Activation and employment support policies in OECD countries. An overview of current approaches

This note examines the balance of activation strategies in OECD countries, where this type of policy approach has a long tradition. Countries share the objective of strengthening employment and reducing benefit dependency and vulnerability among the working-age population, but the balance of policy measures differs widely. While debates on the effectiveness of active labour market policies have tended to focus on individual policy measures, this paper emphasizes the links between policy areas. We argue that an effective policy strategy rests on a finely tuned balance between income support, work incentives, as well as mutual obligations and active labour market programmes. Jel codes: H31, H53, J08, J68


Introduction
Over the past decade, a number of OECD countries have introduced or reinforced strategies to "activate" the unemployed and other benefit recipients, hereafter called "activation strategies". These strategies varied significantly across the board, depending on underlying labour market conditions and the capacity of public employment services. However, the main features of the activation strategies include the enforcement of strict eligibility criteria for benefit recipients together with the provision of effective re-employment services. At the same time, governments have made efforts to encourage economic self-sufficiency while supporting the living standards of low-income families, by providing different forms of in-work benefits.
Evidence suggests that, if well-designed, such strategies have contributed to better labour market outcomes, by ensuring that benefit recipients have a better chance of obtaining employment and minimising the risks that high and/or long-lasting benefits significantly damage work incentives (Reassessment of the OECD Jobs Strategy in OECD, 2006a;OECD 2005).
countries. 1 They comprise measures that: i. Strengthen people's motivation to look for and make use of existing earnings opportunities (e.g., work incentives, job-search requirements and benefit sanctions); While activation strategies were generally launched in the OECD countries in the 1990s with the aim of combating high and often persistent unemployment, their effectiveness in mobilising the unemployed and other benefit recipients was evaluated in particular in the decade prior to the Great Recession of 2007-08a period in which many countries enjoyed relatively strong economic growth and buoyant labour demand and thus when the number of jobseekers was low relative to the number of vacancies. Early activation measures focused on specific groups facing labour-market difficulties in a context of relatively strong overall labour-market performance (such as low-skilled urban youth). More recently, activation principles have been extended to cover a wider range of population groups, and increasingly in a co-ordinated manner across policy domains.
With a rapidly growing number of jobseekers and low labour demand, the recent financial and economic crisis has exposed activation strategies to a hard test. Reemployment services and other active labour market programmes, (ALMPs, henceforth) had to be scaled up to provide support to the greater number of jobseekers; and activation procedures had to be adapted to ensure that jobseekers receive the appropriate supports. While resources for ALMPs have increased in response to the crisis, often more than in previous major downturns (OECD 2012), spending per unemployed has generally declined significantly. This has meant fewer resources available to support jobseekers at a time when their chances of quick reintegration into work were weaker given the limited number of vacancies and high competition among the unemployed.
In this context, policy-makers are facing daunting trade-offs. On the one hand, fiscal consolidation efforts require containing public spending, including for labour market programmes. On the other hand, early successes with activation strategies suggest that they can contribute significantly to mobilise benefit recipients back into employment, which is key both for reducing benefit spending, and for shoring up government revenues now and in the longer term. In addition, the downturn has pushed up the numbers of people living in "low-work intensity" households requiring support, and this has heightened concerns about inequality trends and the situation of the poorest families in particular. In restoring incomes at the bottom, a key policy challenge is therefore to facilitate employment and earnings growth that benefits low-income groups. Moreover, 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 budgets make active labour-market participation of working-age individuals a necessary ingredient of policies that seek to restore the fiscal sustainability of social protection systems.
The paper reviews different elements of the activation strategies adopted by OECD countries and discusses what works under different underlying labour market conditions. After a brief review of key features of activation strategies (Section 2), the paper considers benefit generosity and work incentives (Section 3); eligibility conditions and obligations of job seekers and benefit recipients (Section 4); and ALMPs (Section 5). Section 6 characterizes aspects of countries' overall policy stance by discussing the balance of policies and links between them.

Key features of activation strategies
Activation strategies are generally implemented at the local level by the Public Employment Service (PES), 2 in some cases with the support of private providers of job placement and training services. Actions are generally targeted at people of working age who are not in work, but who could potentially work and are in receipt of an income-replacement benefit, usually an unemployment benefit but sometimes a social assistance, single parent or disability benefit. They make benefit payments conditional on compliance with employment-and job-search related requirements.
Different specific actions are included in the activation strategies depending on labour market conditions but also the capacity of the PES and other service providers. As part of job-search assistance and monitoring, most OECD countries follow a practice of intensive obligatory interviews between the jobseeker and an employment counsellor. However, the frequency of such interviews varies, with five OECD requiring an interview on a monthly basis or less (Czech Republic, France, Korea, Slovak Republic and the UK). Beneficiaries are also required to report regularly on their job-search effort, while the PES refers unemployed clients to vacant jobs. The PES also sets up back-to-work agreements or individual action plans as the joblessness spell lengthens and offer tailored re-employment programmes to prevent loss of motivation, skills and employability. In a few countries -Australia, Denmark, Sweden and the UKparticipation in re-employment programmes is compulsory after a period of unsuccessful job search to reduce the risk of either long-term unemployment or labour market exit.
The impact of these strategies on unemployment arises partly because they ensure that jobseekers participate in employment-related services, and partly because the participation requirements and monitoring of compliance with eligibility conditions, backed up by the threat of temporary financial sanctions, counteract potential disincentives from the payment of benefits.
Activation strategies were primarily developed for people receiving an incomereplacement benefit, to enforce a principle of "mutual obligation" and contain the risks of benefit dependency, loss of human capital among the long-term unemployed and, ultimately, higher public spending on labour market and social policies. Elements of these strategies typically originated in countries with longer-lasting or indefiniteduration insurance or assistance benefits for job seekers. By contrast, countries that pay relatively short-duration UI benefits often associate these with only relatively light activation measures. Some social assistance (or minimum-income) benefit systems distinguish between employable and non-employable recipients. For the former group, work-availability requirements apply. Moreover, in a number of countries workers in low-paid jobs receive in-work benefits, or social assistance benefits at a reduced rate, depending on benefit system parameters. To the extent this occurs, some activation elementswith participation in counselling, training, applying for full-time work, etc. potentially being requiredmay still apply. Finally, employment services can also be made available to jobseekers who do not receive financial support. For these policy clients, enforcing jobsearch and work-availability is more difficult, but arguably also less relevant as those participating voluntarily are likely to be motivated to look for a employment.
Over the past few decades many countries have experienced a sharp increase in recipiency rates for incapacity benefits, lone-parent benefits, non-categorised social assistance benefits or other types of out-of-work benefit. While recipients of such benefits may face specific employment barriers, many of them can work, and they have gradually become another target group for labour market policies. 3 Activation strategies for labour-market inactive benefit recipients may involve making benefits for a particular target group conditional on immediate availability for work and acceptance of referrals to job vacancies, either by transferring them to an existing unemployment benefit or converting their benefit de facto into a type of unemployment benefit. For those with limited work capacity, requirements can include participation in "work preparation" or rehabilitation measures.

Financial work incentives: benefit generosity and make-work-pay policies
A key factor justifying well-designed activation strategies or "back-to-work" policies is the concern about adverse work incentives that may result if out-of-work benefits are "close" to the income a benefit recipient might otherwise earn in the labour market.
Benefit generosity, as well as the tax treatment of in-work and out-of-work incomes indeed vary widely across countries. It is therefore useful to consider the extent of adverse work incentives in some detail (summary tables of the main policy features of unemployment and other benefits can be accessed through www.oecd.org/els/social/ workincentives).

Unemployment benefits
For those entitled to unemployment benefits, one simple way of summarising many of the relevant policy parameters is by means of benefit replacement rates, which express net income of a beneficiary as percentages of net income in a previous job. Table 1 shows benefit replacement rates for different stages of the unemployment spell for prime-age individuals. 4 Results are averages over different earnings levels and family situations and account for taxes and for family-related benefits that are typically available.
During the first year of unemployment, prime-age workers entitled to unemployment benefits had net incomes above 60% in just under half of the countries. Income losses during the first year were smallest in Nordic countries and in continental Europe. On the other end of the spectrum, unemployed entitled to benefits but with no other support in Czech Republic, Korea, Slovak Republic faced income losses of more than 60% during the first year of unemployment.
In countries operating insurance benefits, net replacement rates typically decline during the unemployment spell. For instance, prime-age long-term unemployed in Japan, Italy, Korea, Turkey and several other countries lose their entire unemployment benefit after 12 months or less (prior to recent crisis-related extensions of benefit duration, unemployment insurance benefits in the United States also expired after 26 Net replacement rates in percent, 2010 policy parameters a . a. Countries shown in descending order of the five-year average. Calculations consider cash incomes (including family benefit, which may be available regardless of employment status or income) as well as income taxes and mandatory social security contributions payable by workers. To focus on the role of unemployment benefits, no social assistance or housing-related benefits are considered (see Figure 1 for "all-in" replacement rates including these safety-net benefits) and any entitlements to severance payments are also excluded. Net replacement rates are for a prime-age worker (aged 40) with a "long" and uninterrupted employment record and are averages over 12 months, four different stylised family types (single and one-earner couples, with and without children) and two earnings levels (67% and 100% of average fulltime wage). b. Excluding the retroactive extension of unemployment benefits from three to four years which was passed in December 2010. Source: OECD tax-benefit models (www.oecd.org/els/social/workincentives).
weeks in most states). In several other countries, unemployment benefits are also no longer payable in the second year of unemployment, although families with children can be entitled to child benefits and other family support payments, which maintain a small amount of income for those without any other support. In a number of countries, meanstested unemployment assistance provides continued (and usually lower) benefit entitlements once insurance benefits expire (Austria, Finland, France, Germany, Greece, Portugal, Spain) and four English-speaking countries operate unlimited means-tested unemployment assistance benefits (Australia, Ireland, New Zealand, the United Kingdom), resulting in a flat replacement-rate profile.
Other financial support for low-income groups Housing benefits, as well as social assistance and other benefits of last resort, are important additional components of the overall support package, especially for those running out of unemployment benefit entitlements. Cash rent assistance as well as minimum-income or "welfare" benefits provide a fall-back option for people not, or no longer, receiving unemployment compensation, and an income top-up for others with low incomes. This is illustrated in Figure 1, which shows replacement rates for someone actually receiving these safety-net benefits. But, as argued below, social safety nets are often relatively poorly targeted, in the sense that they reach only a small proportion of the low-income population.

Make-work-pay policies
The twin objectives of supporting the living standards of low-income families and encouraging economic self-sufficiency are often in conflict with one another, so trade-offs  Table 1 for calculation details. "Social assistance" refers to minimum-income transfers. In the United States it also includes the value of a nearcash benefit ("Food Stamps"). Housing-related cash benefits are those intended to cover rent, utilities and associated housing-related expenses. Benefits shown are the maximum amounts to which a family may typically be entitled if there is no other income. Source: OECD tax-benefit models (www.oecd.org/els/social/ workincentives).
have to be tackled. OECD countries have increasingly introduced measures aimed at improving the terms of this trade-off by accentuating financial work incentives while maintaining adequate support for those with no or very low earnings. Today, practically all OECD countries operate such "make-work-pay" policies of one form or another.
Most countries operate gradual benefit phase-outs for individuals who manage to earn only limited amounts (e.g., by working a few hours while looking for a higherpaying job). Disregarding a certain amount of earnings in the income test, or ignoring marginal working hours in the eligibility test of out-of-work benefits, can provide strong incentives to supplement benefit income with a small amount of earnings and to seek or maintain at least some link with the labour market, e.g., in the form of marginal employment of a few hours a week. But, often, they do not provide a genuine incentive to increase employment incomes further; steep benefit phase-outs further up the earnings scale, combined with relatively high tax or social contribution burdens for non-marginal workers can mean that earning more "does not pay" (for instance, OECD calculations indicate that the net income of a German single parent is about the same whether she earns 15% or 60% of the average wage). 5 In this way, it is possible that in-work benefits can act as a disincentive for low-wage workers to work longer hours or invest in their human capital.
Partly to address these problems, an increasing number of countries are considering employment-conditional (or "in-work") benefits or tax credits that support the incomes of workers in non-marginal employment. 6 A few countries "make work pay" by successfully keeping employees' income tax and/or contribution burdens low (most Anglo-Saxon countries, Korea, or the Slovak Republic). In-work benefits often go beyond that by reducing tax burdens below zero for some groups (i.e., the benefit or tax credit exceeds tax/contribution burdens). Depending on how in-work benefits are targeted, they can result in much-improved incentives for non-marginal employment for some groups. In some countries, "transitional" in-work benefits are paid only for a limited period following new employment in a qualifying job (e.g., Australia, Belgium, Canada, Ireland, Japan, Korea and the Slovak Republic).

How important are financial incentives for employment outcomes?
Even though results are not available for all OECD countries, there exists a relatively broad consensus among labour economists on the responsiveness of people's employment decisions to financial work incentives, such as the net income gain of working one hour more or of working at all. Among the main findings are the following: 7 Financial incentives affect overall labour supply mainly through their influence on labour force participation (i.e., the decision on whether or not to work), while changes in the number of hours worked are quantitatively less important; Low-income groups and lone parents react more strongly to financial incentives; and Labour supply is more responsive (or "elastic") for women than for men.
These results are important when considering the potential economic cost of reforming out-of-work support programs. They also provide essential guidance for targeting make-work-pay policies (e.g., for a given amount spent on in-work benefits, targeting these resources on women and low-income groups, especially when children are present, is likely to create the biggest payoff in terms of stronger employment and higher earnings).
But while some general patterns emerge from the international evidence, it is notable that there are widespread differences across countries. For instance, one of the few available cross-country empirical studies reports that single women in Hungary and Poland are only about 1/4 as responsive to financial incentives as in Ireland and the United Kingdom. 8 One important reason for these large country differences is that financial incentives may have limited effects on observed employment outcomes if other barriers prevent people from adjusting their labourmarket status or working hours. For instance, when involuntary unemployment is high during a downturn, many individuals who want to work cannot find a job, or they cannot work as many hours as they would like. Frictions in the labour market (e.g., due to poorly functioning public employment services) can have similar effects. On the other hand, policies that tie benefit receipt to job-search or participation in ALMPs can help to avoid negative employment effects that may result from unconditional out-of-work benefits.

Eligibility criteria and obligations of job seekers and benefit recipients
Simple "textbook" economic models of labour-supply decisions and job search consider out-of-work benefit levels as a de-facto wage floor. In these models, benefit replacement rates assume a central role since wage floors are the main factor determining people's decision to work and exert job-search efforts. In reality, benefit receipt is not simply a choice but is associated with more or less well definedand more or less demanding eligibility conditions. Some of these conditions exclude certain individuals from the group of potential benefit recipients altogether. These provisions, which are sometimes referred to as entitlement conditions, serve as an initial "filter" that targets support measures to certain groups. For instance, individuals with short or interrupted employment records may not be eligible for unemployment insurance benefits, while those with assets may not qualify for means-tested benefits.
In addition, those entitled to receive a benefit in principle may have to comply with specific behavioural requirements which are an integral part of activation strategies, namely, job-search activities, participating in interviews and ALMPs and accepting suitable job offers. These requirements tend to make continued benefit receipt costly for those who are not genuinely seeking to overcome benefit dependency; they act as a "stick" or "hassle" factor in pushing benefit recipients to meet their obligations.
Because of these costs of claiming benefits, the provision or strengthening of outof-work support does not necessarily have to translate into reduced job-search efforts. 9 Well-defined eligibility conditions can therefore help to ease any tradeoffs between adequate out-of-work benefits and maintaining strong labour-market performance.
The importance of entitlement conditions, and the costs associated with benefit receipt, becomes clearer when considering the share of jobseekers who actually receive benefits. Although survey questions differ and data are therefore not fully comparable across countries, labour force surveys do provide some evidence on benefit coverage among the unemployed. 10 Using these data, Figure 2 shows that more than 70% of unemployed Germans and Belgians were in receipt of unemployment benefits in the mid-2000s. But in most countries, coverage rates were much lower. In Italy, Slovakia, Poland, Greece, Estonia and the United States, fewer than one in five job-seekers (those reporting to be available for work and actively looking for a job) received unemployment benefits. 11 Over the time period shown, the shares of unemployed reporting benefit receipt have dropped in more than two thirds of countries, while only a few recorded significant increases. 12 There can be different reasons for these trends, and not all have to do with the strictness of entitlement or eligibility criteria. A closer analysis of the coverage data suggests that shortening average employment spells and reduced employment stability are likely drivers of falling benefit coverage. 13 In some but not all OECD countries, a growing incidence of temporary employment and other types of non-standard work have caused growing shares of workers to remain unprotected (coverage rates decline if entitlement conditions fail to adjust to accommodate rising numbers of non-standard workers). In addition some countries pursuing an activation agenda have tightened conditions, reduced benefit durations, or introduced more demanding behavioural requirements (Immervoll and Richardson, 2011).

How demanding are eligibility criteria for unemployment benefits?
The OECD and other analysts have consistently argued that job-search incentives and requirements can have a significant effect on aggregate unemployment, especially when benefit spells are long (e.g., OECD 2006b). This view is consistent, for example, with the large (25-50%) caseload reductions following welfare-to-work type reforms in the United States, Denmark, the Netherlands and the United Kingdom. To assess countries' policies in this area, this section reproduces selected results from a joint project between the OECD and the European Commission that collected qualitative information on the strictness of eligibility criteria for unemployment benefits for 36 OECD and EU member countries (Venn, 2012). The study covered four categories of eligibility "strictness", each with one or more sub-categories as follows: a. Sanctions for refusing to accept a suitable job or participation in mandatory labour market program b. Sanctions for repeated refusal of a job or the participation in a labour market program Figure 3 reports the overall strictness score. 14 In 2011, formal eligibility conditions for unemployment benefits were rated least strict in Sweden, Canada, Austria, Cyprus and Finland. It is notable that countries with the most demanding eligibility criteria (e.g., Portugal, Romania, Slovakia, Slovenia) tend to have strict rules in each of the four sub-categories. Yet, overall, Venn (2012) finds little cross-country correlation between the different aspects of eligibility strictness. For instance, the pattern across countries does not support the idea that countries trade off individual aspects of eligibility strictness against one another.
Formal eligibility criteria are clearly importantboth for job seekers, and for understanding countries' activation stance. Ultimately, however, the impact of formal rules strongly depends on how they are implemented in practice. For instance, Grubb (2000) argues that the formal strictness of legislation is often an unreliable guide as to its actual impact on job-search behavior: they can be formulated in very general terms and case managers in local PES offices may exercise considerable discretion when applying them to individual clients. 15 This applies even more strongly in the case of lower-tier out-of-work benefits (such as minimum-income transfers). 16 Capacity constraints are crucial as well. Redesigning formal rules may produce little effect in practice if public employment services (PES) do not have the capacity to manage frequent contact with benefit recipients, or are otherwise unable to monitor jobsearch activity in an effective way. More generally, the resources available for active labour market policies determine the extent to which benefit administrations and employment services are able to carry out their legal monitoring and support mandates.

Active labour market programmes
Overall spending levels Spending for active labour market programmes varies significantly across OECD countries, as does its composition across different measures. Spending as a share of GDP ranges from 0.15% to 0.20% in the US and Japan, respectively on the one end of the spectrum, to around 1% in France, Germany and 1.5% or more in the Nordic countries. As shown in Figure 4 for a selected group of countries, differences across countries in the level of spending for active labour market programmes are not closely related to the level of unemployment. While there is correlation between the spending on "passive" income supportas many job losers are entitled to unemployment benefits and other forms of income support-this is not the case for active spending, which in most countries is discretionary. 17 It is also noticeable that, while spending on income support is strongly counter-cyclical, spending on active programmes tends to react only moderately in most countries, with the notable exception of the Nordic countries ( Figure 5, upper panel). The lack of strong responsiveness means that, in most countries, spending per unemployed person tends to decline significantly as unemployment rises during cyclical downturns ( Figure 5, lower panel). 18 This implies that it becomes more difficult to service job seekers effectively. During periods of extended labour-market weakness, as in the weak recovery after the 2008-09 Great Recession, falling resources per jobseeker can be a particular concern as independent job-search is more difficult when unemployment is high. Jobseekers may therefore depend more heavily on job-search assistance and other labour market programmes.  Whether ALMP expenditures are assessed in aggregate or per unemployed, Nordic countries show the highest spending levels. Continental European countries have, however, narrowed the gap and, as a group, spend only slightly less. English-speaking as well as Eastern European OECD countries (Czech Republic, Hungary, Poland, Slovakia) devote the smallest amount of resources to ALMPs, but with opposing trends: spending per unemployed has gone up in the past 10 years, while it has declined in English-speaking countries.

What types of labour market programmes do countries prioritise?
The core matching functions of the PES ("job brokering": placement and job-search assistance) form a fundamental pillar of active labour market policy. Yet, Figure 6 shows that the majority of public ALMP expenditures goes towards programmes that seek to address specific employability issues. 19 In countries with comparatively high overall spending, core PES functions (including job-matching services delivered by private providers but with public financing) typically account for less than 25% of total ALMP resources, and just over 10% in the "Nordic" group. In English-speaking countries, PES spending shares have gone up as total expenditures have declined. In these countries, programs such as training and wage subsidies drove falling ALMP spending levels, while core PES functions were largely protected from spending cuts.
Other than the cost of running the PES, training is the largest spending category in most countries, accounting for one fourth to one third of total spending. However, in Eastern Europe, training measures account for only slightly more than 10% of ALMP expenditures, and English-speaking countries now spend much less on training than they did in 1990. Across the OECD area as a whole, job subsidies and other demandside employment incentives have tended to grow as a spending item, while direct job creation (e.g., public works programs) have seen declining expenditure shares. Start-up incentives for new businesses are only a small part of total spending (but werevery sizable during the early transition phase in Eastern Europe). Finally, increasing spending shares in the "disability" category (including supported employment and rehabilitation) reflect a growing awareness of sickness and disability as a major labour-market issue.

Do active labour market programmes "work"?
There exists a large evaluation literature for OECD countries, and a number of highquality reviews or meta-studies (e.g., Card et al., 2010;Greenberg et al., 2003;OECD, 2005). One general finding is that the impact varies substantially between different types of program but, perhaps more surprisingly, also between different studies looking at the same policy category. In part, this can be explained by different methodologies and outcome measures. But the impact heterogeneity remains even after controlling for the outcome measure. For instance, without distinguishing by type of program, Card et al. (2010) report statistically significant positive short-term effects on employment (12 months after start of program) for just over 40% of the close to 200 evaluation studies they review. Around 30% find significantly negative employment effects, and a further 30% show no discernable effect. The number of studies finding negative effects 24 months after program start is considerably smaller, but this result is less  1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 ALMPspending ( 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999   representative and powerful in a statistical sense as fewer studies look at medium-term outcomes beyond one year. There is less variation across groups of countries. On one hand, this would suggest that there may be no strong reason why ALMPs should work less well in some countries than in others. On the other hand, even in the OECD area, there remains a significant imbalance in terms of the countries covered in evaluation studies. For instance, of the estimates studied by Card et al., 23% are from Germany, 13% from Denmark, 10% from Sweden and 7% from France. As a result, there is no strong basis for making inferences about the effectiveness of programmes in countries that are not included, or under-represented, in the sample of available studies. Subject to these limitations, training measures are generally found to have a positive impact on employment. However, these are not evident immediately after a training program, but take time to manifest themselves. They also tend to be small or insignificant for men and for programmes delivering basic education, while on-the-job training appears to perform best. Public-sector job creation programmes have mostly produced negative labour-market effects (based on assessments after the program has ended). Evidence on private-sector employment subsidies does not give a strong indication either way, with some studies finding positive and others negative effects (displacement/substitution as well as deadweight effects can be very large, although appropriate targeting can reduce these costs). Results are also not conclusive for self-employment support and start-up incentives as there are few evaluations (and take-up tends to be very low).
One limitation of most studies is that they evaluate individual programmes rather than broader activation strategies. As a result, they cannot account for important interactions between policy areas (e.g., between a training program and the enforcement of requirements to participate in the program), which are indeed the essence of activation. Another important consideration when interpreting impact evaluations is that programmes will often impact outcomes other than employment. For instance, even with  negative effects on longer-term employment prospects, public works or other forms of direct job creation might be justified as emergency measures during periods of very high unemployment designed to maintain morale and links to the labour market. Similarly, from an equity perspective, displacement effects are not necessarily a primary concern if a program successfully strengthens employment among the target group. In general, there is, however, still very limited information on the distributional effects of different policies, and even less systematic evaluation of possible equity-efficiency trade-offs (which can for instance arise if strict sanctions increase employment and poverty at the same time). An assessment of distributional effects is especially relevant when policy measures, such as strict benefit sanctions, are targeted towards highly disadvantaged groups.

Overall policy stance on activation and employment support
The discussion above suggests that activation and employment supports are best seen as a package of policy tools, including financial incentives, obligations of job seekers, and programmes that address specific employment barriers on the supply and demand side. To characterise countries' overall activation stance, it is useful to examine how they differ in terms of the balance of these different measures. Figure 7 shows two scatter plots of the indicators presented earlier. Taken at face value, there is no indication that benefit eligibility is generally particularly strict in countries with generous out-of-work benefit amounts (left-hand panel). On the basis of these metrics, eligibility conditions are indeed more demanding in some countries with high net replacement rates (e.g., Slovenia and Portugal), but a number of countries with even more generous out-of-work benefits have fairly light requirements (Canada, Sweden, Austria and Finland). On the other end of the spectrum, benefits in Turkey, Italy, Romania, Estonia and Slovakia are both low and subject to demanding eligibility conditions. Overall, there is no correlation between the two measures: Some countries may seek to offset possible work disincentives of generous benefits (the clients' "rights") with relatively demanding client responsibilities. In others, such a link is not obvious and benefits are generous or tight, both in terms of benefit amounts and in terms of eligibility requirements. By contrast, plotting ALMP spending against benefit replacement rates reveals a fairly strong non-linear correlation across countries (right-hand panel of Figure 7). Active policy "effort", as proxied by ALMP spending, tends to be greater in countries where out-of-work benefits are generous (in part, the correlation is driven by the data points for Denmark, the Netherlands and Italy; but it remains strong even when these three countries are excluded).
Using a similar country regression approach, Venn (2012) finds that ALMP spending is negatively correlated with eligibility strictness. That study also explores links between some of the sub-indicators that are used to construct the summary measures of benefit generosity and eligibility strictness. According to those results, strict entitlement conditions are associated with shorter benefit durations and lower expenditures on PES and training programmes. Countries with more rigorous benefit sanctions also spend less on ALMPs, notably on PES. Finally, Venn reports that strict job-search and availability requirements coexist with more generous benefits.

Conclusions
Although knowledge gaps remain, the experience of a number of OECD countries provides useful pointers for other advanced as well as many emerging and possibly middle-income countries seeking to mobilise benefit recipients back into employment. One uncontroversial conclusion from a review of existing evidence is that welldesigned policies can have a positive impact on employment outcomes for participants. Moreover, the effect of individual policies is strengthened if they are part of a coherent activation strategy in which income support is combined with effective re-employment services tailored to the needs of different beneficiaries. Activation strategies also require significant resources and administrative capacity by public employment services and close coordination of these services with other social services.
It should also be stressed that the implementation of activation strategies in OECD countries has generally involved significant changes in labour market policy institutions, legislation, and management principles, as well as in the design of specific programmes. This has taken time and often required experimentation and testing. But much remains to be done to provide systematic evaluations of co-ordinated policy packages that are the essence of activation strategies. Evaluations of such packages are complicated and much of the literature focuses instead on the effectiveness of specific measures on individuals' labour market outcomes, thereby failing to capture potentially sizeable interactions. As part of a sustained policy commitment to activation, systematic monitoring and evaluation are needed that take into account possible synergies between individual policy elements and help to assess which individual programmes might work best if employed as a package in combination with other measures.
In fact, the impact of individual programmes on aggregate employment or beneficiary caseloads is often fairly modest. This is not necessarily surprising and need not be discouraging given that most programmes are highly targeted, investment of public resources is limited and program durations are short. But it also suggests that there is scope for better co-ordination between policy domains. In part, such co-ordination can be achieved by customizing policy parameters of individual measures. For instance, jobsearch incentives and requirements should arguably be co-ordinated with the provision of job-search assistance. More generally, financial incentives and benefit conditionality can increase participation in employment services. To the extent that these services are effective, this should result in better employment prospects, and a possible virtuous cycle of reduced caseloads and improving service quality.
The large impact of the Great Recession on the labour market of many OECD countries and the modestif anyimprovements in the weak and uncertain recovery in most of them, coupled with the need to tackle large fiscal imbalances, have renewed interest in the role of activation and employment support policies. When fewer vacancies render an effective matching of jobs and jobseekers more difficult, there are good arguments for making, more than ever, adequately resourced labour-market intermediation a policy priority. But in the current environment employers tend to be very selective in their hiring strategies, making it difficult to place some job seekers into jobs quickly. Training programmes and even publically-subsidised work-experience programmes can help prevent this group from becoming discouraged and detached, while preparing them to take advantage of new job opportunities once job creation picks up. But indeed a vigorous recovery in employment is key to tackle the current large imbalances in the labour market. Activation and employment support are best seen as "greasing the wheels" of the labour market. Programmes that ease efficient job reallocation improve labour market outcomes irrespective of the economic cycle. They are, however, not a substitute for job creation.
Endnotes 1 The scope is largely 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 that a short note cannot address properly. Readers are referred to www.oecd.org/els/disability, to OECD (2010) and to OECD (2006c) for discussions of these specific issues.
2 Australia is a notable exception where since 1998 job matching and other employment assistance serviceshave been subcontracted to hundreds of private and community organisations. 3 See e.g. OECD (2007)  ferent surveys, and discuss comparability issues. Some of these issues are probably less problematic when the main interest is in changes over time rather than across countries. 11 These data refer mostly to unemployment insurance and unemployment assistance benefits. Behavioural requirements and other barriers (such as the perceived burden of filing an application) are equally relevant for potential recipients of minimum-income or "welfare" benefits. Cross-nationally comparable coverage data are not available for these benefits of last-resort. But studies on benefit take-up regularly find very high non-take-up rates for means-tested benefits in the order of 40% or more, indicating that the deterrent effect of the various barriers combined is indeed significant (Bargain et al., 2011;Hernanz et al., 2004). 12 Including in Germany, where a merger in 2005 of social assistance and unemployment assistance into the new means-tested Unemployment Benefit II has considerably widened the scope of unemployment benefits. However, the increase shown in the figure is not a net effect as the reform has, at the same time, reduced the number of people receiving social assistance. 13 Other possible explanations are less plausible (see Immervoll and Richardson 2011). For instance, coverage rates have generally moved in the same direction for those with and without prior work experience, so the patterns cannot be explained by a simple change in the composition of the unemployed group such as a greater number of youth unemployed (who are less likely to meet meeting relevant employment or contribution conditions). Across countries, a changing incidence of long-term unemployment is also not a plausible explanation of declining benefit coverage: Between the mid-1990s and the mid-2000s, the proportion of long-term unemployment fell significantly in most countries, with 53 percent (36 percent) unemployed over 6 (12) months in 1994, and 45 percent (32 percent) in 2006 on average in the OECD area. 14 The qualitative information on each of those sub-categories was translated into scores ranging from one (least strict) to five (most strict). The scores were combined into a simple composite indicator using equal weights for each of the four main headings, and equal weights for each of the sub-categories. Venn (2012) explores the sensitivity of the overall strictness indicator to the choice of weights concluding that country rankings remain broadly unaffected. 15 Ireland is a good example of this: according to Figure 3, it has an above-average strictness, yet Grubb et al. (2009) show that this was true only on paper before the 2008 reform that introduced the National Employment Action Plan. 16 Partly for this reason, lower-tier minimum-income programs are not considered in the overview of formal eligibility rules presented in Venn (2012). Typically, these benefit programs are the responsibility of regions or municipalities and legal rules regarding eligibility criteria and sanctions vary between localities and/or take the form of broad guidelines that leave room for interpretation. Subject to these caveats, Immervoll (2009) summarises available information on formal eligibility rules and sanction provisions. 17 In Denmark and Switzerland, spending on active programmes is closely related to the level of unemployment as participation in re-employment programmes is a right and a duty after a certain duration of the joblessness spell and adequate resources are earmarked in the budget every year depending on the expected evolution in unemployment. 18 In the 2008-09 economic downturn, spending for active labour market programs responded more strongly to the rise in unemployment compared with previous downturn, but on average across the OECD countries spending per unemployed declined by 20% (see OECD, Employment Outlook, OECD 2012). Concerning Figure 5, it should be stressed that during the past ten years, two other OECD members -Japan and Koreahave spent less on ALMPs than English-speaking countries, although this is also related to their relatively lower unemployment rates. Unlike the latter group of countries, Japan and Korea have, however, been able to significantly scale up spending per unemployed during the most recent economic crisis. 19 Spending data are influenced by national institutional arrangements and reporting standards and country comparisons are therefore not straightforward. For instance, a minority of countries may report administration costs of active programs in the PES category (see note to Figure 6).