynamics) 9 .

In addition, a number of labour market indicators at the regional level form the EU-LFS (i.e. youth unemployment rate and total unemployment rate) are used for the purpose of allocating funds in the context of the Cohesion Policy.

The EU-LFS also provides indicators for the 2030 Agenda for Sustainable Development 10 . Three of the goals in the Agenda for Sustainable Development encompass indicators stemming from the EU-LFS: quality education, gender equality, and decent work and economic growth.

The EU-LFS-based monthly unemployment rate is an important short-term economic indicator. It is one of the Principal European Economic Indicators 11 , and is also used to build moving annual averages of unemployment rates for the scoreboard of economic and financial indicators. This scoreboard is in turn used to detect macroeconomic and competitiveness imbalances 12 . The EU-LFS provides input for national accounts on employment, working time, and other areas beyond the labour market, such as education.

EU-LFS data are widely used by the Commission DGs. They are also widely used by European Agencies. Eurofound in particular uses the data for its overview report on the characteristics of the workforce in Europe.

Finally, the EU-LFS is one of the largest sources of statistical microdata for researchers in Europe. The large sample allows studies to be carried out on specific groups in the labour market. The EU-LFS also has a very broad coverage of demographic, regional and educational-background variables.

3.Implementation of the labour force survey

This section mainly covers the period 2015-2017, but to highlight improvements since the previous report, information on the 2014 Labour Force Survey is also presented when relevant.

3.1.Accuracy

The accuracy of statistical outputs is the degree to which computations or estimates are close to the exact — or true — values that the statistics are intended to measure. The accuracy of the estimates is subject to two types of errors: sampling errors and non-sampling errors. Sampling errors occur because only a subset of the whole population is surveyed. Non-sampling errors are all other errors that are not the result of sampling (e.g. coverage errors, measurement errors, processing errors, non-response errors).

For sampling errors, the size of the sample is the main determinant. Larger samples improve the accuracy of the results. Table 1 shows the size of the sample of persons aged 15-74 interviewed each quarter in each participating country, in absolute terms and compared with the total population of that age group. The sampling rate 13 ranged in 2017 from 0.14 % (the United Kingdom) to 1.46 % (Luxembourg). On average, the quarterly sample size was about 1.3 million people in 2017, which amounts to 0.29 % of the total population aged 15-74 in the 34 countries taking part.

Official surveys such as the EU-LFS use probability sampling. This makes it possible to quantify sampling errors in terms of confidence intervals. Table 2 provides the estimates and 95 % confidence intervals reached for the seven main indicators on employment and unemployment at aggregated EU-28 level.

These intervals reflect the precision requirements set in Council Regulation (EC) No 577/98.

For non-sampling errors, Eurostat and participating countries regularly monitor relevant sources of non-sampling errors like non-response. Participating countries report the main causes of errors to Eurostat on an annual basis, along with the methods applied nationally to reduce them. A summary of this monitoring is published in the annual quality report of the EU-LFS 14 .

In 2017, participation in the EU-LFS was compulsory in 14 countries (Belgium, Germany, Greece, Spain, France, Italy, Cyprus, Luxembourg, Malta, Austria, Portugal, Slovakia, Norway and Turkey) and voluntary in the other 20 countries. Countries conducting a voluntary survey generally have a higher unit non-response rate than countries conducting a compulsory survey 15 .

Table 1: Achieved sample size and percentage of the population aged 15-74 (quarterly average) in the European Labour Force Survey, 2017 and 2014

Table 2: 95 % confidence interval of the main LFS indicators on employment and unemployment for the EU-28, 2017

Table 3: Participation and non-response rate of the EU-LFS by country, 2017 and 2014

3.2.Timeliness and punctuality

The timeliness of statistics is defined as the length of time between data availability and the event or phenomenon they describe. For the EU-LFS, this length of time depends on: (i) the time Member States need to conduct and process the survey and send the results to Eurostat, and (ii) the time needed by Eurostat to process, validate and publish the results.

Council Regulation (EC) No 577/98 states that Member States must deliver data to Eurostat within 12 weeks of the end of a reference quarter. Eurostat validates the national datasets as soon as they are received. Table 4 shows the time span between the end of the reference quarter and the availability of data for Eurostat’s users for the period 2014-2017. Timeliness of EU-LFS data has slightly improved for the EU-28 in recent years. In 2017, the average number of calendar days needed to disseminate the EU-28 data was 75 days, compared to 77 days in 2014. Eurostat plans to further improve the timeliness of dissemination of EU-LFS results by shortening the deadline for data delivery to Eurostat in the future.

The timeliness of EU-LFS data is also important for estimating monthly unemployment rates. As soon as national data are processed by Eurostat, they are used to calculate the next monthly unemployment rate. The monthly unemployment rate is disseminated approximately 30 days after the end of the reference month.

Table 4: Availability of LFS data to users — Timeliness, 2014-2017

The punctuality of statistics is defined as the time lag between the announced release date and the actual release date. Eurostat publishes the planned quarterly and annual release dates of the main EU-LFS indicators. In 2015-2017, all the announced publication deadlines were met.

3.3.Accessibility and clarity

Eurostat disseminates EU-LFS statistics in several ways. The main tool for dissemination is Eurostat’s online database, which contains more than 400 tables of detailed EU-LFS data (quarterly, annual, household and ad hoc module results). The policy indicators described in paragraph 2.2 are also published in specific sections of the website 16 ; the tables in these sections cover only the most important EU-LFS results. Eurostat produces other combinations of EU-LFS variables as tailor-made tabulations on demand from users.

EU-LFS microdata are also highly relevant for scientific purposes. They are requested by growing numbers of researchers from universities, research institutes and national statistical institutes across Europe and beyond. Eurostat has been providing microdata free of charge since 2011. Access is granted in accordance with the terms set out in Commission Regulation (EC) No 557/2013 17 to ensure the protection of survey respondents’ personal data 18 . The data files, consisting of individual records, do not contain direct identifiers such as names, addresses or ID numbers. Moreover, anonymisation rules are applied.

Detailed documentation (metadata) is provided to users through various dissemination channels. General information is disseminated to the public at large on the EU-LFS section of the Eurostat website. More detailed information can be accessed in the website’s ‘Statistics Explained’ section 19 . This information is updated regularly. Specific information on the content and quality of statistics is released in two annual publications: a methodological publication describing the characteristics of the national surveys, and an annual quality report summarising the results achieved by the EU-LFS 20 . In addition, specific metadata is attached to the EU-LFS statistics in Eurostat’s online database. Since 2014, information from the National Quality Reports of each participating country has also been published online.

Countries regularly improve their LFS, either in its methodology or processes. If these improvements lead to breaks in EU-LFS data, the NSIs involved inform Eurostat. Significant breaks are documented in Eurostat publications 21 . Eurostat also publishes a special dataset called ‘LFS main indicators’, in which previous series are adjusted for breaks and occasional data gaps are filled.

3.4.Comparability

The EU-LFS draws on highly harmonised concepts, definitions, classifications and methodologies. Regulation (EC) No 377/2008 sets out a common coding scheme, ensuring that all participating countries use the same variable definitions. Common classifications are used (e.g. the European standard classification of productive economic activities (NACE) and the International Standard Classification of Occupations (ISCO)), and whenever these classifications are revised, Eurostat ensures that all participating countries coordinate implementation. In addition, Eurostat issues explanatory notes which provide detailed guidelines to national statistical institutes on how to code and implement statistics. The explanatory notes also explain Eurostat’s reasoning for each of the guidelines. Methodological issues are discussed in a dedicated working group, the Labour Market Statistics Working Group. This group fosters the exchange of experience and common practices among all participating countries.

To ensure that the measurement of unemployment is harmonised across all countries participating, Commission Regulation (EC) No 1897/2000 22 lays down an operational definition of unemployment. It also lays down principles to be followed for formulating the survey questions on labour status. The definition of unemployment is consistent with International Labour Organisation (ILO) standards, which were adopted by the 13th and 14th International Conferences of Labour Statisticians 23 . This ensures that EU-LFS statistics are comparable to those from other countries, especially those of other OECD members.

3.5.Coherence

Consistency between population estimates based on EU-LFS and demographic statistics is an important aspect of overall quality. Based on a sample survey, the EU-LFS’s results are calculated from the responses of a subset of the population. The responses are then benchmarked against the entire population. Data for the entire population is based on the best estimates available at the time (broken down by sex and age groups to improve the accuracy of the procedure). In principle, this benchmarking procedure ensures consistency between EU-LFS and demographic statistics. However, differences may occur in exceptional circumstances. For instance, every 10 years, new population census results become available. If a new census produces figures that differ from previous population estimates, a past series may have to be revised. In such a case, the revision policy for demographic statistics and the EU-LFS may differ in length and timing, constrained by the lack of detailed information on the population for the 10 years between censuses.

Most of the countries participating in the EU-LFS carried out a population census in the 2011 round. New census results often lead to new LFS weights, sample frames and/or sample designs. All NSIs in countries where censuses were held in 2011 have checked, and revised when needed, the LFS weights to reflect new population estimates, including re-weighting of back-data series, at least back to 2010.

On the consistency of unemployment estimates, many countries also publish statistics on the number of people registered at the public employment office and looking for work. The LFS unemployment figures and the number of registered job seekers differ because of the different nature of the data collected. While the EU-LFS draws on a harmonised methodology to survey households on their employment activities and availability for work, the public employment offices’ administrative records comprise an exhaustive list of people registered and eligible for unemployment benefits. Because the criteria for registration depend on national social policies, statistics on registered job seekers are not comparable across countries or over time.

Another area in which statistical consistency is important is in comparing employment estimates produced from the LFS and from national accounts, as the estimates are not necessarily the same. This is due to differences in both methodologies (the concepts and coverage of the population are not identical) and compilation processes. National accounts are compiled by comparing and combining all relevant data sources available in a country, and taking the best from each source to obtain comprehensive results. National accounts also seek consistency between measures of employment and production (Gross Domestic Product, GDP). The LFS is one data source in that process. Other data sources in the process include business surveys, employment registers or social security registers. Consistency between the LFS and national accounts has been addressed by the Labour Market Statistics Working Group. Eurostat assesses the differences between the two estimates, and several NSIs have analysed the causes and the size of the differences. In some cases, NSIs have published reconciliation tables between the two datasets.

A comparison between the LFS and national accounts shows that the LFS is more suitable for measuring participation in the labour market (unemployment and activity) or for analysing the situation of specific socioeconomic groups in the population (e.g. by age, gender or educational level). National accounts data are more suitable for comparing employment estimates with aggregate production and income measures at macro level.

4.Initiatives to further improve the Labour Force Survey

4.1.The LFS in a modernised system of social statistics

Building on the Commission’s ‘Communication on the production method of EU statistics: a vision for the next decade’ 24 , the Commission adopted on 26 August 2016 a proposal for a regulation of the European Parliament and of the Council for European statistics relating to persons and households, based on data at individual level collected from samples (the Integrated European Social Statistics (IESS). framework regulation) 25 .

This initiative aimed to modernise European social statistics. Indeed, the proposed framework regulation will increase the comparability and consistency of European social statistics by bringing together seven existing household surveys that are currently carried out in the EU, and harmonising the definitions, concepts and variables that are common to two or more surveys. This will facilitate joint analysis of social phenomena. The proposed framework regulation is expected to lead to a first data collection in 2021.

The EU-LFS is one of the household surveys covered by the IESS framework regulation, and it is being modernised to comply with the framework regulation. This modernisation will align the EU-LFS with the 19th International Conference of Labour Statisticians (ICLS) resolution adopted in October 2013, with a more restricted scope for employment and more precise criteria for parental leave and seasonal work. The modernisation will also improve the relevance of the LFS for: (i) migration analysis and (ii) comparability of information on working time and monthly pay from people’s main job. In addition, the modernisation of EU-LFS will lead to better coverage of: (i) employment (dependent self-employment), (ii) general health information (to compare the labour market situation of disabled people with that of the total target population), and (iii) participation in education and training (lasting at least 12 months).

Furthermore, the revision of the EU-LFS will improve the timeliness of data transmission to Eurostat, revise the precision requirements, and further develop methodological improvements.

4.2.Extending the EU-LFS to additional candidate countries

Currently, 34 countries (the 28 EU Member States, 3 candidate countries and 3 EFTA countries) participate in the EU-LFS. This means that they send data to Eurostat in line with Council Regulation (EC) No 577/98 and that their data are regularly disseminated by Eurostat. The candidate countries already included are Montenegro, the former Yugoslav Republic of Macedonia, and Turkey. Eurostat works closely with the statistical offices from other candidate countries to help them progress towards compliance with Council Regulation (EC) No 577/98 in survey content and quality. Serbia and Albania have already started sending LFS data to Eurostat. As soon as the national surveys of these countries meet the requirements of the European regulations, their data will also be disseminated by Eurostat.

4.3.Developing new statistical products

Statistics to capture labour market dynamics

Changes in the labour status of individual people are of great interest, because they provide additional information on the general situation of the labour market, including its flexibility or rigidity. These changes also provide information on the situation of specific groups. For example, data on labour market dynamics make it possible to analyse how many people became unemployed or found a job over a certain period, and who these people are. This in turn makes it possible to identify groups which have a better chance of finding a job, or groups which are more at risk of becoming or remaining unemployed.

Quarterly labour market analyses showing the movements between employment, unemployment and economic inactivity were first published in 2015. Additional results, giving breakdowns of these movements by sex, age group, unemployment duration, or job-to-job transitions, are published since 2017 as experimental statistics.

Eurostat and national statistical institutes are currently working on developing a methodology to estimate annual labour market transitions, tackling the challenges posed by high attrition rates (decreasing sample sizes across interview waves) as well as population changes over time.

Extension and preparation of the EU-LFS main indicators with the future legal framework

The LFS main indicators are a collection of the most important EU-LFS results disseminated in Eurostat’s online database. Their aim is to provide users with seasonally adjusted time series of the most important indicators describing the labour market (e.g. employment, unemployment and activity rates). These highly relevant indicators have been improved over time and extended to include indicators covering groups such as NEETs (young people not in employment, education or training).

Eurostat and the Labour Market Statistics Working Group have developed a strategy to tackle the breaks in time series caused by the entry into force of the new framework regulation. This strategy will ensure the publication of break-free main indicator series. It requires a duplication of the data, allowing Eurostat to derive back-series aligned with the new LFS in parallel with national dissemination. Due to: (i) the large differences in approaches used by countries to derive information on the breaks in series, and (ii) the large number of indicators that must be derived based on relatively scarce information, Eurostat has already begun working on all implementation steps of this project.

Additional information on the accuracy of EU-LFS results

Surveys such as the EU-LFS provide estimates of indicators on a target population based on a randomly selected sample of that population. As only a subset of the target population is selected, it is important to provide, alongside the estimate, information on the estimate’s accuracy (see paragraph 3.1). Such quality indicators take the form of standard errors, or confidence intervals.

Eurostat and the Member States have produced the variance of annual net changes for a subset of 23 indicators of high policy relevance (as described in paragraph 2.2). This allows users, when disseminating and analysing these EU-LFS indicators for policy purposes, to supplement the information on indicator changes between two consecutive years with information on the statistical significance of these indicator changes.

5.Conclusion

The Commission monitors Member State compliance with Regulation (EC) No 577/98. The Commission considers the implementation of EU-LFS to be very satisfactory, as Member States are fully — or almost fully — complying with this European legislation. Unresolved issues are discussed with Member States, and, if necessary, action plans are jointly drawn up. The overall quality of the EU-LFS is good.

The European Statistical System works to continuously improve EU-LFS processes and methods. It has kept up the momentum of these efforts despite a difficult environment, with scarce resources and deep budget cuts. Constant improvements to EU-LFS are under way, either as part of the modernisation process of social statistics, or as individual EU-LFS projects for adapting to changes in user needs and new challenges (e.g. increased timeliness requirements, estimates on annual labour market transitions). This work will continue in the coming years.

(1)    OJ L 77, 14.3.1998, p. 3.

(2)    According to Article 4 of Regulation (EC) No 223/2009 on European statistics (OJ L 87, 31.3.2009, p. 164), the ESS is the partnership between the European statistical authority (which is the European Commission (Eurostat)), the national statistical institutes, and other national authorities responsible in each Member State for the development, production and dissemination of European statistics.

(3)

   Commission Regulation (EC) No 377/2008 of 25 April 2008 implementing Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community as regards the codification to be used for data transmission from 2009 onwards, the use of a sub-sample for the collection of data on structural variables and the definition of the reference quarters (OJ L 114, 26.4.2008, p. 57).

(4)    Participating countries are the 28 Member States, Iceland, Norway, Switzerland, Montenegro, the former Yugoslav Republic of Macedonia and Turkey.

(5)    In 2015, no ad hoc module was included in the EU-LFS. The ad-hoc module in 2016 was Young people on the labour market. The ad hoc module in 2017 was Self-employment.

(6)    In particular, the EU-LFS follows the resolutions and recommendations of the International Conferences of Labour Statisticians organised by the International Labour Organisation, see
http://www.ilo.org/global/statistics-and-databases/standards-and-guidelines/resolutions-adopted-by-international-conferences-of-labour-statisticians/lang--en/index.htm (available only in English, French and Spanish).

(7)    See Council Decision 2014/322/EU of 6 May 2014 on guidelines for the employment policies of the Member States for 2014, (OJ L 165, 4.6.2014, p. 49).

(8)    Europe 2020 headline targets: http://ec.europa.eu/eurostat/statistics-explained/index.php/Europe_2020_headline_indicators .

(9)     http://ec.europa.eu/eurostat/web/european-pillar-of-social-rights/indicators (available only in English, French and German).

(10)     https://ec.europa.eu/europeaid/policies/sustainable-development-goals_en

(11)    Principal European Economic Indicators — A statistical guide, Eurostat 2009, see http://ec.europa.eu/eurostat/en/web/products-statistical-books/-/KS-81-08-398 (available only in English).

(12)    The MIP Scoreboard is used to identify emerging or persistent macroeconomic imbalances in a country. It is part of an annual procedure, where the first step is the compilation of an Alert Mechanism Report (AMR). See Regulation (EU) No 1176/2011 of the European Parliament and the Council (OJ L 306, 23.11.2011, p. 25).

(13)    The sampling rate is defined as the ratio of the sample size (the number of sampling units in the sample) to the population size (the total number of sampling units in the target population).

(14)     http://ec.europa.eu/eurostat/web/lfs/publications/quality-reporting (available only in English).

(15)    Unit non-response occurs when no data are collected about a population unit designated for data collection. The unit non-response rate is the ratio of the number of units for which no data has been collected to the total number of units designated for data collection.

(16)    Example for EU2020: http://ec.europa.eu/eurostat/web/europe-2020-indicators/europe-2020-strategy/headline-indicators-scoreboard (available only in English, French and German).

(17)    OJ L 164, 18.6.2013.

(18)    See Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/46/EC (General Data Protection Regulation) (OJ L 119, 4.5.2016, p. 1).

(19)     http://ec.europa.eu/eurostat/web/lfs/overview (available only in English, French and German).     http://ec.europa.eu/eurostat/statistics-explained/index.php/EU_labour_force_survey (available only in English).

(20)     http://ec.europa.eu/eurostat/web/lfs/publications/quality-reporting (available only in English).

(21)    The documentation on breaks in series in the EU-LFS is available (only in English) at: http://ec.europa.eu/eurostat/statistics-explained/index.php/EU_labour_force_survey (Chapter 1).

(22)

   Commission Regulation (EC) No 1897/2000 of 7 September 2000 implementing Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community concerning the operational definition of unemployment (OJ L 228, 8.9.2000, p. 18).

(23)     http://www.ilo.org/global/statistics-and-databases/standards-and-guidelines/resolutions-adopted-by-international-conferences-of-labour-statisticians/lang--en/index.htm The implementation of the newly adopted ILO definitions of employment and unemployment by the 19th International Conference of Labour Statisticians is described in Chapter 4 (available only in English, French and Spanish).

(24)    COM(2009) 404 of 10 August 2009.

(25)    COM(2016) 551.