71903 Labor Markets Social Protection & Labor Policy Note June 2012 | Number 14 Highlight Activation Policies in This note gives a brief summary of activation OECD Countries policies in OECD An Overview of Current Approaches countries, where this type of policy approach Herwig Immervoll has some tradition and has been in use for some time. While countries Introduction share the objective of strengthening This note summarizes activation policy strategies in OECD countries.1 Pol- employment and icy makers can choose from a broad menu of options. Each of them comes with potential strengths and weaknesses, but there can also be important reducing dependency links between them. To understand countries’ overall activation stance, it and vulnerability, is useful to look across individual policy measures. For the purpose of this the balance of policy note, activation is therefore defined broadly as a combination of policy tools measures differs widely. that support and incentivize ■■ job search and job finding, ■■ productive participation in society, and ■■ becoming and remaining self-sufficient and less dependent on public support. The menu of policy instruments includes (i) measures that aim at strengthening the motivation for making use of existing earnings opportunities (e.g., work incentives, job- search requirements, benefit sanctions); (ii) training and related active labor market pro- grams (ALMPs) that seek to improve the capabilities of jobseekers and other activation “clients�; and (iii) employment services and programs that support labor demand by expanding the set of earnings opportunities available to jobseekers (e.g., wage subsidies, direct job creation). In high-income and developing countries alike, there is increasing demand for well- designed “active� or “activating� social and labor market policies. However, as a policy approach, activation has a longer tradition in the OECD area. Here, early activation mea- sures have tended to focus on specific groups facing labor-market difficulties in a context of relatively strong overall labor-market performance (such as low-skilled urban youth). More recently, activation principles have been applied more broadly, to a range of different 1 The scope is limited to prime-age recipients of unemployment and minimum-income benefits. Policies for the disabled, for youth and for older jobless people raise additional issues. Readers are referred to www.oecd. org/els/disability, to OECD (2010), Off to a Good Start? Jobs for Youth, and to OECD (2006), Live Longer, Work Longer. 1 Policy Note: Labor Markets June 2012 | Number 14 population groups, and increasingly in a co-ordinated manner across policy domains. Box 1. How important are financial work incentives for employment outcomes? In part, the current interest stems from a largely cyclical, but nonetheless persisting, labor-market weakness. With gov- Although results are not available for all OECD countries, ernment budgets under stress, employment and earnings there exists a wealth of empirical evidence on the growth are essential, both for reducing benefit spending, and sensitivity of people’s employment decisions to financial for shoring up government revenues now and in the longer work incentives, such as the net income gain of working term. In particular, patterns of benefit recipiency following one hour more or of working at all.a Among the main earlier labor market downturns in the OECD area show that findings are the following: expanding “inactive� benefit programs, such as disability or early retirement benefits, is very costly in the longer term. ■■ Financial incentives affect overall labor supply main- But it is also very difficult to cut social expenditures without ly through their influence on labor force participa- further increasing inequality. The downturn has pushed up the tion (the decision whether or not to work), while the numbers of people living in “low-work intensity� households hours worked in a job are typically less responsive; requiring support, and this has heightened concerns about inequality trends and the situation of the poorest families in ■■ Low-income groups and lone parents react strongly particular. In restoring incomes at the bottom, a key policy to financial incentives; and challenge is therefore to facilitate employment and earnings ■■ Labour supply is more responsive for women than growth that benefits low-income groups. In addition to crisis- for men. related factors, longer-term trends also point to a greater role for activation policies. In most of the OECD area, population ageing and the associated pressures on social expenditure These results provide essential guidance for targeting budgets make active labor-market participation of working- make-work-pay policies (e.g., for a given amount spent age individuals a necessary ingredient of policies that seek to on in-work benefits, targeting these resources on women restore the fiscal sustainability of social protection systems. and low-income groups, especially when children are present, is likely to create the biggest payoff in terms of The note considers benefit generosity and work incentives stronger employment and higher earnings). But while (Section 2); eligibility conditions and obligations of job seek- some general patterns can be identified, there are large ers and benefit recipients (Section 3); and ALMPs (Section 4). country differences. One study reports that single women Section 5 characterizes aspects of countries’ overall policy in Hungary and Poland are only about 1/4 as responsive to stance by discussing the balance of policies and links between incentives as women in Ireland and the United Kingdom.b them. A concluding section suggests priorities for knowledge Concerns about work disincentives can therefore be well work in World Bank client countries. justified in some countries and for some groups. In others, different employment barriers may be more pressing. Financial work incentives: Benefit a Surveys are provided by Evers, de Mooij and van Vuuren, (2008), “The generosity and make-work-pay policies Wage Elasticity of Labour Supply: A Synthesis of Empirical Estimates�, De Economist, 156 (1), and by Immervoll, Kleven, Kreiner and Saez (2007), “Welfare Reform in European Countries: A Micro-simulation A frequent starting point for debates on activation is a Analysis�, Economic Journal, 117 (517). concern about adverse work incentives that may result if b Bargain, O., A. Peichl, (2011b), “Labor Supply Elasticities in Europe out-of-work benefits are “close� to the income that someone and the US�, IZA Discussion Paper No. 5820. might otherwise earn in the labor market (Box 1 summarizes available evidence on the relevance of work incentives for people’s labor-market decisions). Benefit generosity, as well as stages during an unemployment spell (not shown here; see the tax treatment of in-work and out-of-work incomes indeed www.oecd.org/els/social/workincentives). This shows that, in vary widely across countries. close to half of OECD countries, those who lost their job early on during the financial crisis and were entitled to UB had Unemployment benefits NRR above 60% in the first year of unemployment. For those entitled to unemployment benefits (UB), a simple But NRR typically decline during the unemployment spell. way of summarising many of the relevant policy parameters Long-term unemployed in Japan, Italy, Korea and Turkey is by means of benefit net replacement rates (NRR), which lose their entire UB after 12 months or less (prior to crisis- express net income of a beneficiary as percentages of net related benefit extensions, insurance benefits in the United income in a previous job. The OECD reports NRR at different States also expired after 26 weeks in most states). In several 2 Social Protection & Labor | Human Development Network | The World Bank Figure 1. Social safety nets are important complements to unemployment benefits Net replacement rates (NRR) resulting from 2009 policy parameters, in percent Unemployment bene�ts only Cash housing-related bene�ts Cash social assistance 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Notes: Replacement rates are an average over five years of unemployment. They are calculated for a prime-age worker with a “long� and uninterrupted employment record. They are averages over four stylized family types (singles, one-earner couples, with and without children) and two earnings levels (67%, 100% of average full-time wage) and consider cash incomes, income taxes and mandatory social contributions paid by employees. Severance payments are excluded. “Social assistance� refers to cash minimum-income transfers. In the United States it also includes the value of a near-cash benefit (“Food Stamps�). Housing-related benefits are those available for rented accommodation. In some countries, housing-related support is covered by social assistance instead. For details, see www.oecd.org/els/social/workincentives. Source: OECD tax-benefit models. other countries, families with children keep a small amount Make-work-pay policies of family benefits when their unemployment support expires in the second year of unemployment. Australia, Ireland, New The twin objectives of supporting the living standards of low- Zealand, the United Kingdom operate unlimited means-tested income families, and encouraging economic self-sufficiency unemployment assistance benefits, resulting in a constant can be in conflict with one another. OECD countries have in- NRR over time. creasingly introduced measures aimed at improving the terms of this trade-off by accentuating financial work incentives Other financial support for low-income groups while maintaining adequate support for those with no or very low earnings. Today, practically all OECD countries operate Housing benefits, as well as social assistance and other ben- “make-work-pay� policies of one form or another: 2 efits of “last resort� provide income top-up for those with low incomes, and a fall-back option for people not (or no longer) ■■ Most countries operate gradual benefit phase-outs for receiving unemployment compensation. Figure 1 shows individuals with low (“marginal�) earnings (e.g., those average NRR during a long unemployment spell for someone working a few hours while looking for a better-paying receiving both unemployment and safety-net benefits accord- job). Disregarding a certain amount of earnings in the ing to formal legal rules. However, in practice, safety nets are income test, or ignoring marginal working hours in often poorly targeted and typically reach only a fraction of the eligibility test of out-of-work benefits, can provide low-income families who would be formally entitled. strong incentives to supplement benefit income with a 2 See Immervoll, H. and M. Pearson (2009), “A good time for making work pay? Taking stock of in-work benefits and related measures across the OECD�, IZA Policy Paper No. 3. 3 Policy Note: Labor Markets June 2012 | Number 14 small amount of earnings and to seek or maintain at least with benefit generosity as central determining factor of people’s some link with the labor market. But, often, steep benefit efforts and behavior. In reality, benefit receipt is not simply a phase-outs for non-marginal work, combined with tax or choice but is associated with more or less well defined—and social contribution burdens can mean that earning more more or less demanding—eligibility conditions. Some of these “does not pay� (e.g., net incomes of a German lone parent conditions exclude certain individuals from the group of are about the same whether she earns 10% or 66% of the benefit claimants at the outset (these are sometimes referred to average wage). Weak incentives to increase employment as entitlement conditions). For instance, individuals with short incomes further can give rise to a low-pay trap. or interrupted employment histories may not qualify for UB, while those with assets may be excluded from receiving means- ■■ Partly to address these problems, an increasing number tested benefits. of countries are operating employment-conditional (or “in-work�) benefits or tax credits for non-marginal work- As part of a “mutual obligations� or “rights and responsibilities� ers. These support measures reduce net tax burdens below strategy, those entitled in principle typically have to comply zero for some groups (i.e., the benefit or tax credit exceeds with specific behavioral requirements, notably active job- tax/contribution burdens). Depending on how they are search, which can make continued benefit receipt costly for targeted, they can help to reduce the extent of in-work individuals who are not genuinely seeking to overcome benefit poverty. Because the strengthened work incentives also dependency. Well-defined eligibility conditions can thus ease reduce the number of people claiming out-of-work benefits, trade-offs between adequate out-of-work benefits and fostering in-work support is a potentially cost-effective redistribution a back-to-work culture, so that strengthening out-of-work measure. Some countries reduce costs further by paying support does not need to translate into reduced job-search in-work benefits only during an initial period following efforts.3 new employment in a qualifying job (Australia, Belgium, Canada, Ireland, Japan, Korea and Slovakia). The importance of eligibility conditions becomes clear when considering how many unemployed actually receive benefits. A recent study reports that more than 70% of unemployed Eligibility criteria and obligations of job Germans and Belgians were in receipt of UB in the mid-2000s. seekers and benefit recipients In most countries, coverage was much lower: Fewer than 20% received UB in Italy, Slovakia, Poland, Greece, Estonia and Simple “textbook� economic models of labor supply and job the United States.4 Prior to the mid-2000s, coverage rates have search regard out-of-work benefit levels as a de-facto wage floor, dropped in two thirds of countries in that study. There can be Figure 2. Unemployment benefits: Strictness of formal eligibility rules, 2011 5 Bene�t sancƟons Monitoring 4.5 Job-search and availability EnƟtlement condiƟons 4 3.5 3 2.5 2 1.5 1 0.5 0 Luxembourg Germany Israel Sweden Greece Portugal Netherlands Slovenia Canada New Zealand Switzerland Austria Ireland Romania Slovak Republic Japan Belgium Australia Korea Malta Hungary United States United Kingdom Lithuania Denmark Norway Spain Cyprus Czech Republic France Turkey Estonia Poland Finland Bulgaria Italy Note: Higher scores for more demanding eligibility criteria. See text for definitions, measurement and calculation details. Source: Adapted from Venn, D, (2011), “Eligibility Criteria for Unemployment Benefits�, OECD Social, Employment and Migration WP No. 131. 3 See Fredriksson, P. and B. Holmlund (2006), “Improving Incentives in Unemployment Insurance: A Review of Recent Research�, Journal of Economic Surveys, 20 (3), for a survey of theoretical models of job search. 4 Immervoll, H. and L. Richardson (2011), “Redistribution policy and inequality reduction in OECD countries. What has changed in two decades?�, OECD Social, Employment and Migration Working Paper No. 122. Results refer to UB. Barriers to claiming benefits can be greater still for potential claimants of minimum- income benefits. Studies on benefit take-up regularly find very high non-take-up rates for means-tested benefits in the order of 40% or more. 4 Social Protection & Labor | Human Development Network | The World Bank different reasons for these trends. Reduced employment stabil- criteria (e.g., Portugal, Romania, Slovakia, Slovenia) tend to ity is a likely driver of falling benefit coverage. But in addition, have strict rules in all four categories. Yet, overall, there is some of the countries pursuing an activation agenda have little cross-country correlation between different aspects of tightened conditions, reduced benefit durations, or introduced strictness (e.g., countries do not appear to trade off individual more demanding behavioral requirements. aspects of eligibility strictness against each other). How demanding are eligibility criteria for unemployment Formal eligibility criteria are clearly important – for job seek- benefits? ers, as well as for understanding countries’ activation stance. Ultimately, however, the significance of formal rules depends The OECD and others have argued that job-search incentives on how they are implemented in practice. The formal strict- and requirements can have a significant effect on aggregate ness of relevant legislation can therefore be an unreliable unemployment.5 This is consistent with large (25-50%) benefit guide to its actual impact on job search behavior. Importantly, caseload reductions following welfare-to-work type reforms resources available for labor market policies determine how in the United States, Denmark, Ireland, the Netherlands and well benefit agencies and employment services can carry out the United Kingdom. To assess countries’ policies in this area, their legal monitoring and support mandates. When there are a joint project between the OECD and the European Com- capacity constraints, redesigning formal rules may produce mission has collected qualitative information on the strictness little effect in practice (as agencies may not be able to main- of eligibility criteria for UB in 36 OECD and EU member tain effective contact with clients or are otherwise unable to countries.6 The study covered four categories of eligibility monitor job-search activity in a suitable way). “strictness�, with one or more sub-categories: In addition, legal rules may be formulated in very general 1. Entitlement conditions terms, meaning that case workers can (or have to) use discre- tion when applying them to individual clients. This applies a. employment/contribution requirements even more strongly in the case of lower-tier out-of-work benefits (such as minimum-income transfers). Typically, these b. sanctions for “voluntary� unemployment benefit programs are the responsibility of regions or munici- palities and with often strongly divergent implementation 2. Job-search requirements practices. These safety-net benefits are also frequently subject to budgetary limits; not everyone who is technically eligible a. Availability requirements during ALMP participation may then receive them.8 b-d. Suitability of job offers: Required occupational, geo- graphic and other mobility Active labor market programs 3. Monitoring of job-search activities (proof of job-search Overall spending levels activity) Like government expenditure on “passive� income support, 4. Sanctions for failing to comply with behavioral require- aggregate spending on ALMPs can be strongly counter-cy- ments clical. However, in the past, spending per unemployed person has in fact often declined very significantly as unemploy- a. Sanctions for refusing to accept a suitable job or par- ment rose during downturns (Figure 3, upper panel).9 When ticipation in ALMP unemployment goes up more quickly than ALMP spending, there is a risk that it becomes more difficult to service job b. Sanctions for repeated refusal of job or labor market seekers effectively. During periods of extended labor-market program weakness, independent job-search becomes more difficult and job-seekers may come to depend more heavily on job-search Figure 2 reports the overall strictness score derived in that assistance and labor market programs. study.7 In 2011, formal eligibility conditions for UB were least strict in Sweden, Canada, Austria, Cyprus and Finland. It is Nordic countries show the highest spending levels, but, as notable that countries with the most demanding eligibility a group, Continental European countries now spend only 5 E.g., OECD (2006a), “General policies to improve employment opportunities for all�, OECD Employment Outlook. 6 Venn, D, (2011), “Eligibility Criteria for Unemployment Benefits�, OECD Social, Employment and Migration WP No. 131. 7 The qualitative information on each of those sub-categories was translated into scores ranging from one (least strict) to five. Scores were then combined into a composite indicator using equal weights for each of the main headings and sub-categories. 8 Lower-tier minimum-income programs are therefore not considered in the overview of formal eligibility rules presented in the EC/OECD study. A recent summary of formal eligibility rules is given by Immervoll, H., (2012), “Minimum-income Benefits in OECD Countries: Policies and Challenges�, in D. Besharov and K. Couch (eds.), Measuring Poverty, Income Inequality, and Social Exclusion. Lessons from Europe. Oxford University Press. 9 During the past decade, two other OECD members—Japan and Korea—have spent less than English-speaking countries. Japan and Korea were, however, able to scale up spending per unemployed during the most recent economic crisis. 5 Policy Note: Labor Markets June 2012 | Number 14 Figure 3: Spending on active labor market programs per unemployed person, in percent of GDP per capita 80 70 OECD Europe, Nordic 60 50 OECD Europe, ConƟnental 40 30 20 OECD Europe, Anglophone 10 OECD Europe, East 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 1992 1993 1994 1995 1996 1997 1998 1999 2000 1985 1986 1987 1988 1989 1990 1991 Spending by policy area prior to the economic downturn OECD All PES OECD-Europe East Training Employment incentives OECD English speaking Disabled OECD-Europe Continental Direct job creation OECD-Europe Nordic Start-up incentives 0% 20% 40% 60% 80% 100% Note: See Statistical Annex of the OECD Employment Outlook and for definitions and limitations. Source: Author calculations based on OECD Labor Market Program Database and OECD Labor Force Statistics; Grubb, D. and A. Puymoyen (2008), “Long time series for public expenditure on labour market programmes“, OECD Social, Employment and Migration WP No. 73. slightly less. Per-client expenditures in English-speaking and expenditures go towards programs that seek to address Eastern European OECD countries are much smaller, but specific employability issues (Figure 3, lower panel). In with opposite trends in recent years (increasing spending in countries with comparatively high overall spending, core PES Eastern Europe). functions (including job-matching services delivered by pri- vate providers but with public financing) typically account for What types of labor market programs do countries prioritise? less than 25% of total ALMP resources. In English-speaking Composition of ALMP spending countries, PES spending shares have, however, gone up as total ALMP expenditures declined (labor market programs The core matching functions of the PES (“job brokering�: drove falling ALMP spending levels, while core PES functions placement and job-search assistance) form a fundamental were apparently largely protected from spending cuts). pillar of active labor market policy. Yet, most of public ALMP 6 Social Protection & Labor | Human Development Network | The World Bank Other than PES, training is the largest spending category time to materialize. They also tend to be small or insignificant in most countries, accounting to one fourth to one third of for men and for basic education. On-the-job training appears total spending. However, in Eastern Europe, just over 10% to perform best. Public-sector job creation programs have of ALMP expenditures go towards training measures and mostly produced negative labor-market effects (based on as- English-speaking countries now spend much less on training sessments after the program has ended). Evidence on private- than they did in 1990. Across the OECD, job subsidies and sector employment subsidies does not give a strong indication other demand-side employment incentives have grow more either way. Results are also inconclusive for self-employment important as a spending item, while direct job creation (e.g., support and start-up. public works programs) have seen declining expenditure shares. Start-up incentives for new businesses are only a An important consideration when interpreting impact evalu- small part of total spending (but have been very sizable ations is that programs will often impact outcomes other during the early transition phase in Eastern Europe). Finally, than employment, e.g., incomes of ‘sanctioned’ individuals spending under the “disability� category (including supported and their families. There is still very limited information on employment and rehabilitation) has grown, reflecting a the distributional effects of different activation approaches. growing awareness of sickness and disability as a major labor- Careful distributional assessments should be a crucial element market issue. of policy evaluations as activation measures typically affect disadvantaged groups. Do active labor market programs “work�? One general finding of existing studies in OECD countries is Overall policy stance on activation that the impact varies considerable between type of pro- gram.10 The meta-study by David Card and colleagues reports To characterise countries’ overall activation stance, it is useful positive short-term effects (12 months after start of program) to examine the balance of the different measures (ALMPs as on employment for less than half of the studies they review well as financial incentives, eligibility criteria and job-seeker (30% find negative employment effects, and a further 30% obligations). show no discernable effect). However, there is some evidence that outcomes “get better� when assessed over a longer time Figure 4 shows two scatter plots of indicators presented period. Training measures are generally found to have a earlier. Taken at face value, there is no indication that benefit positive impact on employment, and positive effects are not eligibility is generally stricter in countries with generous out- evident immediately after a training program, but take some of-work benefits (left-hand panel). Eligibility is strict in some Figure 4. Balance between different activation policy measures 70 4.5 PRT NLD ROM SVK 60 ALMP spending per unemployed Strictness of bene�t eligibility 4 SLO MLT 50 DNK % of GDP per capit TUR ITA EST ESP 3.5 CHE LUX 40 BEL CZE LTU IRL POL AUT GBR NOR POL USA AUS 30 SWE 3 KOR NZL BEL NOR DEU GRC HUN NLD FRA ISR DEU KOR CHE DNK BGR FRA JPN 20 FIN AUT FIN JPN PRT 2.5 CAN SWE HUN LUX IRL 10 AUS NZL GBR ESP CZE CAN USA SVK R² = 0.42 2 0 0 20 40 60 80 25 35 45 55 65 75 85 Bene�t generosity (net replacement rate) Bene�t generosity (net replacement rate) Source: Figures 1, 2, and 3. 10 Card, D., J. Kluve and A. Weber, (2010), “Active labour market policy evaluations: A meta-analysis�, Economic Journal, 120; Greenberg, D. H., C. Michalopoulos and P. K. Robins (2003), “A meta-analysis of government-sponsored training programs“, Industrial and Labor Relations Review, 57; OECD (2005), “Labour market programmes and activation strategies: Evaluating the impacts�, OECD Employment Outlook. 7 Policy Note: Labor Markets June 2012 | Number 14 countries with high NRR (Slovenia, Portugal), but others with have in fact failed to prove effective or cost efficient. Systemat- even more generous out-of-work benefits have fairly light ic evaluation and sustained policy commitment are therefore eligibility requirements (Nordic countries, Japan). Benefits in needed in order to decide which policies should be expanded, Turkey, Italy, Estonia and Slovakia (as well as in non-OECD adapted or ended, and which individual programs might work Romania) are both low and subject to demanding eligibility best if employed as a package in combination with others. conditions. Overall, there is no correlation: Some countries may seek to offset possible work disincentives of generous The impact of individual programs on aggregate employment benefits (the clients’ “rights�) with relatively demanding client or beneficiary caseloads is often fairly modest. This is not responsibilities. In others, benefits are generous (or tight) both necessarily surprising and need not be discouraging given in terms of benefit amounts and in terms of client beneficiary that most measures are highly targeted, investment of public obligations. resources is limited, and program durations are short. But it also suggests that there is scope for better co-ordination By contrast, there is an apparent link between ALMP spend- between policy domains. For instance, job-search incentives ing and NRR across countries (right-hand panel). Active and requirements could be closely co-ordinated with job-search policy “effort� (ALMP spending), tends to be greater in coun- assistance and financial incentives and benefit conditionality tries where benefits are generous (in part, this is driven by can increase participation in employment services. To the Denmark and the Netherlands, but the link remains signifi- extent that these services are effective, this should result in cant when these countries are excluded). better employment prospects, and a possible virtuous cycle of reduced caseloads and improving service quality. Conclusion But institutional contexts, labor-market conditions and potential client groups differ across countries – and these Although knowledge gaps remain, the experience in the differences can be very large between high, middle and lower- OECD area provides useful pointers for World Bank cli- income countries. Knowledge exchange and peer learning are ent countries seeking to strengthen activation policies. One more likely to translate into directly useful policy insights if conclusion from a review of existing evidence is that well- they are anchored in a solid understanding of domestic policy designed policies can have a positive impact on employment challenges and capacities. outcomes for participants, but that many existing policies The findings, interpretations, and conclusions expressed herein are those of the author(s), and do not necessarily reflect the views of the International Bank for Reconstruction and Development/The World Bank and its affiliated organizations, or those of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. For more 8 information, please visit www.worldbank.org/sp.