Introduction

Welfare programs available for low-income households usually have several goals. The most important are distributional objectives such as securing a minimum standard of living, enabling social participation and reducing poverty. Therefore, they are usually targeted at selected needy groups within the population. Due to targeting rules, e.g., income means-testing, these programs are more complex than universal programs. One common problem of countries relying on means-tested benefits to support low-income households is a lack of targeting effectiveness, (1) indicated by an extensive empirical evidence on low levels of benefit participation among European countries and the USA (Eurofound 2015; Currie 2004). These findings are of political concern because low benefit participation implies that programs are ineffective in reaching their target group and thus their distributive goals. The literature on benefit participation assumes that claiming a benefit is costly (Moffitt 1983; Blundell et al. 1988). Costs of claiming can be differentiated into stigma, information and transaction costs. The latter two do not only occur as a result of the complex targeting rules of a particular benefit, but also because the benefit systems in modern welfare states are often characterized by several--often interdependent--means-tested benefits, organizational fragmentation and a division of political responsibilities. This complexity leads to insufficient knowledge or false interpretation of entitlement rules as well as of the claiming process.

Besides distributional goals, another important objective of means-tested welfare programs is to limit their potential adverse effects on labor supply. Individual labor supply decisions are affected through the benefit level, which influences the recipients' utility from non-employment and through the benefit reduction rates inherent in the programs. Benefit systems in advanced welfare states are often characterized by high benefit reduction rates (Blundell 2012), which has motivated an extensive international research on the labor supply incentives of means-tested benefit programs (Moffitt 1983; Friedberg 2000; Meyer 2002; Moffitt 2002; Brewer et al. 2006; Lemieux and Milligan 2008). Moreover, the availability of several uncoordinated means-tested programs bears the risk of excessively high combined benefit reduction rates, which can even lead to falling net incomes for certain ranges of earned income.

This study analyzes the relation among three major means-tested programs available for low-income families in Germany. The three programs share the same objective--providing income support--and address largely the same population. Implicitly, they act as a kind of in-work benefits for low-income, low-paid workers and their families, because all three programs try to encourage recipients to work or expand working time by imposing benefit reduction rates below 100% on earned income. The three benefits are based on different complex income means-tests, whereas the entitlement to one benefit depends on potential entitlements to the other benefits. Despite the dependencies between the three programs, political competencies for the programs as well as their organization are distributed among different policy fields and levels of government. Thus, the system of benefits we analyze is similar to the system in many other modern welfare states.

Understanding the interaction between different benefits and their effects on benefit participation and work incentives may help to inform policy about weaknesses in the current welfare system. Identifying these weaknesses might provide guidance in redesigning a benefit system to better achieve the desired goals of providing income support to the needy population while at the same time integrating low-wage earners into the labor market.

We utilize a static microsimulation model, which consists of a detailed implementation of the German tax and transfer system and is based on data representative for the German population, to simulate the effects of entitlements to the three considered benefits. In order to evaluate the level of non-take-up, we contrast the simulated entitlements with the take-up of the benefits as observed in the data. Furthermore, we discuss the total marginal tax rate resulting from the interaction of the three benefits, as encountered by stylized low-income households. Finally, to illustrate the interdependency among these three means-tested programs, we present the effects of the recent housing allowance reform in Germany, which was implemented in 2016, on the effective marginal tax rates, government expenditures and caseloads for each program.

Existing studies of the German means-tested benefit systems typically either focus on single programs or they look at the hypothetical labor supply incentives of the combined programs while assuming full benefit take-up. The existing evidence points to adverse effects on labor supply (BMFSFJ 2009; Voigtlander et al. 2013; Wiemers and Bruckmeier 2009; Meister 2009; Knabe 2006), and high non-take-up rates (Bruckmeier and Wiemers 2012).

Our analysis makes two contributions. First, we discuss both, benefit take-up and labor supply incentives of all three programs. Looking at both is important because benefit take-up and labor supply should be regarded as a joint decision (Moffitt 1983; Hoynes 1996; Keane and Moffitt 1998; Brewer et al. 2006; Blundell and Shephard 2012). For example, redesigning the welfare system with the aim of motivating households to work longer hours will only be successful if the benefit is actually claimed. A minimum requirement for benefit take-up is that households need to be aware of the existence of the respective benefits. This latter insight was, among others, one rationale for the simplification of the transfer system in the context of the introduction of Universal Credit in the UK (Department for Work and Pensions 2010), and is mostly neglected in the German discussion so far.

Second, focussing on all three programs simultaneously, including the overlap between the programs, allows us to contrast entitlements to the benefits with the observed benefit take-up. For example, a household which would maximize its simulated net income by claiming a particular benefit might be observed to actually claim a benefit which leads to a lower simulated net income. Furthermore, we show the combined work incentive structure of the total tax-benefit system, resulting from the interaction of the income tax, social security contributions and the three benefits. The total marginal tax rate often differs substantially from the benefit reduction rate of one program considered in isolation (Blundell 2012).

Our results point to several weaknesses of the existing system. The simulation of entitlements shows that all programs are characterized by high rates of non-take-up. We also find a significant overlap between the programs. Therefore, the effectiveness of the programs in reaching their target groups could be improved in terms of targeting and program take-up. The analysis of stylized budget constraints shows that low-income households are confronted with a complex benefit structure and high marginal tax rates close to 100% in a wide income range. Finally, because of the strong interdependencies, even minor reforms of one program can have significant effects on the other programs. These results indicate that the benefits need to be substantially better coordinated or even harmonized.

Institutional Background

Social Assistance, Housing Allowance and Supplementary Child Allowance

The relevant institutions for our analysis are the social assistance for employable persons (SA), the housing allowance (HA) and the supplementary child allowance (SCA). Those benefits are the three most important programs for families in which at least one person is capable to work. In what follows, we provide a short description of the three programs. Table 1 summarizes the programs' key features.

Social assistance for employable persons (SGB II) is targeted at employable persons younger than 65 years of age. With more than 6 million recipients in 2014, SA is by far the most important benefit system in Germany. SA includes benefits for living and housing costs. The benefit, which should allow recipients to maintain a legally defined standard of living, is set by the government and is adjusted annually on a legal basis based on the development of prices and wages in the economy. Eligibility for SA is given if the household's total needs exceed the allowable income and the household's wealth remains below the household-specific maximum. The income of the household members is deducted from the total potential benefits, where nearly all types of income are considered in the means test. A proportion of the earned income from dependent employment and selfemployment is exempted when calculating the amount of benefit.

Housing allowance (Wohngeld) is a means-tested program designed to subsidize the housing costs of low-income households. In principle, all households in need can receive HA independent of their labor market status, but HA should not guarantee a minimum income (which is the purpose of SA). In addition, recipients must have a legally defined minimum income to be eligible for HA. Thus, if an HA applicant's income is below this minimum, the HA administration will request that she applies for SA instead. The HA entitlement depends on the household income relevant to the means test, the relevant housing costs and the household size. The benefit reduction rates range between approximately 30 and 40% and are therefore considerably lower than for SA, where the rates vary between 80 and 100%; for gross incomes above 100 per month (see Table 1).

Since 2005, a supplementary child allowance (Kinderzuschlag) has been available for working parents. The SCA is targeted at working parents whose earned income would be sufficient to cover their own needs according to SA, but not the needs of their children. Thus, the goal of the SCA is to provide working parents with a less stigmatizing alternative to receiving SA. (2) More precisely, parents are eligible for SCA if they have income that is sufficient to cover their own living and (partial) housing needs but not the needs of the children. The housing and living costs as well as the household's allowable income and wealth are defined according to the SA rules. HA is not considered in the means test because recipients could receive SCA and HA simultaneously. Because SCA is only an additional transfer to the household income, recipients must have a minimum income. (3) From the design and goals of the SCA, it follows that it is available only in a small income range. If the household income relevant for the means test exceeds the parent's guaranteed income (including partial housing costs) according to SA, the entitlements are reduced by a constant benefit reduction rate of 50% for each additional Euro of relevant income. Additionally, households lose their entitlement completely if their relevant income is above the household's maximum income, which is given by the parent's guaranteed income according to SA plus 140 multiplied by the number of SCA-eligible children in the household. (4)

The comparison of the key features of the three programs shows that the programs differ particularly in the level of entitlements. Households with no own income will typically only be entitled to SA. However, because the three programs are not coordinated and treat earned income differently, complex entitlement structures can occur for low-income workers. The latter holds although several rules are in place to prevent households from arbitrarily combining the benefits:

1. Households can be simultaneously entitled to SA and HA, but they can only claim one of these two benefits.

2. Households can be simultaneously entitled to HA and SCA and claim both of these benefits.

3. Households can never be simultaneously entitled to both, SA and SCA. Therefore, they can claim at most one of these benefits.

4. Households are not entitled to SA, if the pre-benefit income plus its potential HA and possibly SCA entitlement exceed the guaranteed income according to SA. Therefore, they can only claim HA and/or SCA.

The first rule means that, in a certain income range, potential claimants are allowed to choose between HA and SA. If a household chooses HA, its net income will always be lower than when choosing SA, because otherwise the fourth rule would apply, which would eliminate the entitlement to SA. Despite forgoing net income, the household's net utility from claiming HA may be higher than when claiming SA if the (non-pecuniary) costs of claiming SA are considerably higher than the respective costs of claiming HA. (5)

Furthermore, a choice between SA and SCA can never occur due to the third rule because eligibility to SA and SCA is mutually exclusive. This is a result of the intentionally strict targeting of SCA. A household is only entitled to SCA if claiming the benefit averts the household's dependency on SA. Thus, applying for SCA requires an exact means-testing of SA eligibility.

Finally, the third and fourth rule imply that whenever a household applies for SA the administrating agency must at least roughly estimate whether the household's guaranteed income can be covered with the prioritized benefits HA and SCA. Conversely, if a household applies for HA, the case worker is also required to roughly estimate an SA entitlement. If this provisional means test results in a likely entitlement to HA as well as to SA the applicant is typically informed that her net income will be higher when claiming SA.

The administrative organization of the three benefits as well as the division of the financing is divided among different state authorities. The organization and financing of SCA is straightforward in the sense that the federal ministry of family affairs is the only financing institution and is also the supervising authority. On the local level, the administration is organized by the Federal Employment Agency by so-called family offices.

The financing of HA is divided between the Federal Ministry for Environment, Nature Conservation, Building and Nuclear Safety (50%) and the 16 German states (50%). However, the municipal housing offices, which carry out the application procedure, are institutions of a third level of government, the German administrative districts. The 401 German administrative or urban districts are also partially responsible for the financing of the SA benefit, more precisely for the housing costs of the SA recipients. (6) This division of the responsibilities implies that local authorities can avoid financing the housing costs included in SA. They can achieve this by not promoting the take-up of SA for applicants of HA who are able to choose between the two benefits.

The Housing Allowance Reform 2016

As an illustrative example for the interdependency among SA, HA and SCA, we analyze the effects of a recent HA reform in Germany, which came into force in January 2016. The reform's objective is to adjust the benefit to recent housing price developments. This is supposed to lead to an increase the entitlements of existing recipients and to increase the overall number of recipients. New recipients who were not eligible for HA or SA were expected to enter the HA system. Additionally, a small number of SA recipients were expected to shift from SA to the HA program after the reform (Henger 2015; BMUB 2015; Bruckmeier and Wiemers 2015). (7) The HA reform 2016 consists of the following three core elements:

1. Reform of the entitlement calculation formula. The rules regarding how the entitlement is calculated, depending on household size, household income and rent/housing costs to the legally defined maximum have been changed to increase the resulting benefit.

2. Increase in the maximum subsidized rent. Depending on the regional housing price level, the maximum rent considered in the means test has been increased from 7 to 27%.

3. Changes in lump sum deductions from own income. Lump sum deductions for single parents, disabled recipients and employed children from the household income considered in the means test have been increased to between 750 and 1500 per year. The lump sum deduction of 6% of the gross income for recipients who are not obliged to pay income taxes or social security contributions has been abolished.

Methodology

We employ the Tax-Transfer Microsimulation Model of the Institute for Employment Research (IAB) of the German Federal Employment Agency (IAB-MSM) to simulate welfare entitlements. The IAB-MSM is based on the Steuer-Transfer-Mikrosimulations-modell (STSM) of the Centre for European Economic Research (ZEW). (8)

The model is a static microsimulation model that includes a detailed implementation of the German tax and transfer system. It is mainly used for the ex-ante evaluation of social policy reforms directed at low-income households in Germany. The model's validity with regard to official statistics, model assumptions and data selection has been verified in several studies (Arntz et al. 2007; Bios et al. 2007; Wiemers and Bruckmeier 2009; Bruckmeier and Wiemers 2012). The principal task of the IAB-MSM tax and transfer module is the computation of the household net income under varying tax and transfer rules. In this paper, we use the legal status as of January 2016 for the status quo simulation. We use the gross incomes of the household, e.g., labor and capital incomes, as they can be found in the underlying data. All deductions from gross income and public transfers are simulated on the basis of the simulation model. Table 2 describes the incomes, taxes and other income deductions considered in the computation of the net household income. First, gross incomes and pensions as observed in the data enter the model as exogenous inputs (see stage (1) in Table 2). Based on these incomes and household characteristics, deductions from gross incomes are calculated endogenously according to the relevant regulations (stage 2). In stage 3, benefits that are not means-tested are computed based on the results of the previous stages. In the final stage, means-tested benefits are simulated. Finally, incomes and deductions from all stages are added to simulate total net household income.

Figure 1 shows the IAB-MSM's entitlement calculations for the four nationwide means-tested benefits: (1) social assistance for older and not employable persons (SGB XII), (2) SA for employable persons between 15 and 64 years (SGB II), (3) housing allowance (HA) and (4) the supplementary child allowance (SCA).

To determine eligibility for SA, a person first must be classified as employable. The legal definition of employability is vague. (9) Thus, employability with regard to the SA cannot be precisely determined using information from the data. In the model, we categorize a person as employable if he or she is aged between 15 and 64, does not work in a sheltered workshop and either has a degree of disability smaller than 80% or receives earned income. If a household is categorized as employable and passes the eligibility check for SA, the model compares the SA claim to a possible HA plus SCA claim. In a typical simulation, the model assumes that the household will take up the higher benefit, i.e., either SA or HA plus SCA. For this paper, however, we drop this assumption, because we need to retain the potential entitlements to all considered benefits to highlight the overlap, i.e., households that are simultaneously entitled to both, SA and HA.

A detailed description of the calculation of a household's needs and income and, hence, the households's entitlements in the IAB-MSM is provided by Bruckmeier and Wiemers (2011). An important feature of the IAB-MSM is the possibility to consider the take-up behavior of potential SA claimants in a policy simulation. The method is described in Wiemers (2015). While this feature is especially important for the caseload forecasts of reforms, it is not suitable for our analysis, because we explicitly want to show how many households are in an income range that entitles them to one of the benefits, independent from their claiming decision. Therefore, we do not utilize this feature in our analysis.

The IAB-MSM is based on the Geman Socio-Economic Panel (GSOEP), a representative yearly household panel study in Germany. (10) To simulate SA entitlements, information on several socio-demographic characteristics of the household members and on the household incomes are necessary, which are usually provided only in survey data such as the GSOEP in Germany. The GSOEP includes the required demographic variables, information on the incomes of persons and households (e.g., earned income, pensions, and capital income) as well as information on current and past worked hours. Because the tax-transfer module of the IAB-MSM also employs retrospective information (collected in wave t + 1) to compute the net household income, we require two consecutive waves of the GSOEP to run the model. For this paper, we employ the GSOEP waves 2013 and 2014. After sample selection, approximately 12,000 households or 20,000 individuals aged 17 and older remain for use with the IAB-MSM. The most important reasons for excluding households from the simulation sample are missing interviews of partners in couple households (approximately 1700 households) and missing interviews for the household in wave t+ 1, which represent approximately 3000 additionally excluded households. We adjust the weights supplied with the GSOEP to account for the excluded households.

Results

Benefit Take-Up

Table 3 shows the results of the benefit simulation and the benefit take-up observed in the GSOEP data for 2013. The first row shows the non-take-up rates for the individual benefits. The non-take-up rate is defined as the share of households with simulated entitlements that do not claim their entitlements according to the data, relative to all simulated eligible households. For the basic SA benefit, we find a non-take-up rate of 43.1%, which is close to the findings of other studies based on different data (Becker 2012; Bruckmeier and Wiemers 2012).

In comparison, the non-take-up rates of the HA and the SCA are distinctly higher and amount to more than 80%. A higher non-take-up rate for these two benefits is plausible because, in contrast to SA, these benefits do not cover the total minimum income, but instead provide only additional benefits for specific needs (housing costs and children). This finding means that households eligible for these benefits have additional other income resources and thus, cet. par., a lower propensity to claim their entitlements.

Furthermore, the table shows the overlap between SA and HA. There are approximately 1.1 million (weighted) households in an income range that allows them to choose between SA and HA. This number is almost 80% of all households that are only eligible for the HA (1.35 million). Interestingly, approximately 63% of those who are either entitled to SA or to the HA fail to claim their entitlement to at least one of these two benefits. Most of the take-up households that actually claim one of either SA or HA report to have received SA (81%), which is shown in the last two rows. This means that, although it can be assumed that SA as the last safety net is more stigmatizing and exerts higher pressure on recipients to exit benefit receipt than HA, our data reveals a clear preference toward SA.

Table 3 also shows which benefits the non-take-up households actually claim. Most households that do not take-up their entitlements, do not claim other benefits. Only approximately 7.9% of SA non-take-up households receive HA. Reversely, approximately 9.1% of HA non-take-up households receive SA benefits. Again, both figures indicate the overlap of both programs. Concerning targeting effectiveness, it appears that SCA is the most ineffective benefit in that it does not reach its target group. On the one hand, nearly 90% do not claim their entitlements. On the other hand, 15.9% of SCA non-take-up households receive HA and 14.3% receive SA. This ineffectiveness is arguably a result of the small income range in which the SCA is available, as discussed in "Social Assistance, Housing Allowance and Supplementary Child Allowance" section.

The mean simulated entitlement of SA for those non-take-up households that are entitled to SA only amounts to 416 per month. For households which can choose between SA and HA, the mean simulated entitlement to SA amounts to 269 per month, while the corresponding entitlement to HA is 100 per month. Average simulated entitlements of non-take-up households are always below the corresponding entitlements of take-up households.

Overall, the results on the entitlements for take-up and non-take-up households indicate a positive relationship between the entitlements and the take-up. This positive relationship could be explained by the fact that the take-up probability strongly depends on the benefit level, which is a robust finding in the literature on modeling benefit take-up (Whelan 2010; Riphahn 2001; Blundell et al. 1988).

Budget Constraints and Marginal Tax Rates

The IAB-MSM allows us to calculate the stylized budget constraints encountered by typical low-income households. We focus on a single parent household with two dependent children and a couple with two dependent children, since only households with children can be eligible for the SCA. Hence, the interaction among SA, HA and SCA is relevant for households with children only. (11) The calculation of the household's net income builds on the following assumptions: (1)

In each household, only one adult is working in dependent employment and the household has wage income only, (2) the two children in the household are younger than 18 years, and (3) housing costs are approximately 500 for the single parent and 580 for the couple. Furthermore, (4) the budget constraints are constructed around an hourly wage rate of approximately 11.50, (5) the couple is married and (6) the household claims all benefits it is entitled to. Figure 2 first shows the net income corresponding to the gross wage of the single parent household. The two lines show the total net income at various gross wages. The dashed line assumes that the household takes up SA in the income range it can choose between SA and HA, and the solid line assumes that it takes up HA. The shaded areas show the income components of the two alternative net incomes at various gross wages.

For example, without own earned income, the household can only receive the legally defined minimum income of approximately 1585 from SA benefits and the regular child benefit. (12) The household is eligible for SA benefits up to a gross wage of 1500. This corresponds to working time of 30 h per week, which is typical for single parents, who work mostly part time. However, between the wide gross wage range of 1150 and 1500 the household can actually choose between SA or HA (HA alternative in Fig. 2). As explained in "Social Assistance, Housing Allowance and Supplementary Child Allowance" section, the household can choose only between the benefits if HA and pre-benefit income are lower than the SA entitlement. Thus, in the complete income range between 1100 and 1500, the household would always be better off with SA, where the difference between the entitlements decreases. After 1500 monthly gross earnings, the household can only take up the prioritized benefits HA and SCA and must leave SA. Figure 2 further shows that SCA still approximately equals 200 when HA expires at the 1850 gross wage.

In the small income range of 1850 and 2050, the household would be entitled to SCA only. When the SCA expires, the benefit decreases from approximately 200 to 0. The sharp transition is due to the entitlement rules of SCA, which determine that the household is eligible only up to a maximum income. The household would be dependent on SA again from a gross wage starting at 2050and finally exits the SA system at a gross wage of 2200. For higher wages, the increase in net income is determined by the income tax and the social security contributions.

The combined marginal tax rate resulting from the income tax, social security contributions and the benefit reduction rates of SA, HA and SCA corresponding to Fig. 2 is shown in Fig. 3. The marginal tax rate is based on the assumption that the household maximizes its total income and always chooses SA instead of HA and SCA if possible. When the household receives SA only, the impact on the net income is straightforward, as the benefit reduction rates depend on gross income only and are constant over certain income ranges. Only gross incomes up to a maximum of 100 do not reduce the SA entitlement. For higher gross incomes, the benefit reduction rates range from 80 to 90% until the household exits SA at the gross wage of approximately 150013, which corresponds to 30 h of work per week at the assumed gross hourly wage. The figure suggests that SA provides incentives for low-wage incomes to a maximum of 100.

For household incomes above the SA threshold, the marginal tax rate function becomes complicated due to the interaction of HA and SCA and the nonlinear calculation of the earnings exemptions in the HA program. Figure 2 shows that returns to work do not increase immediately after exiting the SA entitlement. Instead, in the wide gross wage range between 1500 and 2200 the household's net income remains nearly constant at 1550. Hence, there are nearly no monetary incentives to increase the labor supply in a wide income range. Figure 3 additionally reveals that the marginal tax rate increases above 100% for a small income range. This finding is a result of the SCA regulations, which imply that the SCA entitlement decreases to zero when the household income exceeds a maximum allowable income, which varies with the household characteristics. At an income of approximately 2200 and above the example single parent household loses all means-tested benefit entitlements, which results in a considerable decrease in the marginal tax rate.

The budget constraint and income components for a couple with two dependent children are shown in Fig. 4 and the respective resulting marginal tax rate is shown in Fig. 5. The legally defined minimum income without the regular child benefit is 1825. Again, the household could choose between SA or HA and SCA in a wide income range between 1250 and 1750. After a gross wage of 1750, HA and SCA exceed SA and the household exits SA, after a gross wage of 2400 (2200), HA (SCA) expires and the household enters the regular tax system. The income loss when the SCA expires also holds for the example couple household.

In general, the findings for the example single parent also apply for the couple with two children. Because the legally defined minimum income is higher for this example, the income range in which the household has the option to choose between entitlements is wider than for the single parent. Additionally, the household benefits more from HA because the housing costs are higher than for the example single parent household, as the benefit level of HA depends on real housing costs. However, the higher the number of children is, the longer the household remains within the benefit system and is confronted with high marginal tax rates from the combined receipt of HA and SCA and income losses when the SCA expires after the gross wage exceeds the corresponding maximum income.

Housing Allowance Reform Effects

Budget Constraints

The budget constraint for the example single parent household with two dependent children after the implementation of the HA reform is shown in Fig. 6. The comparison with the respective budget constraint before the reform (see Fig. 2) shows that the reform has considerable effects on the income and income components of our example household.

The income range in which the household can choose between different entitlements is reduced significantly to a range between 1000 and 1150. This means that the gross wage threshold at which the household can choose between SA and HA is slightly reduced from 1150 before the reform to 1000 after the reform. Thus, the household is able to leave SA earlier after the reform. At a monthly gross wage of 1150, the household is entitled to a HA and the SCA only. This implies that the example household is able to leave the SA system already at 23 working hours per weak, instead of 30 h before the reform. For gross wages starting at approximately 1150, the household is always better off with a combination of HA and SCA. Interestingly, the reform does not affect the income threshold at which the household returns to the regular tax system. Due to the different means test rules of the various benefit programs, the household becomes eligible for SA at a gross wage of approximately 2100 and exits the SA system at a gross wage of 2150, thus, the income thresholds are approximately the same as before the reform.

The purpose of the reform was to increase entitlements, and indeed, the net income of the household increases in a certain range. In the example, the household is better off after the reform when the single parent earns a gross wage between 1200 and 2100, which is the income range in which the household receives the reformed HA combined with SCA instead of SA. At a gross wage of 1850 the net income difference achieves its maximum of approximately 100 after the reform. For gross wages higher than 1850, the income gains through the reform decrease; in addition, at a gross wage of 2100, the household has the same net income as before the reform.

Figure 7 shows the distribution and components of the household net income for the example household of a couple with two dependent children after the reform, which reveals, in comparison with Fig. 4, similar results as for the single parent household. Again, the household can leave SA earlier. At a gross wage of 1150 the household can choose HA, although SA still remains higher at this point. The direct overlap between SA and HA is also slightly reduced to a range of 1150 to 1550. The marginal tax rates are almost unaffected by the reform, as the benefit reduction rate of SA is close to the benefit reduction rate resulting from the interaction of HA and SCA. Income gains achieve a maximum of approxiamtely 65 at a gross wage of 1850 and are lower than the income gains of the reform for single parents, as the latter benefit from the increase in the lump sum deductions from the wage income for single parents.

Fiscal Effects and Caseloads

Table 4 presents the projected effects of the HA reform on the number of recipients receiving a particular benefit and the fiscal costs of the benefit programs. The costs are derived by comparing the projected annual costs of the benefit programs in the status quo (2016) without reformed HA with the scenario of the implementation of the HA reform 2016. The effects are morning after reform effects; thus, behavioral adjustments are not considered. Note that we assume a take-up rate of 100% for the SA, HA and SCA, as explained in "Methodology" section, which results in an overestimation of households with low entitlements. Hence, the figures presented in this section should not be interpreted as forecasts. Instead, the figures provide an illustration of the distribution of households in income positions close to entitlements based on the underlying income distribution and the overlap between the benefit systems.

As expected, the reform's most important single effect is on HA caseloads. According to the IAB-MSM, approximately 900,000 more households or 1.7 million individuals would be entitled to claim HA. Given the high non-take-up of HA, we assume that most of these households at the margin of entitlements would actually not claim HA. Nevertheless, the potential increase in recipients would be associated with additional expenditures of 2.2 billion. Given that only approximately 650,000 households receive HA in Germany, this result demonstrates that far more households are in income positions that would entitle them to HA, at least temporarily or with small entitlements. The HA reform hypothetically would also have large effects on SA and SCA; approximately 274,000 households with 619,000 individuals would be entitled to HA instead of SA after the reform. Due to increased HA entitlements, more households would be able to leave SA in combination with SCA. Hence, the number of SCA recipients would also increase. Our simulation reveals an increase of 126,000 SCA households with 361,000 individuals living in these households. While SA expenditure would decrease by 762,000, SCA expenditures would increase by 301,000. Taken together, although the HA reform in 2016 is only marginal in that it attempts to adjusts the program to recent housing price developments, the simulation reveals that due to the interactions of the programs, SA and SCA would also be significantly affected by the reform.

Conclusion

This paper provides empirical evidence on the effectiveness of interdependent means-tested benefit programs in providing income support and encouraging recipients to take-up employment. We focus on the three major means-tested programs available for low-income households in Germany: Social assistance (SA), housing allowance (HA) and supplementary child allowance (SCA). The institutional set up of these benefits is exemplary for many advanced welfare states.

Our simulation of entitlements based on a static microsimulation model (IABMSM) points to several weaknesses of the existing system. All programs, particularly HA and SCA taken together, are characterized by high rates of non-take-up. Hence, the goals of these programs--securing a certain standard of living and adequate housing for low-income families--are partially not fulfilled. Therefore, the effectiveness of the programs in reaching their target groups could be improved. The analysis of stylized budget constraints for example households shows that low-income households are confronted with a complex benefit structure and high marginal tax rates close to 100% in a wide income range, which negatively effects labor supply. Finally, because of the strong interdependencies between the three considered programs, even minor reforms of one program can have significant effects on the other programs, which is indicated by our results on a recent housing allowance reform in Germany.

Our findings are characteristic for many modern welfare states with a complex benefit structure, organizational fragmentation and a division of political responsibilities. High rates of non-take, the overlap between benefits and entitlements, as well as uncoordinated high benefit reduction rates indicate a lack of transparency and cooperation between the relevant institutions. For modern welfare states, it is therefore a challenge to reform complex systems to function more effectively or even efficient. One starting point is the improved coordination of services and the partial harmonization of different programs, especially if they have similar objectives and target groups.

Acknowledgements We thank the editor and two anonymous referees for helpful comments.

References

Arntz, M., M. Clauss, M. Kraus, R. Schnabel, A. Spermann. and J. Wiemers. 2007. Arbeitsangebotseffekte und Verteilungswirkungen der Hartz-IV-Reform. IAB Forschungsbericht 10/2007.

Becker, I. 2012. Finanzielle Mindestsicherung und Bedurftigkeit im Alter. Zeitschrift fur Sozialreform 58: 123-148.

Bios, K, M. Feil, H. Rudolph, U. Walwei, and J. Wiemers. 2007. Forderung Existenz sichernder Beschaftigung im Niedriglohnbereich * Schatzung von Angebots-, Verteilungs- und fiskalischen Effekten des SMWA-Vorschlags. IAB Forschungsbericht 07/2007.

Blundell, R. 2012. Tax Policy Reform: The Role of Empirical Evidence. Journal of the European Economic Association 10: 43-77.

Blundell, R., and A. Shephard. 2012. Employment, Hours of Work and the Optimal Taxation of Low Income Families. The Review of Economic Studies 2: 481-510.

Blundell, R., V. Fry, and I. Walker. 1988. Modelling the Take-up of Means-Tested Benefits: The Case of Housing Benefits in the United Kingdom. Economic Journal 98: 58-74.

BMFSFJ. 2009. Evaluation des Kinderzuschlags. Ergebnisbericht. Bundesministerium fur Familie, Senioren, Frauen und Jugend.

BMUB. 2015. Hendricks legt Wohngeld- und Mietenbericht vor. Bundesministerium fur Umwelt, Naturschutz, Bau und Reaktorsicherheit. Pressemitteilung Nr. 272/15, Berlin, 28 Oct 2015.

Brewer, M., A. Duncana, A. Shepharda, and M.J. Suarez. 2006. Did Working Families' Tax Credit Work? The Impact of In-work Support on Labor Supply in Great Britain. Labour Economics 13: 699-720.

Bruckmeier, K, and J. Wiemers. 2011. A New Targeting--A New Take-Up? Non-Take-Up of Social Assistance in Germany after Social Policy Reforms. IAB Discussion Paper No. 10/2011, Institute for Employment Research, Nuremberg.

Bruckmeier, K., and J. Wiemers. 2012. A New Targeting--A New Take-up? Non-take-up of Social Assistance in Germany After Social Policy Reforms. Empirical Economics 43 (2): 565-580.

Bruckmeier, K, and J. Wiemers. 2015. Wohngeldreform 2016: Auswirkungen auf Grundsicherungsbezieher. Soziale Sicherheit 64: 442-445.

Bundesgesetzblatt. 2015. Jahrgang 2015 Teil 1, Nr. 38 vom 8. Oktober 2015, S. 1610-1663.

Cutrie, J. 2004. The Take Up Of Social Benefits. National Bureau of Economic Research Working Paner 10488. F

Department for Work and Pensions. 2010, March. Universal Credit: Welfare That Works DWP Policy Paper, 11 Nov 2010.

Eurofound. 2015. Access to Social Benefits: Reducing Non-Take-up. Luxembourg: Publications Office of the European Union.

Friedberg, L. 2000. The Labor Supply Effects of the Social Security Earnings Test. Review of Economics and Statistics 82: 48-63.

Haisken-DeNew, J.P., and J.R. Frick. 2005. DTC Desktop Companion to the German Socio-Economic Panel (SOEP). Version 8.0, DIW Berlin.

Henger, R. 2015. Reform des Wohngeldes--Stellungnahme zur offentlichen Anhorung zum Entwurf eines Gesetzes zur Reform des Wohngeldrechts und zur Anderung des Wohnraumforderungsgesetzes. Koln, 5 June 2015

Hoynes, H.W. 1996. Welfare Transfers in Two-Parent Families: Labor Supply and Welfare Participation Under AFDC-up. Econometrica 64: 295-332.

Jacobebbinghaus, P., and V. Steiner. 2003. Dokumentation des Steuer-Transfer-Mikrosimulationsmodells STSM--Version 1995-1999. Dokumentation Nr. 03-06, Zentrum fur Europaische Wirtschafts forschung.

Keane, M., and R. Moffitt. 1998. A Structural Model of Multiple Welfare Program Participation and Labor Supply. International Economic Review 39 (3): 553-589.

Knabe, A. 2006. Warum Zuverdienstregeln und Kinderzuschlag negative Arbeitsanreize setzen ifo Dresden berichtet 2/2006.

Lemieux, T., and K. Milligan. 2008. Incentive Effects of Social Assistance: A Regression Discontinuity Approach. Journal of Econometrics 142: 807-828.

Meister. W. 2009. Neuer Kinderzuschlag, Wohngeldreform, hohere Hartz-IV-Regelsatze: Insbesondere fur Familien deutliche Einkommenssteigerungen, ifo Schnelldienst 62: 19-31.

Meyer, B.D. 2002. Labor Supply at the Extensive and Intensive Margins: The EITC, Welfare and Hours Worked. American Economic Review 92: 373-379.

Moffitt. R. 1983. An Economic Model of Welfare Stigma. American Economic Review 73 ("5V 1023-1035.

Moffitt, R.A. 2002. Welfare programs and labor supply. In Handbook of Public Economics, ed Auerbach A.J., and M. Feldstein., (1st edn.), vol. 4, Chapter 34, 2393-2430. Amsterdam: Elsevier

Riphahn, R.T. 2001. Rational Poverty of Poor Rationality? The Take-Up of Social Assistance Benefits Review of Income and Wealth 47 (3): 379-398.

Tasseva, I.V. 2016. Evaluating the Performance of Means-Tested Benefits in Bulgaria. Journal of Comparative Economics 44 (4): 919-935.

Voigtlander, M., T. Clamor, R. Henger, and J. Niehues. 2013. Bestandsaufnahme und Wirkungsanalyse des Wohngeldes. Bundesinstitut fiir Bau-, Stadt- und Raumforschung (BBSR) im Bundesamt fur Bauwesen und Raumordnung (BBR) (Hrsg.).

Wagner, G.G., J.R. Frick, and J. Schupp. 2007. The German Socio-Economic Panel Study (SOEP)-Scope, Evolution and Enhancement. Journal of Applied Social Studies 127 (1): 139-169.

Whelan. S. 2010. The Take-Up of Means-Tested Income Support. Empirical Economics 39 (3): 847-875

Wiemers, J. 2015. Endogenizing Take-Up of Social Assistance in a Microsimulation Model: A Case Study for Germany. The International Journal of Microsimulation 8 (2): 4-27.

Wiemers, J., and K. Bruckmeier. 2009. Forecasting Behavioural and Distributional Effects of the Bofinger-Walwei Model Using Microsimulation. Jahrbucher fur Nationalokonomie und Statistik 229 (4): 492-511.

Winkel, R. 2015. Wohngeldreform 2016: Die wesentlichen (Neu-)Regelungen beim Wohngeld Zeitschrift fur Soziale Sicherheit (12/2015), 433-438.

(1) Regarding the targeting effectiveness of means-tested benefit programs, two forms of inefficiencies can be distinguished: Inclusion of non-entitled households ("leakage") and exclusion of entitled households ("non-take-up"), see Tasseva (2016). We restrict our analysis to non-take-up, because of its obvious negative effect on poverty alleviation.

(2) Additionally, according to the official definition of child poverty, all children living in a household that receives SA are considered poor, while they are not considered poor if their needs are covered by other incomes than SA. Therefore, with the introduction of SCA the government tried to reduce the official measure of child poverty by moving households from receiving SA to receiving SCA.

(3) Until the end of 2005 the minimum income exactly coincided with the parents' guaranteed income as provided by SA. Since many parents were unable to calculate their guaranteed income, a large share of the applications for SCA was rejected because the parents' income was below the required minimum income. As a result, the minimum income rules of SCA were simplified at the end of 2005. Since then, the minimum income for SCA is either a gross income of 600 per month for single parents or 900 per month for couples with children. Both minimum incomes are roughly in line with the parents' guaranteed SA income.

(4) SCA was raised to 160 per month and child on 1. July 2016 and further increased to 170 in 2017.

(5) There is reason to believe that for some households the running (non-pecuniary) costs of receiving SA might be considerably higher than the corresponding costs of receiving HA. The Job Centers responsible for managing SA are constantly putting pressure on the recipients to find a job that, ideally, pays well enough to enable the households to overcome their dependency on SA. Households who do not cooperate sufficiently in the search for a job can be sanctioned by the Job Centers. In contrast, when claiming HA, the administrating municipal housing offices do not exert any pressure on recipients to leave the HA system. HA recipients are only required to renew their application to HA once a year (or whenever their income situation has changed). Additionally, it can be assumed that SA as the last safety net is more stigmatizing than HA.

(6) The administrative districts receive an annual grant from the federal budget for their expenditures for the housing costs of SA recipients. The grant is calculated as a fixed proportion of total expenditure.

(7) For a detailed description of the reform elements, also see Bundesgesetzblatt (2015) and Winkel (2015).

(8) For a documentation of the STSM see Jacobebbinghaus and Steiner (2003).

(9) The legal definition given in [section] 8(1) SGB II loosely states that a person is employable if illness or disability does not prevent her from working at least 3 h a day under the regular conditions of the labor market for the foreseeable future. In practice, employability is determined by public health officers.

(10) See Haisken-DeNew and Frick (2005) and Wagner et al. (2007) for documentation on the GSOEP.

(11) A calculation for single households and couples without children shows that for these households SA always exceeds HA and that the household would always be better off with SA.

(12) Notice that the regular child benefit is not a means-tested benefit and therefore constant for all gross incomes.

(13) The benefit reduction rate is 100% for incomes above 1500 for families with children. This is not shown in Fig. 3, as the example single parent household loses its SA eligibility before achieving that income threshold.

https://doi.org/10.1057/s41294-017-0041-5

Kerstin Bruckmeier [1] * Jurgen Wiemers [1]

Published online: 21 December 2017

[mail] Kerstin Bruckmeier

Kerstin.Bruckmeier@iab.de

Jurgen Wiemers

Juergen.Wiemers@iab.de

[1] Institute for Employment Research (IAB) of the German Federal Employment Agency (BA), Regensburger Strasse 104, 90478 Nuremberg, Germany

Caption: Fig. 1 Simulation of welfare entitlements in the IAB-MSM. HH household, SA social assistance for employable (SGB II) or unemployable (SGB XII) persons, HA housing allowance, SCA supplementary child allowance. Source: Bruckmeier and Wiemers (2011)

Caption: Fig. 2 Budget constraint and income components for a single parent with two dependent children by monthly earned gross wage in 2016 without HA reform. Source: Authors' own presentation based on IAB-MSM

Caption: Fig. 3 Marginal tax rate for a single parent with two dependent children by monthly earned gross wage in 2016 without HA reform. Source: Authors' own presentation based on IAB-MSM

Caption: Fig. 4 Budget constraint and income components for a couple with two dependent children by monthly earned gross wage in 2016 without HA reform. Source: Authors' own presentation based on IAB-MSM

Caption: Fig. 5 Marginal tax rate for a couple with two dependent children by monthly earned gross wage in 2016 without HA reform. Source: Authors' own presentation based on IAB-MSM

Caption: Fig. 6 Budget constraint and income components for a single parent with two dependent children by monthly earned gross wage in 2016 with HA reform. Source: Authors' own presentation based on IABMSM

Caption: Fig. 7 Budget constraint and income components for a couple with two dependent children by monthly earned gross wage in 2016 with HA reform. Source: Authors' own presentation based on IAB-MSM

Table 1 Key features of means-tested programs for low-income
households 2016. Source: Authors' own presentation

                       Social assistance (SGB II)

Target group           Households of employable
                       individuals

Legal definition of    Employable individuals, partner and
the 'benefit unit'     children up to the age of 25

Objective              Guaranteeing a minimum income

Calculation of total   (Nationally standardized regular
benefit                benefit + additional benefits due
                       to special needs + adequate housing
                       costs) - income and wealth of the
                       household

Benefit level          Head of the household: 404 [euro]
                       per month, spouse: 324 [euro],
                       children aged < 6 (14, 18, 25): 237
                       [euro] (270 [euro], 306 [euro], 324
                       [euro])

Housing allowance      Adequate housing costs up to a
level                  maximum, depending on housing size
                       and the local differentiated
                       housing price level + heating costs

Incomes exempted       Basic pensions for people in
in the means test      specific situations (Grundrente)

Marginal benefit       y<100 [euro] : 0.0, 100
reduction rate for     [euro]>y<1000 [euro] : 0.8, 1000
earned income          [euro] <y < 1200 [euro] (1500
(y)                    [euro] for recipients with
                       children): 0.9, 1200 [euro](1500
                       [euro]) <y : 1.0, where y is gross
                       monthly income

Maximum wealth         150 [euro] per year of life,
                       minimum: 3,100 [euro], maximum:
                       10,000 [euro], + 750 [euro] per
                       year of live, maximum: 50000 [euro]

                       Housing allowance

Target group           Low-income households

Legal definition of    Low-income households
the 'benefit unit'

Objective              Covering adequate housing costs

Calculation of total   Calculation based on a nonlinear
benefit                formula depending on adequate
                       housing costs, allowable income and
                       household size

Benefit level

Housing allowance      Adequate housing costs up to a
level                  legally defined maximum depending
                       on household size and the regional
                       or local housing price level

Incomes exempted
in the means test
                       Child benefit, child-raising
                       allowance (< 300 [euro])

Marginal benefit       varying rate (between approx. 0.3
reduction rate for     and 0.4 for a single person
earned income          household), depending on gross
(y)                    income, household size, rent amount
                       and local housing price level

Maximum wealth         60,000 [euro] head of the
                       household, 30,000 [euro] for each
                       other household member

                       Supplementary Child
                       Allowance
Target group           Low-income families

Legal definition of    Employable individuals, partner
the 'benefit unit'     and children up to the age of 25

Objective              Avoiding poverty, defined as
                       dependence on social
                       assistance benefits

Calculation of total   Nationally standardized regular
benefit                benefit--income and wealth of
                       the household

Benefit level          140 [euro] per month for each child
                       aged <25

Housing allowance
level

Incomes exempted
in the means test
                       Child benefit, housing allowance,
                       basic pensions maternity
                       benefit, child-raising allowance

Marginal benefit       0.5 on net income
reduction rate for
earned income
(y)

Maximum wealth         Social assistance regulations
                       apply

Table 2 Components of net household income in the IAB-MSM.
Source: Bruckmeier and Wiemers (2011)

Model    Income components                     Determined in tax and
stage                                          transfer module?

1        Earned income                         No
     +   Self-employed income                  No
     +   Capital income                        No
     +   Rental income                         No
     +   Other incomes (pensions)              No
2    -   Social security contributions         Yes
     -   Income tax                            Yes
     -   Alimony payments                      Yes
3    +   Child benefit                         Yes
     +   Child-raising allowance               Yes
     +   Unemployment benefits                 Yesa
     +   Federal student support, stipends,    No
         claims to maintenance, widow's
         allowance, maternity allowance,
         reduced hours compensation
4    +   Housing allowance                     Yes
     +   Supplementary child allowance         Yes
     +   Social assistance for employable      Yes
         persons (SGB II)
     +   Social assistance for unemployable    Yes
         persons (SGB XII)
     =   Net household income                  Yes

(a) Endogenous if labor supply reactions are considered.
Otherwise, we use reported unemployment benefits

Table 3 Individual entitlements and take-up of social assistance,
housing allowance and supplementary child allowance

                                  Entitled to
                                  Social         Housing
                                  assistance     allowance or
                                  (SA)           SA

Non-take-up rate (pet.)           43.1           62.8a
(95% confidence interval)         (38.8; 47.3)   (53.6; 72.1)
Mean simulated entitlement of     416            269 SA /100
  non-take-up hh ([euro]/mth.)                   HA
Percent of non-take-up hh
  receiving
  SA
  HA                              7.9
  SCA                             0.6            0.1
  None                            91.9           99.9
Take-up rate (pet.)               56.9           37.2
Mean simulated entitlement of     794            352 SA/ 119
  take-up hh ([euro]/mth.)                       HA
Percent of take-up hh receiving
  SA                                             80.9
  HA                                             19.1
Observations                      1444           272
Weight. Obs.                      3,838,517      1,057,694

                                  Housing        Supplementary child
                                  allowance      allowance (SCA)
                                  (HA)

Non-take-up rate (pet.)           86.6           88.2
(95% confidence interval)         (81.9; 91.3)   (83.4; 92.9)
Mean simulated entitlement of     85             230
  non-take-up hh ([euro]/mth.)
Percent of non-take-up hh
  receiving
  SA                              9.1            14.3
  HA                                             15.9
  SCA                             1.7
  None                            89.2           69.8
Take-up rate (pet.)               13.4           11.8
Mean simulated entitlement of     111            325
  take-up hh ([euro]/mth.)
Percent of take-up hh receiving
  SA
  HA
Observations                      488            261
Weight. Obs.                      1,353,009      361,780

All figures are based on weighted results
hh households

(a) Non take-up :: Households not claiming SA and HA

Table 4 Effects of the housing allowance reform 2016 on benefit
caseloads and expenditures. Source: IAB-MSM, GSOEP 2013-2014

                              Social assistance   Housing allowance

Households (in 1000)          - 273               901
Individuals (in 1000)         - 619               1711
Expenditures (in 1000 Euro)   - 762               2200

                              Supplementary
                              child allowance

Households (in 1000)          126
Individuals (in 1000)         361
Expenditures (in 1000 Euro)   301