How Reliable Are Social Safety Nets? Value and Accessibility in Situations of Acute Economic Need
Text: HERWIG IMMERVOLL, RAPHAELA HYEE, RODRIGO FERNANDEZ, Organisation for Economic Co-operation and Development (OECD)
Social protection systems in Europe use a range of entitlement criteria – first-tier support typically requires contributions, while safety-net benefits are granted on the basis of need. This article shows stark cross-country variations in both the accessibility and generosity of safety nets in practice. It discusses drivers of these differences and their implications for policy learning and benchmarking.
Social protection systems in Europe employ a broad range of entitlement criteria. For instance, workers acquire entitlements to first-tier earnings-replacement benefits such as unemployment insurance, accident insurance or disability benefits through contributions. Individuals who lack the necessary employment history, who have exhausted their entitlements, or whose first-tier benefits are insufficient to secure their livelihood, may have access to minimum-income benefits. A principal purpose of these “safety-net benefits” is benefits of last resort. They therefore are means-tested and employ a low-income criterion as a central entitlement condition; benefits are not conditional on past employment or contribution histories.
In the policy debate, minimum-income cash benefits (MIB) are occasionally characterised as income floors. But not all low-income individuals receive support. While benefit levels are important, the impact of MIBs on family incomes (and on poverty, inequality and work incentives) depends crucially on their accessibility in practice (Immervoll, Jenkins and Königs 2014, Immervoll, 2010). In some cases, access to services (e.g. employment services) or in-kind support (e.g. social housing) may also be tied to MIB receipt.
Heightened uncertainties about the pace and extent of current and future labour market changes are one reason for renewed attention to MIBs in the policy community. On the one hand, increasingly unstable careers and incomes, paired with a greater availability of “non-standard” employment (self-employment, quasi-self-employment and casual employment) may result in a greater need for income support. On the other hand, non-standard work and an accelerated reallocation of jobs can also make it more difficult to qualify for first-tier insurancebased benefits (OECD 2019). As a result, MIBs may become more crucial as one element of governments’ strategies for stabilising family incomes, tackling inequality and preventing hardship.
This article takes stock of the income-support packages that are available to working-age individuals (18–64 years) in acute economic need. It adopts the perspective of the families concerned and focuses on jobless individuals with no or very limited access to market incomes or to first-tier income support. Drawing on existing data sources and new empirical work, the aim is to compare the accessibility and the levels of income support across European countries.
How many people receive minimum-income benefits?
Securing subsistence and providing povertyalleviation are central objectives of MIB. This article employs commonly-used relative thresholds for identifying “low-income” or “income-poor” households: households whose income (adjusted for household size) is below 50 or 60 per cent of the national median household income. As relative measures, the resulting low-income cut-offs are higher for richer countries. They should, however, not be interpreted as an explicit judgement on the amount of income that would be needed to reach an “acceptable” standard of living.
The significance and role of MIB in the overall social protection system vary across countries. Figure 1 compares MIB recipient numbers, as reported by countries’ benefit administrations, to commonly used poverty headcounts.1 Recipient numbers in France are very close to the number of income-poor households, and they are above 75 per cent in Germany, the Netherlands and the Slovak Republic. In most other countries, the ratios are 50 per cent or lower – and below 20 per cent in Latvia and Estonia. Italy introduced a new programme in September 2016 but did not operate a generally applicable national MIB prior to that. On average across countries, the number of MIB recipients in the database is less than one half compared to the number of low-income households (see figure notes). However, social protection strategies differ across countries and low-income households may have access to other non-contributory benefits in addition to MIB. For instance, Belgium and many other countries operate universal or means-tested child or family benefits. Comparing support across countries therefore requires a broader look at the overall benefit packages that may be available to those in acute economic need.
Statutory entitlements: composition of the benefit package
Families with no or little income from other sources may be entitled to support from a number of different programmes. Their income levels and poverty risks depend on the size of that overall benefit package, including notably MIB but also cash housing benefits (for those living in rented accommodation) and family benefits (for families with children). The statutory rules affecting these entitlements are complex. This section draws on the OECD TaxBEN model to assess statutory entitlements in specific policy-relevant, but hypothetical, household circumstances.2
Support for workless households
Support for single-person households without any other resources is well below commonly used low-income cut-offs in most countries (Figure 2, Panel A). On average across the countries shown, statutory rules translate into entitlements of less than one-third of median income. Only two of the countries – Ireland and the Netherlands – provide support exceeding 50 per cent of the median for this household type. At 11–12 per cent, entitlements in 2019 were lowest in Bulgaria, Italy and Romania.
Most countries provide additional support for families with children.3 MIB (and also housing benefit, if available) are typically higher than for childless adults. Some countries also provide specific top-ups for lone parents (e.g. in the form of a state guarantee for child-support payments or as a specific measure to alleviate child poverty). In most, but not all, countries, child-related top-ups are big enough to cover the extra income needed to support a larger household (that is, “equivalised” income, after adjusting for household size, is higher for a lone parent than for a childless adult). For instance, comparatively generous family benefits allowances boost the benefit packages of lone parents in Poland, Slovenia and Lithuania. But equivalent incomes in a few countries can be lower for lone parents than for childless adults (Figure 2, Panel B).
Support for low-wage, part-time workers
Do existing benefit systems encourage take-up of employment? To reduce the risk of poverty, MIB recipients need to seek income from other sources. But income gains from taking up low-paid work are often small. On average, across the countries listed in Figure 3, a part-time job with an hourly wage at the 10th percentile of the national wage distribution (a value that is often close to the national minimum wage) does move benefit recipients closer to the 50 per cent threshold but with an income gain of 8 per cent of median household income. Market income goes up significantly (+37 per cent of median income before tax, +32 per cent after tax), but this is largely offset by reduced benefits (–23 per cent of median income) when moving into work. When income gains from work are small, this also complicates governments’ efforts to tackle in-work poverty. Net gains in four countries are sufficient to lift the low-paid worker above 50 per cent of median income (Denmark, Finland, France and Germany). In all four, net incomes are close to or above 40 per cent of the median even without any income from work, and taxes and social security contributions are relatively low in this earnings range. Low-paid, part-time workers in these countries can also keep some entitlements to MIB or housing benefits. In Finland and France they receive in-work benefits. Taking up a low-paid part-time job brings particularly large income gains in Italy and Romania. While in-work benefits do not play a role in these cases, labour earnings from even a low-paid job are comparatively high relative to median incomes. And since out-of-work benefits are very low to start with, low-paid workers stand to lose very little from benefit withdrawals.
Benefit recipients in six countries would see no significant income gains from low-paid part-time work. Indeed, Austria, Croatia, Ireland, Latvia and the Netherlands fully withdraw minimum income benefits against employment income at these low earnings levels (i.e., benefits are reduced by one euro for every euro of net earnings). Lone parents experience a much milder decline in benefit payments when taking up low-paid work than households without children. Combined with the fact that their earnings increase adjusted for household size is lower than for workers without children, their overall income gain is broadly the same as for childless households (+7 per cent of median income, Figure 3, Panel B). It should be noted, however, that these calculations do not account for childcare expenses and other workrelated costs, which can be a major barrier for lowincome parents (OECD forthcoming).
Income support in practice: accessibility and generosity
A people-centred policy discussion requires information on the actual support people receive in policy-relevant circumstances. Legal entitlement rules, such as those presented above, are crucial determinants of support payments. But, for a number of reasons, comparing countries’ benefit packages based on statutory provisions alone gives an incomplete picture of the support that is available in practice. First, the strictness ofasset tests and behavioural requirements, such as active job search, differ between and within countries (Immervoll and Knotz 2018, Marchal, et al. forthcoming).
These targeting provisions limit access to MIB for ome low-income people directly; they may also make eligible households less likely to apply for support in the first place. Second, the implementation of these regulations may also vary across countries and groups of (potential) beneficiaries. Finally, the characteristics of low-income households differ between countries, affecting patterns of benefit receipt.
This section complements the information on legal entitlement rules with results on observed benefit payments and support gaps. It presents new results on the likelihood of receiving benefits, and on the amount of benefit payments in selected European countries that employ different types of social protection and for which good-quality household data were available: Austria, Belgium, the Czech Republic, Germany, Greece, Spain, France, Italy, the Slovak Republic, and the United Kingdom.
These countries indeed operate very different social protection regimes (Figure 4). Some rely very strongly on non-contributory benefits for working-age support (e.g. the United Kingdom, where three-quarters of transfers to the poorest 10 per cent of households are means-tested). Others provide the majority of support in the form of contributory insurance-type transfers (e.g. in Italy, where 73 per cent of spending on the lowest-income households is in the form of insurance-based benefits). Many continental European countries use “layered” systems that combine insurance-based first-tier out-of-work benefits with universal support for families with children and MIB as a lower-level safety net (e.g. Austria, France and the Slovak Republic). While this is also true for Germany, support for the lowest-income households is mainly provided by means-tested benefits. The circumstances of households relying on non-contributory benefits can therefore vary considerably across countries.
A model of MIB accessibility and generosity
The empirical approach consists of summarising the complex patterns of observed benefit receipt using a detailed statistical model. The main variable of interest is the value of the total package of non-contributory benefits received during the entire year 2015. It therefore accounts for both the average generosity of monthly benefit payments and the duration of payments during the year. It includes MIB as well as any related means-tested transfers (e.g. housing benefits) or universal transfers (e.g., child benefits), reflecting the fact that countries provide support through a number of different channels and programmes.4 The box summarises the main steps of this empirical analysis.
The results serve as shorthand summaries of support accessibility and generosity in a comparative perspective. In particular, they quantify the support that is available to households in situations of acute economic need, with income from other sources, including market incomes and contributory earnings-replacement benefits, in the bottom 10 per cent or bottom 20 per cent of the population, depending on the scenario.
Benefit levels and accessibility: A statistical model accounting for key policy levers
The empirical work proceeds in two steps. The first step estimates the relationship between household benefit receipt and key structural drivers of income support from the 2016 waves of the German Socioeconomic Panel (GSOEP), and of the European Union Statistics of Income and Living Conditions (EU SILC) for the remaining European countries. The second step uses the estimated relationships for inference on benefit receipt in concrete household circumstances (“vignettes”). All results refer to 2015 incomes and therefore do not account for any reforms implemented post-2015 (e.g. the introduction of a national MIB scheme in Italy and Greece, Bulman et al. (2019), OECD (2020)).
Step 1:Estimating statistical models of benefit receipt
The dependent variable is total non-contributory cash benefits received during the entire year 2015 (net of any old age and survivors’ benefits) by households with at least one non-retired working-age member (“working-age household”). Independent variables include the following key determinants of benefit receipt: family composition, household income (other than non-contributory benefits), work intensity and volatility during the year, housing tenure and costs and health status. Effective sample sizes range from about 4,000 households in Austria, Belgium and the Slovak Republic to more than 14,000 in Italy. Separate models are estimated for benefit receipt (yes/no) and benefit amounts (estimated only on observations with positive benefits) using a generalized Hurdle approach, as the process that determines whether a person receives social benefits is not necessarily the same as the process that determines the amount received (Wooldridge 2010, Cragg 1971).
Step 2:“Predicting” benefit receipt in specific circumstances (“vignettes”)
“Marginal effects” (i.e. statistics computed from model predictions for different values of the independent variables) allow summarising the entire vector of estimated parameters into a single value using the same metric as the dependent variable (here the probability of receipt and the benefit amount). Vignettes are defined by fixing specific values of some key characteristics (e.g. single with no children, couple with children, etc.) and each vignette yields a probability of receiving non-contributory benefits (stage one) and the expected average amount (stage two).
Source and further details: Hyee, Fernández and Immervoll (forthcoming)
Workless, low-resource households in Greece and Italy were least likely to receive non-contributory benefits (receipt probability of 15 to 20 per cent, Figure 5, Panel A, left-hand figure). Both countries had no generally applicable MIB schemes in 2015, but both have since introduced them (Bulman, et al. 2019). Access was also relatively limited in Spain: social assistance recipient numbers are low, and there is significant regional variation (for instance, not all regions operate housing benefits (OECD 2020)). MIB were most accessible in France and the UK.
For those receiving support, benefit levels ranged between under 20 per cent of median household income in Greece, Italy and the Slovak Republic to around 40 per cent in Belgium and the United Kingdom (Figure 5, Panel A, right-hand figure). Overall, there appears to be no general trade-off between accessibility and generosity. While benefit support in the Slovak Republic is comparatively accessible but modest, other countries combine broad access with higher benefit levels (United Kingdom and, to a lesser extent, Austria and France). In Italy and Greece, both accessibility and benefit amounts are low. In some countries, empirical benefit amounts are broadly in line with the statutory entitlements for selected model households (Belgium, Slovak Republic and United Kingdom), whereas in others, even the higher end of the confidence intervals falls short of the results indicated by Figure 2 (e.g., Austria, France and Germany). Reasons for differences between empirically observed support payments and statutory entitlements for model households include (i) average receipt durations under 12 months, as empirical benefit receipt is measured on an annual basis (e.g. if claimants move onto MIB after exhausting their entitlement to first-tier benefits or if time lags delay entitlements for those experiencing lowincome spells), (ii) benefit sanctions (e.g. if claimants do not comply with job-search requirements), and (iii) regional variations in benefit rules or implementation. Regional variations are especially important in the case of housing benefit, as housing costs vary across regions – indeed, in all three above-mentioned countries, empirically observed benefits amounts are broadly in line with statutory entitlements net of housing benefit (Figure 2).
However, in many cases, differences between statutory payments for model households (Figure 2) and empirical amounts (Figure 5) are caused by the withdrawal of benefits for households with some income. Figure 2 shows the statutory entitlement of households with no income except MIB benefits, whereas Figure 5 (Panels A and B) shows expected amounts for low-income households who might nevertheless receive some income from work or insurance-based benefits. Indeed, in Austria and France, about half of all households in the bottom decile have some labour income during the course of the year, whereas in the UK it is less than one-third. Households also receive insurance-based benefits that likely enter the means tests for other benefits (and hence lower overall non-contributory entitlements). This is likely to play a role in Austria (42 per cent of households receive insurance-based benefits), and France (30 per cent).4 In Germany, low-resource households are less likely to work and less likely to receive insurance-based benefits, so lower entitlements are possibly linked to regional variations in housing costs or temporary benefit sanctions or short benefit durations in some cases. Further research is needed to analyse and monitor the factors behind observed benefit amounts across countries.
Working part-time, or part of the year, makes benefit receipt significantly less likely. But a gradual phase-out of MIB (as in Austria, France, Germany) or in-work benefits (as in France and the United Kingdom) can provide continued support to low-paid workers and those in intermittent employment as well as strengthening work incentives (Figure 5, Panel B).
In a number of countries, families with children receive family benefits regardless of their income (Austria, Belgium, France, Germany, the Slovak Republic and for all but very high incomes, the United Kingdom).5 As a result, all low-income families with children received at least some non-contributory benefits in these countries (Figure 5, Panel C). In Spain, a means-tested child allowance provides support to most, but not all, low-income families. Results also indicate that low-income lone parents in Italy had partial access to non-contributory benefit in Italy, even though no universal or means-tested national child benefit or MIB schemes existed in 2015; this likely reflects preferential treatment of families with children in regional MIB schemes.
All EU and OECD countries operate safety-net programmes for working-age individuals, either as lower-tier programmes alongside primary income replacement benefits, or as a principal instrument for delivering social protection. As benefits of last resort, these transfers play a vital role in alleviating poverty and social exclusion and they typically employ strict means-testing to target households in acute economic need. A key objective of these programmes should be to provide meaningful income protection for groups with no or very limited access to other resources.
However, the accessibility and levels of these income transfers differs enormously across countries. The country disparities illustrated in this article are partly due to differences in the overall architecture of social protection systems. But they also indicate substantial scope for upwards convergence and policy learning, through impact assessments and through policy monitoring and benchmarking across and within countries. Peoplecentred policy indicators based on statutory entitlement rules provide a useful starting point for such an exercise. In addition, identifying policy challenges, and solutions, requires granular results on observed patterns of benefit receipt among low-income households.
- Bulman, T., D. Pacifico, M. Pisu, and O. Rastrigina. (2019). “Tax and benefit reforms to support employment and inclusiveness and address poverty in Italy.” OECD, Paris.
- Cragg, John G. (1971). “Some Statistical Models for Limited Dependent Variables with Application to the Demand
for Durable Goods.” Econometrica (JSTOR) 39 (5): 829.
- Hyee, Raphaela, Rodrigo Fernández, and Herwig Immervoll. (forthcoming). “How reliable are ‘safety-net’ benefits? Value and accessibility of minimum-income support in comparative perspective.” OECD, Paris.
- Immervoll, Herwig. (2010). “Minimum Income Benefits in OECD Countries: Policy Design, Effectiveness and
Challenges.” OECD, 1-49.
- Immervoll, Herwig, and Carlo Knotz. (2018). “How demanding are activation requirements for jobseekers?
New evidence on activity-related eligibility criteria for unemployment and social assistance benefits.” OECD.
- Immervoll, Herwig, Stephen P Jenkins, and Sebastian Königs. (2014). “Are Recipients of Social Assistance ‘Benefit Dependent’? Concepts, Measurement and Results for Selected Countries.”
- Marchal, S., S. Kuypers, Ive Marx, and G. Verbist. (forthcoming). “The impact of asset tests in Europe: an application to minimum income schemes in Belgium and Germany.” CSB Working Paper.
- OECD. (forthcoming). “Can parents afford to work? Childcare costs and work incentives in OECD countries.”
- OECD. (2019). “Left on your own? Social protection when labour markets are in flux.” Chap. 7 in OECD Employment Outlook. (2019): The Future of Work, by OECD, 292-335. Paris: OECD publishing.
- OECD. (2020). TaxBEN country policy descriptions.
- Wooldridge, J. (2010). Econometric analysis of cross section and panel data. MIT Press.