RBA Annual Conference – 1998 What Works Among Active Labour Market Policies: Evidence from OECD Countries' Experiences John P. Martin^[*]

1. Introduction

High and persistent unemployment has been a major blot on the economic and social record of most OECD countries since the early 1970s: the OECD average standardised unemployment rate rose from just over 3 per cent in 1973 to 7.3 per cent in 1997. In response to growing political concerns about the seemingly inexorable rise in unemployment, OECD Ministers gave the Organisation a mandate in 1992 to analyse the causes and consequences of high and persistent unemployment and propose effective remedies to deal with the problem.

The first fruits of this work, published in 1994 under the title The OECD Jobs Study, included a list of more than 60 detailed policy recommendations backed up by two volumes of research – see OECD (1994a, b). Ministers then mandated the Organisation to pursue its analytical work in certain areas, including an examination of how to make active labour market policies a more effective weapon in the fight against high and persistent unemployment. They also asked the Organisation to take the detailed policy recommendations and match them to the needs and circumstances of each individual OECD country, and to monitor progress in the implementation of these recommendations and their impacts on labour market performance.

This paper does not aim to report on progress in the implementation of the OECD Jobs Strategy recommendations by individual OECD countries^[1] and the effects on labour market outcomes. Instead, it confines its remit to a narrower topic: what is the potential contribution which active labour market policies can make as part of a strategy to combat high and persistent unemployment and the problems of low pay and poverty among the working-age population? In order to answer this question, it is vital to know what works among active policies and in what circumstances. The OECD Secretariat has been working intensively on these questions in recent years and this paper summarises the main results of our work to date.^[2]

The structure of the paper is as follows. Section 2 provides some factual background on public spending on labour market policies in OECD countries over the past decade, drawing on an internationally comparable data set which the OECD has developed to monitor trends in this field of public spending. Then I summarise the main results of ongoing OECD research into the effectiveness of active labour market policies. My review mainly exploits two sources: (i) the recent literature on the evaluation of labour market programs (Section 3); and (ii) in-depth reviews which the OECD has conducted over the past five years on the interactions between active and passive labour market policies in 16 OECD countries (Section 4). The final section draws some conclusions.

2. Recent Trends in Public Spending on Labour Market Programs

Public spending on labour market programs absorbs significant shares of national resources in many OECD countries, these policies being expected to achieve a variety of economic and social objectives. For analytical and policy purposes, the OECD splits this spending into so-called ‘active’ and ‘passive’ measures. Active measures comprise a wide range of policies aiming at improving the access of the unemployed to the labour market and jobs, job-related skills and the functioning of the labour market. Spending on active measures is, in turn, split into five program areas: public employment services; labour market training; youth measures; subsidised employment; and measures for the disabled. Passive measures cover spending on unemployment and related social benefits and early retirement benefits.

We have been collecting comparable data on public spending on labour market measures since 1985. These data show that the typical OECD country spent almost 3 per cent of its GDP on active and passive labour market measures in 1996 compared with 2 per cent in 1985. There is wide variation across countries in the share of public spending on labour market measures, ranging in 1996 from a low of under 0.5 per cent of GDP in the Czech Republic, Japan and Korea to a high of over 6 per cent in Denmark (Figure 1).

2.1 Indicators of spending effort on active labour market policies

Within the total public spending on labour market policies, the main focus of this paper is on the fraction devoted to active measures. Using the OECD data set on active measures, it is possible to compute three different measures of the ‘spending effort’ of countries: (i) the share of public spending on active measures as a percentage of GDP; (ii) spending on active measures per person unemployed; and (iii) the number of participants on active programs relative to the size of the labour force.

Data on the first indicator are shown in Table 1. These data show a wide disparity in spending on active measures in 1996, ranging from a low of 0.2 per cent of GDP or less in the Czech Republic, Japan, Korea and the United States to a high of 2.4 per cent in Sweden. There has also been a slight upward trend in the spending effort devoted to active measures, on average across the OECD area, especially since 1990.

The second indicator measures public spending on active measures per person unemployed relative to output per member of the labour force. Such normalised measures of public spending on active labour market policies have been used as regressors in cross-country equations attempting to explain OECD unemployment rates by Elmeskov et al. (1998), Layard et al. (1991), Nickell and Layard (1997) and Scarpetta (1996). They can best be thought of as proxies for the so-called ‘replacement rate’ paid to participants in active labour market programs, defined as average compensation per participant relative to expected income in work. However, the data are less than ideal for this purpose since they do not relate to that fraction of the unemployed who participate on active measures, but instead relate active spending to the total stock of the unemployed in a given year irrespective of whether they participated in a program or not; in addition, the spending data include items other than the compensation or training allowances paid to program participants. Be that as it may, the data in Table 2 reveal a high disparity in spending effort per person unemployed relative to output per member of the labour force across countries in 1996, ranging from 5 per cent or less in the Czech Republic, Greece, Hungary, Japan, Korea, Luxembourg, Poland, Spain, the United Kingdom and the United States to a high of over 30 per cent in Sweden.

The third indicator reports data on the numbers engaged in such programs (Table 3). Almost 8 per cent of the labour force in the typical OECD country participated in these programs in 1996, up from 4 per cent in 1985.^[3] Data on inflow rates reveal a similar wide disparity across countries to that shown by the other indicators, ranging from 3 per cent or less of the labour force in 1996 in Canada, the Czech Republic, Greece, Korea, Spain, Switzerland and the United Kingdom to almost 20 per cent in Denmark.^[4]

Looking at all three indicators together, it is clear that there is a strong positive correlation in the country rankings. This is confirmed by computing Spearman rank correlation coefficients using 1996 data:

	ALMP/GDP	ALMP/U
ALMP/GDP	–	0.82
ALMP/U	0.82	–
PART/LF	0.62	0.84
Note: All correlations are significant at the 1 per cent level.

2.2 Has there been a shift from passive to active measures?

In recent years, it has become a common theme in the political debate on remedies to tackle the unemployment problem, that governments should shift the balance of public spending on labour market policies away from passive income support towards more active measures designed to get the unemployed back into work. At first sight, this seems an eminently sensible proposal: why should our societies pay the unemployed to be idle when the public funds in question could be used instead to supply them with a range of labour market services which would raise their chances of getting a job and their future earnings prospects?

The basic principle of shifting public resources from income support to active labour market policies has been endorsed on several occasions in recent years by OECD Labour Ministers, most recently at their meeting in Paris on the 14–15 October 1997.^[5] It also forms part of the current EU Strategy to combat unemployment, as agreed at the Essen Summit in December 1994.

Have countries managed to switch resources into active measures in line with the principle endorsed by Ministers? Progress has been very limited in terms of this goal: for the typical OECD country, Table 1 shows that spending on active measures rose only from 0.7 per cent of GDP in 1985 to 0.9 per cent in 1996. More disappointingly, the share of spending on active measures as a proportion of total public spending on labour market programs was stable or declined between 1985 and 1996 in over one-third of the countries (Figure 1).^[6] Furthermore, Italy, Norway, Portugal, and Sweden were the only OECD countries where spending on active measures was equal to or exceeded spending on passive measures in 1996.

One obvious reason for the very limited success in switching resources into active measures over the past decade is the rising trend in unemployment in many countries. As unemployment and related welfare benefits are entitlement programs, increases in unemployment bring in their wake an automatic increase in public spending on passive income support. Active labour market programs, on the other hand, are discretionary. In addition, as the next section makes clear, the track record of many active programs is patchy in terms of achieving their stated objectives. This has led many policy-makers to be wary of authorising large spending increases on new or existing programs.^[7]

3. Active Policies: What Works and What Does Not

3.1 Macroeconomic evaluations

Since the ultimate aim of active policies is to cut overall unemployment and/or raise earnings, an obvious approach to assessing their effectiveness is to seek to establish robust econometric relationships between key macroeconomic aggregates such as unemployment or real wages and various measures of the size of active policies. There is, indeed, a small but growing empirical literature on this approach: Calmfors (1994, 1995), Calmfors and Skedinger (1995), Forslund and Krueger (1994), Jackman (1994), Jackman et al. (1990), Layard et al. (1991), Nickell (1997) and Nickell and Layard (1997) and have all published articles on this topic recently. The OECD Secretariat has also contributed to this literature – see Chapter 2 in the 1993 edition of the OECD Employment Outlook and recent articles by Elmeskov et al. (1998) and Scarpetta (1996).

I do not intend to review this literature in my paper – there is a good survey of it in Calmfors (1994). However, it is fair to conclude that the jury is still out on this case: the results of these macroeconomic evaluations are inconclusive, some studies appearing to show robust effects of active policies in terms of lowering the natural rate of unemployment or real wage pressures, others appearing to show zero or insignificant correlations.^[8] This literature is bedevilled by a number of data and technical difficulties, notably simultaneity bias since cross-country comparisons reveal that the amount of spending on active programs is positively related to the unemployment rate.^[9] Because of these uncertainties, the rest of this section concentrates on the main findings from the evaluations of individual labour market programs.

3.2 The literature on evaluation of individual programs

There is a large literature which seeks to evaluate the outcomes of individual programs. These evaluations can be divided into two main types. The first type seeks to measure the impact of program participation on individuals' employment and earnings outcomes after they have left the program, judging the outcomes against the experiences of a benchmark or control group of similar individuals who did not participate in the program. This type of evaluation makes sense for programs which attempt to make participants more productive and competitive in the open labour market, e.g. training and job-search assistance.

The second type of evaluation attempts to measure the net effects of programs on aggregate employment and unemployment by estimating what are called in economists' jargon ‘deadweight’, ‘substitution’ and ‘displacement’ effects. These evaluations are mostly relevant for employment programs, i.e. programs which attempt to stimulate job creation in the private sector (including self-employment), as well as direct job creation in the public sector. Since subsidised employment programs have the explicit objective of increasing the number of jobs in the economy at large and/or raising the employment prospects of the target group, evaluations must determine whether the subsidised jobs would have been created anyway in the absence of the subsidy (so-called deadweight effects) or whether the subsidised jobs have displaced, or have been substituted for, unsubsidised jobs elsewhere in the economy.

3.3 Caveats to bear in mind when assessing the literature on program evaluation

Before turning to the main findings from the recent evaluation literature, I think it is important to stress some caveats concerning the reliability and generality of the conclusions that can be drawn from this literature.

First, much of the evaluation literature relates to the United States and Canada where there is a long-standing tradition of evaluating labour market programs. Indeed, in both countries, there is effectively a mandatory requirement on the public authorities to evaluate their programs. Few European countries have carried out rigorous evaluations until recently.^[10] This unsatisfactory situation is changing slowly, as tight fiscal constraints make it imperative to get better value for public spending on active labour market policies. As a result, some European countries (I would single out Norway, Sweden and the United Kingdom in this regard) and Australia are beginning to see the light as regards undertaking rigorous evaluations of their labour market programs. However, in most other countries, the most common method of ‘evaluation’ consists of simply monitoring the labour market status and earnings of participants for a brief period following their spell on a program. While this sort of exercise provides useful information, it cannot answer the vital question of whether the program in question ‘worked’ or not.^[11]

Second, one must recognise that there is almost never a stable set of active programs to evaluate. Countries are continuously chopping and changing the mix of programs. For example, Grubb (1995) highlights the strong tendency on the part of the US Congress to respond to specific new problems with a specific new program, rather than to incorporate new purposes into old programs. This leads to a proliferation of programs, many of which are overlapping, tend to cancel each other out, are costly to administer, and are confusing to firms and the unemployed they are supposed to assist.

Third, there is very little evidence on the long-run effects of active programs. The vast majority of rigorous evaluations only provide evidence on short-run outcomes, covering at best 1 to 2 years after the person has participated in the program.^[12] This may well be too short a period for a full assessment of the private and social returns to public investment in many active measures.

Fourth, ‘outcomes’ in the evaluation literature, are invariably expressed in terms of program impacts on future earnings and/or re-employment prospects of participants, and this stress is reflected in this paper. There is little or no evidence available on social benefits such as reduced crime, less drug abuse or better health.

Fifth, there is an issue about the scale of programs, even those which appear to work. It is unclear from the existing literature how cost-effective such programs would be if they were greatly extended in terms of scale of participation.

Sixth, many evaluations are undertaken by public sector agencies. While there are good reasons for this, it does give rise to concerns about the independence of the findings. Therefore, where evaluations are undertaken by public sector agencies, it is important to check whether there has been any external validation of the evaluation results in question.

Finally, while the evaluation literature, as we shall see, tells us quite a lot about what works, it is not very instructive in answering other equally important and related questions, such as why do certain programs work for some groups (see below) and not for others, and in what circumstances? It is not helpful in explaining what combination of employment services is likely to work. For example, there is almost no evidence on which types and content of training programs work best. Do skill-enhancing activities – e.g. via classroom training and/or on-the-job training – work best or must they be combined with personal counselling, job-search assistance and mentoring services to work? Policy-makers want to know the answers to such questions, but the evidence is simply not there for the moment.

3.4 Findings from the evaluation literature

The OECD has reviewed the available evaluation literature in a chapter which was published in the 1993 edition of the OECD Employment Outlook and a colleague, Bob Fay, has updated this review recently.^[13] What does this latest review of the evaluation literature tell us about what works and what does not? In seeking to answer this question, I will summarise briefly the findings for each major labour market policy measure in turn.^[14] Table 4 summarises the main lessons in terms of what works for which groups.

3.4.1 Public training programs

They usually account for the largest share of spending on active measures: on average, OECD countries devoted 27 per cent of their total public spending on active measures to training programs in 1996, up from 23 per cent in 1985 (Table 5). But evaluations of public training programs in OECD countries suggest a very mixed track record. Some programs in Canada, Sweden and the United States have yielded low or even negative rates of return for participants when the estimated program effects on earnings or employment are compared with the cost of achieving those effects.^[15] However, all is not black on the front of public training programs. A recent comprehensive review of public training programs for disadvantaged groups in the United States by Friedlander et al. (1997) highlights quite a number of successful programs in terms of earnings gains and positive rates of return for participants. It is noticeable that the most consistently positive results were recorded for adult women. The findings were less optimistic with regard to adult men: some programs gave positive results, others not. The most dismal picture emerged with respect to youth: almost no training program worked for them. Even for those groups for whom participation in the programs yielded a positive rate of return, Friedlander et al. (1997) note that the estimated earnings gains are not large enough to lift most families out of poverty.

As noted above, the available evaluation literature can tell us whether training programs work for particular disadvantaged groups or not. However, it does not provide satisfactory answers as to why they appear to work for some target groups (e.g. adult women) and not for others. Until we have answers to this question, it is going to be extremely difficult to design effective public training programs.

Such evidence, as exists, highlights three crucial features in the design of public training programs: (i) the need for tight targeting on participants; (ii) the need to keep the programs relatively small in scale; (iii) the need to have a strong on-the-job component in the program, and hence to establish strong links with local employers. At the same time, it is clear that training programs which foster strong links with local employers are likely to encourage displacement, an outcome which arises when those who participate in the program get jobs at the expense of individuals who did not participate in the program.^[16]

3.4.2 Job-search assistance

Unfortunately, it is not possible in the OECD database at the moment to separate out spending on job-search assistance from the administrative costs of running the public employment service (PES): in 1996, the average OECD country devoted 21 per cent of active spending to PES administration, but much of this comprises the fixed costs of running the service. Job-search assistance comprises many different types of services, for example, initial interviews at the public employment service, in-depth counselling at some stage in the unemployment spell, re-employment bonuses, jobs clubs etc. Such services may also be combined with increased monitoring and enforcement of the job-search requirements for receipt of unemployment benefits.

Job-search assistance is usually the least costly active labour market program and the good news is that evaluations from several countries show consistently positive outcomes for this form of active measure.^[17] It seems that investment in active placement and raising the motivation of the unemployed, as well as taking steps to encourage and monitor their job-search behaviour, pay dividends in terms of getting the unemployed back into work faster.^[18] While the optimal combination of additional job-placement services and increased monitoring of job-seekers and enforcement of the work test is unclear, the evidence suggests that both are required to produce benefits to unemployment insurance claimants and society.

One particularly interesting form of job-search assistance is re-employment bonuses, i.e. cash payments to unemployment insurance recipients who find a job quickly and keep it for a specified length of time. Such a scheme exists in Japan and has been experimented with in several US States. The US evaluations show that the bonus payments did reduce the average duration of unemployment benefit receipt significantly. However, such bonuses can give rise to negative effects too. Their existence may have an effect on the size of the group claiming benefits. In particular, they may induce workers with a high probability of finding a new job quickly to arrange with their employers to be laid off so as to collect the bonus. In order to minimise such abuse, Japan has several safeguards and controls in place involving both the bonus claimant and his former employer.

3.4.3 Special youth measures

On average, OECD countries devoted 12 per cent of spending on active policies to these measures in 1996. One of the most disappointing conclusions is that almost all evaluations show that special measures are not effective for disadvantaged youths. This holds not only for public training programs (see above) but also for targeted wage subsidy measures too. Given the depth of public concern about youth unemployment and the large public spending devoted to special youth measures, a high priority must be assigned to discovering the reasons for the dismal track record of such measures and designing and implementing more effective programs.

Among the large number of negative evaluation results, there are a few hopeful signs. Job Corps in the United States did yield statistically significant earnings gains for disadvantaged youth. However, it had to rely on savings from reduced criminal activity among the target group to produce a net social benefit, given that it is a high-cost program.^[19] In addition, within national demonstrations such as JOBSTART in the United States, it is possible to identify specific sites where the program appeared to work for disadvantaged youth. One such example of a site that appeared to deliver large gains is the Center for Employment Training (CET) in San José, California; it was the only one of the 13 JOBSTART sites which delivered statistically significant earnings gains for youths. However, we do not know precisely what factors distinguished the CET site from the other sites or how feasible it would be to replicate their positive results elsewhere.

The evidence from Canadian and US evaluations suggests that the biggest payoffs for disadvantaged youths come from early interventions. This involves not only intensive efforts to boost their performance in primary and secondary schooling and reduce drop-out rates, it also reaches back to early childhood including the pre-school period. The limited empirical evidence that is available suggests that early childhood interventions of high quality can have lasting effects on the employment and earnings prospects of disadvantaged children.^[20] It is also important to target support not only at the youngsters themselves but also at their families and local communities. It cannot be over-emphasised that if young people leave the schooling system without qualifications and a good grounding in the 3Rs, it is well nigh impossible for labour market programs to overcome these handicaps later on.

Finally, several authors (e.g. Lerman 1997) highlight the importance of poor attitudes towards work among disadvantaged youth as a major factor in explaining the dismal record of special youth measures. It is not easy for many programs to influence attitudes in ways that improve the jobs and earnings prospects of disadvantaged youth. But mentoring programs, by providing for both ongoing contact with an adult over an extended period of time and elements of monitoring as well as support, can help overcome negative attitudes to work.

3.4.4 Subsidies to private sector employment

These measures accounted for 11 per cent of total spending on active measures in the typical OECD country in 1996. Subsidies to private sector jobs may have a number of objectives other than creating additional jobs. They may seek to enhance effective labour supply by helping individuals to keep in contact with the world of work, thereby maintaining their motivation and skills.^[21] For equity reasons they may also be intended to provide the long-term unemployed with jobs, even if this happens at the expense of the short-term unemployed. These other goals of wage-subsidy schemes may still be important even if the net employment gains of these programs are very small or zero.

Indeed, most evaluations show that subsidies to private sector employment have both large deadweight effects (i.e. employers use the subsidy to hire workers they would have hired anyway) and displacement effects (many subsidised hires displace others who would have been hired in the absence of the subsidy). As a result, most such schemes yield small net employment gains, particularly in the short term when aggregate demand and vacancies are fixed. For instance, evaluations of wage subsidies in Australia, Belgium, Ireland and the Netherlands have suggested combined deadweight and substitution effects amounting to around 90 per cent, implying that for every 100 jobs subsidised by these schemes only 10 were net gains in employment. However, such wage-subsidy programs do give an advantage to the target group compared with other job seekers, and the resulting redistribution of job opportunities may be justified on equity grounds.

The evidence suggests it may be possible to raise the size of net employment gains associated with private sector wage subsidies to 20–30 per cent or more via tight targeting of the measures to particular groups among the unemployed and monitoring of employer behaviour in order to curb abuses. However, there is a difficult trade-off for policy-makers here: the evidence also suggests that the more controls are multiplied in order to curb abuse and maximise the net employment gains from wage subsidies, the less willing are firms to participate in such programs and employer take-up drops off sharply, defeating the ultimate goal of the exercise. In addition, the more tightly the program is tied to characteristics of ‘disadvantage’, the greater the risk of so-called ‘stigma’ which may discourage the unemployed from availing themselves of such schemes.

One specific form of wage subsidy that appears to be successful for a small group of unemployed individuals is aid to starting up a small business. Controlled experiments in the United States suggest that such schemes result in employment gains for men, primarily between the ages of 30 and 40, who have relatively high levels of education. Evidence from less rigorous evaluations of such schemes in other countries such as Australia, Ireland, Norway and the United Kingdom tends to confirm longer term survivability, but only for a relatively small share of all enterprises started up in this manner.

3.4.5 Direct job creation in the public sector

Spending on this measure exceeds spending on subsidies to private sector jobs in many countries: on average, the typical OECD country devoted 14 per cent of its spending on active measures to public sector job creation measures in 1996. The evaluation literature shows fairly conclusively that this measure has been of little success in helping unemployed people get permanent jobs in the open labour market. As a result, there has been a trend away from this type of intervention in recent years but it appears to be making a comeback now in some OECD countries, usually as part of a ‘reciprocal obligation’ on the unemployed in return for continued receipt of benefits (see below).

However, OECD countries continue to spend large amounts on public sector job creation programs and the policy debate about the utility of this intervention is still alive. Temporary employment programs in the public sector can be used as a work test for unemployment benefit claimants and as a means of helping the unemployed maintain contact with the labour market, particularly in a recession when aggregate demand is depressed and vacancies are scarce. But since most jobs provided through direct job creation schemes typically have a low marginal product, they should be short in duration and not become a disguised form of heavily subsidised permanent employment.

3.5 The bottom line

In sum, our review of the evaluation research highlights the following five principles which should guide the selection of active policies in order to maximise their effectiveness:

• First, rely as much as possible on in-depth counselling, job-finding incentives (e.g. re-employment bonuses) and job-search assistance programs. But it is vital to ensure that such measures are combined with increased monitoring and enforcement of the work test.
• Second, keep public training programs small in scale and well-targeted to the specific needs of both job-seekers and local employers.
• Third, early interventions, reaching back to pre-school, can pay dividends for disadvantaged youths. This should include steps to reduce early school-leaving targeted on at-risk students combined with policies to ensure that they leave the schooling system equipped with basic skills and competences that are recognised and valued by employers. It is also important to improve poor attitudes to work on the part of such young people and adult mentors can help in this regard.
• Fourth, as the duration of unemployment spells lengthens, various forms of employment subsidies may serve to maintain workers' attachment to the labour force. However, employment subsidies should be of short duration, targeted and closely monitored.
• Fifth, use subsidised business start-ups for the minority among the unemployed who have entrepreneurial skills and the motivation to survive in a competitive environment.

Finally, if we are to expand the range of international knowledge on ‘what works’ and ‘why’ among active labour market policies, it is vital that more countries begin to evaluate their labour market programs systematically. Indeed, evaluation should be built into the design of programs at the beginning rather than being viewed as an ex post exercise. Evaluations should also be undertaken in a rigorous way that allows one to draw useful inferences about the effectiveness of the interventions in terms of their impacts on the employment and earnings prospects of the program participants.

4. Interactions between Active and Passive Policies

Recent OECD research also suggests that it is vital to focus on the interactions between active and passive labour market policies if one seeks to enhance the effectiveness of active labour market policies. I will now turn to this topic drawing on OECD reviews of labour market policies in 16 countries (Austria, Belgium, Denmark, Finland, Germany, Greece, Ireland, Italy, Japan, the Netherlands, Portugal, Norway, Spain, Sweden, Switzerland, and the United Kingdom).^[22]

Why is this an important topic? Unemployment and related welfare benefits provide income support to the unemployed while they are searching for jobs. It is well known that such benefits can have significant effects on work incentives for the unemployed and on the wage-setting behaviour of workers and employers. Active labour market policies aim to help the unemployed get back into work and raise their future earnings prospects by providing them with a range of employment services. But they also provide income support to the unemployed while they participate in an active program and such participation can affect future entitlements to unemployment benefits, thereby influencing the behaviour of labour market actors.^[23] For this reason, it is important to pay attention to the interactions between active measures and unemployment benefit systems.

4.1 Gross and net replacement rates in OECD countries

An obvious starting point to analysing these interactions is the relative generosity of income support to the unemployed via unemployment benefits or the compensation paid while they participate on an active program. Unfortunately, we do not have data on the latter, only on the former. But it is likely that both forms of income support are highly correlated. Indeed, it seems to be the case in many countries that participants on some active measures are paid unemployment benefits, sometimes with a small top-up. Hence, trends in the generosity of unemployment benefit systems are likely to be mirrored closely in the average compensation paid to program participants.

The standard indicator of the generosity of an unemployment benefit system is the so-called ‘replacement rate’, i.e. the proportion of expected income from work which is replaced by unemployment and related welfare benefits. The OECD has devoted much effort in recent years to developing a range of gross and net (i.e. after-tax) replacement rates for the purposes of international comparisons.

Figure 2 presents time-series data on the OECD summary measure of unemployment and related welfare benefit entitlements over the period 1961–95. The summary measure in Figure 2 is an average of 18 separate gross replacement rates covering a variety of household types, a range of earnings possibilities and different durations of an unemployment spell.^[24] It shows that few OECD countries have taken steps to roll back the generosity of their benefit systems in recent years in terms of cutting benefit levels and/or reducing the average duration of benefit payments: the OECD average of the summary measure doubled from 16 per cent in 1961 to 31 in 1995.

At the same time, net replacement rates are typically much higher than gross rates for a variety of reasons described in Martin (1996). Table 6 shows a selection of net replacement rates from the OECD database for two different earnings levels, the earnings of an average production worker (APW) and a worker assumed to earn two-thirds of the APW level. These data show that net replacement rates in excess of 80 per cent are quite common in OECD countries once social assistance benefits, housing benefits and the effect of the tax system are taken into account. While we do not have time-series data on net replacement rates to parallel the data in Figure 2, it seems likely that they too have tended to drift upwards in many OECD countries over the past three decades.

In sum, the available evidence suggests that replacement rates, whether provided through unemployment and related welfare benefit systems or active programs, are sufficiently large to have potentially significant effects on work incentives and on wage-setting behaviour. This, in turn, has led to attempts in recent years to curb the so-called ‘unemployment trap’.

4.2 Actions being taken by OECD countries to curb unemployment traps

The most direct step to curb the unemployment trap is to cut replacement rates. However, where actions were taken to cut replacement rates, they were usually motivated by budget considerations rather than out of concern about the possible emergence of benefit dependency or work disincentives.^[25] Given the political difficulties with dismantling benefit entitlements, the preferred approach to curbing the unemployment trap in the majority of OECD countries has been to make only marginal cuts in the generosity of benefit entitlements, but to tighten up on eligibility conditions for receipt of benefits and to develop ‘activation’ strategies for the unemployed. The aim of the latter is to encourage the unemployed to be more active in job search and keep more in touch with the labour market. Activation strategies range from attempts to provide more effective job-search assistance to the unemployed at one end of the spectrum, to making it obligatory on the unemployed to satisfy work tests or participate in active programs or in education and training if they are to continue to draw benefits. Such activation strategies are becoming quite common for young people in OECD countries (e.g. the Welfare-to-Work initiative in the United Kingdom), and they are even being extended to other groups of the unemployed in some countries.^[26]

The role of active labour market policies changes subtly in the context of an activation strategy. They can then be viewed as a vehicle for enforcing a work test on the unemployed, especially in cases where the supply of job vacancies is low. In such cases, continued receipt of unemployment benefits becomes conditional on program participation, such as is the case in Denmark or Switzerland, and/or by offering a sufficiently wide range of programs so that a maximum number of the unemployed will choose to enter them voluntarily. Related to this, there is a growing interest in the issue of the rules used to control job-search behaviour and curb benefit abuse by claimants of unemployment benefits. As noted in the previous section, the evaluation literature suggests that these rules, if used intelligently and supported by effective sanctions, can help stimulate job search and serve to keep benefit claimants in touch with the labour market.^[27]

However, there is a lack of available evidence on the magnitude and incidence of benefit sanctions across countries; there is also a lack of hard evaluation evidence yet on the outcomes of activation strategies. Our reviews have also revealed that there appears to be a noticeable reluctance on the part of public officials in some countries to enforce the existing rules tightly for a variety of reasons. Hence, it is important not to exaggerate the reality on activation, as opposed to the rhetoric. Since sanctions are a potentially important tool in influencing job-search behaviour, it is important to understand their workings better and assess how they can play a role in enhancing the effectiveness of active measures. We have just launched a new round of country reviews on this subject.

4.3 The importance of integrated management of benefit systems and active labour market policies

There is a close interaction between active and passive measures which is central to the trade-off between equity and efficiency. If the unemployment benefit system is generous and poorly managed, it is very difficult to operate active programs in ways which increase labour market efficiency and reduce structural unemployment. Conversely, if active measures are used on a large scale and mainly serve to re-establish benefit entitlements, they risk becoming a de facto passive measure. They thus need to be better managed and linked more closely to the benefits system.

Our research at the OECD suggests that the public employment service has a central role to play in achieving this better management. It can play this role most effectively if it operates as a fully integrated agency combining the three core functions of job placement, benefit payments and placing participants on active programs. Such integration is desirable for the following reasons:

• a close co-ordination between placement and benefit work is needed in order to apply work tests effectively and hence to fulfil one of the key preconditions for benefit entitlement;
• a close co-ordination between job broking and ALMPs is needed in order to ensure that the unemployed can acquire the attributes necessary to fill available job vacancies; and
• a close co-operation between benefit administration and referral to ALMPs is needed in order to avoid long-term dependency on benefit receipt and program participation for the sole purpose of renewing benefit entitlements.

However, it should be noted that many OECD countries do not have a fully integrated public employment service in this sense, though Australia has taken a large step in this direction recently with the establishment of Centrelink, and New Zealand has announced that it will fully integrate the delivery of income support with the delivery of employment services in a single agency.

4.4 The bottom line

The recent history of active and passive labour market policies in the 16 OECD countries which have been reviewed suggests the following eight lessons for the design of future policies in order to make them more effective:

• First, integrate the referral to active programs as closely as possible with benefit and placement work. Ideally, all three basic functions should be provided by the same front-line public employment office (so-called ‘one stop’ labour offices).
• Second, use ‘profiling’ for new benefit claimants to identify those at risk of becoming long-term unemployed; provide the latter (but not the others) immediately with counselling and job-search assistance.^[28]
• Third, make passive income support as ‘active’ as possible by using instruments like re-employment bonuses, in-work benefits, regular contacts of claimants with the public employment service, job clubs, etc.
• Fourth, use ‘availability for work’ (to be controlled by work tests) and ‘job search initiatives’ (to be confirmed by employers) as independent criteria which must be met in order to qualify for continued benefit receipt (e.g. both criteria have to be fulfilled in Switzerland).
• Fifth, make continued receipt of income support conditional on accepting to participate in active programs after a certain minimum duration of an unemployment spell (say after 6 or 8 months); do not, however, guarantee a slot in a program by that time, but handle the referral flexibly in accordance with the availability of slots which correspond to the needs of the job seeker in question.
• Sixth, ensure that participants in training and public sector employment programs continue to be available for work in the open labour market; and encourage them to engage actively in job search.
• Seventh, ensure that participation in training and public sector employment programs does not serve mainly to establish new benefit entitlements. One way to ensure this is by making the duration of employment subsidies to the private sector shorter than the minimum contribution period required for benefit entitlements. These steps will minimise the so-called ‘carousel effect’, whereby a considerable number of the long-term unemployed move between spells of benefit receipt and program participation. Our research has shown that the carousel effect is a significant problem in many European countries and I am aware that it has also been a problem in Australia. Some countries, e.g. Denmark, Finland, Norway and Switzerland, have taken steps recently to curb this possibility.
• Finally, explore ways of making the public employment service more effective by giving greater play to the role of market signals. For example, many active measures, particularly training programs, are provided by the public sector and this may not be the most efficient form of provision. Some countries are beginning to experiment in this area with a range of initiatives designed to give greater play to private sector agencies in the provision of active measures.^[29]

Australia has gone further in this direction than any other OECD country with its recent initiative designed to make the market for job placements fully contestable, with specific incentives for both private and public employment services to compete to place the most at-risk job-seekers. OECD Labour Ministers were clearly fascinated by this Australian initiative when it was presented to them at their October 1997 meeting, but many of them expressed reservations about imitating it for the moment. They want to see if it will produce better outcomes for job seekers. However, it will be several years before it is possible to evaluate whether the new Employment Services Market improves significantly the employment and earnings prospects of at-risk job-seekers compared with other, more traditional active measures. I hope that a sufficiently high priority has been assigned to rigorous evaluation of this initiative.

5. Conclusions

At first sight, the bottom line from recent OECD research on the effectiveness of active labour market policies is not terribly encouraging. The track record of many active measures is mixed in terms of raising the future employment and earnings prospects of job-seekers and producing benefits to society. In addition, little progress has been made to date in shifting public spending from passive to active labour market measures in most OECD countries, despite the widespread endorsement of this goal by politicians.

While we cannot ignore the undoubted problems with active measures, it would be wrong to draw a pessimistic conclusion about their potential role in the fight against high and persistent unemployment and the problems of low pay and poverty. We now know a great deal more about what works and what does not work among the large array of active measures currently in use across OECD countries. We are also much more aware nowadays of the crucial nature of the various interactions between active and passive measures. Recent OECD research suggests several practical steps which can be taken to enhance the effectiveness of active measures. At the same time, there is a crying need to expand the quantity and quality of evaluations of labour market programs in a wider range of OECD countries so that countries can learn from each other's experiences.

However, even if all these steps were to be implemented, it is important to be realistic about their likely impacts on unemployment; one should not oversell the case for active labour market policies. More effective active labour market policies, as Lars Calmfors has rightly warned, are not a magic bullet on their own to solve the unemployment problem. Since one of the main objectives of active measures is to assist the unemployed to get back into work, they require a reasonably buoyant supply of job vacancies in order to be effective. If an economy is generating few vacancies, one should not be surprised if active measures prove to be relatively ineffective. Aggregate demand matters too. As The OECD Jobs Study has stressed, more effective active policies are only one element in a comprehensive strategy of macroeconomic and microeconomic measures required to cut unemployment significantly. Nonetheless, they remain a potentially important weapon in the fight against unemployment.

Footnotes

Deputy Director, Directorate for Education, Employment, Labour and Social Affairs, OECD, Paris. I am very grateful to Robert Fay, David Grubb, Peter Schwanse and participants at the Reserve Bank conference for helpful comments on an earlier version of the paper, and to Maxime Ladaique and Marco Doudeijns for statistical assistance. The views expressed in this paper are my own and cannot be held to represent those of the OECD or its Member Governments. [*]

For detailed reviews of progress in the implementation of the country-specific recommendations, see OECD (1997d, 1998c). [1]

The results of this work are presented in OECD (1996a, 1997c). [2]

The data in Table 3 relate to annual inflows to slots on various labour market programs. They do not tell us anything about the average length of time which a participant spends on the program nor do they provide any information on repeat spells on programs. We hope to be able to extend our database in the future in order to include information on these important dimensions of participation on labour market programs. We would also like to directly collect data on the average compensation paid to participants on active measures. [3]

The data on inflow rates cover public training for employed adults as well as the unemployed. In a few countries, e.g. Denmark, Belgium, Greece, Ireland and Portugal, this accounts for a large proportion of the total inflow rate to all active measures, ranging from 20 per cent in Ireland to around 50 per cent in Denmark, Greece and Portugal in 1996. [4]

The relevant extract from the Press Communiqué issued at the end of the Ministerial meeting reads as follows: ‘When they last met in 1992, OECD Ministers of Employment and Labour stressed the need to shift public spending on labour market policies from passive to active measures. In most countries, more progress is needed in achieving this objective. Today, Ministers reiterate this policy objective while at the same time underlining the need to enhance the effectiveness of active labour market policies and to design and manage unemployment and related welfare benefits fairly, but tightly. Active measures must not become inadvertently ‘passive’ in that they simply provide parking slots for the unemployed or serve to re-establish benefit entitlements. At the same time, so-called passive measures should be designed and rigorously managed so that active job search by benefit claimants is rewarded, thereby ensuring that they do not become overly dependent on income support’. [5]

However, the fact that inflow rates to programs increased between 1986 and 1996 in many more countries than did spending/GDP ratios suggests that there was a shift to shorter-duration programs over the period. [6]

There is a nice illustration of this in recent US experience. When the national evaluation of the Job Training Partnership Act (JTPA) revealed that it had failed to provide earnings gains to disadvantaged youths, the US Congress eliminated nearly all of the funds for JTPA. [7]

For example, Scarpetta (1996) includes a measure of the intensity of public spending on ALMPs (defined as average spending on ALMPs per unemployed person) in his pooled cross-section/time-series analysis of the determinants of the unemployment rate in a large sample of OECD countries over the period 1983–93. While the estimated impacts of ALMPs were always negative in his regressions, the coefficients were small and in some cases insignificant. However, a sensitivity analysis designed to identify outliers in the data suggested that the data from Sweden should be excluded from the panel. Once this adjustment was made, the magnitude and statistical significance of the estimated coefficient of the ALMP variable increased sharply. Re-estimation of these equations for a larger sample of countries and a somewhat longer time period by Elmeskov et al. (1998) revealed even stronger ALMP effects, once Sweden is excluded. Calmfors and Skedinger (1995), on the other hand, highlight the lack of robustness of their results for Sweden. [8]

See Calmfors and Skedinger (1995) for a discussion of these problems. [9]

One explanation for the lack of evaluations in Europe was put to me frankly about ten years ago by a leading policy-maker who shall remain nameless. If I can paraphrase his explanation, it would run as follows: ‘Most of our programs are lousy! They were dreamed up quickly to give the Minister some good news to announce at a time when unemployment is rising. We do not want evaluations revealing to the general public how bad our programs are; we know this already!’. [10]

There is a large literature on the appropriate methodology to use in evaluating labour market programs. See Friedlander et al. (1997) for a good review of the issues. [11]

There are a few US evaluations which cover longer time periods, following individuals up to 5 to 6 years after their participation in the program. See Grubb (1995) for a review of these studies. [12]

See Fay (1996). See also Katz (1994) for a good review of the recent US literature. [13]

I do not include special employment measures for the disabled since we have not reviewed the evaluation literature in this field. [14]

See Forslund and Krueger (1994) for a review of the Swedish evaluation evidence on training programs; Friedlander et al. (1997) and Grubb (1995) for reviews of the US literature; and Park et al. (1996) for a review of some Canadian programs. [15]

Friedlander et al. (1997) point out that there is no evidence in the rigorous evaluation literature quantifying the size of displacement associated with training programs for disadvantaged groups. [16]

See Meyer (1995) for a review of the US evidence and HRDC (1997) for a review of the Canadian evidence. [17]

However, Canadian evidence, summarised in HRDC (1997), suggests that any earnings gains from job-search assistance are likely to be transitory. [18]

The evaluation results supporting this positive assessment of Job Corps were based on non-experimental methods and were done almost 20 years ago. A rigorous nationwide evaluation of Job Corps is now under way to try to settle the issue of whether it works or not. [19]

See Heckman (1994) and HRDC (1997). [20]

See Richardson (1998) for evidence, using a panel of Australian youth, that participation in subsidised jobs improved their employability. [21]

See OECD (1993b, 1993c, 1996b, 1996c, 1996d, 1997b, 1998b). We are currently undertaking a review of the PES and how it interacts with ALMPs and unemployment benefit systems in two US states, Wisconsin and Connecticut. [22]

See Calmfors (1994) for a detailed exposition of this argument. [23]

See Martin (1996) for a detailed discussion of these data. [24]

It should be noted that the replacement rates in Figure 2 refer to a 40-year-old worker with a long contributions history since this case was considered a good approximation to the average situation of an unemployed worker in most countries. However, this assumption means that most changes in eligibility conditions for receipt of unemployment benefits will not show up in the OECD summary measure. [25]

New Zealand has recently decided to work test not only the unemployed but all other welfare beneficiaries of working age. [26]

Abbring et al. (1996) is one of the few rigorous evaluations of the effect of unemployment insurance sanctions on the transition rate from unemployment to employment. They use a micro data set covering the population of individuals who started collecting benefits in the Netherlands in 1992. Their results show that the transition rates to employment are increased significantly by the imposition of a benefit sanction. [27]

See OECD (1998a) for a review of experiences with different profiling approaches in Australia, Canada, the United Kingdom and the United States. It should be noted that there are strong differences of view about the relevance and reliability of formal profiling methods, and how central a role profiling can play in making active labour market programs more effective. [28]

See Fay (1997) for a review of these issues. [29]

References

Abbring, J., G. van den Berg and J. van Ours (1996), ‘The Effect of Unemployment Insurance Sanctions on the Transition Rate from Unemployment to Employment’, paper presented at the Canadian International Labour Network Conference, Burlington, Ontario, 4–7 September.

Calmfors, L. (1994), ‘Active Labour Market Policies and Unemployment: A Framework for the Analysis of Crucial Design Features’, OECD Economic Studies, 22, pp. 7–49.

Calmfors, L. (1995), ‘What Can We Expect from Active Labour Market Policy?’, Konjunkturpolitik, 43, pp. 1–30.

Calmfors, L. and P. Skedinger (1995), ‘Does Active Labour Market Policy Increase Employment? Theoretical Consideration and Some Empirical Evidence from Sweden’, Oxford Review of Economic Policy, 11(1), pp. 91–109.

Elmeskov, J., J. Martin and S. Scarpetta (1998), Key Lessons for Labour Market Reforms: Evidence from OECD Countries' Experiences, paper presented to the Economic Council of Sweden Conference on ‘The Political Economy of Labour Market Reform’, Stockholm, 25^th May.

Fay, R. (1996), ‘Enhancing the Effectiveness of Active Labour Market Policies: Evidence from Programme Evaluations in OECD Countries’, Labour Market and Social Policy Occasional Papers No. 18, OECD, Paris.

Fay, R. (1997), ‘Making the Public Employment Service More Effective Through the Introduction of Market Signals’, Labour Market and Social Policy Occasional Papers No. 26, OECD, Paris.

Forslund, A. and A. Krueger (1994), ‘An Evaluation of the Swedish Active Labor Market Policy: New and Received Wisdom’, NBER Working Paper No. 4802.

Friedlander, D., D. Greenberg and P. Robins (1997), ‘Evaluating Government Training Programs for the Economically Disadvantaged’, Journal of Economic Literature, 35(4), pp. 1809–1855.

Grubb, W. (1995), ‘Evaluating Job Training Programs in the United States: Evidence and Explanations’, University of California at Berkley National Center for Research in Vocational Education, Technical Assistance Report, May.

Heckman, J. (1994), ‘Commentary: Active Labor Market Policies to Expand Employment and Opportunity’, in Reducing Unemployment: Current Issues and Policy Options, Federal Reserve Bank of Kansas City, Kansas City, Missouri, pp. 291–311.

Human Resources Development Canada (1997), Lessons Learned: Effectiveness of Employment-Related Programs for Youth, Human Resources Development Canada, Ottawa, June.

Jackman, R. (1994), ‘What Can Active Labour Market Policy Do?’, Swedish Economic Policy Review, 1(1–2), pp. 221–257.

Jackman, R., C. Pissarides and S. Savouri (1990), ‘Labour Market Policies and Unemployment in the OECD’, Economic Policy, 11, pp. 449–490.

Katz, L. (1994), ‘Active Labour Market Policies to Expand Employment and Opportunity’, in Reducing Unemployment: Current Issues and Policy Options, Federal Reserve Bank of Kansas City, Kansas City, Missouri, pp. 239–290.

Layard, R., S. Nickell and R. Jackman (1991), Unemployment: Macroeconomic Performance and the Labour Market, Oxford University Press, Oxford.

Lerman, R. (1997), ‘Employment and Training Programs for Out-of-School Youth’, mimeo, Urban Institute, Washington, D.C.

Martin, J. (1996), ‘Measures of Replacement Rates for the Purposes of International Comparisons: A Note’, OECD Economic Studies, 26, pp. 99–116.

Meyer, B. (1995), ‘Lessons from the U.S. Unemployment Insurance Experiments’, Journal of Economic Literature, 33(1), pp. 91–131.

Nickell, S. (1997), ‘Unemployment and Labor Market Rigidities: Europe versus North America’, Journal of Economic Perspectives, 11(3), pp. 55–74.

Nickell, S. and R. Layard (1997), ‘Labour Market Institutions and Economic Performance’, Oxford University Institute of Economics and Statistics Discussion Paper No. 23.

OECD (1993a), Employment Outlook, OECD, Paris.

OECD (1993b), The Labour Market in the Netherlands, OECD, Paris.

OECD (1993c), The Public Employment Service in Japan, Norway, Spain and the United Kingdom, OECD, Paris.

OECD (1994a), The OECD Jobs Study: Facts, Analysis, Strategies, OECD, Paris.

OECD (1994b), The OECD Jobs Study: Evidence and Explanations, Volumes I and II, OECD, Paris.

OECD (1996a), The OECD Jobs Strategy; Enhancing the Effectiveness of Active Labour Market Policies, OECD, Paris.

OECD (1996b), The Public Employment Service: Denmark, Finland, Italy, OECD, Paris.

OECD (1996c), The Public Employment Service: Austria, Germany, Sweden, OECD, Paris.

OECD (1996d), Labour Market Policies in Switzerland, OECD, Paris.

OECD (1997a), Employment Outlook, OECD, Paris.

OECD (1997b), The Public Employment Service: Belgium, OECD, Paris.

OECD (1997c), Enhancing the Effectiveness of Active Labour Market Policies: A Streamlined Public Employment Service, OECD, Paris.

OECD (1997d), Implementing the OECD Jobs Strategy: Member Countries' Experience, OECD, Paris.

OECD (1998a), Early Identification of Jobs Seekers at Risk of Long-term Unemployment: The Role of Profiling, OECD, Paris.

OECD (1998b), The Public Employment Service: Greece, Ireland, Portugal, OECD, Paris, forthcoming.

OECD (1998c), Implementing the OECD Jobs Strategy: Progress Report, OECD, Paris.

Park, N., B. Power, W. Riddell and G. Wong (1996), ‘An Assessment of the Impact of Government-sponsored Training’, Canadian Journal of Economics, 29 (Special issue), Part 1, pp. S93–S98.

Richardson, J. (1998), ‘Do Wage Subsidies Enhance Employability? Evidence from Australian Youth’, London School of Economics Centre for Economic Performance Discussion Paper No. 387.

Scarpetta, S. (1996), ‘Assessing the Role of Labour Market Policies and Institutional Settings on Unemployment: A Cross-country Study’, OECD Economic Studies, 26, pp. 43–48.

US Department of Labor (1995), What's Working (and What's Not): A Summary of Research on the Economic Impacts of Employment and Training Programs, Office of the Chief Economist, January, Washington, D.C.