Official Statistics

Below Average Resources: Background and methodology note

Updated 11 June 2024

Purpose of the statistics

The Department for Work and Pensions (DWP) is developing the ‘Below Average Resources’ (BAR) statistics based on the approach proposed by the Social Metrics Commission (SMC).

The Office for Statistics Regulation (OSR) Review of Income-Based Poverty Statistics recommended that the DWP assess how the SMC’s proposals can be implemented to enhance the public value of our statistics. The report also recognised that a basket of main poverty measures is required to meet varying user needs, but that signposting and coherence between different statistics could be improved to help users navigate the varying measures.

Therefore, these statistics are being developed to add to the understanding of poverty in the UK alongside DWP’s existing poverty-related statistics published in the Households Below Average Income (HBAI) National Statistics publication. The SMC approach provides a more expansive view of available resources (both savings and inescapable costs) than the income measurement adopted under HBAI, and also proposes some methodological changes. A summary of key differences between the SMC approach and HBAI, and how the new measure will relate to existing measures can be found in the main report.

There is significant development work required to test the robustness and further develop different components of the measure. The new measure will be developed by the DWP through a series of Official Statistics in Development (previously called ‘Experimental Statistics’) publications.

This new Official Statistic in Development is named Below Average Resources for consistency with HBAI terminology but to reflect its broader approach. Below Average Resources considers the resources available to families (through income and savings after deductions due to outgoings and inescapable costs) to meet their immediate needs. The DWP will review the suitability of this name and seek user feedback as the measure continues to be developed.

Status of the statistics

These statistics are designated as Official Statistics in Development (previously called ‘Experimental Statistics’).

Official Statistics in Development are Official Statistics that are undergoing development or evaluation; they may be new or existing statistics, and will be tested with users, in line with the standards of trustworthiness, quality, and value in the Code of Practice for Statistics.

The DWP will seek user feedback alongside a series of publications to inform the ongoing development of the statistics to enhance their public value.

Once the development process is complete, the DWP will seek Official Statistics classification for these statistics.

How this data can be used

While still under development these statistics should be viewed as subject to further testing and change. The BAR estimates presented throughout the current publication match those published by the SMC for the period up to FYE (Financial Year Ending) 2022 and are based on the current realisation of the measure. There is significant further development work required to test the robustness of the measure and refine the approach. As this work develops, the estimates presented in the current publication are likely to change and be updated over time until a final approach is agreed.

Users are encouraged to consider the information published here to evaluate the proposed new approach to measuring poverty to provide feedback to DWP for future development. These statistics have been published alongside an initial consultation which is open from 18 January to 11 April 2024. Users should use this consultation period to provide any feedback on the current publication and development of the new measure.

Once fully developed, the new BAR measure will sit alongside and add value to DWP’s existing poverty related statistics. The main report in the current publication outlines the relative strengths and limitations of the BAR approach alongside the existing DWP measures, and the value that each brings to the understanding of poverty. Each measure allows different insights and collectively the measures further enhance the overall understanding of poverty.

The Households Below Average Income (HBAI) statistics are National Statistics and remain the source for official poverty estimates, meeting the DWP’s statutory obligation to publish a measure of relative and absolute low income for children, and low income and material deprivation for children under section 4 of the Welfare Reform and Work Act 2016.

Context of the statistics

The Social Metrics Commission (SMC) is an independent commission formed in 2016 with the aim of developing a new approach to poverty measurement that both better reflects the nature and experiences of poverty that different families in the UK have and that can be used to build a consensus around poverty measurement and action in the UK.

In September 2018, the SMC published the ‘A New Measure of Poverty in the UK’ report in which they proposed a new method for measuring poverty. The SMC have subsequently published further reports incorporating more recent data and proposing further methodological refinements.

The DWP initially started work developing this measure in 2019. This work was paused during the COVID-19 pandemic for two reasons:

1. The impact of the pandemic in 2020 had a huge impact on the data collection required to produce estimates of poverty. DWP had to prioritise work to overcome these challenges and deliver existing poverty estimates for FYE 2021.

2. The government identified that information on material deprivation, a measure of whether people have access to essential items, was not comprehensive so prioritised expanding that suite of measures to meet the needs of users. This work is now in its final stages.

Therefore, in March 2023, the government announced that work on developing this new measure would resume. Since then, the DWP has worked in collaboration with a wide range of stakeholders and users, including the SMC and poverty measurement experts, with oversight by the DWP Chief Statistician.

Source of the statistics

These statistics are developed from existing data sources, primarily the Family Resources Survey (FRS) and the Households Below Average Income statistics.

Detailed methodological information on the surveys and variables used, including coverage, robustness and consistency over time is outlined in accompanying documentation for the associated reports, as well as the reports published by the SMC. These can be found at the following links:

• Family Resources Survey
• Households Below Average Income
• Social Metrics Commission reports

These statistics use data from the FRS and HBAI publications from FYE 1999 to FYE 2022. Changes to the data collection methods and variables across this time period are detailed in the methodological reports above, and in the main report where relevant.

The SMC reports also reference additional data (from Valuation Office Agency, Stats Wales and the Scottish Government) that is used to determine room costs for the overcrowding adjustment. Interest rates from the Bank of England are also used as inputs in the mortgage capital repayment calculations.

For consistency with the SMC’s approach, the data used to produce the BAR estimates have been taken from publicly available data which can be downloaded at the UK Data Service.

These data are slightly different from those used internally by DWP to produce the published statistics, namely:

• these datasets contain no names, addresses, telephone numbers, bank account details, NINOs or any personal details that can be considered as being disclosive under the terms of the ‘ONS Disclosure Control’. Variables are also removed for council tax liability, pensioner flags for individual, head and spouse
• households with 10 or more individuals are removed
• individuals aged over 80 are reset so their age category is recorded as being over 80, instead of either 80 to 84 and 85 and over
• high incomes and pensions are top-coded so that where incomes are over 10 times the mean, they are set to 10 times the mean
• income variables are rounded to the nearest £1

The following diagram outlines how these data sources are used to produce the Below Average Resources statistics:

This diagram shows how the data sources interact to produce the Below Average Resources statistics. It shows that the Family Resources Survey is used for, for example, relative variables for sharing unit creation, demographic information, mortgage costs, childcare costs, benefit receipt, calculation of overcrowding flag. The Family Resources Survey is used to produce the Households Below Average Income data from which income variables are taken. The rooms costs data is used to calculate the overcrowding adjustment. The interest rates data is used to calculate mortgage capital repayments. These data sources all feed into data processing, which is used to produce the Below Average Resources statistics.

This process, including the data processing step, is based on the methodology outlined in the SMC’s reports. This methodology is executed through Stata code, which can be requested from the SMC and is supplied with supporting documentation outlining how the code works in more detail.

Unless stated otherwise, the methodology used in this report remains consistent with that outlined in the above reports and code.

Some additional points of clarification have been outlined below where methodological information is not fully available in the above reports:

• information on household food security and material deprivation status has been taken from the HBAI data

• the TRA calculation is undertaken at the sharing unit level. The sharing unit is defined as individuals within a household who are related. In practice, this is based on anyone self-reporting being related to another individual in the household within the FRS. Two people are considered related if they have the any of the following relationships: spouse, cohabitee, civil partner, son/daughter (including adopted), step-son/daughter, foster child, son-in-law/daughter-in-law, parent, step-parent, foster parent, parent-in-law, brother/sister (including adopted), step-brother/sister, foster brother/sister, grand-child, grand-parent, other relative

• the SMC approach sets the poverty threshold so that the number of people in poverty under the measure matches the number of people in HBAI relative AHC poverty in FYE 2017, currently set at 54% of the three-year smoothed median. However, small methodological improvements detailed in the SMC’s publications since this threshold was set mean that the number of people in poverty under the BAR measure in FYE 2017 do not exactly match the number in HBAI related AHC poverty. The SMC outlined considerations around future revisions to the threshold in their 2019 methodological report. This report notes the intention to revise the threshold only where refinements to the measure result in a significant change in numbers from FYE 2017 HBAI relative AHC levels. The deviations seen in the current statistics have not been regarded as significant enough by the SMC to revise the basis for the threshold

• where estimates require a three-year average due to small sample sizes, the estimates are created using five years of data due to the combination of the three year averaging and the smoothed threshold approach adopted in BAR:
  • three-year rolling averages have been used to report poverty rates/numbers for breakdowns for groups which are subject to smaller sample sizes to smooth out observed variation. This is an established method in the HBAI publication whereby the poverty rate for each group is reported as an average of the calculated poverty rate for the year of interest and the previous two years
  • however, the BAR method uses a smoothed poverty threshold. Under this method, to calculate the poverty threshold, the median is found from the equivalised TRA distributions for the year of interest and the previous two years. The poverty threshold is then set at 54% of this 3-year equivalised median
  • these two factors combined mean that to create a three-year rolling average of the poverty rate for a particular group, five years of data are required in total

  • for example, for an estimate of the poverty rate for the three years to FYE 2018, single year estimates from FYE 2016, FYE 2017 and FYE 2018 are needed. To produce these estimates, data is needed from FYE 2014 to FYE 2018 as outlined below:

    • for a single FYE 2016 estimate, the poverty threshold is calculated from the TRA distributions from FYE 2014, FYE 2015, FYE 2016
    • for a single FYE 2017 estimate, the poverty threshold is calculated from the TRA distributions from FYE 2015, FYE 2016, FYE 2017
    • for a single FYE 2018 estimate, the poverty threshold is calculated from the TRA distributions from FYE 2016, FYE 2017, FYE 2018
  • estimates for FYE 2021 are not presented where there are concerns about the impact of coronavirus (COVID-19) pandemic on data quality and consistency. Detailed analysis of the potential impact on data quality for HBAI for FYE 2021 is provided in a technical report. This means that for periods where the required three-year average would include data from FYE 2021, the poverty rates for these groups are not included in the calculation. However, data from FYE 2021 is used to produce the FYE 2022 poverty threshold, as outlined above:
    • for example, for an average for the three years to FYE 2022:
      • for the single year FYE 2020 estimate, the poverty threshold is calculated from the TRA distributions from FYE 2018, FYE 2019 and FYE 2020
      • for the single year FYE 2022 estimate, the poverty threshold is calculated from the TRA distributions from FYE 2019, FYE 2020 and FYE 2021
      • the poverty rates for the group from FYE 2020 and FYE 2022 are averaged
      • the single year estimate for FYE 2021 is not used due to data quality concerns
• at present, variables used for analysis of poverty rates for benefit unit type in the main report and published data tables (and additionally benefit unit work status in the published data tables only) are consistent with the benefit unit level also used in the HBAI and SMC publications. However, we avoid describing this as relating to the family to avoid any conflation with the sharing unit definition of family used in this publication
• the disability variable is based on disability within the sharing unit/family. The exception to this is for Figure 3.3 (and corresponding table) which presents HBAI data based on disability within the benefit unit, and BAR data based on the sharing unit/family. This is to allow comparison with the published HBAI data
• analysis at the benefit unit level is in contrast with the sharing unit/family level at which Total Resources Available are measured under the BAR approach. Therefore, future work will seek to examine whether new variables should be derived to reflect measurement of different characteristics at the sharing unit/family level, considering to what extent this is possible with the available data and the extent of any impact on current estimates

Comparisons between the statistics

As outlined above, the data underlying these statistics originates from the FRS, HBAI and SMC publications and so are related directly to them.

Diagram 1 in the main report illustrates how the BAR statistics in the present publication relate to other key DWP poverty related publications. Further discussion of measures related to the HBAI poverty statistics, and therefore likely relevant to the interpretation of these statistics, can also be found in this methodological report.

Definitions and terminology within the statistics

Much of the terminology used in this report has been previously defined in the source publications which can be found at the following links:

• Family Resources Survey
• Households Below Average Income
• Social Metrics Commission reports

Common terms used in the report are defined in the following table.

Term	Definition
Absolute low income	Those in absolute low income have their net equivalised income below 60% of the FYE 2011 median income adjusted for inflation.
After Housing Costs	Housing costs (for example, rent and mortgage interest payments) have been deducted.
Before Housing Costs	Housing costs (for example, rent and mortgage interest payments) have not been deducted.
Below Average Resources	The title of the new statistic in development. Below Average Resources considers the resources available to families (through income and savings after deductions due to outgoings and inescapable costs) to meet their immediate needs. Those considered to be in Below Average Resources poverty have Total Resources Available below the poverty threshold.
Benefit Unit	A benefit unit includes a single adult, or a married or cohabiting couple, plus any dependent children.
Childcare Costs	Childcare costs refer to the reported amount spent on childcare in the past seven days for children under the age of 16. Types of childcare covered include: Playgroup or pre school, Day nursery or workplace crèche, Nursery School, Nursery class attached to primary or infants’ school, Reception class at a primary or infants’ school, Breakfast club, After school club/activities, Holiday scheme/club, Special day school or nursery or unit for children with special educational needs, Any other childcare. This includes any costs paid from money received from the childcare element of Working Tax Credit or Universal Credit, salary sacrifice childcare vouchers or Tax-Free Childcare accounts; but excludes any costs paid by others for example, grandparents or paid with non salary sacrifice childcare vouchers. Further information can be found in the Family Resources Survey documentation.
Debt	Obligated debt repayments. No current technical definition as data is not available at present.
Equivalisation	Adjustment for household, or family, composition which allows comparisons to be made of individuals of different ages from different sized households. Each household, or family, member is given a standard weighting which is summed together. Currently, the OECD equivalisation process is used. There are recognised limitations of this approach however, at present, no suitable alternative has been identified. Further work will be conducted to explore alternative approaches. Further information about the equivalisation approach can be found in the HBAI documentation.
Extra Disability Costs	This is currently a proxy measure which defines disability costs as the monetary amount received from extra-cost disability benefits, namely Personal Independence Payment (PIP), Disability Living Allowance (DLA) and Attendance Allowance (AA).
Family	Family is used to refer to the Sharing Unit (see definition).
Household Food Security	Households with high or marginal food security are “food secure”. Food secure households are considered to have sufficient, varied food to facilitate an active and healthy lifestyle. Households with low or very low food security are “food insecure”. Food insecure households have a risk of, or lack of access to, sufficient, varied food. Further information can be found in the HBAI methodology report.
Household	One person living alone or a group of people (not necessarily related) living at the same address who share cooking facilities and share a living room, sitting room, or dining area. A household will consist of one or more benefit units/families.
Inescapable Costs	The inescapable costs that some families face, which reduce the resources they have. In the current realisation of the measure, these include the extra costs of disability, costs of childcare and housing costs.
Liquid Assets	Assets that are freely available to be turned into cash to meet immediate needs. This includes savings, but not non-liquid assets such as housing.
Lived Experience Indicators	A wider set of measures of some of the factors that affect the lived experience of people in poverty. These fall under five domains which are 1) Family, relationships and community; 2) Education; 3) Health; 4) Family Finances; 5) Labour Market Opportunity
Low income	Those considered to be in low income have an income below the poverty threshold, as set by the HBAI methodology. See Absolute low income and Relative low income definitions.
Low resources	Those considered to be in low resources have Total Resources Available below the poverty threshold as set by the BAR methodology. The default poverty threshold is 54% of the median equivalised total resources available smoothed over three years.
Material Deprivation	The inability to afford some items considered by most people to be desirable or even necessary to lead an adequate life (Eurostat Definition). In HBAI, material deprivation is measured differently for different age groups, based on age-appropriate questions asking whether they have access to certain goods and services. Definitions can be found in the HBAI glossary.
Overcrowded Accommodation	Households are defined as living in overcrowded accommodation using the bedroom standard established in the Housing (Overcrowding) Bill of 2003. Where a household is considered to be in overcrowded accommodation, an overcrowding flag is applied to all sharing units in the household.
Poverty Depth	How far below an individual or family’s total resources available are from the poverty line.
Poverty Persistence	An individual being in poverty in the year in question, in addition to two out of three years prior to that. Note that this is the definition proposed in the SMC’s approach, other publications may use different definitions.
Relative Low Income	Those in relative low income have their net equivalised disposable household income below a threshold set at 60% of median income.
Savings	See Liquid Assets.
Sharing Unit	Individuals within a household who are related. The definition of related is taken from the FRS. Two people are considered related if they have the any of the following relationships: spouse, cohabitee, civil partner, son/daughter (including adopted), step-son/daughter, foster child, son-in-law/daughter-in-law, parent, step-parent, foster parent, parent-in-law, brother/sister (including adopted), step-brother/sister, foster brother/sister, grand-child, grand-parent, other relative. ‘Family’ is also used to refer to sharing unit.
Smoothing	Taking a 3-year rolling average of median weekly total resources available to reflect the fact that social norms and expectations will take time to adapt to changes in overall economic conditions.
Total Resources Available	Calculation of net income, additionally accounting for other available resources from liquid assets and deductions due to inescapable costs.

Strengths and limitations of the statistics

As outlined above, the data underlying these statistics is taken from existing sources. The strengths and limitations of these underlying data sources should be considered when interpreting these statistics. The methodology notes for the relevant publications can be found at the following links:

• Family Resources Survey
• Households Below Average Income

The SMC’s published reports also outline strengths and limitations of their proposed measure.

In addition to these strengths and limitations, the present publication has the following strengths (also outlined in Diagram 1 within the main report):

• the BAR approach to poverty measurement is more expansive than that presented in the HBAI publication as it accounts for a greater range of income sources and inescapable costs.
• the Below Average Resources poverty measure has a stronger correlation with other, non-income based measures of poverty such as material deprivation and household food insecurity than does the relative AHC HBAI poverty measure

The statistics should also be interpreted in the context of the following limitations:

• the measure is still under development and so the approach proposed by the SMC has not been fully realised. For example, data on family debt is not yet available and the poverty persistence/lived experience indicators have not been included in the present publication. Further detail of the changes still to be realised can be found in the main report (see Table 1). Further testing and development is also required for the parts of the BAR measure that are included currently
• some BAR components introduce complexity. For example, some measures of costs do not account for unmet need, choice on quality, actual behavioural use of assets/costs
• the present report does not include measures of uncertainty or analyses examining statistical significance. The HBAI publication does report measures of uncertainty using confidence intervals calculated using bootstrapping methodology (further detail can be found in this methodological report). Due to the complexity of this method and the timescales for development, it has not been possible to report this in the present publication. Future work will examine whether it is possible to report measures of statistical uncertainty for the BAR measure
• the main report sets out potential timescales required for developing the measure, highlighting it could be several years before a full BAR measure is agreed and a consistent time series then established (see Section 7 of the report)

Further discussion of the strengths and limitations of these statistics can be found in the main report due to the statistics’ status as an Official Statistic in Development.

Quality assurance process

As outlined above, the data used to produce these statistics have been taken from existing publications which have their own extensive quality assurance processes. Details of these can be found at the following links:

• Family Resources Survey
• Households Below Average Income

These data sources are then processed to produce the BAR estimates in the steps outlined above. This was thoroughly quality assured as follows:

1. Code received from the SMC and quality assured though examining the logic of the code.

2. Code run and outputs quality assured against SMC’s published outputs.

3. Code developed for further breakdowns not published by SMC. Logic and outputs quality assured by colleagues outside the immediate team.

4. Tables and charts produced comparing BAR and HBAI breakdowns. These are quality assured against existing HBAI publications.

5. Throughout these stages, feedback has been invited from and quality assurance undertaken by wider DWP analytical teams and external stakeholders in other government departments and outside government with expertise in poverty measurement.

Where any errors were identified, these were noted and corrected and stakeholders were notified if relevant.

Further quality assurance processes will be established as further developments are made to the methodology underlying these statistics.

The relative strengths and limitations of the statistics have been discussed in this note and in the main report. Additionally, the Official Statistics in Development status and implications of this for further development, testing and change have been stated throughout the publication materials.

Revisions to the statistics

These statistics are Official Statistics in Development and will be subject to a series of publications which will demonstrate the DWP’s work developing the measure and seek user feedback on the development. Once a final approach is agreed, the DWP will seek Official Statistics classification for these statistics.

Further detail on initial proposed development timescales can be found in the main publication and will be revisited as the work progresses to incorporate user feedback on development priorities.

Feedback

The Department for Work and Pensions (DWP) sought user feedback on the new poverty measure through an analytical consultation running from 18 January to 11 April 2024. The consultation has now closed.

We will publish a response to the feedback we received from users by 4 July 2024.

A consultation for users to provide feedback has been opened alongside this publication. We welcome all feedback on the statistics and approach and will use this to inform the future development of this measure. Please use the survey to respond. The consultation is open until 11 April 2024. We will publish a response to the results of the consultation within 12 weeks of the consultation closing.

Any response or queries which cannot be sent through the consultation response should be directed to the BAR statistics team at: [email protected].