Central Statistics Office of Ireland

Earnings Analysis 

Central Statistics Office, Skehard Road, Cork, Ireland

The labour cost index (LCI), as defined by Article 2 and Annex 1 of the Commission Regulation (EC) No 1216/2003 of 7th July 2003, shows short-term trends in the cost of employing labour on an hourly basis. It reflects changes in wages and salaries, non-wage costs, and the quantity of hours worked over the quarter and is important for monitoring inflationary pressures emanating from the labour market.

Statistical Classification of Economic Activities in the European Community, Rev. 2 (2008) (i.e. NACE Rev. 2)

Activity Coverage: NACE sections B to S

Size Coverage: Enterprises with employment size of 3 or more employees

Labour costs are defined as core expenditure borne by employers for the purpose of employing staff. They include employee compensation, with wages and salaries in cash and in kind, employer’s pension contributions and other employer’s contributions including social security contributions minus any subsidies and refunds received by the employers for employees. These labour costs and their elements are defined in Commission Regulation (EC) No. 1737/2005 of 21st October 2005 amending Regulation (EC) No. 1726/1999 as regards the definition and transmission of information on labour costs.

The variable definitions are given in the EHECS instruction leaflet that is provided to surveyed enterprises in conjunction with the paper questionnaire.

As a result of the COIVD-19 pandemic, two major schemes were introduced by the Government in 2020 to support those whose income from employment had been affected due to COVID-19.

The COVID-19 Pandemic Unemployment Payment (PUP) scheme, which is administered by the Department of Social Protection (DSP), operated to provide a social welfare payment to those who lost their employment because of the COVID-19 crisis.

The Revenue Employment Wage Subsidy Scheme (EWSS) was also operational during the COVID-19 pandemic enabling employees, whose employers are affected by the pandemic, to receive significant supports directly from their employer through the payroll system. The EWSS ended for all employers on 31 May 2022. 

Any payments related to the PUP scheme are not collected by EHECS or recorded in the LCI. Such payments are paid by DSP to a person who has lost income from employment and where no attachment in the form of an employment contract is maintained between the employer and employee.

EWSS payments are recorded as follows:

• EWSS payments are included in wages and salaries for employees benefiting from the scheme
• Refunded payments are recorded as subsidies and refunds received, being amounts received by enterprises intended to refund part or all of the cost of wages and salaries.

Individual enterprises in NACE sections B to S with 3 or more employees are the reporting and observation unit.

The target population is approximately 76,500 enterprises employing approximately 2.05 million employees.

The sample is selected each quarter to ensure that it is representative of the population of enterprises in the country. The sampling strata are defined by NACE divisions (i.e. 2-digit level) and employment size class. The sample consists of (i) census of all enterprises with 50+ employees and (ii) a random sample of enterprises with 3 to 49 employees. The employment size classes are (a) 3 to 9 (b) 10 to 19 and (c) 20 to 49 employees. The proportions by size class for all sectors are shown in the table below.

 The sample is approximately 7,800 enterprises which represent around 9.3% of all enterprises and accounts for 62.5% of all employees across NACE sectors B to S.

The geographical area covered is the Republic of Ireland.

Data from the earnings, hours and employment costs survey has been published from Q1 2008 using NACE Rev 2.

The base period for the Labour Cost Index is the reference period of the most recent published Labour Cost Survey, which for 2022 is 2016.

The LCI is based on data generated by the Earnings, Hours and Employment Costs Survey (EHECS) quarterly survey.  It covers enterprises across NACE (Rev2) sectors B-S with 3 or more employees.  Both full-time and part-time employees are covered.  All enterprises with 50 or more employees and a sample of those with 3 to 49 employees are surveyed each quarter. The sample is based on the proportion of companies in each NACE 2 digit economic sectors in the 3 to 49 size classes (3 to 9, 10 to 19 and 20 to 49).  The sample is taken in the first week of the last month of the quarter.   It is updated with the most current information from the previous quarters EHECS return.  

Data is collected on a quarterly basis.

EHECS consists of two questionnaires (Form A and B) which can be returned in paper format or by electronic payroll software system of data collection (.xml returns). The form A questionnaire is issued to all enterprises with registered employment of 100 or more persons. The form B questionnaire is a summarized version of the form and is issued to enterprises with less than 100 persons employees. XML questionnaires have the same format as the form A and are processed via the enterprise’s payroll software system. The return of XML questionnaires is independent of the class size that an enterprise belongs to - the only requisite is that the enterprise avails of the CSO functionality in their payroll software.

The percentage breakdown of issued forms by type is:

[Data refers to Quarter 4 2022]

Extensive edits rules are applied to check unit level data each time a dataset is compiled, with additional quarterly checks against previous returns for all enterprises with 250 or more employees.

The main checks performed on the data include:

• Check that all relevant variables are filled in correctly for each category received:
  • Ensure that employment is present both at beginning and end of the quarter
  • Ensure that both wages and hours variables are present (regular/ overtime)
  • Ensure that both wages and PRSI variables are present
• Check that the hourly rate of wage (regular/irregular/overtime) falls within a reasonable range for the sector
• Check that the average number of hours worked per week (full time/part time/apprentices) fall within a reasonable range for the sector.
• Check that the received figures tally up (e.g. that the PRSI value in the form is within correct parameters for the wages and salaries indicated)
• Check that BIK and pension figures returned are within a reasonable range
• Ensure that the hired and vacancies rates received are correct.

Weighting:

For enterprises with 3 to 49 employees inclusive, a weighting factor (the reciprocal of the sampling fraction) is used to weight the estimates to the total population for both employees and enterprises. The CSO’s Central Business Register (CBR) forms the basis of the sampling frame used for weighting the sample data to the population.

Estimates for non-response:

Imputation is undertaken for non-respondent enterprises with 50+ employees. Where an enterprise responded in any of the previous four quarters ratio-imputation is used to estimate values for the current quarter; replacing any missing value with the proportional change in that variable based on all other respondents in the quarter. Otherwise a stratum average (mean) imputation method is used to estimate the missing values replacing any missing value with the mean of that variable for all other respondents in the stratum in the quarter. Both the ratio-imputation and stratum average (mean) imputation methods are based on respondent enterprises of a similar size and activity.

Final Estimates:

After imputation, all enterprises with 50+ employees are accounted for and included in the final dataset. For those cells where a sample survey of enterprises is used (i.e. enterprises with 3 to 49 employees), the results are expanded using the grossing factors to cover the entire population for the relevant NACE sections in the quarter. Macro edits are carried out at this stage and any outliers are investigated and corrected. Coherence of data is ensured by scrutinising quarter on quarter changes.

The three LCI variables, Wages & Salaries, Labour Costs (excluding Wages & Salaries) and Total Labour Costs (Labour Cost and Wages & Salaries), are transmitted to Eurostat unadjusted, Working Day Adjusted and Seasonally & Working Day Adjusted.

An indirect chain-linked approach is implemented to working day and seasonally adjust the LCI series.  This is completed by applying an X-13-ARIMA model through the use of JDemetra+ seasonal adjustment software.

The managers of the EHECS assess the quality of the survey on an on-going basis and make improvements if and when necessary. The CSO has an internal quality audit team. The production and publication processes for the EHECS were reviewed using a Lean Six Sigma (LSS) approach in 2011 to improve (i) process clarity (ii) efficiency in the data collection process and (iii) timeliness. The EHECS was reviewed again in 2014 using a LSS approach.

In addition to this the EHECS data collection and reporting teams have implemented the Quality Management Framework, which is an office wide initiative to standardise the the quality procudes in the CSO.

The goal of the Quality Manaegement Framework (QMF) in the CSO is to meet the required standard as set out in the European Statistical System Code of Practice (ESCOP). The QMF foundations are based on establishing the UNECE’s Generic Statistical Business Process Model (GSBPM) as the operating statistical production model in the CSO.

The QMSA team have been working on the implementation phase of the QMF since mid- 2016 where they have systematically rolled out the new policies and standards in the form of the quality projects detailed below. The EHECS survey implemented changes required to adhere to the QMF.

• The establishment of the GSBPM as the business process model for the Office. This model is an UNECE standard for statistical production and allows the CSO to advance a more standardised, horizontal approach to quality management. 

• Survey documentation – the improving the level of quality and standardisation of survey documentation across the Office. 

• Process Mapping project – Process mapping is the visual display of steps involved in a business process. It draws a concise picture of the sequence of tasks needed to bring a product or service from start to completion. The main purpose behind business process mapping is to provide clarity on exactly how the process happens, not how it is supposed to happen. 

• Process maintenance project – In order to keep the maps up to date, the process map maintenance policy has been developed which requires business areas to certify that their maps are valid and up to date once a year. 

• Process Metrics and Indicators – In order for staff to make an assessment on how their processes are performing and to better manage the phases of the statistical lifecycle (collection, processing, analysing and the ultimate dissemination of statistical data) appropriate metrics are identified and collected at each phase of the statistical process. These metrics and indicators include response rates, timeliness, edit and imputation rates, precision rates and the degree of revisions. 

• The QMF metadata project designed to establish the standards, based on international best practice, for all relevant parts of the survey life cycle. 

• Quality Review System – This is a self-assessment tool which allows survey owners to review the quality of their statistical processes against the principles of the ESCOP for each phase of the GSBPM they are using.

• Data Management and governance support tools – These include data owners attesting to which data they own and are responsible for, where this data is located and who can access this data. 

The quality of statistics is assessed according to the five quality criteria: relevance, accuracy, timeliness and punctuality, accessibility and clarity, coherence and comparability. Based upon the methodology that is in place and following the documented processes, the EHECS produces high-quality, timely and coherent output that is easily accessible and which meets the needs of its users and the legislative requirements.

A Quality Report: Standard Report on Methods and Quality for Earnings and Labour Costs is updated annually and is available on the CSO website:

The overall quality is deemed to be good, with very limited revisions in the European aggregates.

The European Community, and particularly its economic, employment and monetary authorities, need to have regular and timely labour cost indices for the purpose of monitoring changes in labour costs. A timely and consistent Labour Costs Index is also of utmost importance for the European Central Bank (ECB) to monitor inflation in the European Monetary Union (EMU), and for European partners to use in negotiating pay deals. National users utilise the quarterly Earnings and Labour Costs release based upon data from the Earnings, Hours and Employment Costs Survey (EHECS) for data on earnings, hours and labour costs.  National users include government departments, the Irish Central Bank, social partners such as trade unions and employer’s associations as well as the Central Statistics Office itself (e.g. for the calculation of National Accounts outputs).  

The LCI does not discount the "compositional effect" within an economic sector (i.e. changes in the hourly labour costs due to a change in the employment composition such as the share of low paid/high paid jobs). It is based on average hourly labour costs by NACE section and not on a fixed basket of job profiles. This means that, for instance, the LCI may increase due to the redundancy of low paid workers within one sector as the average of the remaining workers are higher paid than the ones made redundant.

Private companies may use the LCI for indexing contracts, wage agreements and competitiveness analysis.

Eurostat uses the LCI to extrapolate annual labour cost levels from the benchmark data collected every 4 years through the Labour Cost Survey.

There is regular contact with the main users of the data. The range of data disseminated has broadened as a result of requests from users including (i) average weekly earnings, average hourly earnings and average weekly paid hours are now published at the more detailed NACE division (i.e. 2-digit level) on Statbank and (ii) the introduction of the Earnings and Labour Costs - Annual publication.   

Regulation (EC) No 450/2003 of the European Parliament and of the Council concerning the labour cost index is fully implemented.

The EHECS covers enterprises in NACE sections B to S with 3 or more employees. Vocational training costs and other expenditure such as recruitment costs and spending on working clothes are not captured as they only account for 0.5% of labour costs. Paid hours not worked (e.g. annual leave, bank holidays, paid sick leave) are not captured separately by the survey. They are included in the contracted hours variable.

The survey information is collected by the Central Statistics Office (CSO) under the S.I. No 115 of 2018 Statistics (Labour Costs Surveys) Order 2018. The survey results meet the requirements for Labour Costs statistics set out in Council Regulation (EC) 530/1999.

100%

The statistical accuracy and reliability is determined by the accuracy and reliability of the source of information used in preparing the LCI, the Earnings, Hours and Employment Costs Survey.

The design of the sample attempts to minimize sampling errors and the various processes of the survey are intended to eliminate or reduce as far as possible the errors both in the collection phase and in editing, weighting and imputation stages.

A detailed review of preliminary estimates and final data was undertaken to ascertain the extent of change between both sets of results. The review focused on response rates, scale of revision at NACE section level and significant changes in trends. Analysis of the results for individual NACE sections highlighted that the change from preliminary to final data was broadly in the range of +/- 5%.

Labour costs are statistical estimates that are subject to sampling errors because they are based on a sample of enterprises with 3 to 49 employees (in conjunction with a census of enterprises with 50+ employees) which are not the complete universe of all enterprises. In addition, individual enterprises in NACE sections B to S with less than 3 employees are not included in the sample. The CSO tries to reduce sampling errors by using a sample of enterprises that is as large as possible while taking burden on enterprises and time and resource constraints into account.

Coefficients of variation are calculated each quarter to measure variability in each indicator and Nace classification. This highlights any sampling error that may arise and identifies where further quality assessment needs to be focused.

Following a Lean Six Sigma project response rates for preliminary and final data are set to ensure that the change from preliminary to final data remains broadly in the range of plus or minus 5% at a one digit Nace level. Other non-sampling effects such as errors on the Business Register, questionnaire errors or processing errors are all examined as part of the quality assurance techniques used above.

All enterprises are allocated to a NACE according to the NACE recorded on the CSO’s Central Business Register (CBR) and this is assumed to be correct.

All known active enterprises with 3 or more employees are included in the sampling frame so no non-sampling errors outside the minor non-coverage within the CSO’s CBR are known to exist for enterprises in that size class. Enterprises with less than 3 employees were not sampled and are therefore not reflected in the results.

The CSO’s CBR is updated on an ongoing basis to ensure continuing relevance. However, there can be a time lag in updated information being reflected on the CSO’s CBR. As such there can be limited under-coverage where new births of enterprises are not reflected for some time and also over-coverage where closures are not immediately reflected. If such ceased enterprises are included in the quarterly sample and found to be ceased the CSO’s CBR is updated accordingly. The exact level of over and under coverage cannot be estimated.

Over-coverage is analysed at the time that the local register is synchronised with the Central Business Register, while under-coverage following analysis does not appear to be an on-going issue for EHECS.

Not available.

The vast majority of the data is sourced from payroll information thus differences between actual values and those collected by the survey are assumed to be small. For both the scanned and electronically submitted data an extensive range of edit checks were undertaken and respondent enterprises were contacted with follow up queries to identify corrections to the data.

Response Rates for 2022 were as follows:

Unit and item non-response are detailed below.

Unit non-response is dealt with by imputation or weighting.  

Non-response firms with greater than 50 employees will be imputed based upon their previous returns. If they have not returned for 5 or more quarters imputation will be based upon the average of their sample sub-sector.

Non-response firms with less than 50 employees will be accounted for in the weighting process.

If there is item non-response in a survey return, the firm is omitted and considered a unit non-response.

Data is scanned and verified for paper returns, while the electronic returns are uploaded directly to the Data Management System. A dedicated team assists respondents to set up CSO compatible payroll software minimising processing errors.

Imputation is undertaken for non-respondent enterprises with 50 or more employees.  Where an enterprise responded in any of the previous three quarters ratio-imputation is used replacing any missing value with the proportional change in that variable based on all other respondents in the quarter.  Otherwise a stratum average (mean) imputation method is used replacing any missing value with the mean of that variable for all other cases in the quarter (both methods are based on respondent enterprises of a similar size and activity).

Imputation rate for 2022 is illustrated below.

Data model is designed to produce consistent and comparable periods of earnings figures.

Trends in earnings can be affected by the composition of employees in a given sector or group, and characteristics such as length of service, educational attainment and nature of work should be taken into account, but none of these variables are available from the EHECS survey.

LCI is provided to Eurostat unadjusted, working day adjusted and seasonally & working day adjusted.

The LCI time series are adjusted using the X13 ARIMA through JDementra+. The specification and parameters used for each series are reviewed each year and past LCI data is revised on this basis. Seasonal adjustment models continue to be potentially revised, albeit to a limited extent, due to the potential impact of COVID-19 events as outliers.

An indirect chain linked approach is taken to ensure sonsistency between the adjusted series over time, between series and between individual NACEs and combined NACEs.

Provisional data is revised with final data in the subsequent quarter.

The percentage change from preliminary to final data for each indicator and NACE category is published in each EHECS release.

Final data is not subject to revision.

As is the usual practice with short-term data, provisional data is published initially, followed subsequently, at the release of the next quarter’s data, by final data.  For each quarter there will be the current quarter’s provisional data and the previous quarter’s final data.  Thus, there will be a revision to the previous quarter of LCI data each quarter when the current quarter’s LCI data is being sent to Eurostat. Differences between provisional and final data are usually minor.

LCI data sent to Eurostat is seasonally adjusted. The seasonal adjustment specifications for each indicator are updated annually. This update can lead to revisions in historic time series extending back throughout the entire time series. 

Notice about major changes or revisions (e.g. in classification, methodology or base year) are provided in advance of the change. Notice about minor changes or revisions are given at the time the change is introduced. Major revisions can apply to any point in the series. A record is kept in SAS datasets of the old data that has been revised. Minor revision or minor errors that arise are revised for the preceding four quarters.

Each quarter preliminary LCI estimates are produced for the most recent quarter. Preliminary LCI estimates for the previous quarter are replaced with final estimate. Both are transmitted to Eurostat.

The percentage change from preliminary to final data for each indicator and NACE category is published in each EHECS release.

The final data are not subject to revision.

A detailed review of preliminary estimates and final data is undertaken to ascertain the extent of change between both sets of results. The review focuses on response rates, scale of revision at NACE two-digit and one-digit level and significant changes in trends. Analysis of the results for individual NACE sections highlighted that the change from preliminary to final data is broadly in the range of +/- 2%.

Preliminary data versus final data Q3 2022

Provisional Results: The preliminary data are published by the CSO within T + 56 days after the end of the reference quarter. Due to the difficulty of enterprises accessing and sending data to the CSO as a result of the COVID-19 crisis, 2020 results were published within T + 70 days after the end of the reference quarter. Data is transmitted to Eurostat within T + 70 days.

Final Results: The final data are published by the CSO within T + 162 days after the end of the reference quarter and transmitted to Eurostat within T + 162 days. 

Provisional results are provided in T + 70 days after the end of the reference period.

Final results are provided in T + 162 days after the end of the reference period.

See below.

The LCI data has been transmitted to Eurostat within the agreed time frame.

The definitions of the cost components which the labour cost index is based upon correspond to those used in the labour cost survey. These definitions were utilised in the setup of the EHECS survey, thus are comparabile with other member states who also implemente EU regulations relating to the labour cost index. 

After imputation, all enterprises with 50 or more employees are accounted for and included in the final dataset.  For those cells where a sample survey of enterprises is used (enterprises employing between 3 and 49 employees), the results are expanded using the grossing factors to cover the entire population for the relevant industries in the quarter.  Macro edits are carried out at this stage and any outliers are investigated and corrected. Coherence of the data is ensured by scrutinising quarter on quarter changes.

The definitions of the cost components which the labour cost index is based upon correspond to those used in the labour cost survey. These definitions were utilised in the set up of the EHECS survey, thus are comparable with other member states implemented the EU regulations concerning the labour cost index. 

not applicable

Data are comparable over time.

The EHECS was established in 2008 to collect short-term earnings, hours and employment costs statistics for sectors B to S. It is not directly comparable with other discontinued short-term earnings inquiries conducted by the CSO prior to 2008 such as the Quarterly Industrial Inquiry (QII), Quarterly Services Inquiry (QSI) and the Quarterly Earnings and Hours worked in Construction (QEC).

The main differences are:

• The EHECS collects data on the entire reference quarter, the first and last day of the quarter, while the QII, QSI and QEC only collected data for a reference week in the quarter.
• Data on earnings and labour costs per hour is generally presented on the basis of hours paid and worked in the EHECS.  Data on earnings per hour was presented on the basis of hours paid (including paid leave) in the QII.
• The EHECS uses a standardised form for all NACE sectors with a standard occupational classification for all enterprises while the QII, QSI and QEC surveys had their own occupational classifications.  However the EHECS category “Production, Craft and other Manual workers” corresponds broadly to the “Industrial” category in the QII; the EHECS category “Clerical, Sales and Service Workers” also corresponds broadly to the QII category “Clerical and other office staff”; the QII category “Managerial and technical staff” is largely equivalent to the EHECS category “Managers, Professional and Associated Professionals”.
• The EHECS collects data for enterprises while the QII collected data for local units.
• Data on hours is collected for all categories of employees in the EHECS, while such data was limited to the industrial workers in the QII, non-managerial employees in the QEC and not collected at all in the QSI.
• The earnings data collected for the EHECS includes irregular earnings, irregular bonuses etc. while these items were excluded from the QII, QSI and QEC which only collected data on regular earnings (including regular bonuses) and overtime.
• Non-labour costs such as employers PRSI, other social costs, benefit in kind etc., are collected for the EHECS but were not collected for the QII, QSI and QEC surveys.

Comparable time series are available from Q1 2008.

Results are compared to other sources. Public sector employment is compared to data from the Department of Public Expenditure and Reform. Any differences are researched and explained before the results are finalised.

Quarterly earnings, hours, employment and labour costs data are matched across time to create an annual dataset. From this annual estimates of earnings and labour costs are generated and published.

The earnings, hours, employment and labour costs data that is used to calculate the LCI is also utilised in National Accounts calculations. The National Accounts department validate the earnings and labour cost data by scrutinizing the trends in the time series and comparing to admistrative earnings data. Coherence with National Accounts is assessed on a quarterly and annual basis.

Data is tested for coherence at individual enterprise level each quarter and inconsistencies are followed up with respondents. Additionally, specific coherence edits are used for respondent enterprises with 250+ employees by comparing data with the corresponding data from the previous quarter.

Earnings, and employment data from the EHECS survey are systematically checked for coherence with other internal sources such as the Earnings Analysis using Administrative Data Sources (EAADS) and the Structure of Earnings Survey.

Ireland does not publish the LCI index nationally. National users rely on the Earnings and Labour Costs quarterly release for data on earnings, hours, employment and labour costs.  Ireland supplies provisional data by the required deadline and is, therefore, included in the Eurostat LCI related news releases via the Eurostat website.

Eurostat Labour Market Publications

Eurostat Labour Costs Database

Data that are published in the Earnings and Labour Costs - Quarterly and Earnings and Labour Costs - Annual publications are also made available simultaneously on StatBank (CSO Main Data Dissemination Service).

https://data.cso.ie/product/ELCQ

NA

None

None

Information on methodology is available within the (i) Earnings and Labour Costs – Quarterly and (ii) Earnings and Labour Costs - Annual publications (see Background Notes section of publication). A quality report is prepared annually.

The quality report for the EHECS is updated annually and outlines the quality procedures and practices in place. 

Fully complete

NA

In order to reduce the burden on enterprises a review of the EHECS was carried out in 2009. As a result a number of methodological changes were introduced which included reducing the content and complexity of the original questionnaire, introducing a shorter questionnaire for small enterprises and promoting and supporting the EHECS Payroll Project which provides an electronic facility to enable enterprises to extract the required survey data directly from their payroll systems.

In 2013, an annualised estimate of the time burden on respondents - based on the total number of returned forms - was calculated to be 7,800 hours. 

The data collected is treated as strictly confidential in accordance with the Statistics Act, 1993. The provision on statistical confidentiality is regulated by Sections 32 and 33 of the 1993 Statistics Act.

The CSO cannot disseminate, or make available in any way, individual or aggregate data that could lead to the identification of any individual person or entity.

The following confidentiality rules are followed for published data:

1. If an enterprise has 80% or more of the total employment in a division (i.e. 2-digit level) or section (i.e.1-digit level) then the cell is suppressed.
2. If two enterprises have 90% or more of the total employment in a division or section then the cell is suppressed.

A confidential cell that is suppressed (as described above) is aggregated with other confidential and/or non-confidential cells to produce a non-confidential aggregate that can be published.

None