Eurostat, the statistical office of the European Union

F3: Labour market and lifelong learning

2920 Luxembourg LUXEMBOURG

Flow statistics are experimental statistics derived from the longitudinal component of the EU-LFS data. They identify the flows between different labour market statuses between consecutive quarters.

Flow statistics are published in the section 'LFS main indicators', which is a collection of the main statistics on the labour market derived from the EU-Labour Force Survey (EU-LFS). However, the flow indicators are calculated with special methods which justify the present page.

Please note that countries may publish nationally slightly different results due to the use of more sophisticated methods.

This page focuses on the particularities of the estimation of flow statistics. Other information on 'LFS main indicators' can be found in the respective ESMS page, see link in section 'related metadata'.

General information on the EU-LFS can be found in the ESMS page for 'Employment and unemployment (LFS)', see link in related metadata. Detailed information on the main features, the legal basis, the methodology and the data as well as on the historical development of the EU-LFS is available on the EU-LFS (Statistics Explained) webpage.

The 'LFS main indicators' are produced in accordance with the relevant international classification systems.

For more details on classifications, levels of aggregation and transition rules, please consult EU-LFS (Statistics Explained) - Methodology.

Not applicable for the Flows statistics.

Flow statistics quantify the quarter-on-quarter transitions between the labour market states of unemployment, employment and out of labour force.

The definitions of employment and unemployment, as well as other survey characteristics follow the definitions and recommendations of the International Labour Organisation (ILO).

Employed persons are all persons who worked at least one hour for pay or profit during the reference week or were temporarily absent from such work or who found a job to start later, i.e. within a period of at most three months from the end of the reference week.  Figures show the number of persons unemployed in thousands.

Unemployed persons are all persons 15 to 74 years of age (16 to 74 years in ES, IT and the UK) who were not employed during the reference week, had actively sought work during the past four weeks and were ready to begin working immediately or within two weeks. Figures show the number of persons unemployed in thousands. The duration of unemployment is defined as the duration of a search for a job or as the length of the period since the last job was held (if this period is shorter than the duration of search for a job).

Out of labour force (previously: inactive persons) are all persons who were neither unemployed nor employed during the reference week. Figures show the number of persons unemployed in thousandsFor flow statistics, 9 different transitions between the labour market status of unemployment, employment and inactivity between any two consecutive quarters (referred to as initial and final quarter respectively) are calculated; each transition is published under the heading of the final quarter. Individuals used for the calculation of these transitions were surveyed in two consecutive quarters and 15-74 years old in both these quarters.

Transition rates between two labour market states are calculated as the share of a transition in % of the labour market status in the initial quarter, e.g. the transition rate for the flow between unemployment to employment is calculated as:

100*U_E/(U_E+U_U+U_I), which is equivalent to 100*U_E/(Unemployemt of initial quarter).

The relevant codes for transition rates are

For the tables on breakdowns of specific transitions, the following, additional, principles apply:

• All tables are published with the label "experimental statistics". The dedicated page on experimental statistics published by Eurostat in general, and for labour market flows in particular, can be found here.
• All results are estimates performed by Eurostat, derived from a simple regression model.  Information on the methodology used can be found here.
• Breakdowns published by countries on their national website may differ due to differences in methodology. National statistical institutes are usually better placed to tailor the methodology to the data situation in their county.
• The data refer to annual averages of quarterly transitions; the pooling of quarterly data is necessary due to the relatively small relevant sample sizes in most countries.
• Small discrepancies for the same breakdown found in different tables may arise, depending on the covariaties used in the regression as well as due to rounding of results.
• All results refer to transition probabilities, and are rounded to full percentage points -  estimates of levels are not available. Estimates which are statistically insignificantly different from zero (at 99% probability) are not shown.

For more details on LFS data in general, please consult the EU-LFS (Statistics Explained) - Methodology.

Persons.

The EU LFS results cover the total population usually residing in Member States, except for persons living in collective or institutional households. While demographic data are gathered for all age groups, questions relating to labour market status are restricted to persons in the age group of 15 years or older. For more details and exceptions, please consult please consult the EU-LFS (Statistics Explained) - Methodology.

European Union, Euro area, EU Member States, Candidate Countries, EFTA Countries (except for Liechtenstein). Data for Cyprus refer only to the areas of Cyprus controlled by the Government of the Republic of Cyprus. Data for France do include the overseas departments (DOM) from 2014 on.

Available data ranges differ by country and indicator. For reasons of comparison over time, the calculation of flow statistics are at this point restricted to begin in 2010Q2.

Not applicable. 

The EU-LFS implementation is governed by legislative acts of the Council and Parliament, as well as of the Commission. The principal legal act is the new Integrated European Social Statistics Framework Regulation (Regulation (EU) 2019/1700) and its Implementing Regulation (EU) 2019/2240 for the labour market domain entering into force in 2021. This is the main regulation with provisions on design, survey characteristics and decision making processes. For more details on the regulations, please consult EU-LFS (Statistics Explained) - Main features and legal basis.

Not applicable

Regulation (EC) No 223/2009 on European statistics (recital 24 and Article 20(4)) of 11 March 2009 (OJ L 87, p. 164), stipulates the need to establish common principles and guidelines ensuring the confidentiality of data used for the production of European statistics and the access to those confidential data with due account for technical developments and the requirements of users in a democratic society.

EU-LFS microdata as received by Eurostat from the national statistical institutes does not contain any administrative information such as names or addresses that would allow direct identification. Access to this microdata is nevertheless strictly controlled and limited to specified Eurostat staff. After data treatment, records are aggregated for all further use.

The domain 'LFS main indicators' are bound by a release calendar. For further details, please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

Flow statistics are released at the same time as the quarterly LFS data under the "Main Indicator" heading. Dates are announced in the Eurostat release calendar.

For the domain 'LFS main indicators', the precise date of data release is disseminated on Eurostat's website.

In line with the Community legal framework and the European Statistics Code of Practice Eurostat disseminates European statistics on Eurostat's website (see item 10 - 'Accessibility and clarity') respecting professional independence and in an objective, professional and transparent manner in which all users are treated equitably. The detailed arrangements are governed by the Eurostat protocol on impartial access to Eurostat data for users.

In line with this protocol and on a strictly regulated basis, data on unemployment are sent for information to the European Central Bank (ECB) and to the European Commission Directorate General for Employment, Social Affairs and Equal Opportunities (DG EMPL) under embargo the evening before official release of data.

The release format is a regular news item.

Release dates of quarterly Flows follow the LFS Main Indicators release calendar (see here).

Dedicated page in "Statistics Explained" 

Please consult free data on-line or address to ESTAT-LFS-USER-SUPPORT@ec.europa.eu

Micro-data are not applicable to 'LFS main indicators' results, but EU-LFS anonymised microdata are available for research purposes. Please refer to access to microdata. Please note that the microdata does not allow the linking of microdata over time, so the flow statistics cannot be reproduced.

See: http://ec.europa.eu/eurostat

Please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

For a detailed description of methods and concepts used, as well as for other documents related to the EU-LFS, please consult the  EU-LFS (Statistics Explained) webpage.

The EU-LFS disseminates also publications on the methodology of the survey. For more information please consult: Quality reports and methodological publications.

See section 11.1.

Please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata') concerning the quality assurance of the underlying LFS microdata. Flow statistics are calculated based on microdata according to an agreed methodology. Before the first publication, flow statistics result had been checked by all affected countries. Each quarter, new or revised data is checked by Eurostat for internal consistency before publication.

Please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

Relevance is high. Unemployment  and employment statistics are among the most important socio-economic indicators. Flow statistics complement the already existing indicators and give valuable insights into labour market dynamics. They are also part of the European Statistical Recovery Dashboard. 

This is a new data collection, so there is no information on user satisfaction yet.

Quarterly flow statistics cannot be calculated for Germany before 2021,  for Belgium before 2017,  and Luxembourg before 2015, due to the missing overlap in the quarterly samples. EU  aggregates are available for quarterly and annual flows statistics, but levels are published for the quarterly statistics only starting from 2021 due to the missing German data before that year. In cases where country data is missing, Eurostat estimates (not published) are used for the calculation of these aggregates.

There are no measures of accuracy calculated at this point in time for flow statistics.  However, the overall accuracy is considered as high, given the fact that the LFS was not set up as a panel.  Unemployment is arguably the most important variable collected by EU-LFS, the survey design is optimized to measure unemployment.

For accuracy concerning the LFS as a whole, please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

Please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

Please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

The quarterly series are updated 4 times a year approximately 75 days after the end of the reference quarter.

The annual series are published along with quarter 4 data approximately 75 days after the end of the reference year.

Please refer to the ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

No quarterly flows data is available for Germany before 2021 due to the missing overlap of the quarterly samples. No EU aggregates for levels are published before 2021. Rotational sampling designs were introduced in Luxembourg in 2015 and in Belgium in 2017.

The data published for Spain, the Netherlands, Belgium and Portugal  are produced by the respective national institutes  and sent to Eurostat. Small differences in the methodology for the derivation of longitudinal weights used are described in section 18.5.

Generally, geographical comparability of the EU-LFS has historically been very high, and has further improved with the implementation of IESS. However, some countries are still in the process of implementing the Regulation, or have other quality concerns that may impact geographical comparability. Please find details below:

Spain: Pre-IESS, and after the outbreak of the COVID crisis, Spain was including in employment also persons on lay-off paid by a governmental scheme and not directly from the employer. Under IESS, due to the COVID crisis, Spain continued to test for job attachment of employed persons on temporary lay-off as implemented in the pre-IESS situation. 

France: Under IESS, and due to the COVID crisis, France continued to test for job attachment of employed persons on temporary lay-off as implemented in the pre-IESS situation. 

The Netherlands: Data is collected using a rolling reference week instead of a fixed reference week, i.e. interviewed persons are asked about the situation of the week before the interview rather than a pre-selected week (irrespective of the interview time). 

Flow statistics are published beginning with the transitions from Q1 to Q2 2010 for quarterly data, and with the transition from 2010 to 2011 for annual data. Breaks in series are indicated in the data.

There has been no investigation into the cross domain coherence of LFS flows yet.

'LFS main indicators' data have full internal coherence, as they are all based on the same corpus of microdata and they calculated using the same estimation methods. Arithmetic and accounting identities in the production of LFS datasets are observed.

LFS flow statistics are based on a sub-sample of individuals which are surveyed in both, the initial and the final period for which the flows are calculated. This sub-sample is reweighted to be coherent with the levels of employment, unemployment and inactive individuals in the target period as well as employment and unemployment in the initial period. Small divergences may occur in the levels of inactive individuals in the initial period.

Revisions of previously released non-seasonally adjusted data based on EU-LFS are not expected, unless major errors are identified in the data delivered or in their processing.

To further specify the general Eurostat revision policy, the following revision policy has been established for the LFS domain. 

If there is revision to flow data due to revisions to the underlying LFS micro-data, it is published along with the regular quarterly data release. Apart from error corrections (as detailed below), there are no revisions between data releases. Major revisions may occur in the future due to changes in the methodology.

Once the corrected data have been validated, errors meeting the criteria for seriousness trigger a correction of already disseminated data. Reported errors that are deemed to be significant are corrected in the disseminated data as soon as the correct data have been validated. Data are only published once they are deemed to be sufficiently complete for all data providers. New data are only used to update disseminated data if provided according to the provision schedule set by Eurostat, or in the case of reported error.

Whenever new data are provided and validated, the already disseminated data are updated with next regular production cycle.

1. Routine revisions:
Revisions of seasonally adjusted data may occur due to changes in the adjustment models used. Seasonal adjustment is done by Eurostat for most Member States on a disaggregated level (country by gender by age group, indirect approach) using TRAMO/SEATS. Seasonally adjusted data are thus revised 4 times a year for quarterly and annual data. Parameters used in the ARIMA models and for seasonal adjustment are reviewed annually.

2. Major revisions:

The impact of major revisions is analysed.
Major revisions are documented into the metadata and by means of Statistics Explained methodological articles available to all users.
Time series breaks caused by major revisions are flagged. Whenever possible break-free data is provided.
Due to back series revisions it may happen that new estimates of population from a population census are incorporated into the data, or changes are necessary to overcome methodological breaks.
Breaks due to census 2021 revisions: the introduction of new population figures triggered by Census 2021, led to a break in series in several countries. Back data revisions have been also performed in those countries accordingly. In this regard, a “break” flag has been introduced for countries and quarters as follows:
BG: 2022 Q1
IE: 2021 Q1
HR: 2023 Q1
IT: 2018 Q1
PL: 2019 Q1
RS: 2023 Q1

3. Unscheduled revisions:
No unscheduled revisions were known for the reference period up to 2023 Q4.

More information can be found from Employment and unemployment (Labour force survey) metadata page. 

The quarterly EU-LFS is the only source of the data, for details see ESMS page on 'Employment and unemployment (LFS)' (see link below in section 'related metadata').

Quarterly.

EU-LFS data are acquired by interviewing the sampled individuals directly. Three modes of data collection exist for the EU-LFS: personal visits, telephone interviews and self-administered questionnaires. Half of the Participating Countries mix the two first so that the first wave is always or mainly via personal visit while subsequent waves are interviewed with telephone if available. Twenty-one of the countries conduct the interview only with computerised questionnaires. Other two use both computerised and paper questionnaires. The rest rely solely on paper questionnaires.

For more information please consult the corresponding LFS quality reports.

Quarterly LFS data: Eurostat checks the quality and consistency of data transmitted by National Statistical Institutes. Eurostat calculates LFS results and they are then validated by the Member States. Based on these data, Eurostat calculates the available flow statistics.

Eurostat calculates initial quarter-on-quarter flow estimates as 3x3 ILO labour status transition matrices, for the age group 15-74, by sex and for individual countries.

The following general criteria are applied for quarter-on-quarter flow calculations:

• All computations are restricted to persons who are aged 15-74 in the target period;
• For all quarters from the first quarter of 2010 to the current quarter, the quarterly longitudinal flow samples are defined as the overlapping sample of two consecutive quarters.
• For all years from the first year 2010, the annual longitudinal flow samples are defined as the annual average of the overlapping sample of the same quarters of two consecutive years.

The methodology below describes the production of quarter-on-quarter flows. Year-on-year flows are produced in the same way, described below in this section.

In anticipation of possible future consistency requirements for flow statistics and planned more detailed flow estimates, calculation of ILO labour status transitions starts by sex and age group, using 10-year age groups 15-24, 25-34…65-74. For each subgroup and each of the 9 possible quarter-on-quarter transitions between the 3 labour statuses, the respective final quarter sum of coefficients (weights) in the flow sample is computed. Final quarter weights are used as they lead to the correct International Labour Organisation (ILO) labour status distribution in the most recent quarter. As those figures are based on a subset of the final quarter sample only, the resulting grossed-up weights obviously do not provide correct estimates for the underlying population subgroups. They have hence to be calibrated further to known marginal totals for the subgroups in question. In order to do this, the final quarter distribution of the 3 labour statuses in the respective subgroups is taken, and correction factors calculated.

The flow sample weights are then adapted accordingly to match the distribution in the final quarter, namely for each age group x sex x labour status in the final quarter combination. The steps described in the following could be applied to each intermediate matrix produced above. However, in order to avoid empty or poorly populated cells as far as possible and to get more robust results, calculation of the headline indicators for the age group 15-74 starts with a further aggregation of the previous results about age, i.e. all intermediate transition matrix results calculated so far for an individual country are combined into one single matrix, by sex. As for the final quarter, marginal ILO labour status distributions for the initial quarter are available as well. The next step tries to achieve consistency of the transition matrix with both marginal distributions. The procedure applied requires a common population in both quarters – for that, the probably least critical value (population outside the labour force in the initial quarter) is corrected in a way that the total population in both quarters matches that of the final quarter. Afterwards an iterative raking procedure is applied. It starts with the matrix consistent with the final quarter distribution and tries to find matrix values which are as close as possible to the start matrix while ensuring also consistency to the (partly corrected) initial quarter distribution. The iterative raking stops once the deviation of the row and column sums from the marginal distributions is less than 1,005. The results of the iterative raking are the flow estimates published. They are published separately for males and females.

Annual labour market flows between the labour market statuses of employment, unemployment and outside the labour force are derived from cross-sectional EU-Labour Force Survey data by matching microdata from all four quarters of a specific (initial) year to the same quarters of the following (target) year. This pseudo-longitudinal sample is used to derive transition matrices. Reweighting and raking mirrors the methodology used for quarterly labour market flows (see above). Across all countries, the share of individuals surveyed in the initial period but not present in the target period is higher for annual flows than for quarterly flows. This can be due to people moving away, dying, not being at home when surveyed, or refusing to answer the survey. There is some indication that this attrition is non-random and might therefore in some particular cases of significant population changes bias the annual flows estimates.

Estimates for Spain are sent to Eurostat by INE Spain, and make use of additional variables in the weighting (five year age groups, nationality, region). They are restricted to persons 16 to 74 years old in the target quarter due to the legal working age in Spain. Specificites for annual flows are given in the document annexed.

Estimates for the Netherlands are sent to Eurostat by the CBS. The methodology is explained in detail in the document annexed.

Estimates for Portugal (both annual and quarterly flows) are sent to Eurostat by INE Portugal. The methodology is explained in detail in the document annexed.

For quarterly flows: if data is missing (due e.g. to breaks in the series), estimates based on the development in the last year as well as the seasonal component are made for aggregate estimates of all MS which are used for publication in e.g. Statistics Explained. No EU aggregate levels are published in the database due to the missing German data before 2021. Seasonally ajdusted flow series are adjusted separatly for males and females on the country level and aggregated. There is no additional benchmarking to seasonally adjusted levels of unemployment, employment and inactivity done.

For annual flows: if data is missing (due e.g. to breaks in the series), estimates are interpolated based on the preceding and following year's values and used for the EU and EA aggregates.

None of the estimates of missing data is published.

Seasonal adjustment is performed indirectly, i.e. on the lowest available breakdown, and higher aggregates are derived from these series. This also applies to EU and EA aggregates. Models are estimated once a year, while parameters are re-estimated with the inclusion of each quarterly data point.