Income: further information

These Research Outputs are part of our research into the potential use of administrative data and surveys to produce population, household and characteristic information. The Income Research Outputs are not official statistics. They have been published as outputs from research to test the feasibility of a different methodology to that currently used in the production of official income statistics.

The Research Outputs on income from Pay As You Earn (PAYE) and benefits published in 2016 are an overview of what can currently be produced from the income administrative data sources that we have access to. Data tables released include PAYE and benefits income distributions by local authorities for all persons and for England and Wales by 5-year age group and sex.

These Research Outputs do not have National Statistics status and are not to be used as a substitute for official income outputs produced across the Government Statistical Service (GSS). Our Research Outputs series has been developed to keep population data users up to date with our assessments of administrative data and, in the longer term, to show progress towards a possible future census alternative.

These Income Research Outputs have been published as outputs of research to test the feasibility of a different methodology to that currently used in the production of income statistics.

The Income Research Outputs for 2016 use the Statistical Population Dataset (SPD) V1.0 (as published in the 2015 Research Outputs) as a population base. We have access to data on income from the National Benefits Database, Single Housing Benefit Extract, tax credits and Pay As You Earn (PAYE) data supplied by the Department for Work and Pensions (DWP) and HM Revenue and Customs (HMRC). We link these together and to the SPD to derive the income of the population in an aggregated, de-identified manner. For more detailed information, please see the method section of our publication.

Our Income Research Outputs use the Statistical Population Dataset (SPD) V1.0 as a population base. This uses the NHS Patient Register, the Department for Work and Pensions (DWP) Customer Information System and data from the Higher Education Statistics Agency to produce estimates of the population size. In order to measure the income of this population, 4 additional datasets were used: the National Benefits Database, Single Housing Benefit Extract, tax credits and Pay As You Earn (PAYE) data.

We work closely with data suppliers to understand and improve statistical quality issues identified in the data. For this purpose a Data Suppliers Group has been set up, which includes members from government departments across Whitehall. This group is a forum for discussing issues relating to data sharing, sharing research and building stronger relationships between data suppliers and users.

An income working group for census was also set up to support this publication, which includes members from the Department for Work and Pensions (DWP) and HM Revenue and Customs (HMRC).

Income is a variable that users repeatedly ask to be included in the census. The 2007 Census Test showed that the inclusion of income questions reduced overall response rates by a statistically significant 2.7 percentage points. These results were consistent with findings from the 1997 Census Test. There were also concerns raised about the quality of income data collected using the census test questionnaire and whether respondents understood, and were content to answer, income questions. As a result, an income question has never been included in UK censuses.

The 2021 Census topic consultation outlined our intention to explore administrative data on income. The aim is to produce income as an additional census topic by linking administrative data and surveys to the census. We also need to test whether an income variable is possible for the Administrative Data Census project. This would provide evidence to inform any recommendation we make about the future of the census in England and Wales. This initial research is currently within this Administrative Data Census context (that is, using the administrative data as the population base); however, it is also applicable for use with a census population base.

There are a number of income outputs published across the Government Statistical Service (GSS), which are discussed in A guide to sources of data on earnings and income. Most of these statistics are produced directly from surveys except for our small area model-based income estimates (SAIE), which are produced from census, administrative and survey data. Outputs that exist across the GSS are all valuable, but they have limited scope for multivariate analysis, particularly at lower levels of geography, such as income by ethnicity or income by social class, for small areas.

The Census Transformation Programme is conducting research to see if it is feasible to fill this gap in official statistics by using administrative data. This is the first publication towards producing multivariate, small area income outputs. Each year we hope to expand the coverage and geographical breakdown of these outputs (subject to statistical data quality and access). These outputs have been published to share initial research and collect feedback on our definitions and methods. This will help us to establish if it is feasible to produce income outputs from administrative data in the 2021 Census and, in the future, through an Administrative Data Census.

There are a number of income outputs published across the Government Statistical Service (GSS), which are discussed in A guide to sources of data on earnings and income. These are all valuable, but they have limited scope for multivariate analysis, particularly at lower levels of geography.

The 2021 Census topic consultation outlined our intention to explore administrative data on income. The aim is to produce income as an additional census topic by linking administrative data and surveys to the census. We also need to test whether an income variable is possible for the Administrative Data Census project. This would provide evidence to inform any recommendation we make about the future of the census in England and Wales.

Responses to the 2021 Census topic consultation highlighted this information gap within official statistics. Although income data from commercial services exist and were referenced, they were not meeting the needs of all respondents. The consultation found that where such income data has been accessed, it was used to identify small areas of disadvantage which may be hidden when looking at data for larger areas.

The official statistics published from other sources should be used in analysis and decision-making. There are a number of limitations of these Income Research Outputs which are described in the publication. The differences in definitions between these outputs and official statistics are described in Table 1 of the publication.

As with all of the Census Transformation Programme’s Research Outputs, Income Research Outputs do not hold National Statistics status and are not a substitute for other sources. They have been published as outputs of research to test the feasibility of a different methodology to that currently used in the production of official income statistics.

One of the reasons why the Income Research Outputs differ from existing publications is that they use a Statistical Population Dataset (SPD) as the population base. There are also a number of components of income missing, such as income from self-employment. Consequently, because of the different datasets used as well as various population inclusion and exclusion rules, these Research Outputs will not be directly comparable with other publications.

The official small area model-based income estimates (SAIE) for the tax year ending 2014 have been published at the same time as this release of Income Research Outputs. The SAIE publication provides official statistics below the local authority level to be used in analysis and decision-making. The statistics are produced using a model-based method to combine data from the Family Resources Survey with published census and administrative data.

No. We are not interested in the income of identifiable individuals. The Income Research Outputs have been aggregated into 5-year age bands by sex. We have followed our disclosure control policy. No individuals can be identified within these Research Outputs or from the text accompanying the outputs.

We have a formal process for dealing with access requests for the data we have collected; details of this are in How we permit access to data for research. Access requests for data that have been provided by other government departments are directed to the owning department and not answered by us.

All of our employees have a statutory responsibility to ensure the confidentiality of the data they work with. We have published an initial privacy impact assessment to show how this will be done and to demonstrate compliance with our legal obligations.

The Income Research Outputs measure individual gross annual income comprising income from a limited number of sources. At present the income definition used in the Research Outputs does not meet the internationally agreed definition of gross income (as defined in the United Nations Economic Commission for Europe (UNECE) Canberra Group Handbook). However, we aim to work towards this definition with improved availability of administrative data in the future.

The income measure used in the Income Research Outputs includes:

• gross earnings (net of pension contributions) from employment including any benefits in kind paid through Pay As You Earn (PAYE)

• state support – most benefits, including tax credits (see Annex B in the publication for more information)

• income from occupational and personal pensions included on the PAYE dataset

Excluded from the income measure used in the Income Research Outputs – but included in the international gross income definition – are:

• income from self-employment or income from an employer not paid through Pay As You Earn (PAYE)

• investment income including interest from Individual Savings Accounts (ISAs) and other saving accounts, bonds, stocks and shares

• some state support – some benefits including Child Benefit, Winter Fuel Payments, Universal Credit and Personal Independence Payments

Excluded from the income measure used in the Income Research Outputs – and excluded from the international gross income definition – are:

• maintenance payments

• income from educational grants, scholarships and student loans

• the cash value of certain forms of income in kind (free school meals, free school breakfast, free school milk, free school fruit and vegetables, Healthy Start vouchers and free television licence for those aged 75 and over)

Currently, the Research Outputs estimate personal income. This is because a Statistical Population Dataset (SPD) household population base has not yet been developed. However, following the production of the SPD household outputs, it may be possible in the future to estimate household income.

We would like feedback on the income bands presented in this publication. There were other options considered, such as using tax thresholds, for this early research. Feedback on the levels of the bands will be considered in future developments.

Only 2% of the Statistical Population Dataset (SPD) population fall into the £60,000 and over income band based on the data we have access to at the moment. Therefore, there is little value in breaking down this income further, especially while a number of components of income are missing from the outputs.

The Income Research Outputs cover income at a national, regional and local authority level for England and Wales.

Income from such people would be included if they were incorrectly estimated to be resident (according to the Statistical Population Dataset (SPD) definition) and they generated an income within England and Wales in this period. However, if they are not in the SPD, they would not be included.

In our Research Outputs release on 17 November, we highlighted the need for us to have access to more data that can (in combination) more accurately estimate those people who are both resident in the country and abroad.

No. At present we have not done sufficient research to understand the patterns at single year of age, especially while there are a number of components of income missing from the outputs. We would be interested to know whether there is demand for these outputs at single year of age to consider in future developments.

The Income Research Outputs cover some income information for 87% of people aged 16 and over. This percentage varies by age and area. However, it is not possible to know the proportion of people who have an element of income excluded from the Research Outputs. Some information is available in section 6 of the Income Research Outputs publication, such as the proportion of the population who are self-employed and will have their income from self-employment excluded.

Yes. The Income Research Outputs have been aggregated into 5-year age bands at the national level and total population by local authority.

A feedback form is published alongside this publication. In particular we welcome feedback that helps to improve and develop our methodology and definitions. It would be particularly helpful to have feedback relating to this publication by 28 February 2017.

This year our Income Research Outputs are produced directly from administrative data. There are limitations in terms of coverage and they have only been produced down to local authority level (represented by the grid at the bottom of Figure 1). In 2017, dependent on improvements in our population estimates methodology, we are planning to produce these outputs for households, rather than just individuals. Next year we also aim to publish initial research into combining administrative data and survey data. This will be our first exploration into an “integrated sources” output. Our aspiration is to use these developments to expand the coverage of the Income Research Outputs and to produce statistics to a lower geographic level. These improvements are dependent on methodological development and statistical data quality and access.