Findings: Data gaps

Identifying and addressing data gaps will enhance insight

Household surveys contain important data gaps and assumptions

We identified a number of data gaps in the coverage and granularity of income-based poverty statistics. In particular, the homeless and those not resident in private households are systematically excluded from statistics based on surveys that sample on a household basis. A number of users also highlighted concerns that individuals with no recourse to public funds may be more likely to be missed from surveys that sample on a household basis. These users told us that these groups are more likely to be living at the lowest ends of the income distribution and therefore an important omission from the statistics. The UK Statistics Authority’s Inclusive Data Taskforce, who were established in October 2020 by the National Statistician to improve the UK’s inclusive data holdings across a broad range of areas, have an interest in addressing data gaps concerning the non-household population.

Household surveys conducted by ONS and DWP also contain some crucial underlying assumptions about the structure of households that may affect interpretation of the statistics. For example, a number of users raised the issue that household surveys assume equal sharing of income, which might result in household members who receive an unequal distribution of income being ‘hidden’ in the statistics. DWP and ONS should ensure they are clear about the strengths and limitations of household surveys, particularly with regards to missing groups, and clearly set out the implicit and explicit assumptions that underline them.

The presentation of subgroups masks nuances of how these groups are affected by poverty

In the past, conversations around poverty were largely focused on those who have no earnings and rely on benefits for income. Public debate has moved on in recent years to include the concept of ‘in-work poverty’. We found that the statistics published by DWP and ONS provide a good base of information on the number of people in in-work poverty. Users told us, however, that more could be done to differentiate the causes of in-work poverty as this umbrella term masks a variety of different reasons why someone may be in in-work poverty. For example, it would be helpful for the statistics to break down whether the individual is in in-work poverty due to low wages, low hours or because of other factors. DWP and ONS should provide clarity on the different reasons why individuals might be in in-work poverty in their bulletins, to support users’ understanding, and consider whether it is possible to produce breakdowns for the reasons someone is in in-work poverty.

The relationship between ethnicity and poverty is of particular interest to a number of users. The data collected in FRS is presented in some of DWP’s bulletins at a high-level breakdown of Black, Asian, Mixed, Other and White. Separate data is published in the HBAI data tables and on the Race Disparity Unit’s Ethnicity Facts and Figures website for Indian, Pakistani, Bangladeshi, Chinese and Other Asian Groups. Whilst FRS does collect subgroups of ethnicity, the quality and granularity of this data can often be limited by small sample sizes. The limited sample sizes are also a concern for ONS’s data on household income as well.

Users raised concerns that the current presentation of ethnicity in the statistics masks differences in poverty rates for particular ethnic groups, for whom qualitative research points to them being more likely to be in poverty. For example, there are known differences in poverty rates for Pakistani and Indian ethnic groups. We found that many users are increasingly interested in the relationship between poverty and the intersectionality of personal characteristics. However, the limited sample sizes for individual ethnic groups prevents robust multivariate analysis from being carried out. We recommend that DWP and ONS address the ethnicity data gap, as part of the wider GSS response to the findings of the Commission on Race and Ethnic Disparities’ report. We are pleased to hear that ONS is already investigating the feasibility of ethnicity boosts to its household finance surveys and would encourage statistics producers to share their knowledge and approaches in this area. Planned boosts to DWP’s FRS sample in the future may also go some way to improving analysis for ethnic groups.

There are a lack of sub-regional breakdowns in income-based poverty statistics

The UK government’s ‘Levelling Up Agenda’ has created a significant appetite amongst users for data on household income that is disaggregated by geographical location. We found that local authorities and councils, as key users of data on poverty require strong sub-regional data to support local interventions. They feel that sub-regional breakdowns are a significant data gap in income-based poverty statistics as poverty rates can vary considerably between regions and within cities. These gaps impede planning and policymaking at smaller levels of geography. In the absence of good, granular data on income, some users turn to the Indices of Multiple Deprivation to try and understand poverty at a local area level by using deprivation as a proxy for poverty.

In March 2020, DWP and HMRC produced a new joint-release for the first time on children in low-income families (CILIF) at a local area level. These statistics bring together administrative data from HMRC and DWP on benefits and child tax credits, which is then combined with survey data from HBAI to provide a more granular picture of low-income amongst families with children at a local area level. From 2021, this release is now solely produced by DWP. Users are very positive about this release and we found that it is being used widely to understand how low-income varies between regions. We heard from users that they would like this analysis to be rolled out across other groups, such as working-age adults without children and pensioners. We recommend that DWP considers the potential to extend the low-income families at a local area level analyses to working-age adults without children and pensioners. The planned sample boost to the FRS could go some way in addressing gaps in sub-regional analysis going forward. Some users raised concerns about the coherence of regional breakdowns in CILIF with the national breakdowns available in the HBAI statistics. The CILIF statistics are coherent with HBAI data by their construction, however, the feedback we received from users indicated that which statistics they are constrained to could be made clearer.