Office for National Statistics

Labour Market and Households Division

Government Buildings
Cardiff Road
Newport
South Wales
NP10 8XG

[Not requested]

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

ASHE is based on a 1% sample of employee jobs from the UK PAYE (Pay As You Earn) tax register held by HM Revenue and Customs. All employees with National Insurance numbers ending in a particular pair of digits are selected and questionnaires are sent to their employers. This method gives a random 1% sample of the frame population to be selected, across businesses in all industries and of all sizes. Since National Insurance numbers are allocated randomly to individuals, the basis of ASHE sampling is probability sampling. However, the same pair of digits for National Insurance numbers is chosen each year to allow continuity in the data and enable comparisons of earnings year on year.

[Not requested]

[Not requested]

[Not requested]

[Not requested]

[Not requested]

Not available.
New concept added with the migration to SIMS 2.0.
Information (content) will be available after the next collection.

[Not requested]

The data for the UK Structure of Earnings Survey (SES) is taken from the Annual Survey of Hours and Earnings (ASHE).  This is the most comprehensive source of data on the structure of earnings in the UK.  ASHE estimates are used throughout government for policy purposes, such as setting the National Minimum/Living Wage, looking at gender pay issues, pensions policy, transport policy, allocation of local government grants, allocation of health funding and estimating effects of tax changes.  They are also used in the wider community for wage negotiations, research projects, recruitment and a wide variety of other purposes.  Some government departments and academics have access to the microdata through data access agreements.  Other users request ad-hoc analyses, which are produced by the Office for National Statistics (ONS) at a charge.  Some of the main users of ASHE data include HM Treasury, the Low Pay Commission, the Department for Business, Energy and Industrial Strategy, the Department for Work and Pensions and the Department of Health.  Feedback from key ASHE users has shown that the data collected meets most of their needs.

The 2015 provisional ASHE estimates, from which the SES 2014 data are taken, were published in full and were complete.  The published tables are released in the public domain and are available on the ONS website along with articles of interest and methodological information.  Some ministers and nominated individuals from government departments have pre-release access to the published results 24 hours prior to the release.

Government users
ONS has established an ASHE user group which meets annually and includes representatives from 15 government departments. This group discusses changes to the questionnaire, methodology and survey outputs. ONS also meets with some of these users as and when necessary.
ONS conducted a triennial review of ASHE in 2011 which included a user questionnaire. With few exceptions, users said that they regarded the ASHE data and outputs as ‘good’ or better in respect of their accuracy, relevance, timeliness, coherence, comparability and accessibility.

Non-government users
ONS conducted four strands of research into the uses made of ASHE by non-government users between 2012 and 2013. Feedback from these users indicated that the majority were satisfied or very satisfied with the extent to which ASHE outputs met their needs.

[Not requested]

[Not requested]

[Not requested]

Sampling error occurs because estimates are based on a sample rather than a census. ASHE estimates this error through coefficients of variation (CV) which are published alongside all ASHE outputs. The CV is the ratio of the standard error of an estimate to the estimate itself, expressed as a percentage. Generally, if all other factors are constant, the smaller the CV the higher the quality of the estimate.

The tables in the attached document Sampling errors display mean values and coefficients of variation (CV) for male, female, full-time and part-time breakdowns covering hourly and monthly earnings. These statistics are available by age band, level of education (ISCED), major occupation group (ISCO-08), industry (NACE section), region (NUTS Level 1) and enterprise size band.

ASHE statistics are also subject to non-sampling errors. For example, there are known differences between the coverage of the ASHE sample and the target population (i.e. all employee jobs). Further, non-response bias may also affect ASHE estimates. This may happen if the jobs for which respondents do not provide information are different to the jobs for which respondents do provide information.

In addition, ASHE results tables do not account for differences in the composition of different 'slices' of the employee workforce. For example, figures for the public and private sectors include all jobs in those sectors and are not adjusted to account for differences in the age, qualifications or seniority of the employees or the nature of their jobs; all factors which may affect how much employees earn.

The use of the sampling procedure described above results in areas of under-coverage. For example, some low-paid jobs in businesses that do not operate PAYE schemes are not on the PAYE register. An investigation into this in 2004-2005 showed that the impact of this under-coverage on ASHE estimates was very small. In addition, the ASHE sample does not include the self-employed, employees working abroad or working in the armed forces.

The rules covering which employments employers were required to report via PAYE changed in April 2013, effectively extending the coverage of the ASHE sample to include employments that were not covered under the previous rules. The new reporting system is known as “Real Time Information” (or RTI).  Analysis on 2014 results showed that the composition of the ASHE sample was not substantially distorted as a consequence of the move to RTI. This is because the majority of the RTI-type jobs were already being reported via PAYE by employers in previous years. Consequently, ONS judges that the impact of the move to RTI on the estimates for ASHE, and hence SES, is negligible.

All employees who had a period of unpaid absence during the reference month have been excluded from the SES dataset. This will not affect the weights as they have been calculated based on the exclusion of those employees whose pay for the pay period was not affected by absence.

ASHE also selects people outside of the scope of the Structure of Earnings Survey, i.e. those in agricultural industries, but these employees have been removed from the SES dataset.

[Not requested]

The ASHE questionnaire underwent a major redesign in 2004, at which point the questions were fully tested and approved. ONS data validation and survey methodology colleagues review the content and design of the questionnaire annually. Missing data for key variables are imputed on the basis of shared characteristics with imputation ‘donors’.

Although ASHE collects information on bonuses, it is known that some respondents do not have access to this information at the time of responding. Consequently, the levels of gross pay for some individuals on the SES dataset may be lower than they are in reality. It is not possible to quantify the impact of the under-coverage of bonus payments.

The table below displays response rates for the Annual Survey of Hours and Earnings 2015 (provisional).  The unit considered for response purposes is individual employee jobs; therefore response rates are expressed in terms of employee job records.  Note that data from respondents in Northern Ireland are collected via the Northern Ireland Department of Finance on behalf of ONS; hence the table displays response rates for Great Britain, Northern Ireland and the UK. 

Item non-response is dealt with differently for variables relating to hours and pay than other variables. For five main variables (annual gross pay, basic pay, basic hours, overtime pay and overtime hours) imputation is done in full for the Annual Survey of Hours and Earnings, from which SES is derived. Donor imputation is performed by the BANFF imputation tool written in SAS.

Imputation classes are defined with relation to occupation, region, sex, adult rate marker and age band to define donor sets. Donors are then selected from another record within the appropriate donor set with similar responses for other pay and hours variables.

[Not requested]

[Not requested]

ASHE employs a range of validation checks referred to as selective editing to identify potential errors in the data collected. This system ranks all potential errors based on the level of impact that the data item would have on aggregate statistics for key variables if that data item was found to be in error. Errors with the highest selective editing scores, i.e. with the highest potential impact on aggregate statistics, are validated. In these cases, respondents are contacted via telephone to validate the data they have returned. A selective editing approach to data processing does not aspire to achieve accuracy throughout the dataset at microdata level, since it does not flag data that may be in error but that would have little impact on aggregate statistics. Instead, prioritising high-impact data items allows ONS to target validation resources in the most efficient way.

The table below shows the proportion of records where imputation was successful for each of the five main variables in ASHE 2015, based on the total number of valid records on the final dataset (186,673).

Other variables, e.g. classification variables, are not imputed for within the Annual Survey of Hours and Earnings.  The highest level of educational attainment is not collected within the ASHE.  This variable has been imputed based on Labour Force Survey (LFS) figures according to both age groups and 1 digit occupation (using the UK’s Standard Occupation Classification 2010 coding – our equivalent of ISCO-08) using random uniform probability mapping. Hence at the aggregate level, the dataset will provide identical estimates to the LFS.

Imputation has also been used for SES in some cases for the following variables:

• Size of the enterprise – Based on a growth rate multiplied by the number of employees on the 2014 ASHE dataset.
• Collective pay agreement – Based on the agreement of other employees in the same enterprise.
• Length of service – Uses a probability distribution based on age, permanent/temporary status and whether the employee had been in the same job for more than a year.
• Full-time/part-time/apprentice – Uses a probability distribution based on age, full-time/part-time status and service length.
• Form of economic and financial control – If the legal status of the enterprise is ‘non-profit institutions serving households’ then this has been set to ‘private’.  Otherwise it’s based on the status from the 2014 ASHE dataset.

The sample unit for the survey is individual employee jobs. Investigations have shown that not all respondents are equally likely to respond to the survey. For example, respondents in higher earning occupations are less likely to respond that those in lower earning occupations. Therefore, a weighting system is used which takes account of the factors that have been shown to relate most to the likelihood of response and relative earning levels. This weighting system uses 108 weighting classes based on occupation (nine classes), age (three classes), gender (two classes) and geographic location (two classes). Responses within these classes are weighted to population totals taken from the LFS, which in turn is benchmarked against the Census. Note that there is no intermediate stage of weighting to local unit.

[Not requested]

[Not requested]

[Not requested]

[Not requested]

The reference month used for the Annual Survey of Hours and Earnings, that the SES 2014 is based on, is April 2015.  ASHE and its predecessor, the New Earnings Survey, have been conducted in April of each year since 1975.  The April 2015 survey was chosen rather than April 2014 for SES 2014 because data collected for the tax year ending in April 2015 would result in annual data with nine months of relevant coverage.

[Not requested]

[Not requested]

The table below displays a basic timetable of events for the 2015 Annual Survey of Hours and Earnings.

[Not requested]

As far as possible the definitions used in the UK SES meet those specified in the Eurostat regulations and hence will be broadly comparable to other European member states. Any deviations from the definitions have been described elsewhere in this report.

[Not requested]

Since 2011 ASHE has been based on the Standard Occupational Classification 2010 (SOC 2010), which replaced the Standard Occupational Classification 2000 (SOC 2000).  This change affected the calibration weights for individual ASHE records.  At UK level, the difference between the SOC 2000 estimate and the SOC 2010 estimate for full-time median gross weekly earnings in 2011 was 0.5%.

[Not requested]

The current year of ASHE statistics is provisional and is then revised when the following year is published, as this is when information from any late returns is added to the current year.  Therefore the ASHE 2015 microdata, from which the SES 2014 is taken, were provisional at the time of transmission. The number of late returns is small and so there is rarely a large difference in provisional and revised ASHE statistics.

The two main alternative sources of earnings data in the UK are the Average Weekly Earnings (AWE) statistic and the Labour Force Survey. The AWE statistic, based on the Monthly Wages and Salaries Survey of about 9,000 employers, is the lead measure of short-term changes in average earnings in Great Britain and figures are broadly in line with estimates produced from ASHE (although methodological differences between the surveys make comparisons difficult, especially as AWE has more complete coverage of bonus pay than ASHE).  AWE figures are available with industrial breakdowns and public/private sector splits. However no information is available on occupation, hours worked, and other characteristics of the workforce.  The LFS collects information on the earnings and normal and actual hours worked of about 15,000 people aged 16 and over each quarter. In addition it collects data on a wide range of personal characteristics, including education level and ethnic origin. This enables the preparation of statistics on levels and distribution of earnings similar to ASHE but with lower precision due to the much smaller sample size.

[Not requested]

[Not requested]

[Not requested]

[Not requested]

The ASHE results published are not wholly consistent with the SES data, since the requirements of the regulation differ from the normal survey practices.

Full results of the 2015 provisional Annual Survey of Hours and Earnings are published on the ONS website (https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/bulletins/annualsurveyofhoursandearnings/2015provisionalresults) in a series of standard Excel tables.  All statistics have accompanying coefficient of variation levels.

[Not requested]

It is possible, under certain circumstances, for researchers to obtain access to the microdata with a data access agreement in place. Any analysis can be carried out at the Virtual Microdata Laboratories at ONS’s London and Newport sites.

ONS's recommended format for accessible content is a combination of HTML web pages for narrative, charts and graphs, with data being provided in usable formats such as CSV and Excel. The ONS website also offers users the option to download the narrative in PDF format. In some instances other software may be used, or may be available on request.

The statistical bulletin that accompanies the release of the ASHE results contains high-level metadata.  A more detailed Quality and Methodology Information report with additional methodological information is published on the ONS website at https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/qmis/annualsurveyofhoursandearningslowpayandannualsurveyofhoursandearningspensionresultsqmi.  This report describes the intended uses of the statistics presented in the publication, their general quality and the methods used to produce them.  Further information is also available from the ASHE Guidance and Methodology section at https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/methodologies/annualsurveyofhoursandearningsashemethodologyandguidance.  In addition, an ASHE User Group meets regularly to discuss data issues.

[Not requested]

[Not requested]

[Not requested]

[Not requested]

[Not requested]

[Not requested]

Not available