Analytical framework: Implementation failures in minimum income schemes

The literature on public policy implementation has identified different potential sources of implementation failure (Bardach, Reference Bardach1977; May, Reference May2014; Hill and Hupe, Reference Hill and Hupe2002; Mazmanian and Sabatier, Reference Mazmanian and Sabatier1983; Palumbo and Calista, Reference Palumbo and Calista1990; Pressman and Wildavsky, Reference Pressman and Wildavsky1979; Van Meter and Van Horn, Reference Van Meter and Van Horn1975), which are mainly related to how policy goals are defined, how the policy is regulated and designed, and the structure of interests and relationships between the different agents in charge of implementing/governing the policy as well as those whom the policy is targeted at.

In short, these sources of failure may be broadly reduced to two types (Matland, Reference Matland1995). First, the ambiguity in the policy goals, which includes vagueness in the definition of the goals themselves, as well as inadequate specification of what the administration bodies or frontline practitioners should do to implement the policy. This is particularly damaging to the success of a policy when its implementation has high informational costs for the administrative bodies, the citizens, or both. This is the case for MI schemes, which typically involve many eligibility requirements and are strictly targeted.

The second main source of implementation failures, reported in the literature, is a blind design to potential conflicts between the agents and agencies involved, and to their incentives to block or undermine in practice the policy goals. Empirical research has shown how bureaucratic practices and reactions from targeted recipients of a policy may distort, and even turn around, its purported goals. In a similar vein, several features in the design of a policy and its legal regulation may undermine its implementation in particular contexts. Moreover, complex chains of decision-making and the involvement of many administrative bodies can also create coordination challenges and disincentives for the agents in charge of the implementation process.

For the purposes of this article, the above depiction presents certain limitations. First, further concretion is needed to address the specificities of MI schemes, which very often are policies with well-defined goals, but high informational costs in their implementation (Marx and Marchal, Reference Marx and Marchal2024). The objectives of an MI scheme are clearly related to reducing poverty rates and their intensity, and frequently, furthermore, promoting social inclusion and labour activation. However, in the case of the MLI, labour market integration and activation were not salient issues at all, since the policy was negotiated, designed, and implemented primarily as a response to an emergency situation created by the pandemic (Noguera, Reference Noguera, Salido and Massó2021).

Secondly, in the case of highly targeted MI schemes, such as the MLI, informational costs are an important potential source of implementation problems, besides ambiguity, miscoordination, and conflict among agents (LeGrand, Reference LeGrand2003). Therefore, this should be further stressed when analysing failures in these types of policies. Tight targeting implies the need to properly identify eligible beneficiaries while minimising both false positives and false negatives, whose existence tends to undermine the legitimacy of these schemes. Moreover, informational challenges are present here for both the administration and the policy’s beneficiaries. When different layers and levels of the administration are involved, sharing information through fast and compatible interoperable channels is key. Additionally, MI schemes often rely substantially on intensive work from social services and street-level bureaucracy, which is expected to process complex regulations and gather information on eligibility requirements that are frequently difficult to check and certify. For citizens in situations of poverty and social exclusion, in turn, complex requirements and claiming procedures also create informational challenges, which often increase their “cognitive load” in conditions of narrow “bandwidth” (Banerjee and Duflo, Reference Banerjee and Duflo2011; Mullainathan and Shafir Reference Mullainathan and Shafir2013).

In order to address these specificities, we propose an analytical framework that aims to create a typology of the specific implementation failures that may affect MI schemes. This framework uses two dichotomic criteria to produce a typology of problems. The first distinguishes two different types of implementation challenges, depending on whether they are related to the identification of potential recipients or the benefit’s management. The second concerns the main agent whose behaviour may trigger the challenge in the first place. As LeGrand (Reference LeGrand2003) and other scholars have shown, actors involved in the policy process and its implementation are not “knights” nor “pawns”, but active strategic and heuristic-based agents who face different contexts in terms of opportunity sets, available information, and incentives. Based on this, their behaviour may trigger some kind of failure in the policy’s implementation, either from the supply side (administration) or the demand side (potential recipients).

Table 1 presents the four ideal types of implementation failures in minimum income schemes that result from applying these two criteria. Firstly, from the supply side, the behaviour of different administrative bodies, and the layers within them, may trigger problems in the identification of potential recipients when there are complex informational requirements for benefit eligibility. Very strict and/or prejudiced enforcement of the requirements will typically produce false negatives, while more lax enforcement will generate false positives. Consequently, in both scenarios, when numbers reach a specific threshold, the legitimacy of the policy and its public support may be jeopardized. Secondly, administrative capacity and competence may also be structurally limited, cognitively or informationally inadequate to implement the scheme, or motivationally biased against its goals. This will result in incapacity from the supply side to execute the policy.

Table 1. Implementation failures in minimum income schemes

Source: own elaboration.

From the demand side, a similar subjective identification problem arises for eligible citizens, who have to, first, be adequately informed about the policy’s existence and characteristics, and second, be aware of their eligibility. As the high rates of non-take-up in MI schemes often prove, this is far from being the case for many of them. Additionally, when reporting their level of income (and often also assets) to the administration, as well as any changes they may experience, claimants can make mistakes, be delayed, or be unaware of their actual situation, resulting in overpayments and underpayments in the benefits they are entitled to. This typically produces unpleasant reimbursement claims from the administration that harm the economic expectations of vulnerable citizens, and can also affect the legitimacy of the scheme.

Although, of course, there might be other challenges related to the implementation of MI schemes, we are confident that this analytical approach covers the main issues reported by the empirical literature, and is a useful instrument to assess the specific case of the Spanish MLI.

Methodology

To identify both the main challenges and the gaps in research regarding the implementation of the MLI in Spain, we have carried out an exhaustive literature review. The selection process has entailed combining different methods of article recollection: online searches in digital repositories and academic libraries, resourcing to the research team’s own archives, which have resulted from years of tracking this topic, and employing the “snowball technique” on the documents selected through the first two methods. As a result, we are confident that all relevant empirical studies on the implementation of the MLI in Spain have been considered.

The primary criterion for selecting reports and academic papers was the inclusion of empirical evidence addressing the implementation of the MLI scheme, up to December 2024. These studies can use quantitative, qualitative, or mixed methods, and be carried out by either academic or non-academic institutions (such as NGOs, official institutions, etc.), but they are carefully selected for being scientifically minded and using rigorous techniques for data collection and/or analysis. Moreover, our analysis is supported by a review of the legal framework and other resources related to the MLI benefit that are not part of this literature review.

A key challenge encountered during this process was the limited availability of academic literature meeting this criterion. Many academic studies on the MLI tend to adopt a more normative approach or focus primarily on the design of the benefit, rather than its implementation. Consequently, this study has identified and selected only 17 scientific articles and evaluation reports. Table 2 lists these items and describes the types of implementation problems they analyze in the MLI scheme.

Table 2. Studies on the implementation of the minimum vital income in Spain

Within this selection, there are two qualitative, seven mixed methods, and eight quantitative studies. Regarding the topic of these studies, the first three dimensions are analysed in a similar number of articles and reports: administrative capacity (11), identification (11), and non-take-up (10). In contrast, overpayments are covered by only five studies. Moreover, although some studies are centred on one specific issue, others examine multiple dimensions. In particular, the three reports from AIReF (the Spanish Independent Authority for Fiscal Responsibility) provide evidence on all four dimensions. However, this does not imply that the analysis is equally comprehensive for each topic, as the depth of scientific evidence provided on each dimension varies both within and between studies. A final remark is that since the introduction of the CAPI in 2022, most studies include this supplement in their analysis (for example, when estimating non-take-up).

Identification challenges

The MLI has been designed as a highly targeted benefit and entails a series of requirements that create a significant burden for applicants, who must come up with an important amount of documents that are not always easy to access in order to certify residency, household composition, income, assets, and other eligibility conditions. Furthermore, as assessed by some of the reviewed studies, these requirements can lead to exclusion errors, meaning some vulnerable households that require income support may be considered ineligible, and even some errors of inclusion.

On one hand, means-testing relies on outdated and incomplete information on the households’ income, as income assessment is based on the prior fiscal year. There is a subsidiary mechanism for households to apply for MLI based on income from the current year and according to the one of the AIReF reports (2023), 30% of total applications until 2023 asked for this mechanism to be applied. However, given its subsidiary nature, only 2% of beneficiaries receive or have received MLI based on their current income. Ultimately, assessing eligibility based on income from the previous fiscal year entails important issues for households who experience unexpected poverty, and results in many false negatives. It may also create “false positives,” as people who at present have sufficient income may be eligible based on their previous circumstances. This is reflected in the main causes for benefit denial, which have been analysed in the AIReF reports for 2022 and 2023. In both these documents, income was the first cause (42% in 2021, 32% in 2022), followed by residence requirements (22% in 2021, 19% in 2022). Furthermore, FOESSA (2021) gathered information on benefit denial for households in severe poverty through their own survey and found that 52.8% of these households were denied the MLI due to income requirements. For this reason, AIReF has consistently suggested a policy reform that aligns income assessment more closely to the time of application.

On the other hand, both ECAS (2021) and Costas et al. (Reference Costas, Ferrer and Ponce2022) point towards requirements on household composition, which do not adjust to multi-family households or other housing situations vulnerable households may find themselves in. Residence requirements, in turn, exclude homeless people, irregular immigrants, and other social groups, and age requirements make adults below 23 years old non-eligible regardless of their income. Hogar Sí (Reference Hogar2021) identifies specific causes for benefit denial in the case of homeless people, where 32% of homeless applicants stated they did not know or were not informed about the scheme, followed by 17% who stated they did not comply with residency requirements. The previously mentioned FOESSA report (2021) also found that non-compliance with household composition requirements was an important cause of benefit denial, affecting 12.9% of severely poor households that were not granted MLI after applying for it.

Lastly, asset testing may also entail exclusion errors. According to AIReF (2022) and (2023), 9% of application denials were related to non-compliance with asset requirements in 2021 and 6% in 2022. Although this issue is smaller than the previously analysed, Dobrikov (Reference Dobrikov2024) estimates 168.000 households that comply with the remaining MVI requirements are excluded due to asset testing, that is 15% of total eligible households if asset testing were completely eliminated. Although some of these households have high wealth levels, around 20% are only slightly above the threshold, and are driven to either sell their assets or use up savings to survive, further worsening their already vulnerable situation.

Nevertheless, several reports (ECAS, 2021; Hogar Sí, Reference Hogar2021; EAPN, 2024) also point to the fact that, even if the claimants are in fact eligible, complex targeting procedures and identification techniques generate a high administrative burden both for practitioners and applicants, which may result in incomplete or incorrect applications that are denied until claimants give up.

Administrative capacity challenges

The introduction of the MLI in Spain during the pandemic was marked by urgency and unprecedented circumstances. Originally, the MLI was meant to be phased in gradually over an entire legislative period, beginning with child protection initiatives. However, due to the urgency created by the pandemic, it was passed in full within just 6 months, with child supplements only introduced nearly 2 years later. This led to a series of implementation challenges.

Firstly, the Spanish Social Security administration faced significant operational challenges in implementing the MLI. Traditionally, it has managed individual contributory benefits with relatively automated processes for large, homogenous groups with registered contribution records. Instead, administering the MLI required a shift to assessing household-based benefits and checking heterogeneous circumstances on a case-by-case basis. This entailed verifying residency, household composition, and means-testing for households with fluctuating income from different sources.

Secondly, the pandemic strained administrative capacity, with many staff on sick leave or working remotely. As pointed out by ECAS (2021), initially, there was a combined need for training and adaptation on behalf of the personnel in charge, with a rise in vulnerable situations and demand for economic support. Also reflected in this report is the fact that local social services maintained in-person assistance during the pandemic and were under great pressure, as there was an urgent need for support in benefit application. This also links to the findings by Prieto Lobato and De la Rosa Gimeno (Reference Prieto Lobato and de la Rosa Gimeno2023), who describe how the implementation of MLI has led professionals to substitute social intervention for support in benefit application. However, even when families managed to apply, the process of resolution was very slow. AIReF (2022) analyses how most of the 1.5 million applications during 2021 were concentrated during the first months of implementation, leading to an initial resolution period of over 3 months. This means that, although the implementation of MLI was accelerated due to the pandemic, it did not succeed in responding rapidly to an emergency situation. Moreover, mean resolution periods have not decreased, remaining between 120 and 141 days in 2022 and 2023 (AIReF, 2023, 2024).

Thirdly, the MLI application process is cumbersome and largely online, featuring a lengthy and complex form, which proved inefficient and inaccessible to many applicants. This is a shared conclusion of the qualitative analyses carried out by Costas et al. (Reference Costas, Ferrer and Ponce2022), ECAS (2021) and AIReF (2024), while both Hogar Sí (Reference Hogar2021) and Bascones and Álvarez (Reference Bascones and Álvarez2023) mention specific barriers faced by homeless and disabled people. Bascones and Álvarez (Reference Bascones and Álvarez2023) make a special mention of the online application form, as it entails specific barriers for disabled people. Furthermore, Hogar Sí (Reference Hogar2021), in its quantitative analysis, mentions how the great majority of homeless people were not able to fill in the application autonomously (88%), and most (46%) did not receive support from the administration, but from NGOs. Moreover, as detected by Meers et al. (Reference Meers, Halliday and Tomlinson2025), an interface-first claiming process leads to accessibility problems and “glitches” have an important discouraging effect, which can ultimately lead to benefit non-take-up. The authors of this article, for instance, were able to verify that, until recently, the MLI simulator embedded in the Social Security webpage did not include the child supplement, therefore erroneously leading a subset of eligible users to think they were not.

Lastly, another source of complexity was integrating the MLI with pre-existing regional MI schemes. Initially, the Ministry of Social Security considered layering the MLI on top of regional MI schemes if they did not reach the severe poverty threshold, but the final decision was for regional policies to become subordinate to the national scheme, as seems more appropriate for decentralized quasi-federal countries such as Spain. However, this articulation introduced considerable administrative complexity, as regions had varying levels of MI generosity and eligibility criteria, but also different resources and capacity. As found in the qualitative analysis of Noguera and Quintana (Reference Noguera and De Quintana2021), initially, there were important problems of coordination and access to information on behalf of regional administrations to determine the eligibility of regional MI beneficiaries to the national scheme. Moreover, Aguilar and Arriba (Reference Aguilar Hendrickson and Arriba González de Durana2024) also identify different strategies on behalf of the regional governments: although some regions have maintained and even improved their MI schemes (e.g., the Basque Country), others have practically eliminated the regional benefit without allocating additional resources to alternative social benefits (e.g., Madrid).

The problem of non-take-up

Non-take-up is a common issue in MI and other targeted policies, where a proportion of eligible households do not apply for or access benefits despite meeting all the policy’s requirements. Research on this phenomenon mostly revolves around three primary challenges: estimating its scale, identifying its causes, and assessing the effectiveness of different solutions (Marx and Marchal, Reference Marx and Marchal2024).

In the case of Spain, the limitations of administrative data have been a crucial challenge for estimating this phenomenon. Until February 2024, the only available data on MLI recipients was the accumulated number of beneficiaries since the policy’s introduction in June 2020, which included households that had received the MLI but were no longer beneficiaries. Since then, transparency has increased and the Ministry of Inclusion, Social Protection and Migration has begun publishing monthly updates on the current number of MLI recipients. However, the lack of data created important limitations for those attempting to evaluate the policy’s implementation process and its rate of non-take-up. Regarding survey data, the Spanish Living Conditions Survey (source for EU-SILC) also has important data limitations for simulating benefit eligibility (lack of representation of MLI recipients, of information on residence, wealth, etc.). Despite these restrictions, some studies have estimated this phenomenon, obtaining slightly different results but agreeing in their main conclusions: non-take-up is a crucial problem in the MLI scheme, and, although the temporal trend indicated by the studies suggests improvement, take-up is still low.

Looking first at the coverage of impoverished households, OECD (2023) estimates that for the year 2021, only 40% of those living beneath the extreme poverty line (defined as 30% of the national average disposable income) were potentially covered by the MLI. FOESSA (2021), however, with its own survey specifically focused on representing impoverished households, estimated that during the first year of the MLI’s implementation, take-up amongst severely poor households (household disposable income below 40% of the national median income) was only 26%. Lastly, using the same definition of severe poverty as FOESSA (2021), Noguera et al. (Reference Noguera, Álvarez-Cronin and De la Prada2024) estimated that potential coverage of severe poverty in 2023 was lower than 25%, but in some regions could reach almost 50% (Catalonia) or even 100% (Basque Country) when combined with regional MI benefits.

Additionally, other studies and institutional reports offer a broader picture. AIReF, in its most recent report (2024), which refers to the benefit’s implementation in 2023, estimates that 56% of eligible households for the basic MLI do not receive it, and non-take-up of the child supplement (CAPI) rises to 73%. Noguera et al. (Reference Noguera, Álvarez-Cronin and De la Prada2024) estimated a non-take-up rate of up to 74% for the basic MLI benefit that same year, and 88% for CAPI. The difference between these two estimations is partly related to the fact that the AIReF has access to administrative data that the second study lacks, but it is also related to different definitions of non-take-up. The first study does not consider those who have withdrawn their application when calculating the MLI’s non-take-up rate, while they are included in the second. This creates an interesting discussion on which should be the preferred criteria that could be dealt with in future studies, as reasons for withdrawal can be complex and may not necessarily indicate that a household no longer needs the benefit.

Regarding the second challenge, different academic articles have found that non-take-up is a multicausal phenomenon (van Oorschot, Reference Van Oorschot1991; Hernanz et al., Reference Hernanz, Malherbet and Pellizzari2004; EUROFOUND, 2015; Baumberg, Reference Baumberg2016; Janssens and Van Mechelen, Reference Janssens and Van Mechelen2022; Ko and Moffitt, Reference Ko, Moffitt and Zimmermann2022), where lack of information, transaction costs, high conditionality, excess of bureaucracy, administrative complexity and social stigma have been identified as some of the main explanations. In the case of the MLI, AIReF (2022) identified further specific causes of non-take-up when analysing its first year of implementation. On one hand, they found a statistically significant relation between non-take-up and the presence of income from work or other benefits in the household, especially unemployment benefits. On the other hand, they also found that a greater potential gain in income was related to a greater probability of taking up the benefit. This suggests that low benefit levels, paired with the absence of an effective wage supplement – which could increase potential gains for working households – may be connected to non-take-up in top-up benefits like the MLI. Furthermore, AIReF (2023) identifies the slow process of integration with regional MI benefits as another potential cause for persistently high non-take-up rates. That is, eligible households receiving regional MI benefits have not yet been transferred to the national scheme.

In addition, the experimental study conducted by Noguera et al. (Reference Noguera, Álvarez-Cronin and De la Prada2024), showed that the declared reasons for non-take-up of potentially eligible households were related to the presence or absence of children. Two of the main reasons for non-take-up were either not knowing or not believing that the household was eligible. For households with children, these factors ranked as the first and second most common reasons; for childless households, they ranked second and fourth. Lack of awareness (not knowing of the benefit’s existence) was the primary reason for childless households. These differences are mainly led by the presence of CAPI, referred to in previous sections, which boosts the eligibility of households with children beyond the poverty threshold, although most remain unaware that they can apply for this child supplement. FOESSA (2021) also found that 56.8% of severely poor households had no information on the MLI and 10.7% had insufficient information to begin the application process, and AIReF identified lack of information as one of the key factors in its latest report (2024), after carrying out a qualitative analysis with focal groups of experts and representatives of social entities. Ultimately, insufficient awareness of the policy and its requirements is a persisting issue that has not been solved within the first years of the MLI’s implementation. Additionally, FOESSA (2021) analysed the main features of households who had taken up the benefit, observing greater take-up rates in single-parent households and large families, and lower rates in childless households. Low take-up rates were also found in households where the main providing adult had a job, therefore implying that the benefit has not been sufficiently accessible for the working poor, at least in its first year of implementation.

Lastly, solving the problem of non-take-up has been the focus of experimental studies, in which potentially eligible households are typically treated with different types of nudges to evaluate their effectiveness in rising benefit take-up. The Ministry of Inclusion, Social Protection and Migration, together with EAPN (European Anti-Poverty Network), carried out a project named “Acceso Vital” to evaluate different interventions that may help reduce non-take-up in the MLI, finding no significant results when it comes to purely informational campaigns not specifically targeted at potentially eligible households, but important increases in take-up (19.5%) when households are given sufficient support during the application process. The previously mentioned study by Noguera et al. (Reference Noguera, Álvarez-Cronin and De la Prada2024), on the other hand, found that purely informational nudges specifically targeted to potentially eligible households did have a statistically significant impact in households with children (16% increase in take-up), but not in childless households.

Overpayments and requests for refund

MI schemes are traditionally designed as means-tested and top-up benefits, which makes them highly sensitive to changes in households’ income. When this design is combined with slow and rigid mechanisms for adjusting benefit amount and entitlement, it often results in overpayments. The issue of overpayments is particularly serious in these types of policies because they are specifically targeted to very low-income households, which often face income instability – increasing the risk of overpayment – and have limited capacity to confront debt (Millar and Whiteford, Reference Millar and Whiteford2020).

However, this issue is largely understudied in the context of the MLI. The size and evolution of this phenomenon, its conditioning factors, or its possible solutions have not been the central subject of academic research. Within the five studies that include evidence on overpayments, two correspond to the AIReF reports, and the remaining provide limited evidence based on qualitative interviews, mostly directed at experts. Therefore, this section needs special complementary information drawn from legal and institutional documents or communications.

From the available literature, we can draw that the issue of overpayments in the MLI benefit, although mainly related to changes in household income, is caused to a great extent by a particularly rigid design of the scheme. Both benefit amount and eligibility are determined by the household’s income in the previous fiscal year; therefore, they adapt more to the circumstances of households in chronic poverty than to those with unstable incomes, who face specific periods of unexpected poverty.

The AIReF report (2023) explains how untimely mechanisms for benefit update, based on previous yearly earnings, systematically generate either underpayments, which are compensated retroactively, or overpayments, which may entail benefit discounts (if the household still holds entitlement) or refund claims (if the household has lost entitlement). According to the report, although benefit updates are scheduled to be applied on January 1 (as mandated by Article 16.3 of the MLI Law; Congreso de los Diputados, 2021), they are based on fiscal information that is only available for the National Institute of Social Security starting from the last quarter of the year. According to the same report, benefit updates negatively affected 43% of households that were receiving the benefit in 2022, as 27% saw their benefit reduced and 16% lost the benefit, therefore incurring debt. The AIReF report for the following year (2024) shows a slight decrease in benefit reduction (25%) and a more significant one for benefit loss (7%). The cause of this decrease is not specifically analysed, but it may be the result of more drastic changes in income during the first period, as benefit updates in the year 2022 affected households that had been granted MLI based on their income during the pandemic in 2020. The exact number of households that incurred debt is only available in the first report (2023), and it amounts to 47.932 households. Moreover, both reports coincide that households who lost the benefit were also those who accumulated greater amounts of overpayment, with a median debt of 2.500€ in 2022 and 1.400€ in the year 2023. Furthermore, maximum debt amounts rise above 6.000€ in both periods. This involves a great amount of financial pressure on households that, in all probability, are still in a vulnerable economic situation.

Qualitative results from Costas et al. (Reference Costas, Álvarez, Aumaitre, Martínez García, Muñoz and Sánchez2024) emphasise that households are only given 30 days to pay these debts and are not properly informed about the possibilities of fractionating payments. Bascones and Álvarez (Reference Bascones and Álvarez2023) stress that, moreover, payments are only fractionated within a 3-month period. This same study points to delays in benefit updates when changes in the household’s circumstances are specifically communicated, for example, after accepting a job. According to interviewees, including both social entities and beneficiaries (which, in the case of this report, are all people with disabilities), this creates fear and uncertainty, and often disincentives people to accept temporary jobs, as this may involve losing the benefit and incurring in debts they may not be able to pay off in the future.

An additional source of overpayments is mentioned in Costas et al. (Reference Costas, Álvarez, Aumaitre, Martínez García, Muñoz and Sánchez2024) and briefly covered in Noguera and Quintana (Reference Noguera and De Quintana2021): the integration process between the new national scheme and the previous regional MI benefits. As mentioned in previous sections, regional benefits have become complementary to the MLI. However, during the process of integrating both benefits and transferring beneficiaries of regional schemes to the national MLI, many households have experienced an overlap in payments and have been required to reimburse these overpayments. Also, in Noguera and Quintana (Reference Noguera and De Quintana2021), professionals who were interviewed describe how the regional administrations have been burdened with the responsibility of solving this problem of coordination and finding solutions for the households affected.

Given the potential consequences these debts may have on already vulnerable households, from January 2023, MLI overpayments below a certain amount are pardoned, but only for households with children. The Real Decreto-ley 20/2022 modified Article 19 of the MLI law (Congreso de los Diputados, 2021), which now states that “The amounts that do not exceed 65 per cent of the monthly amount of non-contributory pensions shall not be subject to claim when at least one beneficiary in the household unit is a minor.” While this represents a potential improvement, it only applies to small debts, which are mostly those who maintain benefit eligibility. This leaves households with larger accumulated debts facing significant challenges in repaying both the overpayments and the associated interest. Regarding overpayments in regional MI benefits caused primarily by their integration with the MLI, different initiatives have been put in place on behalf of regional governments to alleviate or solve this situation. This was covered by the 2022 annual report from the Spanish Ombudsperson, which mentioned practices such as making the regional MI compatible with the MLI for a period of time (Aragon), but also cases where action was still to be taken.

Overall, overpayments related to the implementation of the MLI are a significant, yet understudied, phenomenon. There is still an important literature gap related to understanding the magnitude, the short- and long-term effects on households, the determining factors and, lastly, the possible solutions to this problem.

Concluding remarks

After reviewing the available literature on four sources of implementation failures, we have identified several issues concerning the implementation of the Spanish MLI. The exhaustive literature review presented above suggests that the study of implementation challenges in MI schemes should pay more relative attention to informational factors that affect the supply and demand of these policies, and that are key in addressing targeting as well as management challenges. However, coordination between different administrative levels has also been shown to have an important role in decentralized states such as Spain.

First, given its strict eligibility criteria, the identification challenges this policy faces are mainly related to errors of exclusion. Its design may be suitable for specific situations of long-term poverty, but its requirements create exclusion errors for those facing either temporary or unexpected periods of economic vulnerability or special circumstances (such as homelessness, irregular status, sharing residency with other household units, etc.). Although the existing research has already given steps in quantifying these errors, including for specific groups, some of the literature refers to the initial phases of the benefits implementation. Also, the time gap between income level at the moment of claiming the benefit and the income used to determine eligibility (which refers to the previous fiscal year) has proven to be a clear source of identification problems. Further research is needed to assess how the inclusion of severely poor households and vulnerable social groups has evolved, and which factors still lead to benefit exclusion for households who are within the policy’s target population.

Second, administrative challenges were especially important during the first year of the MLI’s implementation, but persisted in the following years, including long waiting periods for benefit resolution, personnel leaving aside assessment related to social and labour market inclusion to support applicants, NGOs having an important role in assessing vulnerable households due to insufficient administrative capacity, and digital and cognitive barriers that suppose an impediment for households and especially vulnerable groups to apply for the benefit autonomously. These are some of the identified persisting challenges that have been analysed in different quantitative and, more especially, qualitative studies. This dimension has been one of the most broadly assessed within the sources of implementation error. However, the challenges related to the integration of MLI and regional benefits will be a central issue in the following years, and the problems or possibilities this may involve have yet to be studied in greater depth.

Third, take-up has been identified as a central problem in the implementation of the MLI. Estimates that aim at quantifying this phenomenon have given an important insight into its magnitude. However, greater access to data is essential to understand the main factors involved, the access to MLI of specific groups, such as migrants or working poor adults, the central causes of non-take-up (administrative, informational, behavioural, etc.) and, more especially, potential solutions to this problem. The available evidence points towards insufficient information and lack of institutional support, paired with administrative complexity, as fundamental causes for non-take-up. The evidence also shows that there may be an insufficient adaptability to the changing income of impoverished households. The higher non-take-up rates of the CAPI supplement (which, as said, has higher income eligibility thresholds than the basic benefit) suggest that many eligible households with income over the basic threshold do not consider themselves as “poor” (often rightly) and therefore lack awareness of their own eligibility.

Lastly, overpayments and the potential problems refund claims may cause on the families affected (economic, psychological, and in their attitudes towards public institutions and social benefits) are the most understudied implementation issues. The existing evidence, although scarce, points towards administrative problems related to benefit updates and the MLI’s integration with regional MI schemes. However, other factors deserve more attention in future research, such as the long-term consequences of overpayments and reimbursement claims on the economic and mental well-being of families, the possible disincentives created by the fear of incurring in debt, which may affect benefit take-up or job acceptance by recipients, or the potential solutions to this problem.

All the dynamics analysed suggest that a highly targeted design – shaped by budget constraints, limited administrative capacity, and concerns over social legitimacy – has led to an institutional framework that ultimately hinders the rapid and effective implementation of the MLI. Virtually all the reports reviewed agree that some reforms might improve this implementation: adjusting the income assessment period to the present, reducing administrative burdens, raising MLI thresholds, or automating adjustments of the benefit to changes in income or households’ conditions, are probably not the only necessary steps forward, but they might be key advances for the effectivity of the MLI in supporting vulnerable families and significantly reducing poverty in Spain.