Employment and unemployment

Information on employment and unemployment is obtained from the continuous Labour Force Survey (LFS), which collects data on labour status of population of Latvia and labour force characteristics by sex, age and educational attainment. The LFS also provides information on the main kind of economic activity of the main job and occupation in the current job (for employed persons) or the last job (for unemployed persons), as well as other labour market indicators.

Data are compiled, calculated and published using:

• Statistical Classification of Economic Activities in the European Community (NACE Rev.1.1);
• Statistical Classification of Economic Activities in the European Community (NACE Rev. 2 from 2008 to 2025, NACE Rev. 2.1 as of 2026);
• Classification of Administrative Territories and Territorial Units (ATVK);
• National classification of occupations, which is based on the International Standard Classification of Occupations (ISCO-88 (COM), ISCO-08 as of 2011);
• National classification of education, which is based on the International Standard Classification of Education (ISCED 1997 until 2013, ISCED 2011 as of 2014);
• Nomenclature of Territorial Units for Statistics (NUTS);
• International Classification of Status in Employment (ICSE).

N/A

Person

The LFS has two target populations:

1. usual residents of Latvia aged 15–89 (aged 15–74 up to and including 2020) living in private households during the reference period;
2. private households with at least one usual resident aged 15–89 (aged 15–74 up to and including 2020) during the reference period.

As of 1 January 2024, there are five statistical regions in Latvia (Riga, Vidzeme, Kurzeme, Zemgale and Latgale) and the territories thereof coincide with the territories of the planning regions. Read more.

Before 1 January 2024, there were six statistical regions (Riga, Pierīga, Vidzeme, Kurzeme, Zemgale and Latgale).

2019–2023 data are recalculated based on the territories in force as of the beginning of 2024.

• Annual data – from 1996 
• Quarterly data – from 2002
• Monthly data – from 2002 to 2025

N/A

N/A

Month, quarter and year.

Regulations in force since the 2021 data collection

Framework regulation

• Regulation (EU) 2019/1700 of the European Parliament and of the Council  of 10 October 2019 establishing a common framework for European statistics relating to persons and households, based on data at individual level collected from samples, amending Regulations (EC) No 808/2004, (EC) No 452/2008 and (EC) No 1338/2008 of the European Parliament and of the Council, and repealing Regulation (EC) No 1177/2003 of the European Parliament and of the Council and Council Regulation (EC) No 577/98.

Delegated acts

• Commission Delegated Regulation (EU) 2020/257  of 16 December 2019 supplementing Regulation (EU) 2019/1700 of the European Parliament and of the Council by specifying the number and the title of the variables for the labour force domain.
• Commission Delegated Regulation (EU) 2020/256  of 16 December 2019 supplementing Regulation (EU) 2019/1700 of the European Parliament and of the Council by establishing a multiannual rolling planning.
• Commission Delegated Regulation (EU) 2022/2447 of 30 September 2022 supplementing Regulation (EU) 2019/1700 of the European Parliament and of the Council by specifying the number and the title of the eight-yearly variables in the labour force domain on ‘young people on the labour market’, ‘educational attainment – details, including education interrupted or abandoned’ and ‘reconciliation of work and family life’.
• Commission Delegated Regulation (EU) 2025/668 of 15 November 2024 supplementing Regulation (EU) 2019/1700 of the European Parliament and of the Council by specifying the number and the titles of the variables in the labour force domain for the 2026 ad hoc subject on Digital Platform Employment.

Implementing acts

• Commission Implementing Regulation (EU) 2019/2240 of 16 December 2019 specifying the technical items of the data set, establishing the technical formats for transmission of information and specifying the detailed arrangements and content of the quality reports on the organisation of a sample survey in the labour force domain in accordance with Regulation (EU) 2019/1700 of the European Parliament and of the Council.
• Commission Implementing Regulation (EU) 2019/2180 of 16 December 2019 specifying the detailed arrangements and content for the quality reports pursuant to Regulation (EU) 2019/1700 of the European Parliament and of the Council.
• Commission Implementing Regulation (EU) 2019/2181 of 16 December 2019 specifying technical characteristics as regards items common to several datasets pursuant to Regulation (EU) 2019/1700 of the European Parliament and of the Council.
• Commission Implementing Regulation (EU) 2019/2241 of 16 December 2019 describing the variables and the length, quality requirements and level of detail of the time series for the transmission of monthly unemployment data pursuant to Regulation (EU) 2019/1700 of the European Parliament and of the Council.
• Commission Implementing Regulation (EU) 2022/2312 of 25 November 2022 on eight-yearly variables in the labour force domain on ‘young people on the labour market’, ‘educational attainment – details, including education interrupted or abandoned’ and ‘reconciliation of work and family life’ pursuant to Regulation (EU) 2019/1700 of the European Parliament and of the Council.
• Commission Implementing Regulation (EU) 2024/2887 of 15 November 2024 specifying the technical items of the data set for the 2026 ad hoc subject on Digital Platform Employment in the labour force domain in accordance with Regulation (EU) 2019/1700 of the European Parliament and of the Council.

Regulation concerning general classification of the EU-LFS

• Regulation (EU) No 317/2013 of 8 April 2013 amending the Annexes to Regulations (EC) No 1983/2003, (EC) No 1738/2005, (EC) No 698/2006, (EC) No 377/2008 and (EU) No 823/2010 as regards the International Standard Classification of Education. This regulation stipulates the use of the ISCED 2011 in the LFS from 2014 onwards.
• Regulation (EC) No 1022/2009 of 29 October 2009 amending Regulations (EC) No 1738/2005, (EC) No 698/2006 and (EC) No 377/2008 as regards the International Standard Classification of Occupations (ISCO). This regulation stipulates the use of the ISCO-08 in the LFS from 2011 onwards.
• Commission Delegated Regulation (EU) 2023/137 of 10 October 2022 amending Regulation (EC) No 1893/2006 of the European Parliament and of the Council establishing the statistical classification of economic activities NACE Revision 2. Article 2 of this regulation stipulates the use of the NACE Rev. 2.1 in the LFS from 2026 onwards.
• Regulation (EC) No 1059/2003 of the European Parliament and of the Council of 26 May 2003 on the establishment of a common classification of territorial units for statistics (NUTS). This regulation provides the codification of NUTS regions. The changes in codification can be found in the respective amendments.

Regulations on statistical confidentiality

• Regulation (EU) No 557/2013 17 June 2013 implementing Regulation (EC) No 223/2009 of the European Parliament and of the Council on European Statistics as regards access to confidential data for scientific purposes and repealing Commission Regulation (EC) No 831/2002.

Regulations applicable to the data collections between 1998 and 2020

General regulations

• Regulation (EC) No 577/98 of 9 March 1998 on the organisation of a labour force sample survey in the Community.

This is the main regulation with provisions on design, survey characteristics and decision making processes.

• Regulation (EU) No 545/2014 of the European Parliament and of the Council of 15 May 2014 amending Council Regulation (EC) No 577/98.

This regulation specifies the financing provision and the sample conditions for ad hoc modules.

• Regulation (EC) No 596/2009 of the European Parliament and of the Council of 18 June 2009 adapting a number of instruments subject to the procedure referred to in Article 251 of the Treaty to Council Decision 1999/468/EC with regard to the regulatory procedure with scrutiny — Adaptation to the regulatory procedure with scrutiny — Part Four.

This regulation specifies the organisation of the ad hoc module within the LFS.

• Regulation (EC) No 1372/2007 of the European Parliament and of the Council of 23 October 2007 amending Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community.

This regulation changes the status of the survey characteristic 'income' from optional to mandatory.

• Regulation (EC) No 2257/2003 of the European Parliament and of the Council of 25 November 2003 amending Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community to adapt the list of survey characteristics.

This regulation introduces 6 new variables and allows the wave approach for structural variables.

• Regulation (EC) No 1991/2002 of the European Parliament and of the Council of 8 October 2002 amending Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community.

This regulation puts a time limit on the adoption of the continuous LFS.

This is a documentation is the latest consolidated version with all amendments for the main regulations ((EC) No 577/98) on the organisation of a labour force survey in the Community.

Implementation regulations of the core survey

• 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.

The regulation implements the codification to be used for data transmission from 2009 onwards including the compulsory survey characteristic 'income', the use of a sub-sample for the collection of data on structural variables and the definition of the reference quarters.

• Commission Regulation (EC) No 430/2005 of 15 March 2005 implementing Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community concerning the codification to be used for data transmission from 2006 onwards and the use of a sub-sample for the collection of data on structural variables.

The regulation implements the codification to be used for data transmission from 2006 onwards and the use of a sub-sample for the collection of data on structural variables.

• 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.

This regulation implements the operational definition of unemployment and contains the 12 principles for constructing the questionnaire.

• Commission Regulation (EC) No 1575/2000 of 19 July 2000 implementing Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community concerning the codification to be used for data transmission from 2001 onwards.

This regulation provides the codification to be used for data transmission from 2001-2005. It was corrected twice: Corrigendum to Commission Regulation (EC) No 1575/2000, Corrigendum to Commission Regulation (EC) No 1575/2000.

• Commission Regulation (EC) No 1571/98 of 20 July 1998 implementing Council Regulation (EC) No 577/98 on the organisation of a labour force sample survey in the Community.

The Annex I of the regulation defines the reference quarters for the first two years of the continuous survey and Annex IV defines the codification in force for 1998 to 2000.

N/A

Confidentiality of the information provided is protected by Statistics Law: 

• Section 7, Paragraph two, Clause 8, which sets out the obligation of the statistical institution to ensure statistical confidentiality;
• Section 17, which defines the procedures for data processing and the requirements for data protection;
• Section 19, Paragraph one, which stipulates that official statistics must be disseminated in a way that does not allow either directly or indirectly identify a private individual or a State institution;
• Section 19, Paragraph two, which stipulates that the statistical institution shall publish the official statistics which have been produced within the framework of the Official Statistics Programme in a publicly available form and by a predetermined deadline on the portal of official statistics. Until the moment of publication of official statistics this statistics shall not be published

In social statistics, the protection of individual data begins with an assessment of the table content (or other summary information) to determine whether it contains cells that disclose confidential data. Disclosure risk is evaluated using several factors, including the sensitivity of the indicators and the precision and timeliness of the information. Primary confidential cells are then suppressed.

To protect these primary confidential cells when publishing summary information, the principle of secondary confidentiality is also applied. This involves suppressing additional cell values to ensure that primary confidential values cannot be derived through arithmetic operations.

All official statistics are released according to the data release calendar at 13:00.

Data release calendar

Statistical release dates and times are pre-announced in the data dissemination calendar.

Data are published every quarter.

Press release is published every quarter.

Annual informative leaflet Labour Force Survey: key indicators (available in Latvian only).

Statistics database

For research purposes, anonymized individual data are available through remote access. Depending on the data processing methods required, datasets may be used on the researcher's own infrastructure (OffSite) or via the CSB remote access system (OnSite). Data are provided upon submission of application and, if the CSB issues a positive decision, the conclusion of a contract. Anonymized individual data may be used only for scientific or research purposes, and the research results must provide a benefit to society.

Individual data, or microdata, relate to individual persons, households or enterprises and are obtained from surveys, population censuses and administrative registers.

Research data catalogue

Not available

Methodological information of the Labour Force Survey available on Eurostat website.

N/A

To achieve high user satisfaction and ensure compliance with regulatory requirements, the CSB has introduced a Quality Management System (QMS). The system defines and, at the procedural level, describes processes of statistical production and identifies the persons responsible for their monitoring throughout all production stages. Its structure follows the principles of the Generic Statistical Business Process Model (GSBPM).

The QMS sets out the sequence in which processes are implemented – that is, the activities to be performed, including verifications of processes and produced statistics, the order and implementation requirements of these activities, and the persons responsible for their execution. It also defines the approach to evaluating production processes and their outcomes, and to implementing necessary improvements.

The CSB quality management system is certified to the ISO 9001:2015 standard Quality management systems — Requirements since 2018 (scope of certification: development, production and dissemination of official statistics). The original certification audit was performed by BM Trada Latvija SIA and a recertification audit, in 2024, was performed by Bureau Veritas Latvia SIA.

The CSB information security management system is certified to the ISO/IEC 27001:2022 standard Information security, cybersecurity and privacy protection — Information security management systems — Requirements since 2017 (scope of certification: collection, processing and storage of information and data for functions of the Central Statistical Bureau of Latvia. Provision of statistical information for inland and foreign users). The original certification audit was performed by BM Certification SIA and a recertification audit, in 2024, was performed by Bureau Veritas Latvia SIA.

The quality of statistics is assessed in accordance with the existing requirements of both external and internal regulatory enactments, as well as the established quality criteria.

Regulation (EC) No 223/2009 of the European Parliament and of the Council on European statistics stipulates that European statistics shall be developed, produced and disseminated on the basis of uniform standards and harmonised methods. In this context, the following quality criteria shall apply: relevance, accuracy, timeliness, punctuality, accessibility, clarity, comparability and coherence.

As the national statistical institute and the principal authority of the national statistical system, the CSB has set common general institutional-level quality requirements for authorities responsible for producing or providing national statistics. These requirements are based on the European Statistics Code of Practice, which comprises 16 principles.

The overall assessment of data quality is good.

N/A

The task of the CSB is to produce reliable statistics to support the analysis of socio-economic processes and to inform future decision-making.

Send feedback on data quality to pasts@csp.gov.lv

The sample framework was built for a target population defined based on the best information available at the sampling moment by ensuring that all survey units, according to the sample plan, have the same chances to be included in the sample.

Even though non-response and over-coverage are the main causes of error, there is no substantiated evidence of their impact on the resulting estimates, moreover these errors are addressed when adjusting the design weights.

Sampling errors are calculated for the key indicators in various breakdowns.

Standard error – an indicator characterising sample error, which serves also as an output value for the calculation of other quality indicators, for example, for relative standard error or coefficient of variation and absolute margin of error. By calculating coefficient of variation, it can be established whether the obtained result is sufficiently reliable.

Example: If an estimation is 90.0 and standard error is 6.14, relative standard error or coefficient of variation (CV) can be obtained as follows:

CV = 6.14 / 90 * 100 = 6.82

To calculate the absolute margin of error, standard error must be multiplied by 1.96 (coefficient used at 95% confidence).

Absolute margin of error (+/–) = 6.14 * 1.96 = 12.03

Since the Labour Force Survey is a sample survey, its estimates, generalised for the entire population, may differ from the results that would be obtained in a full-scale survey.

Absolute margin of error – a precision indicator which expresses the maximum possible difference between the true population parameter and a sample estimate of that parameter with 95 % confidence.

When comparing estimates, it is important to use absolute margin of error to determine whether differences between values are statistically significant.

Example: if estimate is 90.0 and absolute margin of error (+/–) 12.03, confidence limits are (77.97; 102.03), with 95% confidence that the true value lies there.

Currently, data accuracy indicators (standard error and absolute margin of error) are available for the following tables:

NBA010. Population by labour status and sex

NBL040. Employed by economic activity and sex (NACE Rev. 2.)

NBB160c. Unemployed persons and unemployment rate by age group and sex

NBA050c. Active population, inactive population and activity rate by age group, sex

NBL020c. Employed and employment rate by age group, sex

Reliability limits

Quarterly data (thousand)

Annual data (thousand)

Structural (annual) data from the first wave (thousand)

Additional information:

A – quarterly and annual estimates are not published because they are too uncertain for presentation.

B – quarterly and annual estimates are based on small number of respondent answers.

Information about structural (annual) data from the first survey wave is available in the reference metadata, section 19. Statistical processing (data source, etc.), sub-section Data compilation / Wave approach.

The sampling frame is designed to ensure the best possible coverage of the target population while minimising non-sampling errors and their impact.

N/A

Several units with over-coverage were identified during the data collection. The units were accordingly marked, their impact limited, and the weights were accordingly adjusted.

N/A

N/A

N/A

Monthly gross earnings in the main job are imputed as of 2021.

N/A

N/A

• Q1, Q4 and annual data are published within 55 days after the end of the reference quarter;
• Q2 and Q3 data are published within 50 days after the end of the reference quarter.

N/A

EU data on Eurostat website in section: Employment and unemployment

From 2019 onwards, data by statistical region are not comparable with previous periods.

• Annual data – from 1996
• Quarterly data – from 2002
• Monthly data – from 2002 to 2025

Differences in employment data between the NBA/NBL and EKA tables (e.g., NBA010 and EKA011) as well as in employment breakdowns by occupation (e.g., NBL080 and EKA051), economic activity (e.g., NBL040 and EKA071), and status in employment (e.g. NBL070 and EKA041) ­– arise from the data collection and compilation methods applied.

In the tables NBA and NBL, data are obtained from the continuous Labour Force Survey (LFS), which provides information on the labour status of population (employment and unemployment).     

In the EKA tables, data on labour status, occupation, economic activity, and status in employment are compiled annually from administrative data sources.

Main reasons for differences

• Administrative sources provide monthly, quarterly and annual data used to calculate employment in November, while the LFS asks respondents about their employment in the reference week.
• Administrative data cover persons aged 15 and over, including those in collective households, while LFS covers persons aged 15–89 and excludes collective households.
• Administrative sources do not capture information on unregistered employment or unregistered unemployment.
• Administrative sources also do not cover unpaid family workers, i.e., persons assisting a family member in a business, private practice, or family farm.

N/A

N/A

N/A

One of the CSB priorities, in line with the strategic directions of the European Statistics System and current approaches to producing statistics, is to collect data through broader use of administrative sources together with regular CSB surveys, while proportionately reducing response burden.

In cooperation with administrative data holders and within the competences set out in the Statistics Law, CSB is addressing the issues related to the use of administrative data to ensure that the sources used are as complete and reliable as possible, helping to reduce the administrative burden on businesses and households.

CSB measures to improve the use of administrative data and reduce response burden (2024) – available in Latvian only.

Revision policy is an important component of good governance practice. The aim of the CSB Revision Policy is to define how the statistics produced and published by the CSB are reviewed and revised. The first chapter explains the main terms used in the document, the second chapter provides a brief description of the CSB Revision Policy, and the third chapter outlined the revision cycle for the statistics produced by the CSB.

Revision policy guidelines (8 September 2022, PDF, 597 KB)

Seasonally adjusted monthly indicators were revised monthly, while unadjusted time series were revised quarterly (January 2002–December 2025).

N/A

Data collection methods used in the Labour Force Survey:

• 1995–2005: face-to-face interviews using paper questionnaires (Paper-and-Pencil Interviewing – PAPI);
• starting from 2006: face-to-face interviews using portable computers (Computer-Assisted Personal Interviewing – CAPI);
• starting from 2007: CAPI and telephone interviews (Computer-Assisted Telephone Interviewing – CATI);
• starting from 2018: CAPI, CATI and online surveys (Computer-Assisted Web Interviewing – CAWI);
• starting from 13 March 2020, with aim to limit spread of Covid-19: CATI and CAWI;
• starting from May 2022: CAPI, CATI and CAWI.

The survey covers all persons living in the surveyed household, and questions on activity status are asked to persons aged 15–89 (to persons aged 15 and over before 2001 and to persons aged 15–74 from 2002 to 2020 incl.).

The survey questionnaires include key questions on the labour market activity of the population and are developed in line with the internationally accepted methodology for labour force surveys established by the International Labour Organisation (ILO), ensuring international comparability.

The LFS provides data on the population (including active (employed and unemployed) and inactive population) broken down by sex, age, educational attainment, place of residence, employment status, and other characteristics.

The results of the Population and Housing Census 2011 revealed that the actual population was smaller than the estimate used to extrapolate LFS data. Considering the population decline since the Census 2000 and the updated population figures based on the Census 2011, LFS data for the period 2001–2011 were revised accordingly.

In 2025, CSB changed the method for producing population estimates. Since 2012, a logistic regression model was used, but going forward, the Sol-Logit model will be applied. The methods are similar, with the main difference being that the logistic regression model is a supervised model trained on data from the Population and Housing Census 2011, whereas the Sol-Logit model is unsupervised model and does not require training data.

Data for the period 1996–2000 are published according to the national data calculation approach, which includes all surveyed persons as well as those in compulsory military service.

To comply with the recommendations of Eurostat Task Force on LFS quality, LFS data are extrapolated solely to the population living in private households, excluding persons living in collective households (such as care homes for elderly, student hostels, hospitals, prisons etc.). Information on the number of persons in collective households is updated regularly.

Starting with 2014, the quarterly average number of residents in private households has been used to generalise LFS quarterly data (previously, population figures at the beginning of the year). These methodological changes in the calculation of weights do not have significant impact on the time series.

Starting with 2021, several changes were introduced in the generalisation of quarterly LFS data (see section DATA COMPILATION).

Monthly estimates

From 2022 to 2025, unadjusted and seasonally adjusted monthly estimates for two groups – employed and unemployed aged 15–74 – were published on the official statistics portal.

Three monthly indicators by sex were published (January 2002–December 2025):

• number of unemployed;
• unemployment rate;
• number of employed.

Unadjusted and seasonally adjusted monthly indicators were estimated based on the ILO methodology.

Monthly indicators were calculated indirectly using quarterly LFS estimates and monthly data on registered unemployment from the State Employment Agency (SEA). To derive unadjusted monthly time series for the number of employed and unemployed, the Chow–Lin temporal disaggregation method was applied¹. Calculations were performed using the R package tempdisagg.

From February 2026 onwards, monthly figures for the unemployment rate and the number of unemployed persons are available only from Eurostat (table une_rt_m).

This approach is consistent with Commission Implementing Regulation (EU) 2019/2241, which provides that Member States that do not produce monthly unemployment statistics may transmit input data to Eurostat, which then compiles the monthly unemployment estimates on their behalf.

Figures produced by the CSB cover the period from January 2002 to December 2025, and from January 2026 onwards, the figures are produced by Eurostat.

Please note that, in line with the methodology, monthly estimates of actual unemployment are revised when quarterly data are published. As a result, when new monthly data are released in table une_rt_m, indicators for previous months may also be revised and may no longer coincide with the data in table NBB150m, where figures are fixed as at the time of their publication (20 January 2026).

¹ Chow, G.C. and Lin, A.-L. (1971) Best Linear Unbiased Interpolation, Distribution, and Extrapolation of Time Series by Related Series.

Until 2001, LFS was conducted twice a year, in May and November.

From 2002, LFS has been conducted continuously throughout the year, with data collected every week.

Annual LFS sample:

N/A

1995–2001

Sample

The first Latvian LFS was launched in November 1995. Thereafter, it was conducted twice a year – in May and November.

The sample represents the usual resident population aged 15–74 living in private households.

A sample of dwellings was used. Where a response was obtained, one household in each sampled dwelling participated in the survey, and all household members aged 15 and over during reference period were interviewed.

The sample was designed as a rotating panel. Each selected dwelling participated in the survey three consecutive times and was then replaced by another dwelling, ensuring regular rotation within the panel.

Weighting

From 1995 to 2000, design weights were adjusted using the response homogeneity group method. Each primary sampling unit served as a response homogeneity group for households.

Final weights were obtained through post-stratification, using data on the usual resident population at the beginning of the reference year as auxiliary information. Auxiliary information on Riga, six cities and 26 administrative territory towns and rural areas was broken down by sex and age group. Weight adjustment was performed at the individual level.

In 2001, design weights were again adjusted using the response homogeneity group method. Each primary sampling unit served as a response homogeneity group for households. In addition, strata and survey waves were defined as response homogeneity groups.

In 2001, final weights were obtained using weight calibration (ranking-ratio). Weight adjustment was carried out at the household level, with households classified into three age groups (0–14, 15–74, 74+). Calibration was based on auxiliary information about population living in private households, taking into account the results of Population and Housing Census 2011 at the beginning of the period. This information was broken down by:

• 14 age groups;
• sex;
• territory type of place of residence (Riga, eight cities, towns and rural areas);
• region of the place of residence (Riga, Pierīga, Vidzeme, Latgale, Kurzeme and Zemgale).

The number of unemployed persons registered with the State Employment Agency, broken down by sex and age group, was also used as auxiliary information for weight calibration.

Weight calibration was implemented using the statistical computing environment R and the Sampling package.

2002–2006

Sample

In 2000, CSB started developing a new LFS methodology with the following main objectives:

1. to obtain quarterly estimates;
2. to organise LFS as a continuous survey;
3. to introduce efficient dispersion estimators.

CSB developed a new sample design based on information about the Population and Housing Census 2000 enumeration areas, resulting in the creation of a list of territories covering all private households. These territories were subsequently used as primary sampling units. The territories were stratified into four strata: Riga, cities under state jurisdiction, towns and rural territories.

A two-stage sampling design was applied. In the first stage, territories were selected as primary sampling units using stratified systematic πps sampling, in which elements are selected with probabilities proportional to their size. In the second stage, dwellings were selected within each sampled territory by simple random sampling method.

Where a response was obtained, one household in each sampled dwelling participated in the survey, and all household members aged 15–74 on the Sunday of the reference week were included in the LFS.

The main aim for applying two-stage sampling was to reduce survey costs. The design effect, which increases when using two-stage sampling, was minimized by selecting a large number of primary sampling units.

The LFS was organised as a rotating panel survey. The household rotation scheme was similar to that used before 2002. Each dwelling participated in the survey three times, at intervals of 26-weeks.

Weighting

Design weights were adjusted using the response homogeneity group method. Each primary sampling unit was used as a response homogeneity group for households. Stratum and survey wave breakdowns were also defined as response homogeneity groups.

From 2002 to 2006, final weights were obtained using the wight calibration method (ranking-ratio). Weight adjustment was carried out at the household level, dividing households into three age groups (0–14, 15–74, 74+).

For calibration, data on population living in private households were used as auxiliary information, taking into account the results of the Population and Housing Census 201 at the beginning of the period. The information was broken down by:

• 14 age groups;
• sex;
• territory type of place of residence (Riga, eight cities, towns and rural areas);
• region of the place of residence (Riga, Pierīga, Vidzeme, Latgale, Kurzeme and Zemgale).

The number of unemployed persons registered with the State Employment Agency, broken down by sex and age group, was also used as auxiliary information for weight calibration.

Weight calibration was implemented using the statistical computing environment R and the Sampling package.

2007–2012

To obtain more precise estimates of employed and unemployed persons by region, sex and age group, the LFS sample size was increased 2.4 times as of 2007. The dwelling rotation scheme was revised to ensure overlap between samples of successive quarters.

From 2007, dwellings participated in the LFS four times, with intervals of 13 weeks, 39 weeks and 13 weeks. This rotation scheme ensured overlap between samples of successive quarters and consecutive years.

Between 2002 and 2009, the list of territories used for selecting LFS sample was not updated. At the end of 2009, a study concluded that the territorial frame had become outdated. Due to population migration, it no longer adequately represented LFS target population. As a result, a new list for sampling territories was developed at the end of 2009 (based on the previous list of territories). A decision was made to update LFS sample gradually.

The new sample of territories was introduced for 2010–2011. Dwellings entering the LFS in 2010 were selected from the updated frame, while earlier entrants remained from the previous frame. Consequently, two territorial samples were used in 2010.

Until 2010, LFS sample represented usual resident population aged 15–74 living in private households. From Q4 2010, changes to the sampling frame were introduced to ensure that LFS sample represented the entire usual resident population living in private households.

As the LFS is a rotating panel survey, a transition period of five quarters (from Q4 2010 to Q4 2011) was required for the full sample to become representative of the entire usual resident population living in private households. As of Q1 2012, LFS sample has represented the entire usual resident population living in private households.

Weighting

Design weights were adjusted using response homogeneity group method. Each primary sampling unit was used as a response homogeneity group for households. Stratum and survey wave breakdown by were also defined as response homogeneity groups.

From 2007 to 2009, final weights were obtained using the weight calibration method (ranking-ratio). Weight adjustment was carried out at the household level, dividing a household into three age groups (0–14, 15–74, 74+). Calibration was based on auxiliary information about population living in private households, taking into account the results of Population and Housing Census 2011 at the beginning of the period. This information was broken down by:

• 14 age groups;
• sex;
• territory type of place of residence (Riga, eight cities, towns and rural areas);
• region of the place of residence (Riga, Pierīga, Vidzeme, Latgale, Kurzeme and Zemgale).

The number of unemployed persons registered with the State Employment Agency, broken down by sex and age group, was also used as auxiliary information.

Weight calibration method was implemented using the statistical computation environment R and Sampling package.

As of Q1 2010, weight adjustment was performed at the household level, meaning that all persons within a household were assigned the same weight, equal to the household weight. The remaining weighting methodology remained unchanged ­– final weights were obtained with the same calibration method (ranking-ratio) and with the same auxiliary information and breakdowns as described above.

2013-2020

Sample

To improve quality of estimates in more detailed breakdowns, the sample size was gradually increased between 2013 and 2014. By 2015, sample size was 1.2 times larger than in 2012.

From 2013 to 2015, LFS sample design and weighting methodology did not change significantly. The LFS used a rotating panel survey and represented the entire usual resident population living in private households. Dwellings participated in LFS four times, with intervals of 13 weeks, 39 weeks and 13 weeks.

Weighting

To improve precision of statistical estimates, the weighting methodology was refined.

In 2013, calibration of weights used data on the usual resident population living in private households at the beginning of the reference year as auxiliary information. The information was broken down by:

• 14 age groups;
• sex;
• territory type of place of residence (Riga, eight cities, towns and rural areas);
• region of the place of residence (Riga, Pierīga, Vidzeme, Latgale, Kurzeme and Zemgale) and eight cities (Daugavpils, Jelgava, Jēkabpils, Jūrmala, Liepāja, Rēzekne, Valmiera and Ventspils).

The number of unemployed persons registered with the State Employment Agency, broken down by sex and age group, was also used as auxiliary information for weight calibration.

2015–2020

In 2015, the weighting methodology was improved by introducing an additional non-response adjustment based on the mode of data collection (face-to-face interviewing [CAPI] or telephone interviewing [CATI]).

Starting from 2014

As of 2014, quarterly average data about usual resident population living in private households were used as auxiliary information for weight calibration, maintaining the above-mentioned breakdown by age, sex, type of territory and region of residence.

From Q3 2014, data on the number of employed persons from the State Revenue Service database were also added as auxiliary information for weight calibration, while continuing to use the number of unemployed persons registered by the State Employment Agency, broken down by sex and age group.

Starting from 2017

As of 2017, State Employment Agency (SAE) data on persons registered as unemployed have been linked with data obtained from the LFS; therefore, this information is no longer collected directly from respondents.

Since then, these data have been used for weight calibration by linking them with SEA data on the number of registered unemployed persons broken down by gender and age group.

The remaining weighting methodology has not changed and remains comparable with previous years.

Starting from 2021

Sample

Between 2009 and 2019, the list of territories used to select LFS sample was not updated. In 2018, it was concluded that the list had become outdated. As a result address system reorganisation, demographic changes and new construction, it no longer adequately reflected the target population of LFS. Therefore, in 2019 a new list of territories was developed for sampling purposes.

The new territories were constructed as defined polygons with clearly specified boundaries. This ensures that, for each selected territory, the list of dwellings and corresponding building within that territory is known.

The new list of territories was tested during 2020, and a decision was taken to introduce it in the LFS starting from 2021.

Dwellings entering the LFS in 2021 were selected from the updated frame, while earlier entrants remained from the previous frame. Consequently, two territorial samples were used in 2021 and Q1 2022.

As of 2021, LFS sample is drawn independently between years and negatively coordinated within the year. Territories surveyed during a given year are selected once, together with the first quarter sample, and do not overlap within the year.

Dwellings participate in LFS four times, with intervals of 13 weeks, 39 weeks and 13 weeks. A transition period of five quarters (from Q1 2021 to Q1 2022) was required for the full replacement of the sample. Starting with Q2 2022, LFS sample is based exclusively on the new territorial frame.

Weights

Taking into account amendments to Regulation (EU) No 2019/2240, which entered into force on 1 January 2021, and in order to improve statistical accuracy, the methodology of calculating weights was revised. The following changes were introduced:

• Weights were calculated using the registered place of residence (until 2021, the actual place of residence was used). The change was necessary because, starting with 2021, CSB no longer produces statistics on the resident population by actual place of residence. However, LFS territorial statistics is still compiled according to the actual place of residence.
• Following the administrative-territorial reform that entered into force on 1 July 2021, Ogre is classified as a State city. In addition, the number of residents in Ogre is also used in the weighting procedure to improve the accuracy of statistics for Ogre.
• A more detailed breakdown of age groups is used for weighting purposes. The age group 0–14 is divided into three five-year groups (0–4, 5–9, 10–14), and the age group 75+ is divided into four groups (75–79, 80–84, 85–89, 90+).
• Additional auxiliary information is used for weighting, including the number of employees and employers aged 75–89, registered with the State Revenue Service, broken down by sex.
• Based on the Eurostat guidelines, from Q2 2022 LFS data are extrapolated to a population that includes citizens of Ukraine who have been granted temporary protection in Latvia and who live in private households.

The preparation of weighting information has also been revised. Weighting data are now compiled on a weekly basis, and quarterly figures are derived as average of weekly estimates. This approach is consistent with the LFS methodology, as the target population is observed on a weekly basis.

Starting from 2024

As of 1 January 2024, Latvia has five statistical regions (Riga, Vidzeme, Kurzeme, Zemgale, and Latgale), which has resulted in corresponding changes to the regional breakdowns used weight calibration.

Wave approach

Sample

As of 2017, annual LFS indicators are produced using a wave approach. In accordance with Commission Implementing Regulation (EU) 2019/2240 (until 2020 – Commission Regulation (EC) No 377/2008) indicators with annual periodicity are collected only during the first wave.

In line with the sample design, annual estimates in a given quarter are therefore based on information collected from one quarter of the quarterly sample.

Weights

In 2017, additional weights were introduced for indicators with annual periodicity.

In line with the sample design, quarterly design weights are first calculated for the first-wave respondents. These design weights are adjusted for each quarter using response homogeneity group method.

The quarterly weights acquired are combined prior to calibration and divided by four. Calibration is then applied to annual data, using the raking-ratio method.

For calibration purposes, auxiliary information includes annual average statistics on the usual resident population living in private households, broken down by sex, five-year age groups and in territory, as well as State Employment Agency (SEA) data on registered unemployed persons, broken down by age groups.

To comply with Article 9 (4)(b) and 9 (5)(b) of Commission Implementing Regulation (EU) 2019/2240 (until 2020 – Article 3 in Annex I to the Regulation No 377/2008) regarding compliance of the totals, additional auxiliary information is used in calibration: full-scope sample estimates on employment, unemployment and inactive population, broken down by sex and ten-year age groups.

Monthly gross earnings in the main job are imputed as of 2021.

The methodology for calculating wages and salaries was revised in February 2025, and corrections were introduced to the gross wage and salary data for 2021 and 2022.

Until 2025, gross wages and salaries from part-time work were recalculated into full-time units, whereas as of 2025 part-time variables show the amounts reported by respondents.

A time series is a sequence of observations collected at regular time intervals, for example, a monthly time series. It characterises indicator changes or development thereof.  Seasonality and calendar effects are present in a large number of economic time series.

To produce monthly results for the period 2002–2025, CSB carried out seasonal adjustment that followed the European Statistical System (ESS) seasonal adjustment guidelines.

Software: JDemetra+
Method: TRAMO/SEATS

A time series is a sequence of observations collected at regular time intervals, for example, a monthly /quarterly time series. It characterises indicator changes or development thereof.  Seasonality and calendar effects are present in a large number of economic time series.

Seasonality or seasonal fluctuations of time series are those movements, which recur with similar intensity in the same season each year. Seasonally adjusted data do not include seasonal fluctuations and calendar effects, so it is possible to compare, for example, current quarterly data with previous quarterly data.

Calendar effects cover influence of calendar on time series. It is the number of different working days in months/quarters; distribution of days of the week; the impact of the leap year on changes of the indicator. The CSB uses Latvian-specific calendar regressors, which are calculated in accordance with the law On Public Holidays, Commemoration Days and Celebration Days. Calendar adjustment is performed only for those time series for which the calendar effect is statistically significant and economically explainable. Calendar adjusted data do not include calendar effects and are used to compare, for example, data for the current quarter with data for the corresponding quarter of the previous year.

Revisions are expected when new data are added to the time series, as the entire seasonally adjusted time series is recalculated. The time series models are reviewed once a year – that is, they are checked for compliance with the data and adjusted if necessary. Larger revisions may occur in the month or quarter following the model review.