IMF：欧洲最低收入保障计划：景观与设计【英文版】

WP/21/179 Guaranteed Minimum Income Schemes in Europe: Landscape and Design by David Coady, Samir Jahan, Riki Matsumoto, and Baoping Shang © 2021 International Monetary Fund WP/21/179 IMF Working Paper Fiscal Affairs Department Guaranteed Minimum Income Schemes in Europe: Landscape and Design Prepared by David Coady, Samir Jahan, Riki Matsumoto, and Baoping Shang July 2021 IMF Working Papers describe research in progress by the author(s) and are published to elicit comments and to encourage debate. The views expressed in IMF Working Papers are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management. Abstract This paper provides an overview of the design of means-tested Guaranteed Minimum Income schemes, which constitute an important component of social protection systems in European countries. It discusses how key design features differ across countries, including how countries balance the primary objective of poverty alleviation against the desire to both manage the work disincentives inherent in such programs and contain fiscal cost. The analysis finds a clear trade-off between both concerns in practice, with many countries combining low generosity with low benefit withdrawal rates (BWRs) thus prioritizing employment incentives over the primary objective of poverty alleviation. Many countries can reduce this trade off by combining higher generosity with higher BWRs. Countries with very high BWRs should consider reducing these, including through allowing income disregards and time dependent (rather than income-dependent) benefit withdrawal. The work disincentives associated with higher BWRs can also be attenuated through strengthening complementary activation policies that incentivize and support participation in the labor market. JEL Classification Numbers: H21, H53, I38 Keywords: means testing, poverty alleviation, benefit generosity, work disincentives. Author’s E-Mail Address: DCoady@imf.org, SJahan2@imf.org, RMatsumoto@imf.org, BShang@imf.org 3 Content Page Abstract ____________________________________________________________________________________________2 I. Introduction ______________________________________________________________________________________4 II Overview of Spending on Income Support _______________________________________________________5 III. Key Design Features_____________________________________________________________________________7 IV. GMI and Work Disincentives __________________________________________________________________ 10 V. Conclusions____________________________________________________________________________________ 17 References _______________________________________________________________________________________ 37 Figures 1. Spending on Income Support in European Countries, 2018 _____________________________________6 2. Means-tested Income Support in European Countries, 2018 ____________________________________6 3. Evolution of Means-tested Income Support Spending in European Countries, 1990–2018______7 4. Distribution of PTRs and METRs Across Countries by Household Type, 2019 _________________ 12 5. Relationship Between Generosity and PTRs, 2019 _____________________________________________ 15 6. Relationship Between Generosity and METRs, 2019 ___________________________________________ 15 Table 1. Variation in Key GMI Design Features in European Countries _________________________________ 19 Annexes 1. Estimation of PTRs and METRs_________________________________________________________________ 20 2. PTRs and MTRs in European GMI Schemes by Household Type _______________________________ 22 3. Activation Policies _____________________________________________________________________________ 25 4. Reform Experiences in the U.K. and Germany _________________________________________________ 34 Annex Figures A1.1 Comparison of PTR and METR Estimates, 2017 _____________________________________________ 21 A2.1. Marginal Effective Tax Rates by Household Type, 2019 ____________________________________ 22 A2.2. Participation Tax Rates by Household Type, 2019 __________________________________________ 23 A3.1. Strictness of Activation Requirements in European Countries, 2017 _______________________ 31 A3.2. Labor Disincentives and Strictness of Benefit Conditionality _______________________________ 31 Annex Tables A2.1. Key Features of GMI Schemes in Europe ___________________________________________________ 24 A3.1 Strictness of Activation Requirements and Components, 2017 _____________________________ 33 4 I. INTRODUCTION Most European countries have a comprehensive set of income support programs to protect households from income shocks and poverty. These include both social insurance and social assistance transfers. Social insurance programs, such as unemployment benefits, protect workers from loss of income due to unemployment, and eligibility is typically time-limited and requires individuals to have made sufficient contributions while employed. Social assistance (or income support) transfers are typically not linked to contributions but tend to be categorical, such as universal child benefits, or means tested, and aim to protect households from poverty or compensate for higher expenditures (e.g., child benefits). At the center of means-tested benefits are Guaranteed Minimum Income (GMI) schemes, which play the role of last-resort income support programs aimed at protecting working-age households from poverty. Such schemes become increasingly important in the context of economic shocks and have played an important role in in protecting households from the economic consequences of lockdowns in the context of the current COVID-19 pandemic.1 The primary objective of GMI schemes is to provide households with enough income to prevent them falling into poverty.2 Benefit levels are typically determined by comparing household “needs” with household “means”. Needs reflect both household size and composition, as well as other characteristics such as disability status and local housing costs. Means reflect income both from work and assets. The means-tested nature of GMI schemes, with benefits withdrawn as other household income increases, can result in strong work disincentives. While eligibility is typically not time limited at the outset, it may be conditioned on working adults being available and searching for work to help mitigate the work disincentives inherent in means-tested schemes. Indeed, many schemes embed periodic reassessment of eligibility which, in practice, could result in recipients becoming ineligible for GMI payments. This paper presents an overview of the design of European GMI schemes and is intended to provide a starting point for a country-level dialogue on potential reforms to enhance the performance of these schemes. Section II provides a brief overview of the level of overall public spending on income support programs in European countries, and its composition in terms of means-tested and non-means-tested benefits. Section III describes the broad landscape of GMI schemes and how key design features vary across countries. Section IV reviews the empirical evidence on the generosity of benefits and the strength of work disincentives, and discusses measures that countries use to mitigate the latter—including through reducing benefit generosity, tapering benefit withdrawal, providing employment incentives through in-work support, targeting income support to the most disadvantaged workers, making benefit 1 For a discussion of the performance of the U.K. GMI scheme in the context of COVID-19, see Brewer and Handscomb (2020). For an assessment of the measures taken by advanced economies to tackle pandemic-related unemployment and challenges posed for recovery (OECD, 2021). 2 Such transfer programs can also have important beneficial impacts in other dimensions such as health (Evans and Garthwaite, 2014; Hoynes et al. 2015; Averett and Wang, 2018), children’s test scores (Chetty et al., 2011; Dahl and Lochner, 2012); and long-term educational attainment (Bastian and Michelmore, 2018; Manoli and Turner, 2018). 5 withdrawal time (rather than income) dependent, and attaching conditions to receipt of benefits.3 Variation in key design parameters across countries can be suggestive of opportunities for welfare-improving reforms, although precise policy recommendations would require more detailed country-specific analysis to capture the nuances of their tax and transfer systems, social preferences, and political and labor market realities. Section V concludes. The analysis presented in the paper suggests some key takeaways. First, there would appear to be ample room in many European countries to enhance the generosity of benefits since low benefits undermine the primary objective of these schemes. Second, countries with relatively high benefit withdrawal rates can consider reducing these to strengthen work incentives, especially where poor employment outcomes for beneficiaries reflects weak demand for their labor rather than their strong preference for leisure. Third, incentives to work can be further strengthened through requiring beneficiaries to participate in effective job-search activities and supporting them in this activity. Fourth, where weak labor demand is a key factor in determining employment outcomes, strengthening active labor market policies focused on training for skills development is crucial. II. OVERVIEW OF SPENDING ON INCOME SUPPORT European countries allocate a large share of resources to income support. In 2018, average income support spending across 35 European countries stood at 3.4 percent of GDP (Figure 1). Spending varies significantly across countries, with 5 countries having spending above 5 percent, and 5 countries with spending below 2 percent. On average, the highest spending is on family benefits (at 1.86 percent), followed by unemployment benefits (0.81 percent), social exclusion benefits (0.50 percent) and housing benefits (0.25 percent). Around two-thirds of income support is not means tested (Figure 2); in 2018, average means-tested income support stood at 1.07 percent of GDP compared to 2.35 percent of GDP on non-means-tested benefits. But the relative importance of means-tested benefits also varies across countries, with this share being above 30 percent of total income support in 15 (out of 35) countries, and below 15 percent in 6 countries. Many high-spending countries increased their use of means testing over recent decades. Figure 3 shows that most countries with relatively high initial levels of income support spending (i.e., countries with relatively large bubbles) around the early 1990s increased their reliance on meanstested spending between over subsequent decades, presumably as a way of containing spending levels and maximizing the poverty impact of spending. In countries where income support spending data are available from 1990 (Denmark, Finland, France, Norway, Switzerland, Germany, and the U.K.), average spending increased sharply up to around 7 percent of GDP in 1995 before falling sharply to around 5–6 percent of GDP thereafter. Over the same period, the share of means-tested spending in these countries increased on average from around 26 percent in 1990 to around 37 percent in 2015. 3 In many countries, minimum wage policy is also a key policy lever for addressing low incomes. However, in this paper, we focus solely on design features within the benefits system. 6 Percent of GDP Figure 1. Spending on Income Support in European Countries, 2018 Unemployment Social exclusion Housing Family/Children 7 6 5 4 AAvevreargaeg,e,3.34.242 3 2 1 0 DK FI FR LU SE AT DE BE NO UK IS NL CY IT ES CH EE SI HR HU IE PL LT CZ BG LV RS SK PT RO BA ME MT TR Country Notes: 2018 is the latest year with broad coverage of countries reporting spending. Categories are ultimately shaped by countries’ specific benefit architecture and so some types of support may be categorized differently across countries. Sources: Eurostat (2021); Authors’ calculations. Figure 2. Means-tested Income Support in European Countries, 2018 Means-tested Non means-tested 7 6 5 4 3 AAveve. .nnoonnmmeeaannss--tteesstteedd,, 22..3355 2 AAvvee.. mmeeaannss--tteesstteedd, ,11.0.707 1 0 DK FI FR LU SE AT DE BE NO UK IS NL CY IT ES CH EE SI HR HU IE PL LT CZ LV RS BG SK PT RO BA ME MT TR Country Notes: Country total spending is the same as in Figure 1. Sources: Eurostat (2021); Authors’ calculations. Percent of GDP 7 Figure 3. Evolution of Means-tested Income Support Spending in European Countries, 1990–2018 0.5 2.5 6.5 10.2 - Initial total income support, percent of GDP Latest (2018) percent share of means-tested 75 NL UK IE CY DE SI 50 PT TR IT 25 0 DK FI RS FR AT CH ES BE ME SE LU BA HU IS HR LT NO LV EE PL RO BG CZ SK 0 25 50 Initial percent share of means-tested MT 75 Notes: Data point size is proportional to the size of a countries initial spending on income support as a share of GDP. In Eurostat (2021), the earliest date for which spending levels and components are available differs across countries. Spending is available from 1990 for 7 countries (Denmark, Finland, France, Norway, Switzerland, Germany, and U.K.); from 1995 for 13 countries (Italy, Ireland, Netherlands, Luxembourg, Austria, Sweden, Portugal, Belgium, Spain, Czech Republic, Malta, Slovak Republic, Iceland); from 2000 for 8 countries (Lithuania, Slovenia, Latvia, Estonia, Hungary, Cyprus, Romania, and Turkey); and from 2005 onwards for 5 countries (Bulgaria, 2005; Croatia, 2008; Serbia, 2008; Bosnia and Herzegovina, 2013; and Montenegro, 2016). Sources: Eurostat (2021); Authors’ calculations. III. KEY DESIGN FEATURES The design of GMI schemes varies with country approaches to social protection. Some countries offer one comprehensive benefit while others combine a single GMI payment with other payments that eligible households can receive to augment their income. Or GMI schemes may be a collection of separately administered means-tested payments targeting different areas of need. In categorizing GMI schemes in Europe, we follow the typology set out in Frazer and Marlier (2016):4  Simple and comprehensive schemes open to all with insufficient means to support themselves. Countries deemed to operate simple (or unified) and comprehensive schemes include Belgium, Cyprus, Switzerland, Estonia, Italy, Iceland, Lichtenstein, Luxembourg, the 4 Frazer and Marlier (2016) define GMI schemes as “income support schemes for people of working age (whether in or out of work) that provide a means-tested safety net for those ineligible for social insurance payments or those whose entitlement to these payments has expired”. In countries where social protection policy is devolved, GMI schemes do not necessarily fall into one category. For example, the authors deemed regional schemes in Spain to fall into the first two categories. The allocation of countries to different types also reflects the situation around 2015, and therefore may not reflect the current situation where countries have since implemented relevant reforms. 8 Netherlands, Norway, Sweden, Slovenia, and Slovakia.5 These types of schemes typically operate by specifying a subsistence level that incorporates a range of relevant expenditures and which are adjusted to reflect household composition. Payments are then intended to fill the gap between household income and this explicitly specified subsistence level. In Cyprus, for example, the minimum income standard is calculated in reference to a minimum consumption basket, while in the Netherlands it is in reference to the minimum wage.  Simple schemes but with somewhat restricted eligibility and coverage. Countries with this variant of GMI scheme include Austria, Greece, Croatia, Hungary, Lithuania, Portugal, and Serbia. Restrictions on eligibility and coverage vary greatly by country and for a variety of reasons. For instance, the targeting mechanisms in Portugal’s Rendimento Social de Inserção (RSI) (Arnold and Farinha Rodrigues, 2015) and Croatia’s Zajamčena Minimalna Naknada (ZMN) (European Commission, 2018a) exclude large proportions of those in or at-risk-of poverty; the low coverage of Greece’s Social Solidarity Income (SSI) (Vardarmatou and Pertsinidou, 2018) reflects an under-resourced administration; and eligibility for GMI in Austria is restricted by residency requirements.  General schemes of last resort with additional categorical benefits covering most people in need of support. This group includes Germany, the U.K.,6 Poland, Latvia, and North Macedonia. Categorical support targets households with specific characteristics—for example, households where a member may have a disability or lone parent households. A particularly important categorical payment in countries with these schemes is support for families, including child benefits. While such support is present in all EU countries, it tends to be higher in these countries. For example, in 2017 it stood at between 65–87 percent of all income support expenditure in Germany, Poland, and Latvia compared to an EU average of 57 percent. Approaches to family benefits vary, with Germany and Latvia providing largely universal payments and Poland and the U.K. opting to means test these benefits. In the context of the U.K., the introduction of means testing for child benefits in 2013 marked a break from a previously universal approach.  Complex network of schemes, often categorical and sometimes overlapping, which cover most people in need of support. Countries with such schemes include France, Ireland, Malta, and Romania. For example, while France’s main minimum income payment is the Revenu de Solidarité Active (RSA), additional means-tested minimum income payments exist in the form of the Allocation de Solidarité Spécifique (which provides additional support to those that have exhausted their unemployment insurance and meet certain criteria), the Allocation de Solidarité aux Personnes Agées (which provides income support for the over65s), and the Allocation aux Adultes Handicapés (which provides income support for those with disabilities). 5 Spain’s launched its national Ingreso Mínimo Vital scheme in 2020 and this can be considered as a simple (or unified) and comprehensive scheme. 6 The U.K. is currently undergoing the transition to a new “universal credit” payment aimed at consolidating various existing payments into a single payment, although this payment is still a composite of different components based on various categorical criteria. 9  Very limited, partial or piecemeal schemes that are restricted to narrow categories of people and fail to cover many of those in need of support. The only country in this category is Bulgaria. Although Bulgaria operates a single, universal GMI scheme incorporating means testing and a system of reference values capturing different household needs, the absence of mechanisms for indexing key policy parameters to economic conditions combined with one-for-one reductions in payments with employment income has led to erosion in coverage and adequacy of the scheme over time (European Commission, 2017). Restrictive eligibility criteria, such as stipulations on the number of rooms in a home per household member, also result in large numbers of poor people being inadequately covered by the scheme. The complexity of GMI schemes can have important implications for their coverage, adequacy, and work incentives. Schemes consisting of multiple, separately administered payments with different objectives and conditions can create benefit distributions that vary widely with household circumstances as well as significant work disincentives for some population subgroups that can be difficult to determine a priori. This can also result in unintended consequences, especially when different benefits are interlinked. For example, where eligibility for one payment is conditioned on eligibility for another (so-called “passporting”), changes to one can have knock-on consequences to the receipt of the other. Furthermore, schemes consisting of numerous distinct payments can often result in low levels of take-up by those eligible due to the complexity of assessing entitlement for each payment and the administrative burden of making separate applications. For instance, recent reforms in the U.K. consolidate separate payments into a single payment which is expected to produce an “automatic take-up” effect whereby households receive benefits for which they were previously entitled to under the old system but did not claim. While all GMI schemes have some form of conditionality associated with them, the extent and nature of conditionality differs. Common forms of conditionality across all schemes include the requirement to register with employment offices, demonstrate engagement in active job search, accept suitable job offers, and engage in activities that increase the chance of employment (such as training). Breach of these conditions typically results in partial or full suspension of benefits. Stricter conditionality often involves a more rigid implementation of these common conditionalities. For example, recipients may be required to simply register and demonstrate job search activity at their employment office every week as a pre-requisite for the weekly payment. Or they may be required to accept any job offered even if it does not directly match skills or career aspirations. In some schemes, beneficiaries may be directed to public works programs or other state-sponsored employment as in Denmark, Bulgaria, Croatia, and Hungary. Immervoll and Knotz (2018) assess the relative strictness of conditionality measures for jobseekers receiving unemployment and social assistance benefits and produce an overall strictness indicator that incorporates sub-indices on the severity of availability requirements and suitable work criteria, job search requirement and monitoring, and sanctions. For social assistance payments, they find that Malta, Switzerland, and Slovenia have the strictest conditions for jobseekers while Iceland, Bulgaria, and Greece have the most relaxed. While GMI schemes tend not to be time limited, regular reassessment is common. Although time limiting of payments runs counter to the notion of a GMI as a payment of last resort, many 10 countries require regular reassessments to ensure that recipients are still eligible. This is especially the case in schemes that include categorical payments conditioned on household characteristics that change over time, such as the number of adults and children in the household, and housing costs. Over half the countries reviewed by Frazer and Marlier (2016) had periodic reassessment and some schemes require households to report changes that may affect their eligibility as soon as possible. In practice, in many countries, this periodic reassessment can result in payments ceasing for a sizeable proportion of recipients. IV. GMI AND WORK DISINCENTIVES Both the generosity and design of social benefits can create in-built work disincentives. Work disincentives arise through both an income effect (higher out-of-work income) and a substitution effect (lower return to increasing earnings due to benefit withdrawal). These work disincentives can manifest themselves both in terms of the reduced incentive to take-up employment in the first instance (participation, or the “extensive” margin) and to increase earnings once employed (primarily through increasing hours worked, or the “intensive” margin). However, faster benefit withdrawal also reduces the fiscal cost of benefits and thus the need to finance higher benefits through higher distortionary taxation elsewhere, including higher income taxation further up the income distribution. The magnitude of these disincentives is determined by the generosity of benefits, how quickly generosity is reduced as income increases, and how earnings are taxed through income tax systems and social insurance contributions. Of course, whether these financial disincentives actually affect employment outcomes will depend on both labor demand and supply conditions.7 Empirical evidence finds that such work disincentives are particularly important for low-income and secondary (typically female) workers (Meghir and Phillips, 2010). A key challenge then for the design of GMI schemes is how to best manage this inherent trade-off between equity (poverty) and efficiency (work incentives) objectives. The strength of work disincentives inherent in GMI schemes can be seen by examining Participation Tax Rates and Marginal Effective Tax Rates. The Participation Tax Rate (PTR) is calculated as the share of additional household income from moving into employment that is lost due to reductions in benefits and income taxation. The Marginal Effective Tax Rate (METR) is calculated as the share of additional household income from an increase in earnings that is lost due to reductions in benefits and income taxation. In general, PTRs and METRs will depend on household composition, the level of earnings, the generosity of GMI payments, the rate at which support is tapered, and the interaction with the tax system, and estimates can be especially sensitive to the assumptions underpinning the estimation methodology, including the time horizon used and the size of the earnings increase considered (Annex I). The estimates presented below come from the OECD’s Social and Welfare Statistics database, which simulates PTRs and METRs for selected European countries in 2019.8 Work is assumed to 7 For example, where unemployment reflects low labor demand then financial incentives may not influence employment outcomes much. Where it reflects supply disincentives (e.g., due to high out-of-work benefit generosity) then reducing financial disincentives to work may have sizeable employment impacts. 8 In some cases, the income support provided on moving into employment is time-limited, ending at some point after moving into employment. Where PTRs concern the income lost on the first day of moving into employment, 11 be remunerated at two-thirds the average wage9 and estimates also allow for the presence of housing-related income support and in-work support to capture the overall incentives of the social protection system while also incorporating the effects of each country’s direct income tax system. Labor disincentives are based on a working-age adult changing their labor supply in four types of low-income households typically considered in the literature: (i) Single adults with no children (Single); (ii) Single adults with two children (Lone Parent); (iii) Couples with two children where both adults are unemployed and one adult changes their labor supply (Couple/Unemployed); and (iv) Couples with two children and one adult employed earning twothirds the average wage and a second adult who changes their labor supply (Couple/Employed). For the calculation of PTRs, it is assumed that one unemployed adult moves into full-time work. For Couple/Unemployed this involves one of the two unemployed adults moving into work fulltime at two-thirds of the average wage (the other adult remaining unemployed), while for Couple/Employed this involves the only unemployed adult moving into full-time work at twothirds the average wage. For METRs, it is assumed that the working adult moves from part-time work (50 percent) to full-time work. For Couple/Unemployed this involves one adult moving from part-time to full-time work (the other adult remaining unemployed), while for Couple/Employed it involves the second adult worker moving from part-time to full-time work. While PTRs and METRs vary across household types, variation across countries is much greater. Figure 4 presents the distribution of PTRs and METRs across countries for different household types. Country estimates are presented in Annex II for PTRs (Figure A2.1) and METRs (Figure A2.2).  Variation in PTRs. The highest median PTR, at 66 percent, is faced by Couple/Unemployed households, followed closely by Lone Parent (at 58 percent) and Single (at 55 percent) households. Couple/Employed households have a substantially lower average PTR at 33 percent, in part reflecting that their initial benefits are typically less generous due to the presence of an earning adult in the household. The variation in PTRs across countries for given household types is more substantial.  Variation in METRs. Focusing on the median across countries for each household type, Lone Parent households have the highest METR at close to 51 percent. Couple/Unemployed households have the second highest average at around 50 percent, with Single households following at 44 percent. Couple/Employed households face the lowest average METRs at 34 percent, again reflecting the lower benefit levels due to the presence of an earner. they are unlikely to reflect that support will be phased out over time and so they may not be fully reflective of the costs of moving into employment. To the authors’ knowledge, the PTR estimates used in this paper do not reflect these dynamics. 9 Wages at or below two-thirds of the average wage are commonly considered to be low, although this may still prove to be a substantial jump for some households entering the workforce. A 2020 report by the European Foundation for the Improvement of Living and Working Conditions (Eurofound) found that legal minimum wage levels in the EU fell below 60 percent of the average for the majority of countries and it may therefore be reasonable to assume that a large number of returning individuals find employment well below two-thirds of the average wage. 12 Figure 4. Distribution of PTRs and METRs Across Countries by Household Type, 2019 PTR METR 90 PTR/METR 60 30 0 Couple (Employed) Couple (UE) Lone Parent Single Couple (Employed) Couple (UE) Lone Parent Household type Single Notes: Horizontal line is median (50th percentile quartile), open circle is mean, upper/lower box notches are 25th/75th percentile, whiskers are min/max, and outliers are values that exceed 1.5 * IQR from the upper/lower notches and represented by closed circles. Sources: OECD; Authors’ calculations. Countries adopt various approaches to limit the magnitude of work disincentives. These include restricting the generosity of benefits, tapering the withdrawal of benefits, and even providing additional in-work income support (i.e., wage subsidies).  Restricted generosity. Low benefit levels can provide strong incentives to take up employment, although this can undermine the primary objective of poverty alleviation. Since different household characteristics can create very heterogeneous needs and benefit levels, it is difficult to provide a comprehensive assessment of the generosity of GMI benefits across countries and population groups. Nonetheless, Immervoll (2009) shows that the maximum benefits entitlements for the majority of household types falls below the income that could be gained from a 40-hour work week at the country-specific statutory minimum wage in the majority of OECD countries. These disparities also vary widely across countries and the household types considered.10 Similarly, Pena-Casas and others (2013) find that around 70 percent of the 25 EU member countries they study have an average maximum benefit less than or equal to 40 percent of median income, which is a common poverty threshold. The 10 For example, the author finds income for a single adult in France was around 70 percent of median income for those working at minimum wage compared to only around 40 percent for those relying on minimum income benefits. The corresponding figures for couples with two children and one employed adult are 40 and 30 percent of median income, respectively. In contrast, household income for a single adult in the Netherlands is around 55 percent of median income for those working at minimum wage but close to 45 percent for those relying on minimum income benefits, while for couples with two children income from working at the minimum wage and income from relying on minimum income benefits are both equal to 35 percent of median income. 13 authors interpret this as governments prioritizing the need to incentivize employment at the expense of adequacy.  Tapering of benefits. Many countries “taper” the withdrawal of benefits as recipients earn more income by reducing entitlements gradually as income rises or by disregarding a fixed amount or proportion of earned income from the standard means test. In some instances, benefit tapering is indexed to time rather than income, with benefits reduced over fixed durations largely irrespective of income earned.11 Tapering helps avoid so-called “cliff-edge” effects where earning a marginal amount of income could lead to the complete loss of a benefit and therefore very high disincentives for entering employment and increasing earnings. The extent of tapering should also account for the effect of taxation and social insurance contributions as earned income increases. Indeed, tapering can also take the form of reductions in tax and social insurance contributions.  In-work support. As with tapering, in-work support serves to reduce disincentives to take up employment or earn more income. However, whereas tapering smooths disincentives by allowing recipients to retain some amount from their existing benefits, in-work support tends to reward recipients with additional support (“wage subsidies”) for moving into work or increasing their labor supply. A notable example of this is the U.S. Earned Income Tax Credit, but elements of this can also be seen in European schemes. For example, the pre-cursor to the universal credit policy in the U.K. (tax credits) provided premia for those working certain hours, and France’s RSA provides tax credits to workers whose earned income falls below a minimum level. The theoretical and empirical literature on optimal income taxation suggests that moderate-tolow PTRs and METRs are optimal when trading off equity (poverty) and efficiency (employment) considerations.12 Simulations within a social welfare framework suggest optimal METRs of (60, 70, 90) percent for lower-income groups, corresponding to (low, middle, high) inequality aversion parameters, and optimal transfers equal to (36, 54, 67) percent of median income (Coady and Le, 2020).13 Interpretation of these estimates should incorporate the existence of consumption taxes, which vary around 17 percent in many European countries (European Commission, 2020) and would imply lower METRs associated with income tax and benefit withdrawal rates. Therefore, for moderate aversion to inequality, METRs in the order of 50 percent would appear optimal. Saez (2002) shows that much lower, even negative, METRs may be desirable for unemployed groups who respond strongly to financial incentives and where their market wage is low, e.g., low-skilled (low earnings) workers without children who can find job matches. So, for example, for such 11 By breaking the link between earnings progression and benefits withdrawal, time-based tapering can help to remove disincentives associated with earning additional income. However, where individuals fail to find steady work or substantially progress their earnings, such a feature could result in individuals being both unemployed or on very low incomes and without access to income support. Mitigation of this scenario requires careful design. 12 Since PTRs and METRs reflect a combination of benefit withdrawal rates and other direct income taxes such as personal income taxes and social contributions, they can be adjusted through changes in any of these components. 13 Similar estimates were found for the U.K. by Brewer, Saez, and Shephard (2010). Simulations within a poverty alleviation framework also suggest optimal METRs of around 60–70 percent (Kanbur, Keen, and Tuomala, 1994). 14 workers it may be desirable to combine lower out-of-work benefits with wage subsidies. Based on our earlier analysis, this suggests that METRs may be too high in many European countries, which is also consistent with previous findings (OECD, 1997). Also, it is likely that optimal METRs are lower when the generosity of the income grant is set below the optimal level. Similarly, if work is deemed to have a social value then optimal METRs will likely be lower. On the other hand, if eligibility for benefits are conditioned on work search or participation in active labor market programs (Annex 3) then this can mitigate the adverse employment impact of high generosity and high financial work disincentives. In practice, many countries seem to reduce work disincentives at a high cost in terms of achieving the primary objective of poverty alleviation. Adequate generosity is required to address poverty14 while containing fiscal cost (and avoiding higher income taxes further up the income distribution) requires the withdrawal of benefits as income increases. Figures 5 and 6 show that in practice there is a strong positive correlation between generosity of benefits and PTRs/METRs, with a significant number of countries having very low levels of both, consistent with them prioritizing employment incentives over poverty alleviation. Addressing poverty concerns through their GMI will therefore likely require a significant increase in generosity in many cases.15 To contain the fiscal cost (and thus avoid the need for higher income taxes further up the income distribution), PTRs and MTRs can be increased from their current low levels. But there are also that countries with very high PTRs and MTRs that could consider reducing these. Some countries achieve this through the use of income disregards and fixed-duration benefits to lower work disincentives. Both features attempt to target support to those moving into lower wage jobs who typically face the greatest work disincentives from payment withdrawal. Targeting can also help contain fiscal cost. Such designs are used in Ireland, Malta and Estonia, where income disregards are combined with a fixed duration for benefits (i.e., benefits are tapered not with increased earnings but with time in employment) to create low effective taper rates:16 14 Van Lancker and Farrell (2018) identify adequacy of benefits as one of the three key principles that should be used to evaluate the effectiveness of GMI programs. The others were (a) clear eligibility rules and application procedures, and (b) the integration of support with a set of active labor market programs that help enhance employment prospects. 15 Note that country-specific recommendations should also allow for the fact that low generosity could reflect low pre-transfer poverty. 16 At the other end of the spectrum, three countries combine high work disincentives with low generosity. In Romania, a high effective taper rate of 100 percent contributes to high PTRs, with main GMI payments decreased one-for-one with employment income. In Croatia, the ZMN payment has a default 100 percent taper system. However, provisions exist for gradually tapering payments to long-term unemployed recipients who find employment—payments are tapered by duration over three months with no loss of payment in the first month, a 25 percent reduction in the second month, a 50 percent reduction in the third month, and a 100 percent reduction thereafter. In Italy, the Reddito di Cittadinaza (RdC) (Jessoula, Natili, and Raitano, 2019) came into operation in 2019 with the aim of improving coverage and take-up amongst the most vulnerable as well as increasing generosity. The RdC has a high effective taper rate, with only 20 percent of employment income disregarded from the means test for benefits and this has resulted in a high METR for low-earning families. 15 Figure 5. Relationship Between Generosity and PTRs, 2019 a Couple (UE) a Lone Parent a Single 100 75 50 25 0 NO CY FI SE AT SI LU DE LU DK DK DK SI HR IS BE IS SE IE NL RO RO LV HU BG ES BG LT PT CZ IT HU IT HU BG IT SK SK CZ SK NO RO PT ES PT LT LV ES CZ LV EL EL EL EE DE CY CY FR UK NL LT IE MT MT UK IE MT EE EE 0 25 50 75 100 Generosity (Percent of Median Income) PTR Sources: OECD; Authors’ calculations. Figure 6. Relationship Between Generosity and METRs, 2019 a Couple (UE) a Lone Parent a Single 100 75 50 25 NO LT SE FI IE LU LU DE SI UK AT UK SI DK DK DK MT RO CZ IT LT HU IT IT RO RO EL EL BG PL NO SI CZ EE PT LV FR BE IS NO BE FR NL BE FI DE AT AT SE DE IS IS NL IE FI UK PL MT ES HR HR CY MT NL IE BG SK EE PT CY ES LV EE 0 0 25 50 75 100 Generosity (Percent of Median Income) METR Sources: OECD; Authors’ calculations. 16  Ireland. In Ireland, €20 of earnings per day of permitted work (beneficiaries are allowed up to three days per week) is disregarded from employment income when calculating Jobseekers’ Allowance entitlement and deductions are calculated as 60 percent of earnings less this income disregard.17 In addition, the Part-time Job Incentive Scheme and Back to Work Family Dividend are fixed-duration payments offered to the long-term unemployed to incentivize moving into work. In return for relinquishing claims to primary assistance benefits, both schemes provide benefits for a fixed duration that are slightly lower than household GMI entitlements, but which are not tapered with employment income, subject to certain eligibility requirements.18 Ireland’s relatively generous tapering system serves to smooth disincentives to increase income and work and contributes to their lower measured PTRs and METRs.  Malta. The tapering design of Malta’s Social/Unemployment Assistance (SUA) payment is equivalent to a fixed-duration payment incentive to move into work. Malta’s SUA payment is tapered in proportion to time elapsed since finding employment. Beneficiaries moving into work have their payments tapered by 35 percent for the first year of employment, 55 percent for the second year, 75 percent for the third year, and 100 percent thereafter. The use of a fixed duration taper contributes to the lower PTRs and MTRs in Malta. In particular, by extending the duration taper for a number of years after moving into employment, beneficiaries have a strong incentive in the short term to increase earnings once employed.  Estonia. The GMI system has slightly lower than average generosity for all household types and makes use of duration-based tapering. Upon moving into employment, beneficiaries can keep all earned income during the first two months of employment. Subsequently, beneficiaries are allowed 50 of earnings as an income disregard for the following four months after which benefits are withdrawn one for one (i.e., a 100 percent taper). While these approaches reduce work disincentives in the short term, their gradual withdrawal is needed to avoid “cliff edges” where benefits decrease sharply at some future point in time after entering employment. Careful monitoring of labor market developments of such benefit recipients can also be used to signal the need for further tailored public support or benefit conditionality such as training or specific job placements. Countries can also use complementary “activation policies” to attenuate the work disincentives created by higher PTRs and METRs. These include (Annex III): 17 In Ireland, Jobseekers’ Allowance (JA) is the main payment for those that are unemployed and have exhausted any unemployment insurance (known as Jobseekers’ Benefit) that they might have been entitled to. 18 The year-long Part-time Job Incentive Scheme is targeted at those that have been unemployed and in receipt of JA for at least 15 months. Under the scheme, participants may work for under 24 hours a week and receive a payment that is slightly lower than their JA payment, but which is not subject to any tapering for earned income. Similarly, in return for relinquishing primary income replacement benefits, the Back to Work Family Dividend guarantees payments for two years for families that have been claiming JA for at least one year and that move into employment. Payments are not subject to tapering and are equal to the JA rate for qualifying children in the first year and half that in the second. 17  Conditionality. Recent reforms to GMI schemes in many countries have focused on strengthening conditionality for receipt of GMI benefits. These conditions typically relate to mandated job search activity and acceptance of employment offers but can also include participation in ALMPs, which aim to ensure that recipients are both equipped and incentivized to participate in the labor market. Failure to meet conditions is typically met with sanctions.  Direct labor market interventions. Although somewhat rare, some countries undertake direct intervention in the labor market. These include interventions that change the legal or regulatory structure of the labor market, create public employment, or directly influence the cost of employment to employers. These interventions are often particularly focused on promoting employment of the long-term unemployed who may face more barriers to entry than those unemployed for shorter durations. Germany’s Hartz reforms between 2002 and 2005 contained numerous labor market interventions, including wage subsidies, deregulation of temporary jobs, and the creation of public jobs (Annex IV). The large variation in combinations of PTRs/METRs and strictness suggest that many countries may have opportunities for strengthening strictness either to attenuate the work disincentive effects from high rates or to facilitate an increase in rates where these are low (Table A3.1). V. CONCLUSIONS Means-tested GMI schemes play an important role in combating poverty in most European countries. Over recent decades, the scope of means-tested benefits has expanded significantly in European countries, especially those where social benefit spending had increased to relatively high levels. Such schemes can enable countries to adequately protect households from individual and economy-wide income shocks while containing fiscal cost but can also result in strong work disincentives. This paper describes how the scope and design of GMI schemes differ across European countries. It also discusses the inherent trade-offs that exist between key design features, namely, benefit generosity, benefit withdrawal, and benefit eligibility. Adequate benefit levels are required to achieve the poverty alleviation objectives of GMI programs. However, containing the fiscal cost of generous benefits (and thus avoiding the need for other distortionary taxes) requires that high generosity is accompanied by high benefit withdrawal rates, which can introduce strong work disincentives that can undermine the primary objective of poverty alleviation if they result in lower employment earnings. This creates an inherent equity and efficiency trade-off for countries. In practice, many countries address this equity-efficiency tradeoff by combining low benefit generosity with low benefit withdrawal rates (BWRs) thus prioritizing employment incentives over the primary objective of poverty alleviation. These countries can reduce this trade off by combining higher generosity with higher BWRs. Countries with very high BWRs should consider reducing these, including through allowing income disregards and time dependent (rather than income-dependent) benefit withdrawal. The adverse work disincentives associated with higher BWRs can also be attenuated through strengthening complementary activation policies that incentivize and support participation in the labor market. 18 Opportunities for enhancing the design of GMI schemes to achieve poverty objectives at lower efficiency and fiscal cost will depend on existing design features. Table 1 presents information on how various design features differ across countries. This can be used as a starting point to identify possible welfare-improving reforms in specific countries. For instance, such a discussion can be initially informed by the following sequence of questions using the cross-country data used in Table 1 and elsewhere in this paper:  Are benefits generous enough to achieve poverty objectives? How does generosity vary across different household types?  Is there room to increase benefit withdrawal rates to contain the fiscal cost of more generous benefits without introducing unduly strong work disincentives?  Where work disincentives are large, can these be reduced in a fiscally sustainable way, e.g., through greater use of income disregards and time-dependent benefit withdrawal, possibly targeted at the most disadvantaged groups?  Is there room to strengthen activation policies to contain the work incentives associated with generous benefits and higher BWRs? Of course, developing more refined policy reform options would require more country-specific and program-specific analyses since the design features displayed in Table 1 (and more disaggregated data available elsewhere in the paper) are summary in nature and may not adequately capture important subtleties for specific household types.19 In addition, detailed analysis of the dynamics of beneficiary work and benefit histories can be used to better tailor the design of benefit and conditionality rules to individual work and unemployment histories and the specific challenges they face in becoming and remaining attached to the workplace. Design features will also need to reflect social preferences (e.g., the social valuation of work) as well as fiscal and political realities. 19 More detailed country-level information on how structural features and parameters determine PTRs and METRs is available in the OECD’s country-level policy documents. Where possible, these should be complemented by country-level analyses by national and international researchers. 19 Table 1. Variation in Key GMI Design Features in European Countries Total Social Share of Total Protection Social Country Spending Protection Average PTR, as Percent Spending that 2017 of GDP, is Means- 2017 Tested, 2017 Austria 5.2 26.9 72.5 Belgium 4.8 16.7 66.5 Bulgaria 2.4 16.7 29.5 Switzerland 3.3 30.3 76.0 Cyprus 3.8 57.9 56.5 Czech Republic 2.5 16.0 33.5 Germany 5.1 54.9 65.0 Denmark 7.0 30.0 82.0 Estonia 2.7 7.4 51.0 Greece 0.0 42.0 Spain 3.3 33.3 44.5 Finland 6.8 29.4 66.5 France 6.1 39.3 63.0 Croatia 2.9 24.1 56.5 Hungary 3.0 20.0 32.5 Ireland 3.2 62.5 49.5 Iceland 3.9 23.1 69.0 Italy 3.4 41.2 18.0 Lithuania 2.1 19.0 27.0 Luxembourg 5.4 16.7 70.0 Latvia 2.4 4.2 43.5 Malta 1.5 46.7 33.5 Netherlands 4.2 59.5 65.0 Norway 5.0 10.0 68.0 Poland 3.0 26.7 59.0 Portugal 2.2 50.0 45.0 Romania 1.3 23.1 42.5 Sweden 5.3 11.3 64.0 Slovenia 3.0 56.7 73.5 Slovak Republic 2.4 12.5 20.0 United Kingdom 4.5 64.4 65.5 Average METR, 2017 0.0 60.0 22.0 51.5 31.5 39.5 61.5 79.5 23.0 29.0 19.0 62.0 67.0 24.0 34.0 55.5 51.0 35.0 26.5 61.0 32.0 31.0 57.0 57.5 51.5 30.5 33.0 43.0 71.5 32.0 61.0 Average Generosity, 2017 46.0 40.0 17.0 44.0 23.0 48.0 61.0 36.0 27.0 27.0 49.0 40.0 30.0 17.0 53.0 47.0 9.0 26.0 41.0 29.0 44.0 50.0 39.0 53.0 31.0 27.0 38.0 44.0 22.0 56.0 Overall Strictness 3.1 1.6 4.0 2.8 3.1 2.7 3.4 0.2 2.7 2.8 2.8 2.5 2.8 0.0 3.6 1.9 3.3 3.6 3.2 2.4 0.9 3.8 3.4 4.0 0.9 3.8 Notes: Colors denote the location of the country in the top (red), middle (orange) or bottom (red) of the distribution. Average PTR, METR, and Generosity refer to the median value of the respective metric across the different household types that we consider—Single, Lone Parent, Couple/Unemployed, and Couple/Employed. PTRs and METRs are in percent; Generosity is percent of median income; and Strictness ranges from 1 (least strict) to 5 (most strict). Sources: Authors’ classification based on rankings of the sample. Overall Strictness is taken from Immervoll and Knotz (2018) and is a composite index of separate strictness indices including: availability for activation policies, expected occupational and geographical mobility for employment, frequency and evidence required of job search activity, and reasons accepted for non-participation in activation measures and turning down job offers. 20 Annex 1. Estimation of PTRs and METRs A comprehensive understanding of work disincentives arising from the of design tax and benefit systems requires knowledge of how benefit, tax and social security systems interact as well as how disincentives change over time. Jara and others (2019)—henceforth J&O—use the Euromod tax-benefit simulation model and 2017 policy rules to estimate and decompose Participation Tax Rates (PTRs) and Marginal Effective Tax Rates (METRs) for 28 European countries. They also capture dynamics by measuring both short-term and long-term PTRs, where the former captures incentives to move into work when individuals are still eligible for (typically more generous) unemployment insurance (UI) and the latter when eligibility for UI is exhausted and social assistance (typically less generous GMI benefits) is the main source of support. Looking first at the differences between short-term PTRs (SPTRs) and long-term PTRs (LPTRs), LPTRs are on average 21 percentage points lower than SPTRs (Figure A1.1). This is expected as UI benefits are typically far more generous than the social assistance transfers that replace them once UI eligibility has expired. The authors find that benefits (principally UI) account for more than half of the SPTR in all but four countries. Luxembourg, Sweden, France, and Finland are examples of countries with relatively generous unemployment benefits and high SPTRs. By comparison, the relatively low generosity of social assistance payments means that LPTRs are dominated by the impact of the tax and social contribution system, which accounts for more than half of LPTRs in most countries. Nonetheless, there is substantial heterogeneity in the contribution of social assistance to LPTRs, and in countries with established comprehensive GMI schemes (such as Greece, Denmark, Cyprus, and the U.K.) social assistance benefits make a relatively higher contribution to work disincentives. Although LPTRs are systematically smaller than SPTRs, both are highly correlated with countries that have high LPTRs also typically having high SPTRs. With respect to METRs, tax and social contributions explain well over half of the disincentives to increase labor supply for all countries studied. However, the loss of benefits is also an important driver of METRs in some countries such as the U.K., the Netherlands, France, Cyprus, Ireland, Romania, Slovenia, and Luxembourg. Comparing the PTRs and METRs derived by J&O with those produced by the OECD helps to highlight the sensitivity of estimates to the estimation methodology used (Table A1.1). Averaging over OECD estimates of PTRs for different household types, we find that the OECD estimates are on average 5.3 percentage points lower when compared to SPTRs, but 14.9 percentage points higher compared to LPTRs. Conducting an analogous exercise for METRs, the OECD estimates are 5.2 percentage points higher. Although both the OECD and J&O use data derived from Eurostat and employ broadly similar microsimulation techniques, slight differences in assumptions about implementation of policy rules and how results are averaged over different household types can clearly result in different estimates. Despite these level differences, the estimates produced by J&O and the OECD are closely correlated.  For PTRs, the correlation coefficient between the average OECD PTRs and SPTRs is 0.56, rising very slightly to 0.57 for LPTRs. While, on average, SPTRs are higher than OECD PTRs for 21 countries with relatively low OECD PTRs, long-term PTRs are on average lower for countries with high OECD PTRs. This results in lower variation in SPTRs and LPTRs across countries compared to OECD PTRs.  For METRs, the correlation coefficient is much stronger at 0.83. However, while both METRs coincide for countries with low OECD METRs, countries with high OECD PTRs systematically have relatively lower J&O METRs. Again, this results in lower variation across countries for J&O METRs relative to OECD METRs. Figure A1.1 Comparison of PTR and METR Estimates, 2017 Panel A 90 80 70 60 50 40 30 20 10 0 PTR Italy Slovak Republic Lithuania Bulgaria Hungary Malta Czech Republic Greece Romania Latvia Spain Portugal Ireland Estonia Croatia Cyprus Poland France Sweden Germany Netherlands United Kingdom Belgium Finland Luxembourg Austria Slovenia Denmark METR OECD Average Jara and others (2019) Long-Term Jara and others (2019) Short-Term Panel B 90 80 70 60 50 40 30 20 10 0 OECD Average Jara and others (2019) Sources: OECD; Jara and others (2019); Authors’ calculations. Couple, one adult unemployed Median - couple, one adult employed Sources: OECD; Authors’ calculations. Median - couple, one adult unemployed Couple, one adult employed Spain Slovak Republic Croatia Italy Bulgaria Malta Greece Hungary Czech Republic Latvia Belgium Portugal Poland Romania Estonia Iceland France Netherlands Ireland Austria United Kingdom Sweden Lithuania Slovenia Germany Denmark Luxembourg Norway Finland Cyprus METR Two adult households 140 120 100 80 60 40 20 0 Median - single, no children Lone parent Single, no children Median - lone parent Estonia Bulgaria Greece Spain Croatia Cyprus Portugal Slovak Republic Latvia Hungary Sweden Italy Lithuania Ireland Austria Romania Norway Iceland United Kingdom Germany Czech Republic France Luxembourg Slovenia Poland Finland Belgium Netherlands Malta Denmark METR 22 Annex 2. PTRs and MTRs in European GMI Schemes by Household Type Figure A2.1. Marginal Effective Tax Rates by Household Type, 2019 Single adult households 140 120 100 80 60 40 20 0 23 Figure A2.2. Participation Tax Rates by Household Type, 2019 Single adult households 140 120 100 80 60 40 20 0 PTR Estonia Greece Bulgaria Lithuania Spain Croatia Latvia Cyprus Malta United Kingdom Belgium Germany Sweden Switzerland Slovenia Denmark PTR Estonia Czech Republic Slovak Republic Malta Latvia Spain Portugal United… Belgium Ireland Germany Slovenia Austria Denmark Finland Cyprus Single, no children Median - lone parent Median - single, no children Lone parent Two adult households 140 120 100 80 60 40 20 0 Couple, one adult unemployed Median - couple, one adult unemployed Median - couple, one adult employed Couple, one adult employed Sources: OECD; Authors’ calculations. 24 Table A2.1. Key Features of GMI Schemes in Europe Simple and comprehensive scheme open to all with insufficient means to support themselves. Austria Belgium X Bulgaria /2 Croatia Cyprus X Czech Republic X Denmark X Estonia X Finland France Germany Greece Hungary Iceland X Ireland Italy /3 X Latvia Lithuania Luxembourg X Malta Netherlands X Norway X Poland Portugal Romania Slovak Republic X Slovenia X Spain /4 X Sweden X Switzerland X United Kingdom /5 Organisation of support /1 Simple and noncategorical scheme but with somewhat restricted eligibility and coverage General schemes of last resort with additional categorical benefits which cover most people in need of support. X X X X X X X X X X X X Complex network of schemes, often categorical and Tapering of benefits sometimes overlapping, Restricted generosity /6 which cover most or income disregards /7 people in need of support X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X Features of support In-work support /7 Regular job search monitoring and enforcement (non-discretionary) /8 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X Active Labour Market Participation (ALMP) Programs /9 X X X X X X X X X X X X X X X X X X X X X X X X X Direct labour market intervention /10 X X X X X X X X X X X X X X X X X X X X X X X X X X X /1 Categorizations in this table largely draw on the categorizations in Frazer and Marlier (2016) but for the impact of recent reforms not covered by this paper. In these cases, the authors of this note have made their own assessment. /2 For Bulgaria, the authors break with Frazer and Marlier's (2016) assessment based on Peer Review on Minimum income benefits—securing a life in dignity, enabling access to services and integration into the labour market, European Commission (2018b). /3 Italy's Reddito di Cittadinaza was introduced in 2019 and the categorization in this table is the authors' assessment of this program, rather than that of Frazer and Marlier (2016). /4 Categorization for Spain is based on Frazer and Marlier (2016), which in turn is based on assessment of regional schemes. However, the Spanish government have recently announced a national minimum income scheme in response to the COVID-19 pandemic. /5 The U.K. is currently still undergoing large-scale welfare reform as it moves from a network of categorical payments to its universal credit payment. The assessment here is the authors' and based on the reforms. /6 Here defined as adequacy less than 40 percent of median income for all three households types for which OECD adequacy data exists. /7 Based on the OECD's benefits and wages country descriptions, plus country-specific documents where available. /8 Based on Immervoll and Knotz (2018) the OECD's benefits and wages country comparative policy table 2018 and authors' assessment. /9 Based on the OECD's benefits and wages country comparative policy table 2018 and authors' assessment. May also include direct labour market intervention, such as wage subsidies and public works. /10 Assessment is the authors' and predominantly based on spending greater than 0.1 percent of GDP on the sum of "40: Employment Incentives", "60: Direct Job Creation", and "70: Start-Up Incentives" programs recorded in the OECD's Public Expenditure on Labour Market Programs database. As outlined by the OECD, data should be interpreted with caution as spending may include transfers to individuals for some countries given the complexity of classification. 25 Annex 3. Activation Policies Most countries implement activation policies to mitigate financial work disincentives caused by the tax and benefits system, although the range and strictness of such conditions vary. The potential roles for such conditions can also differ, including the screening out of unwarranted benefit claims, mitigating the adverse labor market incentives generated by generous benefits and high benefit withdrawal rates, and providing support for individuals with difficulty finding or maintaining employment (especially the long-term unemployed).20 Reflecting this, activation policies take many forms, including: requiring GMI recipients to engage in closely monitored job search activity as a pre-condition for benefit receipt; policies to reduce the cost of employing GMI recipients (such as wage and hiring subsidies); and increasing employability of recipients through state-sponsored training opportunities. Consistent with this, the literature on activation policies often ascribes their effectiveness to either “threat/motivation” effects (penalties for not engaging in a mandatory and time-consuming activity designed to increase the chance of employment) or “treatment” effects (interventions that enhance employability due to training or better job search techniques). A. Job Search, Monitoring and Sanctions Perhaps the most common form of activation policy is the requirement that benefit recipients commit to job search activity. This is often accompanied by assistance with job search activity, monitoring of job search effort, and sanctions for failure to conduct adequate search activity. While almost all European countries condition GMI payments on job search requirements, the monitoring of this condition can vary extensively. In some countries, GMI recipients may be required to simply confirm that they are still looking for employment on a regular basis (Cyprus), whereas in others they may be required to submit proof of having applied for jobs on a regular basis (Austria). While job search might be simply left to the individual in some countries, others may require participation in a more structured job search activity, serving to act as an additional form of effort monitoring as well as to increase the quality of job search activity. Although sanctions for insufficient search effort are typically payment-related, they can also take the form of increased monitoring. Increasing the frequency of reporting to authorities creates a time cost that acts as a further disincentive to remain unemployed and makes it difficult to undertake unreported paid work. In some instances, conditionality is tailored to a beneficiary’s relative risk of remaining unemployed and their assessed capacity for work.21 20 Barriers that may prevent the unemployed from moving into employment include skill gaps, work inexperience, difficulty in conducting effective job searches, labor market discrimination, and the financial disincentives created by the tax and benefits system. Policies aimed at lowering these barriers include training programs, work counselling, employment subsidies, and job creation programs. 21 Evidence on the effectiveness of strict job search monitoring and sanctions is mixed depending on the outcome being assessed. Studying Swiss data, Lalive and others (2005) find that sanctions and warnings tend to lower unemployment duration at a lower cost compared to other forms of activation policy, a finding generally confirmed by studies conducted elsewhere. However, also looking at Swiss data, Arni and others (2009) find that while warnings and sanctions decrease the duration of unemployment, they also depress wages over the long term. They also find a negative relationship between sanctions and job stability. This raises the (continued) 26 Examples of structured approaches to job search commitment, monitoring, and sanctions include:  The U.K. Under the Universal Credit scheme recipients are required to sign a personalized commitment with their work coach. This commitment outlines their responsibilities in looking for work (or working more to increase their incomes) and the penalties for failing to meet the agreed conditions. Recipients are placed in one of four groups with each having differing job search requirements. Those in the “all work-related requirements” group are required to undertake all possible activity to gain employment or raise their income. This can entail having to demonstrate up to 35 hours of job search activity a week. Meanwhile, those in the “work-focused interview and work preparation requirements only” group are required to regularly phone their work coach while also preparing for work by writing resumés and attending training. Finally, members of the “work-focused interview” group are required to meet their work coach regularly and members of the “no-work related activity” group face no work-related conditionality. Group membership is normally determined at the outset of a claim but can be challenged or reviewed in the event of a change of circumstance (e.g., having a child) or on the grounds of an initial misclassification (e.g., not accounting for a relevant health condition). Recipients who fail to meet conditions are subject to tiered sanctions, with the most severe resulting in a cessation of payments for up to 26 weeks.  Denmark. Benefit recipients are assessed and divided into two groups corresponding to those who are deemed to require limited support in finding a job and those that might require activation policies. However, reforms implemented in 2015 have marked a move away from the tight requirements on job search activity. Prior to reform, unemployed recipients were required to upload a resumés to the job center website within one month of becoming unemployed and to attend in-person interviews on their job search activity at least once every three months. Post-reform, interventions are initially centered on counselling before progressing on to so-called ”right and duty” program participation. As such, the impact of the reforms has been to introduce more flexibility into job search requirements in the earlier stages of unemployment.  Italy. The recently introduced Reddito di Cittadinaza (RdC) program is notable for its strict conditionality. Aside from the requirement to sign a “work pact” (a feature seen in many other countries), recipients are also required to accept one of three “suitable” job offers within the first 18 months of receiving the payment. The “suitability” of offers is determined by their wage level and the distance that the recipient would have to travel from their residence to get to work. But while the minimum suitable wage level is fixed at €858 per month, the notion of “suitable” travelling distance is widened with each job offer declined, and recipients can be expected to accept a job anywhere in the Italian territory following the rejection of two previous offers. In addition to the above requirements, recipients must also be prepared to make themselves available for eight hours a week for activities mandated by local municipalities. However, there is some variation in these requirements and exemptions and relaxations can be obtained for people with disabilities and individuals with caring concern that strict monitoring and sanctions creates large disincentives to continue claiming assistance and beneficiaries therefore opt for worse job matches in order to exit the benefit system. 27 responsibilities (including for minors). In assessing the possible impact of the stricter conditionality embedded in the RdC on labor market activation, OECD (2019) noted that improving the quality and capacity of public employment centers would be critical, highlighting the primacy of administration in underpinning structured approaches to activation. B. Wage and Hiring Subsidies Whereas job search monitoring and sanctions increase the cost of unemployment for benefit recipients, wage and hiring subsidies can increase the benefit of employment for both recipients and employers. These payments typically take the form of a reduction in social contributions or payroll taxes paid by employers or workers but can also take the form of direct payments to employers or employees. Where employees receive direct payments, the separation between inwork benefits and subsidies can become blurred as, without an effective minimum wage policy, in-work benefits can subsidize employment by enabling employers to pay lower wages.22 For example, Rothstein (2008) estimates that the net transfer to low-skill workers is $0.28 of every $1 spent on the U.S. Earned Income Tax Credit (EITC), with the remainder captured by employers. The design and evaluation of subsidies (and other ALMPs more generally) must account for various challenges that determine program effectiveness (Card and others, 2018; Brown, 2015). For instance, these programs tend to be more cost effective at increasing aggregate (as opposed to just individual) employment when targeted at the most disadvantaged groups in the labor market (e.g., older and longer-term unemployed low-skill workers). Broader targeting may simply result in employers replacing existing workers with the subsidized unemployed by “cream skimming” the unemployed who are closer substitutes for existing workers. At the same time, too narrow targeting can result in a “stigma effect” whereby eligibility for subsidies signals the subsidy beneficiary as potentially being less productive with employers subsequently unwilling to employ them. The use of subsidies varies greatly across European countries and is often motivated by one of two objectives. These aim at securing existing employment links or creating new links.  Safeguarding pre-established links between workers and jobs. The use of subsidies to avert mass layoffs and job destruction is something that has been seen both in response to the Great Recession and, more recently, to the economic disruption caused by the COVID19 pandemic. Perhaps the most widely discussed of these is Germany’s Kurzarbeit scheme (Annex 4). Under the scheme, employers who cannot afford to maintain current staffing levels can reduce the hours offered to employees while the government pays a fixed percentage of the earnings lost as a result of the reduction in hours. Employees are therefore cushioned from large reductions in their earnings and firms can retain their workforce, averting layoffs and increases in unemployment. Kurzarbeit support is typically for a fixed duration, although the length of support is a flexible policy parameter that can be adapted to economic conditions (as it has been in the context of the COVID-19 pandemic). 22 By creating a floor on acceptable wages, minimum wage policies limit the ability of employers to capture the benefits of in-work benefits by paying lower wages to recipients of them. 28  Providing access to the labor market for hard-to-reach or disadvantaged groups. Wage subsidies are often used to encourage employers to hire those that are traditionally disadvantaged or hard-to-reach.23 For example, Estonia’s Palgatoetus scheme is aimed at providing access to the labor market for the long-term unemployed and ex-prisoners. The conditions for the subsidy, which is paid for up to 12 months and which cannot exceed 50 percent of wages, are agreed between the Unemployment Insurance Fund and the employer, while working conditions are agreed between employee and employer. In many cases, such subsidies are not permanent, but serve to enable recipients to obtain a foothold in the labor market and develop workplace skills that might improve their employment prospects. Other countries that currently provide some form of subsidy or have done so in the past include Germany, Sweden, and Denmark. Sweden uses various subsidies aimed at increasing employment of the long-term unemployed. One such subsidy is the Nystartjobb (NSJ) subsidy program aimed at the long-term unemployed who are already subject to activation. Paid to employers, the subsidy is proportional to the employers’ social security contributions, and its duration linked to the length of unemployment of the individual, subject to maximum limits. In order to prevent abuse of the subsidy, numerous conditions are stipulated for participating employers. In particular, employers cannot take part in the program if they have laid off staff in the previous nine months for lack of work, and wages and work conditions must be in line with the prevailing collective agreement at the industry level. Final agreement on wage subsidies is jointly decided between the public employment service, the employer, and the relevant trade union body. While evaluation of the NSJ is limited, Anxo (2014) cites research that suggests that it was effective in increasing the probability of employment. But the same research also suggested that 63 percent of jobs would have occurred regardless of the subsidy—highlighting the large risk of deadweight associated with subsidy programs. Evidence that the subsidy program is well-targeted to those that would otherwise have had difficulty in entering the labor market is also limited and issues such as “cream skimming”, whereby employers can benefit from subsidies for employment of workers that they would in any case hire, remain a real concern. Start-up subsidies provide an alternative to employee subsidies. Although most subsidy programs in Europe focus on subsidizing employee positions, some countries have implemented programs designed to assist the unemployed in becoming self-employed. A notable example of this is Germany, where start-up subsidies were included as part of the Hartz IV reforms in the early-2000s (the well-known ‘Ich-AG’ and ‘bridging’ subsidies) and have continued through successor schemes. Under the current subsidy scheme, Gründungszuschuss, unemployed people who wish to become self-employed can apply to receive a start-up grant equal to the value of unemployment benefits plus an additional amount to cover social contributions. The duration of 23 Creating new employment links (as opposed to securing existing ones) is more pertinent for GMI schemes during normal economic conditions. Disadvantaged groups that are typically targeted by wage and hiring subsidies include the long-term unemployed, the young and elderly, those with illnesses and disabilities, and migrants. However, the coverage of these subsidy programs does not always overlap exactly with GMI eligibility. For example, some young people may be eligible for wage subsidies but may be included in their parents household for GMI purposes while residence requirements may rule out GMI eligibility for migrants in many European countries. Long-term unemployment, which is closely correlated with receipt of GMI benefits, is often a precondition for eligibility for subsidies. 29 the subsidy is six months in the first instance, but some support can be extended by a further nine months. In applying for the subsidy, individuals must present a viable business plan to the employment office as well as demonstrate that they have sufficient skill to run a business. For the long-term unemployed receiving social assistance, vouchers provided under activation measures can be used to obtain counselling to assist with both producing a viable business plan and developing sufficient competency for running a business. The impact of the Hartz IV subsidies in Germany has been found to produce highly positive results. For example, Caliendo and Kunn (2011) found recipients of these subsidies to have a significantly higher probability of labor market participation relative to non-recipients for up to five years later. The same study also found earnings to be higher for those that received subsidies. C. Strictness of Conditionality Immervoll and Knotz (2018) assess the relative strictness of conditionality measures for jobseekers receiving unemployment and social assistance benefits. They produce an overall strictness indicator that aggregates scores across the following areas:  Requirements on the availability for work. While some countries relax the requirement to find and accept job offers for those participating in activation programs (such as training), others maintain these requirements for the duration of the program. For example, while participation in full-time training means that an individual is permitted to be “unavailable for work” in Ireland and Spain, those in training must continue to search for work and accept suitable offers in both Sweden and Austria.  Occupational mobility. A common way in which many countries increase the strictness of work availability rules is to require that individuals search for, and accept, jobs in occupations that they have not worked in previously or for which they have limited experience. Similar restrictions can also cover wages, with individuals not permitted to refuse jobs that pay a wage lower than what they previously earned. While the authors find that all countries in their sample require occupational mobility, strictness varies according to when such requirements come into effect. For example, unemployment assistance recipients in Spain may restrict their job search to their previous occupation for a limited initial period whereas many countries require individuals to be mobile from the outset.  Geographical mobility. Another dimension of the strictness of activation policies is the distance over which individuals must look for work. Depending on the maximum distance that is deemed acceptable, this may even necessitate relocating. In practice, few countries mandate relocation as part of their activation requirements and acceptable commuting times and distances vary. For example, individuals are expected to spend up to two hours commuting for a full-time job in Austria, compared to three hours in the Netherlands. In France, jobseekers must look for jobs up to distance of 30 kilometers from their home.  Other valid reasons for refusing job offers. Beyond considerations of occupation, wages, and geographical mobility, the strictness of activation policies can also vary according to what is considered a valid reason for not accepting a job offer. Incapacity due to health circumstances is considered a valid reason for job offer refusal in most countries, with caring responsibilities another consideration deemed valid in a number of countries. 30  Documentation of job search activity. Requirements on reporting job search activity can increase the strictness of activation either by increasing the amount of evidence that one must submit or by increasing the frequency with which individuals must submit evidence. However, requirements on the documentation of job search activity can be driven just as much by the quality of public employment services (who must administer and verify evidence) as by the desire to increase strictness. As such, there is notable variation across countries—for example, while in Malta individuals are required to keep a job search “diary” and provide the contact details of those to whom they applied for a job, in Cyprus GMI recipients are not required to submit evidence but must “sign on” with the Social Insurance Office at regular intervals in order to continue to receive payment.  Sanctions for refusal of job offers and ALMP participation as well as voluntary unemployment. In the event that a GMI recipient receives a suitable offer of employment, many countries introduce sanctions to ensure that this is taken up. The strictness of these requirements varies according to the severity of the sanctions. In some cases, sanctions are a time-limited reduction in payments (Belgium) whereas in others they can result in the termination of a claim (Portugal). Strictness can also vary in the number of suitable job offers one can refuse before being subject to sanctions—for example, while in France sanctions are only applied to recipients after a second refusal of a suitable job offer, in Slovenia they are applied at the point of the first refusal. Similar sanction mechanisms are also in place in many countries for refusals to participate in ALMPs. Where GMI recipients are already engaged in some form of employment, most countries have also adopted sanctions for choices that result in voluntary unemployment, such as resignation or poor workplace behavior. The strictness of these sanctions varies across similar dimensions as for the refusal of job offers and ALMP participation. Based on an aggregation of scores for the above criteria (0 being the least strict and 5 being very strict) in 2017, the authors find that Malta, Switzerland and Slovenia have the strictest conditions for jobseekers, while Greece, Turkey, and the Slovak Republic have the most relaxed (Figure A3.1). Strictness Index Score 31 Figure A3.1. Strictness of Activation Requirements in European Countries, 2017 5 4 3 2 1 0 Malta Switzerland Slovenia Portugal United Kingdom Latvia Italy Estonia Sweden Luxembourg Netherlands Romania Germany Austria Canada Finland Ireland Czech Republic France Denmark Spain Cyprus Hungary Norway Lithuania Belgium Croatia Poland Slovak Republic Turkey Greece Bulgaria Iceland Israel 2017 Notes: Zero values are interpreted as “non-response”. Sources: Immervoll and Knotz (2018). Figure A3.2. Labor Disincentives and Strictness of Benefit Conditionality A. PTRs and Strictness a Couple (UE) a Lone Parent a Single 100 75 50 25 0 0 CY PL PL PL SK SK BE HR HR BE BE LT HR LT LT NO FI DK AT DK NO DK IE FI DE AT DE NO FI AT RO ES IE HU ES CZ CZ HU CZ SE LU LU LU SE DE SE CH SI SI SI CH CH UK UK RO UK PT MT NL PT RO LV IT LV PT MT IT LV MT EE IT SK EE EE 2 4 Strictness (Overall 2017) y=52+1.6⋅(x), R2=0.0044 6 PTR 32 METR B. METRs and Strictness a Couple (UE) a Lone Parent a Single 100 75 50 25 0 0 PL PL SK SK SK FI NO IE DK DE LT DK DK AT BE FR IE DE NO FR FI BE FI RO LT FR BE NO CZ IE CZ LT HU HU HR HU CY CZ HR HR ES CY ES ES LU SI LU NL UK SI SE UK NL LU SE SI UK EE IT SE LV IT PT LV PT EE IT PT EE LV MT MT MT 2 4 Strictness (Overall 2017) Sources: OECD; Immervoll and Knotz (2018); Authors’ calculations. y=49+1.2⋅(x), R2=0.002 6 33 Table A3.1 Strictness of Activation Requirements and Components, 2017 Country Country code Valid Frequency Jobsearch Sanctions for Overall ALMP Occupation Geographic Reasons for of Monitoring Voluntary Strictness Availability al Mobility al Mobility Refusal of Jobsearch Documentatio Unemployme Job Offer Monitoring n nt Sanction for first refusal of Job Offer Sanctions for Subsequen t Refusal of Job Offers Sanction for first refusal of ALMP Sanctions for Subsequen t Refusal of ALMP Austria AT 3.1 4.0 5.0 2.0 3.0 4.0 4.0 1.0 2.0 2.0 2.0 2.0 Belgium BE 1.6 0.0 0.0 0.0 0.0 2.0 3.0 2.0 2.0 2.0 2.0 2.0 Bulgaria BG - - - - - - - - - - - - Switzerland CH 4.0 0.0 5.0 3.0 3.0 4.0 4.0 5.0 5.0 5.0 5.0 5.0 Cyprus CY - - - - - - - - - - - - Czech Republic CZ 2.8 5.0 4.0 1.0 3.0 1.0 1.0 4.0 4.0 4.0 4.0 4.0 Germany DE 3.1 5.0 5.0 5.0 3.0 3.0 4.0 1.0 1.0 2.0 1.0 2.0 Denmark DK 2.7 4.0 5.0 3.5 3.0 2.0 4.0 1.0 1.0 3.0 1.0 1.0 Estonia EE 3.4 5.0 3.0 2.0 5.0 4.0 4.0 1.0 1.0 5.0 1.0 5.0 Greece EL 0.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 1.0 1.0 1.0 0.0 Spain ES 2.7 1.0 2.0 2.0 5.0 1.0 2.0 5.0 3.0 4.5 2.0 4.5 Finland FI 2.8 3.0 3.0 3.0 1.0 3.0 2.0 3.0 2.5 4.5 2.0 4.5 France FR 2.8 0.0 3.0 1.5 5.0 4.0 3.0 3.0 1.0 3.0 1.0 3.0 Croatia HR - - - - - - - - - - - - Hungary HU 2.5 2.0 5.0 3.0 5.0 3.0 1.0 1.0 2.0 2.0 2.0 2.0 Ireland IE 2.8 2.0 4.0 1.0 5.0 3.0 4.0 2.0 2.0 2.0 1.0 2.0 Iceland IS 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Italy IT 3.6 4.0 4.0 3.0 3.0 3.0 1.0 5.0 5.0 5.0 5.0 5.0 Lithuania LT 1.9 0.0 0.0 0.0 5.0 4.0 3.0 1.0 1.0 1.0 1.0 1.0 Luxembourg LU 3.3 3.0 4.0 3.0 3.0 4.0 4.5 3.0 1.0 3.0 1.0 3.0 Latvia LV 3.6 2.0 3.0 2.5 3.0 4.0 4.0 3.0 3.0 5.0 5.0 5.0 Malta MT - - - - - - - - - - - - Netherlands NL 3.2 5.0 5.0 3.0 5.0 3.0 1.0 3.0 3.0 3.0 3.0 3.0 Norway NO 2.4 5.0 5.0 4.0 5.0 2.0 1.0 1.0 1.0 1.0 1.0 1.0 Poland PL 0.9 0.0 0.0 0.0 5.0 1.0 1.0 1.0 1.0 0.0 0.0 0.0 Portugal PT 3.8 3.0 2.0 3.0 5.0 3.0 3.0 5.0 5.0 5.0 5.0 5.0 Romania RO - - - - - - - - - - - - Sweden SE 3.4 5.0 4.0 3.0 3.0 4.0 4.0 3.0 1.0 3.0 1.0 3.0 Slovenia SI 4.0 3.0 3.0 3.0 5.0 3.0 4.0 4.0 5.0 5.0 5.0 5.0 Slovak Republic SK 0.9 0.0 0.0 0.0 5.0 1.0 0.0 1.0 1.0 1.0 1.0 0.0 United Kingdom UK 3.8 4.0 4.0 3.0 3.0 5.0 4.0 3.0 3.0 4.5 3.0 3.0 Notes: Colors denote whether country is in the top (green), middle (orange) or bottom (red) third of the strictness distribution. Sources: Immervoll and Knotz (2018). 34 Annex 4. Reform Experiences in the U.K. and Germany Both the U.K. and Germany have undertaken large-scale reforms of their welfare systems in recent years, with changes to each country’s GMI programs at the core of these efforts. Germany’s 2002–2005 Hartz reforms are considered by some to be “the most far-reaching reform endeavor in the history of the German welfare state” (Jacobi and Kluve, 2006), while the U.K.’s ongoing Universal Credit (UC) reform has been labelled one of the “most significant reforms since the Beveridge Report” (OBR, 2018), the post-war reforms that created the modern British welfare state. Here we outline the rationale and features behind these reform efforts, examine performance against their stated objectives, and discuss implementation challenges. Germany The Hartz Commission was established by the German government in 2002 in response to a sustained rise in unemployment throughout the late 1990s and early 2000s. The goal of the Commission was to set out a strategy for lowering unemployment and improving the flexibility of the German labor market by addressing three priorities: i) improving employment services; (ii) activating the unemployed; and (iii) fostering employment demand by deregulating the labor market. In doing this, the Commission identified a wide range of labor market reforms, with each reform taking place in one of four waves. In 2003, the Hartz I and II reforms aimed to increase labor market flexibility by lowering taxation and legal requirements on temporary and low hours work, while also introducing a range of subsidies targeted at the unemployed. Prior to 2003, strict guidelines on fixed-term contracts, the maximum duration of temporary work, and the sectors in which temporary work was permitted (notably, in the construction sector) were deemed to be causing labor market inflexibility. The Hartz I and II reforms removed many of these restrictions but required that temporary workers receive either equal pay and treatment to non-temporary workers or that their wages were determined by bargaining agreements between unions, employers, and temporary work agencies. Subsidies aimed at reducing the cost of “hard to place” workers were also a core feature. Social security exemptions were granted to businesses that employed workers over the age of 55, “integration subsidies” provided employers with a subsidy of up to 50 percent of wages for 6– 24 months for older and disabled workers, and top-up subsidies were offered directly to older workers who took up a job that paid less than their previous job. Support for self-employment was also provided, notably with the “Ich-AG” subsidy, which was paid directly to the selfemployed with income below a threshold and could be received for up to three years. These policies were followed in 2004 by Hartz III, which restructured the German employment agency system to make it more service oriented. Finally, the 2005 Hartz IV reforms were passed, and completely restructured the unemployment benefit system by combining social assistance and unemployment assistance for the long-term unemployed into one benefit (Unemployment Benefit II). Overall generosity was drastically cut for many households, continued benefit receipt became subject to means-testing, and strict activation measures were introduced (e.g., sanctions for refusing job offers). 35 Evidence on the impact of the Hartz reforms broadly suggests that they fulfilled their key aims. Writing shortly after the introduction of Hartz IV, Jacobi and Kluve (2006) found positive impacts on employment probabilities for the various wage subsidies introduced and for deregulation of “mini and midi-jobs”. However, the impacts of reforms to the employment agency system (including services offered) were found to be limited and employment agency-mandated temporary work was found to actually reduce the probability of employment for those required to participate. Krebs and Scheffel (2013) find that the Hartz reforms lowered the non-cyclical unemployment rate by 1.4 percentage points—thereby achieving a main objective of the reform—while also contributing to long-run growth. But the same study also finds an initial reduction in real wages due to higher levels of employment and a reduction in the welfare of unemployed households, consistent with the restrictions to GMI generosity. Although Bradley and Kugler (2019) also find evidence that the Hartz reforms lowered wages, they find that the reforms had a limited impact on aggregate unemployment as the reforms increased both the job finding and separation rates in roughly equal proportion. However, the authors argue that the increase in labor market churn may be indicative of improved labor market flexibility. More generally, the Hartz reforms are taken by some to highlight a trade-off between labor market flexibility and wage inequality. Although evidence suggests that the reforms improved flexibility, the class of marginal jobs created tend to be low-paying, not unionized, and provide little opportunities for skill build-up. As a result, people who are marginally employed tend to be stuck in low-paid segments and to rely on social assistance to reach a minimum income level. United Kingdom With its roots in the aftermath of the Great Recession, the U.K.’s UC payment was first proposed by the British government in 2010 as a reform that would make the welfare system more cost effective, increase its simplicity, and improve work incentives amongst benefit recipients. At the heart of the policy is the combination of five pre-existing, and hitherto separately administered, working-age benefits. These “predecessor benefits” cover unemployment assistance, in-work support for low earners, income support for those with sicknesses and disabilities, and housing support. However, rather than simply administer a combined payment through a single system, UC also includes fundamental structural reforms. A maximum entitlement is calculated for each recipient based on their eligibility for various components of the UC benefit (roughly mirroring the predecessor benefits the UC replaces) and then subjected to a single means-test and taper. In contrast to the previous in-work support (tax credits), only those with responsibility for children or with caring responsibilities benefit from an income disregard under UC. Under UC, the conditionality and sanction regime is also stricter and extended to groups previously not subject to conditionality. Recipients are required to sign a personalized commitment that outlines their work-related responsibilities and the penalties for failing to meet the agreed conditions. Recipients are placed in one of four groups with each having differing job search requirements. For example, those in the “all work-related requirements” group are required to demonstrate up to 35 hours of job search activity a week while those in the “workfocused interview” group are required to simply meet their work coach regularly. Recipients who fail to meet conditions are subject to tiered sanctions, with the most severe resulting in a cessation of payments for up to 26 weeks. Finally, the monthly reporting mandated under UC 36 represents a break from the annual reporting that was required of large numbers of recipients on in-work support under the previous system. With reform starting in 2012, the UC rollout was initially expected to be completed in 2015 but is now projected to complete in 2024. Reform delays reflect the administrative difficulties in simultaneously combining and reforming a large number of disparate payments as well as concerns regarding the number of households losing out under the policy and entering poverty. Given the incomplete nature of the UC reform in the U.K., there is limited evidence on its impact. Instead, studies to date have relied on small-scale investigation in geographical areas where the payment has been rolled out and microsimulation analyses using current policy parameters and ignoring labor supply responses. Qualitative analysis of the impact of UC in the Liverpool region by Gardiner and Finch (2020) found that while “UC removes the very highest benefit withdrawal rates that can disincentivize earnings progression…it creates the risk that some recipients are incentivized to reduce their hours”, with weaker incentives for second earners to increase labor supply in couples. The study also found that the ability of the benefit’s in-work conditionality to reduce these disincentives “seems far from assured”. Reflecting on the payment’s performance during the COVID-19 pandemic, Brewer and Handscomb (2020) suggest that, while administrative systems have held up to a surge in claims, the fall in disposable income for recipients after accounting for UC is markedly higher than that for those benefitting from the U.K.’s furlough scheme and will likely constitute a shock to household living standards. 37 REFERENCES Anxo, Dominique, 2014, "Stimulating Job Demand: The Design of Effective Hiring Subsidies in Europe–Sweden," European Employment Policy Observatory (Brussels: European Commission). Arni, Patrick, Rafael Lalive, and Jan C. van Ours, 2009, How Effective Are Unemployment Benefit Sanctions? Looking Beyond the Unemployment Exit Rate, IZA Discussion Paper Series No. 4509 (Bonn: IZA Institute for the Study of Labor). Arnold, Jens Matthias and Carlos Farinha Rodrigues, 2005, “Reducing Inequality and Poverty in Portugal,” OECD Economics Department Working Papers No. 1258 (Paris: OECD Publications). Averett, Susan L. and Yang Wang, 2018, “The Effects of the Earned Income Tax Credit on Children’s Health, Quality of Home Environment, and Non-Cognitive Skills,” Public Finance Review, Vol. 46, No. 4, pp. 519–57. Bastian, Jacob, and Katherine Michelmore, 2018, “The Long-Term Impact of the Earned Income Tax Credit on Children’s Education and Employment Outcomes,” Journal of Labor Economics, Vol. 36, No. 4, 1127–63. Bradley, Jake, and Alice Kugler, 2019, "Labor Market Reforms: An Evaluation of the Hartz Policies in Germany," European Economic Review, Vol. 113 (April), pp. 108–35. Brewer, Mike, and Karl Handscomb, 2020, “This Time Is Different–Universal Credit’s First Recession: Assessing the Welfare System and its Effect on Living Standards during the Coronavirus Epidemic,” Resolution Foundation (London). Brewer, Mike, Emmanuel Saez, and Andrew Shephard, 2010, “Chapter 2: Means-Testing and Tax Rates on Earnings,” in Dimensions of Tax Design: The Mirrlees Review, ed. by Stuart Adam, Timothy Besley, Richard Blundell, Stephen Bond, Robert CHote, Malcolm Gammie, Paul Johnson, Gareth Myles, and James Poter (Oxford University Press). Brown, Alessio J.G, 2015, "Can Hiring Subsidies Benefit the Unemployed?," IZA World of Labor, (June) pp. 163 (Bonn: IZA Institute of Labor Economics). Caliendo, Marco, and Steffen Kunn, 2011,"Start-up Subsidies for the Unemployed: Long-Term Evidence and Effect Heterogeneity," Journal of Public Economics, Vol. 95 (April), pp. 311– 31. Card, David, Jochen Kluve, and Andrea Weber, 2018, “What Works? A Meta Analysis of Recent Active Labor Market Program Evaluations,” Journal of the European Economic Association, European Economic Association, Vol. 16, No. 3, pp. 894–931. Chetty, Raj, John N. Friedman, and Jonah Rockoff, 2011, “New Evidence on the Long-Term Impacts of Tax Credits,” National Tax Association Proceedings, pp. 116–24 (New Orleans: 104th Annual Conference Proceedings, 2011). Coady, David and Le, Piotr-Nghia, 2020, “Designing Fiscal Redistribution: The Role of Universal and Targeted Transfers,” IMF Working Paper No.20/105 (Washington: International Monetary Fund). 38 Dahl, Gordon B., and Lance Lochner, 2012, “The Impact of Family Income on Child Achievement: Evidence from the Earned Income Tax Credit,” American Economic Review, Vol.102 (5), pp. 1927–56. Eurofound, 2020, “Minimum Wages in 2020: Annual Review - Minimum Wages in the EU series,” Publications Office of the European Union (Luxembourg: European Foundation for the Improvement of Living and Working Conditions). European Commission, 2017, "Country Report 2017: Bulgaria" (Brussels: European Commission). ______, 2018a, “Reaching for More: Challenges of Adequacy and Coverage in Croatian Minimum Income System,” from Peer Review on “Minimum Income Benefits – Securing a Life in Dignity, Enabling Access to Services and Integration into the Labour Market” (Brussels: European Commission). ______, 2018b, “Minimum Income Schemes in Bulgaria (Design, Strengths, Weaknesses, Comparison with German Policy Example),” from Peer Review on “Minimum Income Benefits – Securing a Life in Dignity, Enabling Access to Services and Integration into the Labour Market” (Brussels: European Commission). ______, 2020, “Taxation Trends in the European Union,” Data for the EU Member States, Iceland and Norway (Luxembourg: Publications Office of the European Union). Eurostat, 2021, “European System of integrated Social PROtection Statistics (ESSPROS)” (Dataset). Evans, William N. and Craig L. Garthwaite, 2014, “Giving Mom a Break: The Impact of Higher EITC Payments on Maternal Health,” American Economic Journal: Economic Policy, Vol. 6 No. 2, pp. 258–90. Frazer, Hugh and Eric Marlier, 2016, “Minimum Income Schemes in Europe: A Study of National Policies, 2015," Directorate-General for Employment, Social Affairs and Inclusion (Brussels: European Commission). Gardiner, Laura and David Finch, 2020, “The Long And Winding Road: The Introduction and Impact of Universal Credit in Liverpool City Region and the U.K.,” Resolution Foundation (London). Hoynes, Hilary, Doug Miller, and David Simon, 2015, “Income, the Earned Income Tax Credit, and Infant Health,” American Economic Journal: Economic Policy, Vol. 7, No. 1, pp. 172–211. Immervoll, Herwig, 2009, “Minimum-Income Benefits in OECD Countries: Policy Design, Effectiveness and Challenges,” IZA Discussion Paper Series No. 4627 (Bonn: IZA Institute of Labor Economics). ______, and Carlo Knotz, 2018, “How Demanding Are Activation Requirements for Jobseekers?,” IZA Discussion Paper Series No. 11704 (Bonn: IZA Institute of Labor Economics). Jacobi, Lena and Jochen Kluve, 2006, Before and After the Hartz Reforms: The Performance of Active Labour Market Policy in Germany, IZA Discussion Paper Series No. 2100 (Bonn: IZA Institute of Labor Economics). Jara, Xavier H., Katrin Gasior, and Mattia Makovec, 2019,”Work Incentives at the Extensive and Intensive Margin in Europe: The Role of Taxes, Benefits and Population Characteristics," Social Indicators Research, Vol. 152, pp. 705–78. 39 Jessoula, Matteo, Marcello Natili, and Michele Raitano, 2019, “Italy: Implementing the New Minimum Income Scheme,” ESPN Flash Report 2019/35, European Social Policy Network (ESPN) (Brussels: European Commission). Kanbur, Ravi, Michael Keen, and Matti Tuomala, 1994, “Optimal Non-linear Income Taxation for the Alleviation of Income-Poverty,” European Economic Review, Vol. 38, No. 8, pp. 1613– 32. Krebs, Tom, and Martin Scheffel, 2013, "Macroeconomic Evaluation of Labor Market Reform in Germany," IMF Working Paper No.13/42 (Washington: International Monetary Fund). Lalive, Rafael, Jan C. van Ours, and Josef Zweimuller, 2005, "The Effect of Benefit Sanctions on the Duration of Unemployment," Journal of the European Economic Association, Vol. 3, pp. 1386–1417. Manoli, Day, and Nicholas Turner, 2018, “Cash-on-Hand & College Enrollment: Evidence from Population Tax Data and the Earned Income Tax Credit,” American Economic Journal: Economic Policy, Vol.10, No. 2, pp. 242–71. Meghir, Costas and David Phillips, 2010, “Chapter 3: Labour Supply and Taxes,” in Dimensions of Tax Design: The Mirrlees Review, ed. by Stuart Adam, Timothy Besley, Richard Blundell, Stephen Bond, Robert CHote, Malcolm Gammie, Paul Johnson, Gareth Myles, and James Poter (Oxford University Press). Organization for Economic Co-operation and Development (OECD), 1997, The OECD Jobs Strategy: Making Work Pay: Taxation, Benefits, Employment and Unemployment, ISBN 978-9264156661 (Paris: OECD Publications). ______, 2019, “OECD Economic Surveys: Italy” (Paris: OECD Publications). ______, 2021, “Scaling Policies that Connect People with Jobs in the Recovery from COVID-19,” OECD Policy Responses to Coronavirus (COVID-19) (Paris: OECD Publications). Office for Budget Responsibility (OBR), 2018, “Office for Budget Responsibility: Welfare Trends Report, London–January 2018,” Presented to Parliament by the Exchequer Secretary to the Treasury by Command of Her Majesty: Office for Budget Responsibility (UK: APS Group on behalf of the Controller of Her Majesty’s Stationery Office). Pena-Casas, Ramon, Dalila Ghailani, Sebastiano Sabato, and Ides Nicaise, 2013, "Towards a European Minimum Income," European Economic and Social Committee Report No. EESC/COMM/03/2013 (Brussels: European Union). Rothstein, Jesse, 2008, “The Unintended Consequences of Encouraging Work: Tax Incidence and the EITC,” Working Papers 1049 (Princeton University, Department of Economics, and Center for Economic Policy Studies, CEPS). Saez, Emmanuel, 2002, “Optimal Income Transfer Programs: Intensive Versus Extensive Labor Supply Responses,” The Quarterly Journal of Economics, Vol. 117, No. 3, pp. 1039–73 Van Lancker, Anne and Fintan Farrell, 2018, “Guaranteed Minimum Income: Nobody Deserves Less, Everyone Benefits,” EMIN2 Final Report (Brussels: European Minimum Income Network, EMIN). 40 Vardarmatou, Dina and Ioanna Pertsinidou, 2018, “EMIN Context Report: Greece-Developments in Relation to Minimum Income Schemes” (Brussels: European Minimum Income Network, EMIN).