Social security: benefit take-up strategy

Chapter 4: Measuring Take-up

Take-up rate measures

Take-up rate reflects the extent to which people claim the benefits they are eligible for. The simplest measure of take-up rate is in terms of caseload. Caseload take-up compares the number of benefit recipients with the eligible population. The eligible population includes both those who claim the benefit and those who should claim the benefit but do not.

An alternative measure of take-up rate is in terms of expenditure. Expenditure take-up compares the total amount of benefit payments received with the total amount of benefit payments that would have been received if everyone eligible for the benefit had claimed it. This measure is useful to consider for benefits that can be paid at different levels due to the different elements they may include or benefits that are tapered. Expenditure take-up is often higher because households who are eligible for smaller payments are generally less likely to apply. DWP publishes estimates of caseload and expenditure take-up for a number of low income benefits.^[10] For the remainder of this section, all references to take-up, unless otherwise specified, should be interpreted as caseload take-up.

Review of available data

Generally, measuring the take-up rate of benefits can be challenging. Where eligibility is largely linked to household type and income, estimating the eligible population is possible. However, where eligibility determined by less rigid criteria (e.g. for disability benefits) or where it is less easily observable in the population until an application is made (e.g. for carer related benefits), estimating take-up becomes more challenging. For a number of devolved benefits, there is no established method of estimating take-up. Furthermore, there are limited examples of existing methods in the literature.

To estimate the take-up rates, we need:

• data on the number of benefit recipients; and
• an estimate of the total eligible population.

The data on number of benefit recipients would be collected by Social Security Scotland, whereas eligible population depending on the benefit can be estimated based on a range of data sources.

Data on benefit recipients

Administrative data on the number of benefits recipients can be sourced from Social Security Scotland statistics and management information data once the benefit has started being paid.

Data to estimate eligible population

For benefits where eligibility is linked to household income and characteristics (such as Best Start Grant and Funeral Support Payment, for example), population-level surveys, population statistics and microsimulation modelling can be used to estimate the eligible population.

The relevant population survey is the Family Resources Survey (FRS) which collects data on a representative sample of private households in the United Kingdom. This includes information on income from all sources, housing tenure, household characteristics and employment status. Other demographic data can be sourced from statistics produced by the National Records Scotland (NRS) and NHS Scotland. Microsimulation modelling using the FRS Scottish dataset to estimate benefit entitlement can supplement administrative data to estimate the eligible population.

For benefits, such as Disability Assistance, where eligibility is determined by a face-to-face or paper-based assessment on a case-by-case basis against set but less rigid criteria, the estimation of the eligible population is more difficult. Difficulties also arise where the eligibility criteria are specific but difficult to observe until an application is made – for example the number of hours cared for where the person cared for is in receipt of disability benefits. This is covered in more detail in following sections.

Estimate of take-up of benefits currently delivered

Social Security Scotland administrative data on benefits currently delivered are only available for Best Start Grant. Whilst Social Security Scotland currently delivers a suite of benefits, including BSG Early Learning and School Age payments and Funeral Support Payment, at this stage only data for BSG Pregnancy and Baby Payment cover a sufficient period of time to allow us to provide an estimate of take-up. This is due to there not having been enough time since the launch of other Social Security payments for there to be sufficient number of applications for data analysis.

Carer's Allowance Supplement (CAS) has also been paid to Scottish carers since 2018. However, because eligibility for the payment is reliant on receipt of Carer's Allowance and the payment is automatic, the take-up is by definition 100%.

Take-up estimates of BSG first and subsequent births

Take-up rates have been estimated by comparing the number of payments made by month of birth to the number of estimated eligible births. This section provides estimates of take-up of BSG for eligible first and subsequent births.

We note that the results presented here are provisional. We have used available data for only one set of eligible births between 1 December 2018 and 31 March 2019 where only births reported in March were subject to a full application window (9 months long). This means that more robust estimates will be possible once we have data on eligible births that cover more than one full application window. As a result we will have more confidence in the take-up estimates provided in the next take-up strategy. More detail on the methodology and limitations on estimating take-up rates for BSG Pregnancy and Baby Payment are provided at Annex C.

For eligible births occurring in the three months after the BSG launch in December 2018 (i.e. births in the period between December 2018 and March 2019), the application window will close by September 2019. Using data on payments made for these children up until September 2019, the estimated take-up rate is 67% as presented in table 1. More specifically, the estimated take-up rate is 53% for first and 77% for subsequent births respectively.

Table 1: Take-up estimate for births occurring between 1 December 2018 and 31 March 2019

	Estimated take-up rate
First births	53%
Subsequent births	77%
Total	67%

Comparison of BSG first births with SSMG take-up estimate

Table 2 compares the take-up rate estimates of Sure Start Maternity Grant (SSMG) in 2017/18 in Scotland and BSG for first births in 2018/19.^[11] It shows that that take-up rate of BSG for first births is higher than SSMG. This comparison provides useful context in interpreting the take-up rate of BSG.

Table 2: Comparing take-up estimates

	Estimated take-up rate
SSMG	41%
BSG first births	53%

The difference could be driven by the following differences in the two policies:

• the payment amount - with SSMG payment at £500 and BSG for first births at £600.
• application window - 6 months for SSMG compared to 9 months for BSG
• Service design and application – wider range of application channels and service design improving ease of claim.

Data limitations

The main limitations for estimating take-up rates of carers and disability benefits are around data that could be used to estimate eligible population. Administrative data on benefit recipients include only those who apply for benefits. Since the eligible population also includes those who are eligible but do not apply, less is known about the size and profile of that group. Whilst it is possible that some may be in receipt of benefits but are not eligible for support, this number is expected to be small.

Carer Benefits

Identifying the size of the carer population who do not claim is particularly challenging due to the complexity of eligibility criteria which are based on the hours of care given, the cared for person's receipt of disability benefit and the level of carer's earnings.

The DWP do not produce estimates of take-up for Carer's Allowance but commissioned research in 2010 which looked at feasibility of estimating the take-up rate in the UK using FRS data.^[12] The total eligible population is estimated based on the responses of people who meet all eligibility criteria for Carer's Allowance. The reported analysis suggested that of the people providing high levels of care within their own household to people in receipt of AA or DLA and earning below the Carer's Allowance earnings threshold, 65% receive Carer's Allowance. The authors cautioned that this approach is likely to result in underestimate of take-up due to misreporting by respondents in the survey. This covers misreporting related to receipt of Carer's Allowance as well as the eligibility criteria such as hours of care provided, earnings and receipt of disability benefits by cared for person. This study concluded that misreporting could be corrected through linking survey responses with administrative data.

As with DWP's Carer's Allowance similar challenges apply to the estimation of take-up rate for Scottish Carer's Assistance and Young Carer Grant. For Young Carer Grant these challenges would likely be more pronounced because the assistance targets a much smaller sub-group of general population (aged 16 to 18) and therefore the sample sizes in the existing population surveys asking caring-related questions are unlikely to be large enough to identify the total eligible population.

Disability benefits

Eligibility for disability benefits is a complex concept. This is due to the fact that eligibility is currently determined by the outcome of an assessment that is person-centred as opposed to strict eligibility criteria, such as, for example, household income or family composition. Existing data on disability are not comprehensive or granular enough to allow us to produce robust estimates of the number of people who would be eligible but do not apply.

A number of population surveys ask Scottish households disability-related questions. For example, the FRS includes questions about receipt of disability benefits or whether respondents have any physical or mental health conditions or illnesses lasting or expected to last for 12 months or more. Respondents are then asked to identify specific impairments that apply to their circumstances from a list and whether this affects their ability to carry-out day-to-day activities. Similar questions are asked as part of the Scottish Surveys Core Questions (SSCQ) which pools responses from the Scottish Crime and Justice Survey, the Scottish Health Survey and the Scottish Household Survey into one output.

While there is a range of survey questions asking about disability that may more or less capture the same concepts, these data alone may be insufficient to determine eligibility for a disability benefit. Since disability is self-reported in the surveys, there is no objective measure of the eligible population. For example, in the 2017 SSCQ publication^[13] it was estimated that there were nearly 1.1 million adults (aged 16+) who reported a long-term^[14] physical or mental condition that limited their day-to-day activities. This number is significantly higher than the total number of people in receipt of Personal Independence Payment, Attendance Allowance and Disability Living Allowance in Scotland in 2017/18 (478,000)^[15].

An additional consideration is related to the way disability related questions are asked. A common approach in the current surveys is that a question on the existence of a limiting long-term condition is asked first. Those who provide a positive response are then asked follow up questions around difficulties with activities of daily life. Jackle and Pudney (2015)^[16] find no evidence that asking first about a long-term health condition helps to filter out those with trivial or mild disability and they stress that this approach may miss substantial disabilities by only capturing needs and limitations for those who first self-identify existence of a limiting condition.

Finally, despite the survey information helping identifying some of those potentially eligible for disability benefits, there is an element of discretion in how the actual assessments are undertaken and benefit award decisions are made. As a result, those determined to be eligible through the surveys may not match those determined to be eligible through the assessments. While it is reasonable to expect that all assessors should be adhering to standard guidance and rules, the element of discretion and the person-centred nature of the assessment would mean eligibility would be difficult to infer without the actual assessment.

Can these limitations be improved?

Researchers have been assessing feasibility of using survey data to identify disabled population. Whilst DWP do not produce a take-up estimate for disability benefits, they commissioned a study^[17] (published in 2007) that developed a methodology to predict the eligible population for Disability Living Allowance (DLA) and Attendance Allowance (AA). According to this methodology, DWP administrative data can be used to understand what kind of characteristics could make people eligible for receipt of disability benefits. It was envisaged that these characteristics would in turn inform the development of questions in a new survey targeted at potentially eligible people with limiting health conditions to estimate the total eligible population for disability benefits. In a subsequent study, Kasparova et al (2011)^[18] began to explore the proposed methodology by examining whether data from the DLA claim forms and additional evidence submitted with the application could be used to predict award decisions. This method proved to be relatively accurate, correctly predicting award decisions in 71% of care cases and 73% of mobility cases. However, the final step of undertaking a larger survey to ask about limiting health conditions to estimate take-up has not been undertaken.

A similar approach that has been proposed in literature is concerned with the idea of using a broader set of survey indicators to determine eligibility for disability benefits. This approach is based on the idea that since the information on which an award decision has been made is not observable directly in survey questions on disability status, a wider set of indicators provided in surveys can be used to measure eligible population for disability benefits. This wider set of different indicators can be concerned with socio-economic status, for example level of education or income, and also limitations with day-to-day activities such as those related to difficulties with walking or personal care. Hancock et al. (2015)^[19] have explored this approach and they suggest that survey indicators used to measure disability should focus on potential difficulties with everyday activities rather than health or disease. Using data on older (aged 65+) people from the Family Resources Survey (FRS); the British Household Panel Survey (BHPS) and the English Longitudinal Survey of Ageing (ELSA) the study investigates whether the different indicators of disability collected in these surveys are consistent with receipt of Attendance Allowance (AA). In general, the three surveys show a similar relationship between the different indicators and receipt of AA.

Finally, there may be merit in using data linkage as an approach to improve the measurement of eligible population for disability and carers benefits. The key advantage of this method is that it corrects for misreporting that is common in surveys by linking survey responses to administrative data. This can improve the accuracy of estimates of the eligible population.

Considering the various limitations and challenges around estimating the take-up rate of devolved benefits, we will continue to explore methods in order to develop a systematic approach to enhancing the available data sources.