Introduction
Low coverage rates of conditional and means-tested benefits negatively affect social protection systems (EESC, 2013). One of the factors behind them is non-take-up, i.e. potential claimants not applying for a benefit despite being eligible for it as meeting all requirements. Merely having policies in place is not enough; they must be thoroughly implemented and reach the neediest groups (EMIN, 2014).
The magnitude of the problem is alarming. In the United States, the Earned Income Tax Credit reports a meagre take-up rate of 25 per cent, amounting to 6,7 million non-claimants each year (Bhargava and Manoli, Reference Bhargava and Manoli2015). In OECD countries, non-take-up rates for social assistance programmes range between 40 and 80 per cent (Hernanz et al., Reference Hernanz, Malherbet and Pellizzari2004: 4) and stand at 50 per cent for means-tested benefits across all European countries (Matsaganis et al., Reference Matsaganis, Paulus and Sutherland2008: 12). In Finland, those not claiming assistance benefits stands at between 40 and 50 per cent (Bargain et al., Reference Bargain, Immervoll and Viitamäki2012), while the French Revenue de Solidarité Active is just taken up by 36 per cent of its potential recipients (Domingo and Pucci, Reference Domingo and Pucci2014: 119). In Spain, the Rentas Mínimas de Inserción do not even reach 10 per cent of their target population (Ayala et al., Reference Ayala, Arranz, García-Serrano and Martínez-Virto2021). Other research reveals even worse figures across Europe (Eurofound, 2014: 20). The concerns this raises for national and regional institutions is evident. According to the European Commission, “Non-take-up is a matter of concern and needs to be carefully monitored” (EC, 2013: 89). Similarly, the European Economic and Social Committee fears that most “minimum income schemes […] fall short of alleviating poverty” and is thus “concerned that non-take-up of such schemes tests their effectiveness still further” (EESC, 2013: 9).
Beyond these concerns, any attempt to improve the performance of cash transfer policies requires an understanding of why people do not claim them. This paper addresses this phenomenon and aims to pinpoint some of its possible causal factors. To do so, it analyses the B-MINCOME pilot scheme, an experimental cash transfer programme implemented in Barcelona from 2017 to 2019. The paper presents the main factors underpinning non-take-up highlighted in the literature to analyse them in light of this pilot scheme. The data collected allows us to study and compare the two main dimensions of analysis the literature usually identifies as most relevant: i) the individual characteristics of the applicants in relation to the information barriers they face, and ii) the administrative and institutional functioning in terms of communications strategies. Which of these two dimensions is more relevant to addressing the issue of non-take-up? Responding to this question may be of help for other similar programmes, and is therefore the central goal of this paper.
The first section reviews the most relevant factors affecting non-take-up according to the literature and defines the hypotheses to test. The second offers a summary of this pilot scheme and the data used. Based on different logistic regression models, the third section compares the variables and hypotheses determining the likelihood of not applying for this benefit. The findings in section four reveal that beyond potential claimants’ individual characteristics, the administrative functioning and communications strategy employed by public institutions are key in explaining why an individual may or may not apply for the B-MINCOME benefit. Conclusions suggest that the problem of non-take-up does not only affect the efficiency and effectiveness of this kind of benefits, but also raises moral concerns on how social protection systems perform in relation to the most vulnerable individuals and social groups.
Why do people not take up a benefit?
Reducing the number of people not applying for a benefit “requires more than just making them eligible” (Heckman and Smith, Reference Heckman and Smith2004: 287). Attention must therefore be given to factors that beyond establishing de jure eligibility requirements lead to de facto applications for a benefit. In the 1970s, scholars mostly looked at ignorance, stigma, and administrative complexities. During the following decade explicative models of decision-making processes gained importance, first focusing on econometric explanations based on gender, age, education, or income; and second prioritizing a more psychological approach considering the claimants’ attitudes and beliefs (Huby and Whyley Reference Huby and Whyley1996: 3). Based on the literature, the main factors behind non-take-up can be classified according to two levels of analysis.
1) Individual and claimant level
Application rates are usually explained on the basis of two premises. First, the applicant’s sociodemographic variables (education, gender, age, origin, household composition, etc.) in combination with socioeconomic ones (income, receiving other benefits, financial assets, etc.). In the United States, black individuals with low levels of education from poor families and with no professional experience show lower-than-average application rates (Heckman and Smith, Reference Heckman and Smith2004: 282). In Spain, 55.89 per cent of homeless people do not apply for income support (Khalifi et al., Reference Khalifi, Bermejo, Pérez, Hernández and Abellán2016: 290). In sum, “many groups of special interest to policymakers […] have high rates of eligibility but low probabilities of program participation” (Heckman and Smith, Reference Heckman and Smith2004: 246).
The second premise refers to the observed behaviour that is attributed to certain mechanisms of instrumental and parametric rationality (Kerr, Reference Kerr1982; Kleven and Kopczuk, Reference Kleven and Kopczuk2011). This explanation emphasises the private transaction costs incurred by an individual making rational choices between the expected utility of a benefit and the effort required to apply for it (Fuchs et al., Reference Fuchs, Gasior, Premroc, Hollan and Scoppetta2020). These costs “may be quite important, especially […] when access to the program is conditioned” (Coady, Crosh and Hoddinott, Reference Coady, Crosh and Hoddinott2004: 8). Among these costs, “there is strong evidence that information barriers play a major role in determining differences in program participation rates” (Heckman and Smith, Reference Heckman and Smith2004: 286). People may be deterred from claiming a benefit because these information barriers entail excessively high personal costs. First, the fact that one is aware of the existence of the benefit, of the eligibility requirements and of the application process (Ranci and Arlotti, Reference Ranci and Arlotti2019: 575; Eurofound, 2014: 30). Second, the shame (Walker et al., Reference Walker, Kyomuhendo, Chase, Choudhry, Gubrium, Jo, Lødemel, Mathew, Mwiine, Pellissery and Ming2013), the stigma a person experiences (Currie, Reference Currie2004: 11), or the perception of stigmatization they feel when others devalue their identity (Baumberg, Reference Baumberg2016: 183) may impose insurmountable costs. Third, the level of participation in social or community networks based on ethnicity, language or affinity reduces these costs as distributing the information about the benefit and its application procedure (Currie, Reference Currie2004: 7; Mäkinen, Reference Mäkinen2018: 13). Finally, the household composition: single-person households usually report lower rates of take-up because the information and opportunity costs are very high as being assumed by a single person (Eurofound, 2014; Tempelman and Houkes-Hommes, Reference Tempelman and Houkes-Hommes2016). These costs are lower in households with several members as they can be shared. The presence of minors increases the likelihood of claiming a benefit due to the higher perceived need of parents when the welfare of children is considered (Hernanz et al., Reference Hernanz, Malherbet and Pellizzari2004), particularly in single-mother households (Brady, Reference Brady2018).
The literature attributes significant explanatory weight to these information costs, especially when correlated with variables such as education (associated with reading and writing skills), place of origin (related to proficiency in the language and knowledge of the legal framework and other benefits), household composition (with multiple members and minors), and place of residence (associated with stigma or networks). The weight of these individual sociodemographic variables tends to increase when controlled by two further economic factors: the family’s income (Gustafsson, Reference Gustafsson2002; Huby and Whyley, Reference Huby and Whyley1996) and being in receipt of other benefits (Blundell, Fry and Walker, Reference Blundell, Fry and Walker1988; Finn and Goodship, Reference Finn and Goodship2014). Given the profile of individuals applying for the B-MINCOME benefit, these variables seem particularly relevant in explaining the pilot’s take-up rates. Accordingly, it can be predicted that:
Hypothesis 1: the more information barriers caused by sociodemographic and socioeconomic characteristics, the less likely take-up for the benefit became.
However, these kinds of explanations tend to omit the fact that the lack of information – and thus the bias in calculating cost-benefit trade-offs – depends to a great extent on factors beyond the rational individual scope and particular characteristics (Buckland and Dawson, Reference Buckland and Dawson1989; van Oorschot, Reference van Oorschot1995). Informational barriers “do not fully account for the low participation rate” (Heckman and Smith, Reference Heckman and Smith2004: 245-6). Effectively, “People may not claim even where perfectly informed: imperfect information is not the only reason for non-take-up” (Atkinson, Reference Atkinson1995: 252). Hence, other non-individual factors must exist.
2) Administrative and institutional level
The administrative logic and the functioning of the institutions do also have an effect (Deacon and Bradshaw, Reference Deacon and Bradshaw1983; van Oorschot, Reference van Oorschot and van Oorschot1996). A key element to study this is the communications strategy employed by the public institutions (Fuchs, Reference Fuchs2007: 7) when informing about the existence of a benefit (Craig, 1991; Chetty, Friedman and Saez, Reference Chetty, Friedman and Saez2013) and its application process (Mäkinen, Reference Mäkinen2018; Eurofound, 2014). Regarding the former, the effect of a publicity campaign will differ depending on whether it is merely publicised on an institutional website, or through more direct methods such as letters, phone calls or face-to-face contact with social workers (Daponte, Sanders and Taylor, Reference Daponte, Sanders and Taylor1999; Warin, Reference Warin2018). The same occurs with the application process. A long or complex process may increase opportunity costs due to the time or travel requirements necessary to complete the application form (Anne and Chareyron, Reference Anne and Chareyron2017). Likewise, processes requiring a large number of supporting certificates, deeds, or records to accredit the applicant’s eligibility present another obstacle (Currie, Reference Currie2004: 11). Digital tools and online administration offer many advantages, but the digital divide remains an obstacle for the most vulnerable populations (Finn and Goodship, Reference Finn and Goodship2014: 64).
As detailed below, the communications strategy of the B-MINCOME consisted of i) sending out letters to all potential applicants with a detailed explanation and the application form, and ii) holding 400 information sessions where they were invited by phone and mail. This communications strategy seems fundamental, as observed in similar experiments (Betkó et al., Reference Betkó, Spierings, Gesthuizen, Scheepers and Delsen2019). The fact that some potential claimants’ cohabitants also received the letter of invitation (due to also being social service users and meeting all the requirements) may have encouraged them to regard the project as important and thus to increase the pilot’s take-up rate. As described below, the number of application forms collected during the information sessions in comparison with those received by post would also demonstrate the impact of these sessions. This leads us to assume that:
Hypothesis 2: receiving more than one letter of invitation per household increased the individuals’ likelihood of taking-up the benefit.
Hypothesis 3: attending an information session dramatically increases the likelihood of taking-up the benefit.
There are some additional variables related to how a benefit is implemented that should be considered. First, the fact that the application processes or its requirements may not be sufficiently detailed can lead to a certain “administrative discretion” (Eurofound, 2014: 35, Mäkinen, Reference Mäkinen2018: 21), especially under conditions of minimal institutional accountability (Brodkin, Reference Brodkin1997: 3). When the implementation favours professional practice over formal procedures, the task of social workers becomes “a matter of assessment […] of users’ behaviours, where this behaviour and no longer only users’ conditions and/or status, determines their eligibility” (Warin, Reference Warin2018: 64). Administrative discretion undoubtedly affects take-up rates (Elster, Reference Elster and Gutman1988: 60). Second, the duration and the amount of a benefit are also crucial, since a low take-up rate is expected when it is short-lived and low value (Riphahn, Reference Riphahn2001; Whelan, Reference Whelan2010; Domingo and Pucci, Reference Domingo and Pucci2014). Third, the length of time a benefit has been available and the changes it may have undergone are also important (Mäkinen, Reference Mäkinen2018: 13; Hernanz et al., Reference Hernanz, Malherbet and Pellizzari2004: 17). High take-up rates are observed when it is operational over a long period and has experienced few changes, but lower when it is implemented for the first time or its design is altered (Ranci and Arlotti, Reference Ranci and Arlotti2019). Finally, there are always people or certain social groups who tend to be excluded by default. Nomadic or irregular populations may lack a postal address or bank account which make proving settled status, income or family ties more difficult and further hinder their eligibility (EMIN, 2014; ESPN, 2016). “Hard-to-reach social groups – such as the homeless, people with a disability or mental illness, or immigrants – may still fall through the cracks of the system” (De Wispelaere and Stirton, Reference De Wispelaere and Stirton2011: 117).
Nevertheless, in the case of the B-MINCOME pilot schema, stringent implementation rules and clear application process were exhaustively explained from the outset. Being newly launched and having a temporary nature may have reduced its potential claimants’ confidence in the program, and then have reduced their likelihood to apply. However, considering the potential claimants’ profile, the incentives of applying (24 months of duration and clearly defined amount of the benefitFootnote 1 , the formative and relational opportunities offered by the social policies, the permanent-tailored advice of social workers, etc.) seem attractive enough to exceed their possible lack of trust on such a programme. Furthermore, the number of potential claimants who may have been excluded was deemed irrelevant considering the exhaustive work carried out by the social servicesFootnote 2 . In sum, it seems reasonable to assume that the possible aggregate effect of these four additional variables had a negligible impact, and therefore they are not considered in our statistical analysis.
Barcelona’s B-MINCOME experiment
B-MINCOME was an experimental project co-funded by the EU’s Urban Innovative Actions programme, implemented by Barcelona City Council’s Social Rights Area in collaboration with five universities and research centres (Laín, Riutort and Julià, Reference Laín, Riutort and Julià2019; Laín and Torrens, Reference Laín and Torrens2019). Its goal was to reduce poverty and social exclusion rates in the ten poorest neighbourhoods of the city. It aimed to test the effectiveness and efficiency of combining an economic benefit with active social-inclusion policies. From November 2017 to November 2019 and based on a randomised control trial, 1,000 individuals from a universe of 4,824 were randomly selected and assigned to ten treatment groups (based on whether the benefit was conditional or unconditional on participation in active policies) while a further 500 individuals were also selected as a control group. The benefit of up to € 1,676 was granted individually although it was calculated according to the beneficiaries’ household income, members, living costs (food, clothing, basic suppliers, etc.) and housing costs (rent or mortgage). The benefit was paid monthly into a credit card with no maintenance costs and it could be accumulated, so permitting beneficiaries to use it as they pleased (withdrawing money or paying with the same credit card).
Participation only required the completion of an application form which then entered a draw to select the final participants and to assign them into the different treatment groups. Only those individuals not exceeding the severe poverty threshold (below 40 per cent of the median equivalent income) were able to participate. Eligibility requirements were: 1) to be registered as a resident of Barcelona for two years and living in the treatment area; 2) to be a user of social services; 3) the participant (or one member of their household) must be aged between 25 and 60; 4) to accept the assigned group and the type of participation; 5) to authorise the city council to obtain personal information and commit to completing the assessment surveys; and 6) that the joint assets of the participants’ household did not exceed four times the value of their estimated benefit.
The study was carried out by a consortium made up of the Catalan Institute for the Evaluation of Public Policies (IVALUA), the Barcelona Institute of Regional and Metropolitan Studies (IERMB), the Youth Foundation, the Institute of Environmental Science and Technology (ICTA) and the Institute of Government and Public Policy (IGOP) from the Autonomous University of Barcelona, the Polytechnic University of Catalonia, and a municipal team including six social workers. An Advisory Board of thirty professional, academic and social institutions was created to follow up the design and implementation. Ethical clearance was provided by an Ethic Committee made up of members of the City council, juridical services, and other public servants with ethical expertise in the public administration, in charge of assisting participants’ complaints and guiding the consortium in any potentially ethically-conflicting scenarioFootnote 3 .
Data and method
Sample
Analyses are based on records of social services users which provided information on their sociodemographic characteristics, household composition, addresses and telephone numbers for contact purposes. The study sample comprised 4,824 potential candidates meeting all eligibility criteria. They were firstly informed of the pilot scheme and its application process through postal letters inviting them to attend one of the 400 information sessions held in their neighbourhoods. 2,525 applications (52.34 per cent of all candidates) were collected, of which 2,039 (80.8 per cent) were personally delivered during the sessions, and 486 (19.2 per cent) were later received by post. Application forms were then refined by means of the interoperable records of public administrations to verify the accuracy of the data provided and to ensure that there were no duplicate applications nor more than one per household since, in some cases, various members of the same household were identified as eligible.
During the various rounds of phone calls, 2,383 (49.4 per cent) of candidates were finally contactedFootnote 4 : 91.2 per cent said they were about to attend the information sessions, 5.8 per cent were not interested in participating, and 3 per cent had already sent the application form. 2,203 (92.4 per cent) of contacted candidates attended the information sessions. Thus, a total of 45.7 per cent of the total 4,824 candidates who were sent a letter finally attended the face-to-face sessions where the project was described, all questions answered, and support provided to anyone who needed help to complete the application form. These sessions lasted approximately 45 minutes, with an average of 10 individuals in each session, involving around 80 professionals, social workers and social scientists. The difference between those who attended and who did not had a remarkable impact, since 92.3 per cent of the former went on to apply for the benefit (Figure 1).
Diagram of eligible candidates sample evolution
Variables and method
In order to analyse the factors affecting the likelihood of take-up, a hierarchical logistic regression was designed with different models to observe the specific and cumulative incidence of each of the variables presented in the theoretical framework in the first section of this paper. The sociodemographic variables identified at the individual analytical level were entered in model one (M1) including the educational level of the letter addressee (no qualifications; level 1 = compulsory secondary education not completed; level 2 = compulsory secondary education; level 3 = higher secondary education or professional training; level 4 = university); gender (female; male); age; country of origin (Spain; foreign); number of cohabitants (one; two; three; four; five or more; ‘no information’Footnote 5 ); if there were children under 16 in the household; and the neighbourhood of residence sorted by its level of average income in 2017 (with the highest one, € 29,881 taken as a reference in comparison with € 21,604 of the lowest one).
The socioeconomic variables identified in the first level of our theoretical framework were added in model two (M2). First, we incorporated the claimant’s household income based on their 2017 personal income tax return. As we did not have this information in 34.7 per cent of the cases, we conducted a multiple imputation to complete the missing values based on the other values entered in the model. Average income per claimant’s household was € 11,510 (Table 1). We also added the annual income received from other public benefits and the variable of having received the benefit for children under 16 available in Barcelona since 2013. These two economic variables captured the relationship between formal inclusion in the public benefit system and the value of the benefits received, with the individual likelihood of taking-up of the B-MINCOME benefit. Thus, M1 and M2 assess hypothesis 1 as they describe the impact of individual factors which may entail greater or lesser information costs and consequently may go on to affect take-up rates.
Descriptive statistics of sociodemographic and economic variables
In models three (M3) and four (M4) of the hierarchical regression, we added the factors identified in our second level of theoretical analysis: the administrative and institutional functioning regarding the communications strategy. In M3 we focus on communications regarding the existence of the B-MINCOME benefit to test hypothesis 2 by entering the variable of an individual whose household had received more than one invitation letter (27.2 per cent) (Table 1). In M4 we look at the communications offered on the application processes to test hypothesis 3 by using the variable of having taken part in the information sessions that, as indicated above, appears to be essential.
Results and discussion
Individual factors: sociodemographic variables
The results of the hierarchical logistic model show that the individual sociodemographic variables identified in the first level of our theoretical framework had an irregular impact. For example, contrary to other findings (Bargain et al., Reference Bargain, Immervoll and Viitamäki2012: 11), level of education did not affect the likelihood of take-up. We did not find significant differences in any model between people with a high level of education (who may be expected to have a greater understanding of the programme and face lower information barriers overall) and those with a low level or no education. Age seems of much relevance. According to Kayser and Frick (Reference Kayser and Frick2001), higher rates of non-take-up among the elderly were expected as they often have lower levels of literacy and education and consequently face higher information barriers. Results in all models in Table 2 seem to confirm this. Nevertheless, this result may also be due to the fact that older people may have had fewer incentives to apply as they knew they could be randomly assigned to an inclusion police they may not have been willing to participate in.
Hierarchical logistic regressions of requesting to apply
Note: OR=Odds Ratio; S.E.=Standard Error.
*** p < 0.001 ** p < 0.01 * p < 0.05 † p < 0.1
Regarding gender, results indicate that men were more likely to apply when they were the letter addresses in M2. This is consistent with other findings (Kemp and Davidson, Reference Kemp and Davidson2009), despite this difference being not statistically significant in the other models. Contrary to other cases (Tempelman and Houkes-Hommes, Reference Tempelman and Houkes-Hommes2016), the number of household members had no effect (Table 2). However, potential claimants living with children under 16 were more interested in applying, which corroborates our hypothesis in line with previous research (Currie, Reference Currie2004; Hernanz et al., Reference Hernanz, Malherbet and Pellizzari2004).
The neighbourhood had an impact, partly due to its sociodemographic characteristics, though this is not conclusive. Unlike other observations (Goodin et al., Reference Goodin, Headey, Muffels and Direvn2003: 51), those from the lowest income areas were more likely to apply than those from richer ones. However, the likelihood was also higher in the second highest average income areas. This paradoxical polarisation leads us to induce that, although the economic profile of a neighbourhood may be influential, other cultural or ethnographic variables could explain this relationship more accurately. As described in our theoretical analysis, literature explaining that cultural networks and ethnic ties can contribute to reducing stigma (Calnitsky, Reference Calnitsky2016) and information costs (Sharkey and Faber, Reference Sharkey and Faber2013) would support the idea that potential claimants living in the poorest areas would have shown higher rates of take-up. However, although the qualitative and ethnographic research performed during the project revealed that community ties were established throughout the project, it cannot be assumed they existed before it, nor even to what extent they may have affected take-up ratesFootnote 6 .
In this regard, previous exploratory analyses pointed out a certain association between the typology of neighbourhoods and their ethnic and cultural composition. Although there is not a correlation between the country of origin in each neighbourhood and the income level, we found a relationship between the average of their level of education and their income: the highest income areas (the reference category in the regression model) had more targeted population with tertiary and post-secondary studies (8.1% and 24.4% respectively) and less with no studies (19.8%). Therefore, socio-cultural and educational factors seem influential, though we should be cautious as the variable ‘maximum level of studies’ in Table 2 does not support this assumption. Further qualitative research on the effect of education, ethnic and cultural composition is therefore required in this respect.
In line with the importance of community, cultural and social networks in lowering stigma and information costs and consistent with other studies (Bertrand, Luttmer and Mullainathan, Reference Bertrand, Luttmer and Mullainathan2000), country of origin is one of the most relevant sociodemographic factors. Those born outside Spain were more likely to apply. Therefore, language and/or cultural barriers do not seem to have caused foreign-born people to be less interested in applying. This would contradict Dragos et al. (Reference Dragos, Hudson and Phillips2011) who sustained that in some contexts, language constitutes a barrier. Nevertheless, if we differentiate by country of birth, the results are markedly different. As seen in the analysis of the marginal effects of regression models (Figure 2 of the Appendix), those living with people from North or Sub-Saharan Africa were more likely to apply than those living with local-born people. Those born in Latin America (with Spanish as a native language) and ‘the rest of the world’ (primarily Asia) were also more likely to apply but only when the model is not controlled by attending the face-to-face information sessions.
Individual factors: socioeconomic variables
Model two of the hierarchical regression (M2 of Table 2) shows the results of entering the claimants’ household income and the public benefits previously received. The results show that receiving other benefits and their total amount somewhat determines the likelihood of taking-up. This trend increases in those who received the municipal benefit for families with children under 16, although this variable loses statistical significance in model M4 when the communications strategy is introduced. Therefore, our analysis partially endorses the first hypothesis: individual sociodemographic characteristics do contribute to explaining the information costs potential claimants must face and they determine to a great extent the likelihood of applying for the benefit. Moreover, when socioeconomic variables such as the claimants’ household income and being in receipt of other benefits are also considered, personal factors seem to gain much explanatory power. In sum, some expected information barriers (being elderly or not having previously received other public benefits) determine the probability of non-take-up to a great extent; some are not statistically significant (education); although other counterintuitive factors arise at the same individual level (being native-born).
Administrative and institutional factors: the communication strategy
Regarding the administrative and institutional level of our theoretical framework, the variable of having received more than one information letter in the potential claimant’s household is statistically significant in M3, but not in M4. Thus, the hypothesis two is partially confirmed. This is because the factor with the greatest explanatory weight was attendance at information sessions (the Nagelkerke R2 increased by 0.36 points from M3 to M4). Those who attended had an OR of 42.46 of applying for the benefit compared to those who did not (confirming therefore hypothesis 3). The magnitude of the results on how information sessions affect may be conditional on the fact that applicants could have similar characteristics differentiating them from those who did not attend the sessions. Table 3 (Appendix) shows the results of the logistic regression on the effects of some sociodemographic and socioeconomic characteristics that could make the target population take part in the sessions to a greater or lesser extent. Those born outside Spain were more likely to attend, as were those who received a higher amount of public benefits, those who received the municipal benefits for children under 16 and, in a higher proportion, those whose households had received more than one letter of invitation.
Despite the higher explanatory power of attendance to the face-to-face sessions, some individual characteristics (such as age, country of origin, having children under 16 years old, the neighbourhood of residence, or having received other public benefits) remain statistically significant and thus relevant to understand the phenomenon. Nevertheless, according to the power of the sociodemographic and socioeconomic variables at the individual level of our theoretical model (R2 Nagelkerke=0.28, M2 Table 1), and of those belonging to the administrative and institutional functioning (added R2 Nagelkerke=0.37, M4 - M2 Table 1), results indicate that factors associated to this second level of analysis have a higher explanatory power.
Conclusions
Results are consistent with some causal factors described in the literature. In line with our theoretical approach and with our first hypothesis, sociodemographic and socioeconomic variables and the information barriers they entail are of paramount importance. However, they do not fully account for the entire phenomenon. The logic and functioning of institutions when communicating about a benefit seem to have a greater influence. As our two additional hypotheses maintain, the impact of the communications strategy is very different when only informing about the existence of a benefit through letters or phone calls, or when personally informing, accompanying and counselling potential claimants throughout the entire application process with personalized information sessions. The way public institutions address people seems therefore key in reducing non-take-up rates, especially when direct communications strategies are used.
We signalled four factors belonging to the second level of theoretical analysis that may also discourage people from applying for a benefit (an insufficiently detailed application process; the duration and the amount of the benefit; the time it has been available and the changes it has undergone; or the existence of hard-to-reach, vulnerable population). Although we explained the reasons why they do not seem significant in this case, these and other additional factors (Castro and West, Reference Castro and West2022) beyond the scope of this article may certainly play a role in explaining non-take-up rates. Further research would contribute to refining these findings and reassessing our understanding of the entire phenomenon.
As seen, appropriate and exhaustive communications are essential to reach all potential claimants. As a result, the cost of the policy will increase due to the tasks and the human resources necessary to implement a comprehensive communications strategy. This raises the technical question of whether this cost can be assumed for other, larger-scale programmes (Hudson, Hunter and Peckham, Reference Hudson, Hunter and Peckham2019). Moreover, this cost is expected to increase, thus decreasing their efficiency, as the benefit is more focused and conditional, especially when targeting the most vulnerable, hard-to-reach groups. Thus, conditional benefits easily fall into a circular paradox: the more focused they are, the greater the risk of non-take-up, and hence, the higher the cost of implementation.
This research revolved around the simple question of whether individual or institutional factors were more relevant in explaining non-take-up. The answer is more complex though, since it also involves some moral concerns (Laín, Reference Laín2021, Reference Laín2022). As it is shown, not claiming a benefit does not solely (nor primarily) respond to individual factors but to the administrative functioning of public institutions. Whether or not a person decides to apply for a benefit does not seem an especially controversial question. However, from a moral standpoint, it is very much a matter of concern if the person is not claiming the benefit because they are unaware of its very existence, they have not been properly informed about a too complex application process, or this is only accessible through an online procedure they are not able to navigate. The phenomenon of non-take-up must be addressed not only by those invested in the adequate technical performance of social protection systems but also by those morally concerned about how contemporary societies treat their most vulnerable members.
Supplementary material
To view the reports on the pilot results and other supplementary material, please visit: https://ajuntament.barcelona.cat/bmincome/en/financial-aids-information.
Acknowledgements
The authors would like to thank the reviewers and editors for helpful comments on this paper.
Competing interests
The authors declare none.
Appendix
Logistic regressions of participate into face-to-face information sessions
Note: OR=Odds Ratio; S.E.=Standard Error.
*** p < 0.001 ** p < 0.01 * p < 0.05 † p < 0.1.
Logistic regressions of requesting to apply with the sample that do not participate to face-to-face information sessions
Note: OR=Odds Ratio; S.E.=Standard Error.
*** p < 0.001 ** p < 0.01 * p < 0.05 † p < 0.1.
Margins effects of the type of country of origin on requesting to apply
Note: M1: controlled by household level of studies, sex; age; household members; children < 16 years old; neighborhood; household income; level of public benefits; municipal benefit children < 16 years; more than one letter. M2: controlled by the same M1 variables plus the participation of the face-to-face information sessions.
Sample description by country of origin: Spain (56.4%), Rest Europe (3.2%); North Af. (10.4%); Sub-Saharan Af. (3.8%); Latin Am. (21.1); Rest World (5.1%).
Open access
