The meaning and significance of (non-)take-up

Take-up is different from uptake, which just means the actual numbers of claimants of any benefit or other provision. Take-up instead refers to the proportion of those eligible for a benefit (in cash or in kind) or other provision who receive it, and/or the proportion of the potential total expenditure claimed. This review article follows the usual practice in the literature, in which the commonest use of ‘take-up’/‘non-take-up’ refers to the first two categories of the typology developed by Warin (2016). He included: non-knowledge (lack of awareness of a benefit); non-request (awareness of, but no claim for, a benefit); non-receipt (eligible individuals request but do not obtain a benefit); and non-proposal or non-proposition (when a benefit is not offered, despite the applicant’s eligibility) (also see Lucas et al, 2021: 165).

Several authors have also discussed the relationship between coverage and take-up (Goedemé and Janssens, 2020) or rates of coverage, eligibility and take-up (Nelson and Nieuwenhuis, 2021), with the latter distinguishing between ‘formal’ (legal) and ‘effective’ (in practice) coverage, a distinction often employed in European Commission documents. They also note that empirically the focus on potential recipients – exposed to a risk, and with a legal right to a benefit – used to dominate, even though many might not get the benefit in practice, perhaps because of not meeting the conditions attached. But these terms are used more in relation to social insurance. In what follows, we focus instead on means-tested benefits, in part because these are central to the take-up literature and in part because they increasingly dominate UK provision.

Later, we first discuss the significance of non-take-up. As Baumberg Geiger (2021) notes, ‘it is sometimes said that non-take-up is not a problem, because the people that really need benefits claim them. According to this view, non-take-up improves the efficiency of the system’. But, as he points out, those not claiming benefits are often in real need. Bucelli and McKnight (2022) cite one estimate, by Matsaganis et al (2008), that imperfect take-up increased the UK poverty rate by 4 per cent, and highlight (p 14) the greater impact of non-take-up on deeper poverty. Non-take-up may result in damage to the welfare of the person concerned even if they are not in poverty as a result.

In addition, take-up of social security entitlements is an important issue of principle for social policy because, as Eurofound (2015: 1) puts it: ‘benefits are implemented by governments for specific reasons, such as preventing poverty or guaranteeing access to basic goods and services. If benefits do not reach the people they are meant for, they fail to fulfil these aims’ (and see Baumberg Geiger, 2021). As Van Gestel et al (2022) argue, non-take-up therefore undermines the effectiveness and fairness of social policies. In the UK, means-tested benefits are also increasingly acting as the gateway to other provisions, including most recently one-off payments to help with the rising cost of living. This additional function of such benefits increases the significance of non-take-up, because of its knock-on effect on access to other policy instruments.

The right to social security can also be seen to encompass the right to access and maintain benefits without discrimination (Alston with Van den Reen, 2019: 17). A workshop stream at a recent ESPANet conference linked structural inequalities, discrimination and non-take-up as related barriers to social rights. This relationship in turn highlights ‘the central role of potential beneficiaries of social rights in the definition and characterisation of… contemporary welfare states and in the experience and meaning of social citizenship’ (Lucas et al, 2021: 162). In addition to the deleterious effects of non-take-up on individuals themselves, these are the key reasons for its significance.

Measuring (non-)take-up

Non-take-up across European Union (EU) member states was examined by Eurofound (2015) and Ko and Moffitt (2022) provide a more recent review of benefits take-up in a wider range of countries. The most recent overview of the UK situation was given by Hoynes et al (2023), using 2014–15 take-up estimates, for the Deaton Review of Inequality organised by the Institute for Fiscal Studies.¹ As Bangham and Corlett (2018) noted, however, no regular official estimates are, or have previously been, available for the take-up rates of many UK benefits, including non-means-tested benefits to help with additional costs for disabled people and Carer’s Allowance for unpaid carers.

In Great Britain (GB),² take-up estimates of ‘income-related’ (means-tested) benefits have generally been published annually. In 2012, there was a proposal by the then government to halt publication, which was opposed and subsequently dropped. This led to analysis demonstrating that official estimates of take-up rates for means-tested benefits had declined over time.³ These figures have traditionally been presented in two ways (DWP, 2020): by caseload (the proportion of those who are eligible taking up a benefit); and by expenditure (how much of what should be spent on a benefit is actually spent). For 2014–15, Hoynes et al (2023) concluded that £1 in every £7 of means-tested and contributory benefits in GB remained unclaimed. While only half of those eligible for means-tested Jobseeker’s Allowance claimed it in 2014–15, 96 per cent of those entitled to Child Benefit claimed it in 2011–12.⁴ They also showed (Figure 23) that take-up generally increases with age and is higher for those with children than those without. Benefits in kind can also suffer take-up problems: Forsey (2023), for example, reports that 37 per cent of potential claimants miss out on Healthy Start vouchers, which give eligible pregnant women and low-income families with young children access to certain healthy foods. As explained later, the introduction of Universal Credit in the UK from 2013, replacing six working-age means-tested benefits and tax credits, has meant the recent suspension of take-up data for these. (‘Universal Credit’ is a misnomer: it is not universal but means tested and conditional; and it is not a credit but paid monthly in arrears.)

Estimates of take-up are uncertain (DWP, 2020; section 7; DWP, 2022a), in part because information on benefit receipt is taken from a general household survey (currently the Family Resources Survey (FRS)) not designed specifically for this purpose, known to under-report benefit receipt, and only covering people in private households. From fiscal year end 2014, and retrospectively to 2010 and 2013, more use was made of administrative data to address under-reporting of benefits (Brewer et al, 2017; Corlett, 2020), alongside the continuing use of the FRS and policy simulation (DWP, 2022a), with further changes in 2017 (Marc et al, 2022); but some information in the FRS cannot currently be linked to DWP administrative data to confirm benefit receipt. Take-up totals for certain benefits in GB are therefore usually presented as falling within a range.

HM Revenue and Customs (HMRC) provided similar estimates for the benefits and tax credits for which it is responsible until 2017–18 (HMRC, 2019), as well as publishing research about reasons for non-take-up of tax credits (Breese et al, 2011). Subsequently, however, it has only produced figures for Child Benefit, as tax credits take-up rates were affected by the introduction of Universal Credit; and it does not adjust for under-reporting of income in household surveys, as the DWP now does, as noted (Bangham and Corlett, 2018).

Countries in the EU, however, do not have such developed official practices and admire the annual estimates for GB produced by the statistical services (Marc et al, 2022).⁵ Marc and colleagues also discuss the challenges in achieving this, and potential alternative data sources; out of five European countries examined, only the Netherlands estimates take-up from local administrative data sources. They describe the UK as using take-up estimates as a social policy monitoring and management tool, and as not including secondary non-take-up (understood as not claiming all relevant benefits) or a dynamic perspective (whether non-take-up is temporary or longer term). They recommend using take-up figures for evaluation purposes. They also suggest moving from a focus on effectiveness and efficiency (the provider viewpoint) to the examination of take-up figures from the recipient perspective. This is the emphasis given here.

The annual DWP publication cycle has recently been disrupted due to the COVID-19 pandemic. But, as explained later, the DWP has also produced no take-up estimates for Universal Credit and has now stopped publishing them for other working-age means-tested benefits too (DWP, 2022b; 2022c). Future estimates will currently therefore cover only pensioner benefits. Bangham and Corlett (2018) called for a review of benefit take-up statistics.

A growing focus on ‘automation’ for take-up

The UK has been particularly interested in take-up, perhaps because of its emphasis on means-tested benefits, steadily increasing recently (Hoynes et al, 2023). But the further development of such benefits, including in-work supplements,⁶ in other European countries has resulted in more attention being paid to this issue in EU debates as well.

These debates sometimes seem to imply that automation will be the solution to take-up problems. It is often unclear what this means, although in some countries more administrative information is held about individuals (for example, for the Netherlands, see Considine et al, 2022; for Nordic countries, see Van Gerven, 2022) and could be better used to alert them to potential entitlement.

It is argued here, however, that it is crucial not to focus only on claimant behaviours but also on the content of the offer and on how it is implemented. In other words, here we argue for taking more account of the reasons for non-take-up relating to (potential) claimants’ relationship with the state, and the characteristics of benefits left unclaimed, rather than assuming that administrative information and automation will overcome all obstacles.

This argument is developed later. But first we present an overview of the causes of non-take-up proposed in the literature and discuss the potential impact of the introduction of Universal Credit in the UK.

Overview: reasons for non-take-up

Lucas et al (2021) give a valuable overview of the research agenda on non-take-up, albeit confined to western societies. Olivier de Schutter, UN Special Rapporteur on Extreme Poverty and Human Rights since March 2020, issued a report on non-take-up of social protection rights, based in part on an international survey with 421 responses from 36 countries, and input from individuals, experts and organisations (2022: 3–5). Primary non-take-up is defined here as not applying for a benefit; secondary non-take-up is applying for but not obtaining it (though it can sometimes mean not claiming all the benefits to which someone is entitled). Following Van Mechelen and Janssens (2017), Lucas et al (2021: 165) add tertiary non-take-up, referring to vulnerable people excluded by eligibility conditions, who may meet financial rules but lack civil rights.

Although non-take-up may affect different kinds of benefits (Eurofound, 2015: 1), non-means-tested benefits seem relatively neglected in the take-up literature. Baumberg Geiger et al (2021) estimated that early in the COVID-19 pandemic some 80,000 people eligible for the UK’s New Style (contributory) Jobseeker’s Allowance (NS-JSA) did not claim it, with awareness lower for contributory benefits than for Universal Credit. Even the overwhelming majority of those whose Universal Credit application was rejected because income and/or savings were too high said that they had not considered applying for NS-JSA (for unemployed people) or New Style Employment and Support Allowance (for ill-health/disability). This may be partly explained by the lack of public and policy focus on non-means-tested UK working-age benefits (Harrop et al, 2023).

Evidence suggests, however, that means-tested benefits often suffer most from non-take-up, and the remainder of this article therefore focuses on these. They may also cause most concern, given their targeting on individuals and households with fewest resources. And non-take-up may occur for different reasons for benefits that are or are not means tested.

De Schutter (2022), like Van Mechelen and Janssens (2017), divides the major issues into obstacles for beneficiaries; problems within the administration; and the design of the system itself. Lucas et al (2021) argue for an integrated analysis of these micro, meso and macro levels within a systemic framework. In addition (pp 166–8), they draw attention to time and timing as essential in grasping the mechanisms and dynamics of non-take-up. (Entitlement is also affected by the permitted backdating period.)⁷ Crucially, Bucelli and McKnight (2022: 11), reviewing international evidence on take-up, add to these three levels – individual, administration and scheme – society itself (see ‘Macro level reasons’).

Micro level reasons

Lack of information is often the starting-point in analysing reasons for non-take-up (de Schutter, 2022). However, some studies (for example, Breese et al, 2011) find lack of information not to be the main reason for non-take-up – here, of tax credits; instead, they found people’s beliefs about (non-)eligibility, even when they knew about tax credits, leading to failure to claim. In GB, families with children are much more likely to claim Working Tax Credit than those without children (92 compared with 41 per cent in terms of expenditure in 2014–15 (Hoynes et al, 2023)); families may be more inclined to see themselves as in need, and those without children or disabilities on low incomes were only recently included in in-work tax credits. Social housing tenants are also more likely to claim housing benefit than private tenants (cited in Bucelli and McKnight, 2022: 6). Other analyses (for example, OMB, 2022, for the US) argue that provision of information by itself is not as effective in stimulating take-up by itself as when it is combined with other forms of assistance.

Another focus in investigating reasons for non-take-up is the costs of applying, in terms of time, money and/or psychological burdens (for example, Bennett et al, 2009); OMB (2022: 3), for the US, divides the processes involved into learning about, applying for and using public services and benefits. Ko and Moffitt (2022) analyse the tension between utility and amount of money gained. Cost/benefit analysis can assess how potential claimants may see trade-offs (for example, OMB, 2022; and Finn and Goodship, 2014, who include the effort involved in continuing claims, as well as original applications). The ‘claiming costs scale’, including information, process and stigma, has recently been developed by Janssens et al (2021). There is a debate about whether it is the more needy or less needy who are excluded (Hoynes et al, 2023): if little is to be gained and the hassle significant, those who need less may not claim; but if complex needs pose insuperable obstacles to claiming, the most vulnerable may miss out. Both these factors are likely to play a part.

If someone thinks that their entitlement will not just be low but also potentially short term, or fluctuating unpredictably, they may be particularly reluctant to enter the claiming process (de Schutter, 2022: 10). This may be due to ‘optimism bias’, with people thinking their situation will soon improve (Hoynes et al, 2023), or linked to possibly having to repay overpaid benefits, which many find off-putting (see, for example, Breese et al, 2011).

Meso level reasons

The complexity of administrative procedures is also suggested as a reason for non-take-up (Ko and Moffitt, 2022). In some cases, specific documents must be produced – such as a special exemption certificate for free prescriptions in pregnancy in the UK. Tasks have been redistributed recently between the state and claimants, with more work in applying for and maintaining benefit claims by claimants themselves.⁸ Channels of application are a topical focus, with arguments that digital application is more straightforward but not suited to all potential applicants’ needs, either in general or at certain times – and, as a US Memo to Federal agencies points out (OMB, 2022), moving away from paper- or post-based applications may not appreciably reduce administrative burdens in the absence of simplification measures and/or assistance. A developing US literature suggests that ‘ordeals’ imposed on claimants may comprise deliberate policy tools (for example, Heinrich et al, 2022).

Non-means-tested benefits, especially without repeated checks of entitlement and/or demanding conditionality, tend to have higher take-up (for example, Hoynes et al, 2023, on the UK’s Child Benefit). Maintaining the claim, the need to report changes of circumstances and continuing to comply with conditions may be under-reported reasons why benefits may not be taken up or claims may lapse. Those benefits requiring additional proofs of entitlement, not just resources relative to needs – such as regular health assessments or work requirements – are likely to figure prominently here.

The quality of human contact is also a key issue; as ATD Fourth World emphasised in an international study (Bray et al, 2019), institutional and social maltreatment can be central to the experience of those living on low incomes and is commonly highlighted by benefit claimants themselves. Patrick (2023) also discusses ‘misrecognition’ and ‘disrespect’ and talks of the ‘relational harms’ done to people, often embedded in policy processes. Sarygulov and Arslanagić-Wakefield (2023) found participants in their research more likely to identify stigma due to benefits staff’s inflexibility and negative behaviour than complexity in applying as an obstacle to claiming (though providing proof of entitlement was also a barrier). De Schutter (2022) argues that erosion of trust in the state is both a reason for non-take-up of benefits and a result of it. These last points bring us on to the macro level reasons for non-take-up, which relate even more clearly to (potential) claimants’ relationship with the state and with society more widely. This is explored in the following section and discussed again in the concluding sections.

Macro level reasons

Some people may not claim benefits because of fear – of conditionality, and/or sanctions for failure; of triggering problems with their migration status (de Schutter, 2022: 7); or of penalties for claiming wrongly (House of Commons Public Accounts Committee, 2019). Codreanu and Waters (2023), analysing the ‘lone parent obligations’ introduced in the UK from 2008 to 2012 – obliging parents of progressively younger children to actively seek employment in return for benefit – found that ‘the reform reduced the number of single parents claiming welfare by a quarter, partly by discouraging eligible individuals from beginning a claim’. Lower take-up for means-tested benefits may also result partly from stigma in claiming these benefits (Baumberg et al, 2012; Baumberg, 2015; Patrick, 2023) that in marketised societies appear to be a badge of failure and therefore shame (Bucelli and McKnight, 2022: 9; Walker, 2014: chapter 4). More focus on tackling social security fraud is also likely to lead to lower take-up (de Schutter, 2022).

Some people may wish to remain independent of government help (Breese et al, 2011). This may be truer of certain groups. Lucas et al (2021), building on Warin (2016), suggest going beyond seeing shortcomings in potential claimants as reasons for non-take-up; instead, they should be seen as capable – and for some, non-take-up may mean putting their values into practice. For example, male breadwinners may be more reluctant to claim in-work benefits because they see this as reflecting on their (in)ability to support their family, or as social disqualification; lone parents may be seeking emancipation; migrants may wish to protect their family; and some ‘ethical intellectuals’ may want to live in a certain way.⁹ However, non-take-up should not just be assumed to represent a choice – in particular because the highest poverty is likely among not these groups but those without social networks or internet access or information.¹⁰

Some people, on the other hand, may not claim because of a refusal to accept the content of the offer from the state, or the conditions demanded – which, as Marc et al (2022: 5) argue, reintroduces ‘critical perspectives on non-take-up’. (See also Lucas et al (2021: 171) on the increasing focus on targeting and activation of claimants, which may lead to non-take-up.) ATD Fourth World, in recent work on social philosophy,¹¹ included in ‘resistance’ by people living in poverty the bypassing of the authorities by not engaging with them to claim benefits. And Tabin and Leresche (2019) describe people living in poverty as not necessarily sharing the hegemonic positive view of social rights; non-take-up can thus sometimes represent a response to the (negative) power of the state, rather than just resulting from various barriers and obstacles. Similarly, some claimants may think it futile to challenge a refusal (Fitzpatrick et al, 2019).

In addition, given that couples apply for UK means-tested benefits via a joint means test, some incomplete take-up may be due to individuals wishing to retain their financial independence, rather than depending on a partner. This is an under-explored topic (though see Griffiths, 2017), but one which may be increasingly important as independent money management by partners in couples increases (Bennett, 2013), and as the UK’s social security system shifts ever further towards means testing (Hoynes et al, 2023).

Implications of integration for take-up?

When the UK government was introducing Universal Credit (from 2013), it argued that having one integrated means-tested benefit, and thus one application process rather than several, would result in higher take-up. Those who had previously, for example, claimed Tax Credits but not Housing Benefit would now (have to) claim all elements of this one benefit in one go (DWP, 2018a: 10, updated in Quince, 2021). (However, statements by ministers or commentators sometimes seemed to suggest that people were claiming all six ‘legacy’ means-tested benefits and tax credits, which was vanishingly unlikely.) The higher cost of Universal Credit was due to this estimate of improved take-up (DWP, 2018a), which would also thereby reduce poverty. The White Paper on Universal Credit suggested wide agreement that it would mean higher take-up than other alternatives (DWP, 2010: 64). The lack of complexity, it argued, would also obviate the need for the existing advice services network.¹²

On the other hand, if there is a problem with Universal Credit there is no fall-back provision (EAC, 2020), as might be available in the legacy system. The main exception to this is Council Tax Benefit (now Council Tax Reduction). This means-tested benefit helping with local taxation is still claimed separately, with local authorities’ and devolved administrations’ rules varying across the UK. There is some evidence that its take-up, already low, is falling further (Sarygulov and Arslanagić-Wakefield, 2023), but no official estimates are published.

Integrating benefits for different groups, as Universal Credit does, also has implications for ‘passported’ benefits – those formerly available to claimants of certain benefits/tax credits without further conditions. Given the range of categories entitled to Universal Credit, qualifying for such benefits was clearly not going to be straightforward. In introducing Universal Credit, the DWP first postponed devising solutions and then defined it as other departments’ responsibility (while the Social Security Advisory Committee (2015: 8) suggested a cross-departmental approach). Both ‘passported’ free prescriptions and free school meals were complicated by the addition of earnings limits, making it more likely these will not be claimed. This limitation, caused by the integrated design of Universal Credit, goes in the opposite direction from simplification.

Universal Credit rollout

The first few years of Universal Credit rollout saw mounting criticisms (EAC, 2020), particularly of the (at least) five weeks’ wait caused by monthly assessment and payment in arrears. This reputational damage was only made good, at least in the government’s eyes, by Universal Credit’s performance in dealing with high volumes of claims during the COVID-19 pandemic (EAC, 2020).

One problem for the government is that, despite the importance of improved take-up to its argument for Universal Credit (Bangham and Corlett, 2018), there are no official estimates. As Baumberg Geiger (2021) put it, ‘non-take-up is therefore becoming an invisible problem once more’. In addition, as noted, take-up figures for other means-tested working-age benefits and tax credits are no longer published either (DWP, 2022a), because they are made unreliable by the phased introduction of Universal Credit. The argument from the UK’s Statistics Authority is that it is much more complex, given Universal Credit’s wider coverage and varied claimant base, to estimate eligible non-recipients. This means that the integrated nature of Universal Credit poses an obstacle to this, which is further complicated by the ongoing transition of claimants, because there remain large numbers on legacy benefits, and an estimate for Universal Credit only would therefore not give the whole picture. ‘Managed migration’ (compulsory transfer) of claimants to Universal Credit will now not be completed until at least the end of 2028. So it is unclear whether or when official take-up figures for those under pension age will be available.

Does Universal Credit increase take-up in practice?

As Hoynes et al (2023) point out, all else being equal, transaction costs are reduced compared with legacy benefits/tax credits when claiming Universal Credit as one benefit and only reporting changes in circumstances to one authority. However, all else is not equal. We explore the operation and design of Universal Credit later.

The reduction in administrative costs for the state resulting from using automated ‘real time information’ about earnings submitted by employers to calculate Universal Credit proved a turning point for support within government; this also clearly reduces the administrative burden of reporting changes of circumstances, at least for employees, which may help with continuing take-up over time. However, in terms of maintaining claims, Universal Credit unlike tax credits involves monthly means tests; and for the self-employed this also includes reporting monthly on their income from work.

The significant investment in the Universal Credit technology has not been matched by resources for legacy benefits; so the assertion that Universal Credit is digitally superior is not surprising. But the new ‘digital by design’ application process also causes problems for many, with one investigation finding only just over half of claimants able to register their claim online unassisted (DWP, 2018b). Overall, over four in ten claimants reported needing more support with this and some three in ten needed more ongoing support using their digital account. (Experience in the COVID-19 pandemic may have differed, but was atypical, as somewhat different groups claimed Universal Credit then.) Such problems may lead to people either giving up their applications or taking longer to claim and then receive benefit, with both affecting take-up. In the year to February 2023, about a third of the nearly 3 million registrations for Universal Credit were abandoned before a claim was submitted (Mears and Howes, 2023).¹³

It is not just digital literacy at issue, however; there are also problems for potential claimants of (lack of) availability of equipment, access to broadband and credit on smartphones. And at stake is not just an initial application but also maintenance of the claim. Income volatility resulting from the workings of the monthly assessment period of Universal Credit may persuade some either to work enough hours, with sufficient wages, to move beyond its scope (if they can), or to decide to do without it (if they can).¹⁴ Paying Universal Credit in arrears was intended to reduce over- and under-payments; but fear of owing money to the state may still put people off claiming (Bucelli and McKnight, 2022: 12), as other benefits and tax credits debts are added to a new claim – and there is no exemption for overpayments due to official error. Evidence in the discovery phase of ‘managed migration’ (compulsory transfer) of claimants to Universal Credit suggests that some people on tax credits in particular decided not to claim because they believed that they were ineligible or the application was not worth the effort (DWP, 2023).

The administration of Universal Credit is not the only important issue here, therefore. Its design is also key. For example, combining an in-work support scheme with low stigma (tax credits) with out-of-work benefits, some with significant stigma (for example, means-tested Jobseeker’s Allowance), is uncertain in its implications. Welfare at a (Social) Distance, a research project undertaken during the COVID-19 pandemic, estimated that some half a million people eligible for Universal Credit early in the pandemic did not claim it; of these, 27 per cent said that this was due to stigma (Baumberg Geiger et al, 2021). Although some may have claimed later, this would still represent some unclaimed benefit – though, as noted, the (potential) claimant population during this pandemic was likely to differ from the usual mix. Conditionality will also be increased for many as they move from tax credits to Universal Credit (Hoynes et al, 2023), which again does not suggest the likelihood of increased take-up.

Official evidence is not (yet) available about how these factors interact in practice for Universal Credit take-up. And unofficial estimates by entitledto (2023) and Policy in Practice (Clegg et al, 2023), while using somewhat different methodologies, are based on previous take-up estimates for working-age means-tested benefits and tax credits, rather than taking account of the specific features of Universal Credit noted earlier. On this basis, they both estimate that some £7.5–8 billion of Universal Credit is unclaimed by about 1.2 million households.

Increasing interest in improving take-up

As Eurofound (2015: 2) points out, the variety of causes of non-take-up suggests that approaches to address it need to be similarly diverse and apply multiple strategies simultaneously. Bucelli and McKnight (2022: 22–23) analyse challenges and facilitating factors for policies aiming to address poverty and social exclusion by increasing benefit take-up. Here, we discuss current UK and other initiatives to improve take-up and issues likely to affect such initiatives and drive take-up in future.

The UK government launched a take-up drive for Pension Credit in April 2022 (DWP and Guy Opperman MP, 2022), which appears to have had some success, according to official figures.¹⁵ However, this drive has not been matched by a similar campaign for people of working age to claim Universal Credit and/or other benefits – although the Money and Pensions Service now provides a benefit calculator (in addition to those of independent organisations) to help people find out whether they might qualify.¹⁶ However, as described by Bucelli and McKnight (2022: 8), the devolved administrations issued a joint statement on take-up in 2020 and Scotland has developed a benefit take-up strategy (see Scottish Government, 2021). They also note that promoting take-up has been part of the Welsh government’s poverty strategy; it uses a system called Dangos¹⁷ to encourage take-up of benefits and other financial support by giving frontline workers information and resources about help available.

In France, President Macron’s promise to improve take-up of minimum income benefits has resulted in a flurry of activity and publications (for example, Gonzalez and Nauze-Fichet, 2020; Marc et al, 2022). This appears to be an alternative to integrating several means-tested benefits under the banner of ‘revenu universel d’activite’ (universal activity income), which was under consideration before the COVID-19 pandemic – including through officials’ examination of Universal Credit – but now seems to be no longer on the cards. Instead, the current focus is on implementing ‘solidarite a la source’ in 2024 (automatic benefit payments, to mirror the recent introduction of more automated tax payments, like PAYE) and creating ‘zero non-take-up’ territories. One in three eligible claimants in France are reported not to claim the ‘revenu de solidarite active’ (RSA, or active solidarity income) and the ‘prime d’activite’ (activity bonus), and one in five do not claim help with housing costs¹⁸ – although over a longer period (nine months in this instance), more people are likely to take up benefits (with an estimated one in five not claiming the RSA) (Hannafi et al, 2022).

The suggestion was to base automatic benefit payments on tax returns (completed by a far higher proportion of the population in France than the UK and including some information about household composition) as well as information from employers. Analysts stated that the calculation could therefore be automatic and the beneficiary would not have to do anything themselves. However, President Macron’s announcement referred to combating not only non-take-up but also fraud.¹⁹

More widely, in the EU, the Recommendation on minimum income, recently agreed by member states,²⁰ included a statement about non-take-up of minimum income, which it said typically ranged from 30 to 50 per cent – although as noted only a few member states gather regular official information on this. The Recommendation includes an objective to improve the coverage and take-up of minimum income, thus prompting further interest in this across European countries.²¹ In the US, a recent Memo (OMB, 2022) urged Federal agencies to accurately estimate burdens experienced by the public when accessing benefit programmes and to implement initiatives to minimise such burdens – which could be expected to lead to better access.

Data plus

Here, we look at proposals now being made for using existing data and automation to promote higher take-up – but also pose some questions about this approach. Van Gerven (2022: 251) noted that ‘governments are keen to replace old methods of welfare provision with automation and datafication, with the aim of developing effective and ‘better’ public policies’, and that digitalisation of governance may solve non-take-up problems (p 256). And it has been argued that ‘the digitalisation of the welfare state creates new opportunities for proactively contacting people who are potentially entitled to benefits’ (Van Gestel et al, 2022: 1). However, potentially negative outcomes of this approach have also been identified (see, for example, Eubanks, 2018, on monitoring and control, and Alston with Van den Reen (2019) on the implications for human rights).

We examine these advantages and disadvantage by using the helpful outline from de Schutter (2022) about what ‘automation’ of benefit claims might mean in practice: identification of who is eligible and providing a benefit without them having to claim (a common interpretation of ‘automation’, and the strongest); eligibility for one benefit leading to receipt of another (perhaps more commonly called ‘passporting’); or using documentation provided for one benefit to support a claim for another. For the US, the OMB (2022) lists ‘using existing administrative records to automatically enrol[l] individuals, populate application information, or determine eligibility’ as methods shown to increase participation in programmes.

An alternative typology was suggested by the TAKE project in Belgium, investigating routes to improve take-up (Goedemé et al, 2022). This identified four main approaches to automating benefit access (p 83): administrative simplification (for example, introducing derived rights); automatic ‘actualisation’ (making it unnecessary to reapply frequently); proactive identification and outreach to those potentially eligible; and fully automated benefit access (such that no action is required from the beneficiary) (see also Bucelli and McKnight, 2022). In the TAKE project, ‘proactive identification and outreach’ was found to have resulted in a significant increase in take-up of a Belgian government programme. The ‘proactive identification’ of potential claimants provided the element of automation here, with subsequent outreach instead of automatic granting of benefit. However, the researchers warn that the inclusiveness of such interventions must be balanced against increased applications by ineligible people. And Lucas has argued that automation initiatives must recognise that some people wish to remain autonomous and not claim benefits.²²

There are some proposals for the first, and strongest, interpretation of automation from de Schutter (2022) listed earlier. The New Economics Foundation in the UK, for example, has proposed a ‘national living income’ for those of working age (jointly means-tested for couples) (Tims and Stirling, 2022). They suggest that this could be obtained by ‘auto enrolment’ via a small, individually based, flat-rate ‘national allowance’ to replace part of the tax-free allowances for income tax and national insurance contributions.

Eurofound (2015: 2) has argued that ideally benefits should be paid automatically, with no need for applications, with databases linked as necessary in ways that respect privacy. But when the DWP did experiment with paying estimated Pension Credit automatically for 12 weeks, there were only marginal increases in take-up at the end of the experiment – and at ‘significant cost’ compared with a similar sample invited to be helped with claiming (Radford et al, 2012: 3). And a decade on from this, the UK government stated that automatically awarding Pension Credit is impossible (House of Commons Work and Pensions Committee, 2022: 5): ‘eligibility for Pension Credit depends on a household’s composition and financial circumstances, which can be complex. It is not possible to determine eligibility with sufficient accuracy to enable Pension Credit to be awarded automatically, based solely on the data held by Government’. This is likely to hold true, therefore, for other core means-tested benefits, including Universal Credit. Other issues include subtly different definitions of elements of the calculation which have labels in common (such as forms of capital); the date of data capture; the notional period of entitlement assessment; and data ownership (such as, for example, who has the authority to correct errors), as well as technical issues with data transfer and validation.²³

In addition, if automation is interpreted strictly to mean that claims would not be necessary, this raises various more fundamental issues, which again concern the claimant’s relationship with the state. In the case of means-tested benefits, as suggested by the UK government’s comment about Pension Credit earlier, these include whether it is feasible to track not only the circumstances of different individuals which might entitle them to a benefit (including their status as the basis of entitlement, as well as their capital and income, and their needs) but also, in case they are one of a couple, who is living with whom and in what kind of relationship, as well as similar information about the other partner. In addition, if a benefit is conditional, it would not be possible to automatically award payment unless the person concerned had agreed to the conditions (such as actively looking for work for up to 35 hours per week, which applies to many currently claiming Universal Credit in the UK).

The second of the meanings listed earlier by de Schutter (2022) – eligibility for one benefit leading to entitlement to another – was evident in UK practice recently, when, as noted, the DWP used data already in its possession and automaticity to make widespread one-off emergency payments to claimants of existing benefits during the energy prices crisis.²⁴ Such practices can clearly result in extremely effective take-up, as no action is required from the potential recipient. (When action was required – for households with pre-payment meters, to obtain a voucher towards their costs – it was reported that many did not claim.)²⁵ But one-off emergency payments are designed precisely to avoid the more complex information demands about other income and household circumstances needed to assess ongoing support needs; and they are not usually attached to conditions (such as work requirements) which potential claimants may have to fulfil to get a payment. Thus, they avoid some of the key elements known to affect take-up negatively. In addition, as they are one-off payments, they do not rely on circumstances being stable enough to qualify for benefit (or on changes being reported accurately and in a timely fashion). The converse is also true, however: if someone’s circumstances are not appropriate at the time of the automatic award, they do not qualify; this has led, for example, to many Universal Credit claimants in the UK missing out on such one-off payments because they had a nil award in the qualifying month, thus ruling them out of entitlement. Efficiency of delivery is therefore achieved, but at the expense of crude entitlement rules and ‘cliff edges’.²⁶

Bucelli and McKnight (2022) argue in favour of ‘streamlined or automated cross enrolment’, or ‘auto enrolment’, because this focuses on the failure of the administration rather than the claimant – thus putting the onus on the provider rather than the recipient – and also signifies endorsement of entitlement by the authorities, which may combat stigma. They give an example from the US; but this also appears to be a case of ‘passporting’ to another benefit, rather than the third option given here, of using documentation for one benefit to support a claim for another. A US Memo (OMB, 2022), containing advice to Federal agencies to reduce administrative burdens for applicants for public services or benefits, also suggests shifting from opting in to opting out and setting defaults in such programmes to be the most favourable possible for beneficiaries.

Hoynes et al (2023) suggest that experiments could be conducted at least to automatically identify people likely to be eligible for certain benefits, which fits the third of de Schutter’s (2022) meanings of automation explained earlier. Scotland is experimenting with using pre-completed forms to be signed (or not) in order to pay claimants the various additional payments to some groups now available under devolved arrangements there. Others have also suggested forms pre-filled with at least some data that are then complemented by additional information from potential claimants. It may be unclear in some proposals, however, what would happen if the original data in the form were inaccurate or out of date (as Alston with Van den Reen, 2019: 9–10, warn). And, as de Schutter (2022) also points out, the most vulnerable may not be registered to begin with, meaning that no information will be available about them as a starting point for this ‘automation’.

Also following the third meaning of ‘automation’ listed earlier from de Schutter (2022), as one element of its benefit take-up strategy the Scottish Government (2021) has initiated an ‘invite to apply’ letter sent to those already on some benefits who appear eligible for certain others (Bucelli and McKnight, 2022). Policy in Practice, a UK social policy and analytics company, would like the DWP to allow Universal Credit data to be used for more purposes and is working with it on allowing local authorities to make their own decisions on how to use these. There is a data protection issue here, about obtaining permission from individuals to pass on information about them to others; and the GDPR is currently quite strict about data being used solely for the purpose for which they were originally collected. (See also OMB (2022), with an equivalent warning about data matching and populating forms with data in the US.) All this discussion omits benefits administered and delivered locally. Eurofound (2015: 3) argues that such benefits are at risk of non-take-up when they are part of a complex, fragmented benefits structure. Sarygulov and Arslanagić-Wakefield (2023) recommend one digital portal for claimants in the UK to access both national and local benefits; but this seems unrealistic, given the variation in many local rules of entitlement and other provisions.

Alternative approaches?

Bucelli and McKnight (2022: 16) argue that automation of means-tested benefits can be effective – if adequate data sharing protocols and legal frameworks exist; alternative claim routes are offered; individual and social barriers are addressed; and claimants understand the necessary procedures. This mirrors the focus on the recipient rather than the provider suggested earlier, suggesting that public services should adapt to user needs and realities, rather than trying to fit these into administrations’ expectations and priorities (see also Defender of Rights, 2022, especially page 5). These are demanding requirements, however, and take us some distance from a simple statement that automation is the solution to take-up problems.

Application processes should therefore be as simple as possible, and alternatives to digital channels easy to access. Baumberg Geiger (2021), discussing those who claimed the ‘wrong’ benefit at the start of the COVID-19 pandemic, were rejected, and did not try for an alternative benefit, suggests that ‘more useful rejection letters, which provide clear details as to why an application was rejected and what alternative support might be available’ would be helpful (see also Baumberg Geiger et al, 2021.) More generally, Eurofound (2015: 2) usefully suggests that non-claimants themselves could give important information on making application procedures more accessible.

There is some agreement that adding assistance with form filling to information will achieve more significant increases in take-up (see, for example, Bucelli and McKnight, 2022; Hoynes et al, 2023). De Schutter²⁷ has also suggested facilitating access not just by online formulae but also by agents offering help. He referred to the retreat of the state in rural areas, resulting in social and symbolic distancing, creating longer journeys between people and their rights. (OMB (2022) makes a similar point about the importance of geography in its Memo on reducing administrative burdens in the US.) Morgan (2023)²⁸ also puts forward a ‘blended approach’, arguing that using automation is insufficient without also ensuring that local organisations can advise and support potential claimants. This is clearly only possible, as Morgan argues, if advice organisations are funded adequately.

These authors therefore all point to human contact as an important ingredient in actualising take-up, even if digitalisation and/or better use of data can reveal potential beneficiaries and overcome some obstacles (see also Spicker, 2017). Bucelli and McKnight (2022) also discuss local, integrated approaches, involving partnerships with civil society and trusted relationships, while noting that such initiatives are often hard to evaluate.

Design of benefits and attitudes to claimants

Therefore, as noted earlier about Universal Credit, the nature of benefits and how they are administered, as well as the views of the general public and policymakers about claimants, are critical to successful take-up. Universal Credit itself is an object lesson in how integration and automation may create problems and policy dead ends; and designing a benefit with a priority goal of automating its calculation is not necessarily the best way forward in terms of improving take-up (Bennett and Millar, 2022).

On the other hand, Holler and Benish (2022) argue that benefits with high selectivity and discretion are very dependent on personal relationships, feelings and power relations. Taking this to its extreme, de Schutter (2022), based on the global survey of take-up carried out by his office, also points to corruption as one reason for non-access to social rights. Pelissery (2006) forensically examined social assistance in two rural sites in India, and showed some households gaining formal social protection and others failing to, because the ‘local practices and informal rules underlying these public policy processes are purposively guided by the private interests of the local elites’.

Even those who emphasise the importance of human contact, and/or who assist claimants to gain access to benefits, should therefore not be tempted into proposing more discretion. Discretionary benefits may give more scope for them to help people obtain additional income. While this may be very satisfying, however, success for claimants in relation to such benefits depends on knowing where to seek help and access to the right gatekeeper and can also involve stigma (Fitzpatrick et al, 2019). A recent thinktank report about discretionary support in the UK (Handscomb, 2022) also pointed to the postcode lottery frequently involved, in that over six times as much support per head is provided in Scotland as in England, where there is no local welfare assistance in a quarter of council areas. Another report revealed that discretionary payments administered by local authorities were much less effective than the UK government’s more automatic support for the cost of living for low-income working families, due to bureaucratic barriers and burdensome eligibility criteria (Griffiths, 2022). Discretion should only be used to award additions at the edges of a non-cash-limited system grounded in unambiguous legal rights to decent levels of income.

Patrick (2023) argues that ‘there is an urgent need to both increase incomes at the bottom of the distribution and radically improve interactions and policy interventions in order to generate respectful and dignified encounters between citizens and the state’. As Hignell (2023) argues, however: ‘we cannot continue with an approach that only takes account of people’s real lives when they happen to speak to an empathetic and experienced individual in a Jobcentre. We have to integrate the same level of empathy and understanding into the system as we redesign it.’ This is not solely about individuals’ actions, therefore, but centrally about reshaping the relationship between (potential) claimants and the state.

Thus, the areas requiring more systemic thinking are the design and structure of social security on the one hand and the attitudes to claimants that may be embedded in these policies on the other (macro level) – not just obstacles for claimants (micro level) and problems within the administration (meso level) (de Schutter, 2022). As de Schutter also argues, involving claimants themselves in this rethinking is also crucial (and see earlier, about involving non-claimants).

Babic (2018) is clear that benefits with low take-up tend to be means tested, discretionary, with conditionality and paid at low levels. On the other hand, benefits with consistently high take-up in the UK, including the State Pension and Child Benefit, have few conditions, are not means tested, are paid for a long time, have a simple application process and are not stigmatised. Many would argue that Universal Basic Income (a regular payment to each citizen/resident without means tests or other conditions) also embodies these characteristics and so would be likely to have high take-up if it were introduced – although it might need to be complemented by other benefits in any realistic scenario. Integrating inherently complex means-tested systems, as with Universal Credit, has been shown earlier not to provide a straightforward solution to non-take-up. More policy energy should instead be put into non-means-tested benefits in the UK’s social security system if policymakers wish to effect radical improvement in take-up.

The ministerial foreword to the latest benefit take-up strategy from the Scottish Government (2021) roots this in commitment to a positive vision of benefits and claimants, by declaring: ‘Social security is a collective investment in building a better and fairer society. Part of that is ensuring people are aware of, and enabled to access, the financial support that they are eligible for and entitled to.’ This references the relationship between (potential) claimants and the state but also their relationship with wider society. As recommended by Baumberg Geiger et al (2021), we would urge the UK government to publish its own benefit take-up strategy with a similar perspective and commitment.