Contracting employment services to outcomes: is this the promised land?

Posted on 13/11/2018 by pdavidson

Learning to swim at the deep end

Posted on 13/11/2018 by pdavidson

The end of Work for the Dole?

Work for the Dole was the flagship employment program of the Abbott Government (Abbott reprised his role as the program’s founder as a Minister in the Howard Government). From July 2015, working for benefits for at least 15 hours a week became the ‘default activity’ for 6 months of every year for unemployed people. One fifth of employment services spending is devoted to the program – almost $300 million a year. Aboriginal and Torres Strait Islander peoples in remote communities face continuous Work for the Dole for 25 hours a week.

The Turnbull Government seems less enthusiastic: it has cut the program back: instead of Work for the Dole after 6 month’s unemployment, from April 2017 unemployed young people will be enrolled in the ‘Youth Jobs Path’ program which combines work-readiness training with private sector internships.

Work for the Dole has always been controversial, but does it work? This blog looks at recent evidence of the impact of ‘work for benefits’ schemes on transitions to paid employment in the UK and a 2014 evaluation of Work for the Dole in Australia.

The birth, decline and re-birth of Work for the Dole

Prime Minister Tony Abbott’s first Ministerial gig in the Howard Coalition Government was in employment services. One of his first programs was ‘Work for the Dole’. Abbott made it clear that one purpose of the program was to increase incentives for people to leave benefits.

‘If they’re not working for wages they’ll be working for the Dole’ (Tony Abbott, Minister for Employment Services)

Work for the Dole – in which long term unemployed people were compelled to work for their benefits for community organisations for 15 hours a week for 6 months –  was the Howard Government’s flagship employment program. By the mid 2000s, it was the default compulsory activity for long term unemployed people for 6 months of every year. ‘Job avoiders’ had to work for their benefits for 25 hours a week for 11 months a year.

The Rudd Labor Government kept Work for the Dole as one of a series of compulsory program options organised by Job Services Australia (JSA) providers for long term unemployed people, but it was no longer the ‘preferred’ option. This meant it was no longer the ‘default’ activity for people who failed to choose another program, and was no longer separately funded (instead JSA providers had budgets to draw upon to finance a range of options). Providers mainly chose to refer long term unemployed people to vocational training programs instead and participation in Work for the Dole plummeted.

Now it’s back with a vengeance! The program was expanded across the country from July 2015. People on Newstart Allowance for 12 months must undertake 15 hours a week’s unpaid work for 6 months, unless they’ve already enrolled in an alternative employment or training program approved by their ‘Jobactive’ (the new JSA) provider. People under 30 years will work unpaid for 25 hours a week after 6 months’ unemployment.

In remote Indigenous communities, people on unemployment payments face 11 months a year of working for the dole for 25 hours a week – the same as the ‘job avoiders’ of yore. Unlike the previous Community Development Employment Program in which they worked in community-controlled services and received ‘top up’ payments (so that it was more like waged employment), the new ‘Community Development Program’ (CDP) offer unpaid work and other unspecified ‘activities’ that are mainly coordinated by for-profit employment services companies. The upshot has been a steep rise in financial penalties for Aboriginal and Torres Strait Islander people for breaching their 25 hour a week requirement.

Controversy

From its beginning in 1997 Work for the Dole attracted controversy.

Some argued that unemployed workers were being exploited in work for which they were paid less than the hourly minimum wage. Former Opposition leader Kim Beazley called the program a “chain gang”.

In anticipation of these claims, the original scheme was limited to 15 hours a week – roughly equivalent to an hourly ‘training wage’. Later, a $10 a week allowance was paid for travelling expenses. In 2006, a ‘full time’ variant of Work for the Dole requiring 25 hour’s work each week (about $10 per hour in current values) was introduced as a punishment for people identified by Job Network providers as ‘job avoiders’. Interestingly, it was rarely used because providers preferred to keep unemployed people under their own close supervision while they searched for jobs.

Others criticised the nature of the work on offer (usually ‘make work’ activity in low-productivity jobs) and the lack of training. Supporters argued the program improved work motivation and ‘soft skills’ like punctuality and communication and improved self esteem. However, in an early official evaluation of its impact on self esteem. Dr Winefield found that it had no impact and that participants were already strongly motivated to work:

“Although, WFD had no impact on participants’ self-esteem, it had clear ongoing benefits on other measures of psychological well-being, although whether these benefits were lasting or short-lived remains to be seen. Research has shown that self-esteem is sensitive to changes in employment status. The current finding implies that participants did not perceive WFD as a change in their employment status.  Similarly, although WDF had no impact on work involvement, the level of work involvement was so high that there was little room for improvement.” (Winefield, A 1999, “Measuring the impact of the Work for the Dole pilot projects on work ethic and self esteem.”)

The views of participants varied: many objected to being forced to undertake work they felt wasn’t helping them find paid work while others enjoyed the work or companionship with fellow unemployed workers.

Crucially, Work for the Dole is not a labour market program. It does not have an explicit goal of helping people secure paid employment. Its official goals are mixed. They range from fulfilling a ‘moral’ requirement that unemployed people should ‘give something back’ for their benefits, to discouraging people from staying on benefits, to building soft skills and raising self esteem, and undertaking work of benefit to local communities. Depending on the audience and political mood, it has been portrayed either as punishment for ‘bludgers’ or a balm for the psychological harms of prolonged unemployment.

The program’s chameleon-like character makes it hard to evaluate objectively. To my way of thinking, the litmus test for any employment program is whether it improves people’s job prospects. If not, then there’s no good reason to force people to undertake it. Whether this is the case has been debated among experts from the outset.

Evaluating employment programs

The best way to assess this is a ‘net impact evaluation’, in which employment and ‘off-benefit’ outcomes are compared over time with outcomes achieved by a similar ‘control group’ of unemployed people not referred to Work for the Dole. Employment programs have three different effects:

1. Referral or ‘threat’ effects: Some people leave benefits rather than participate in the program
2. Attachment or ‘lock-in’ effects: Participation in some programs reduces job search activity
3. Program effects: Ideally, participation in a program improves employment prospects, for example by improving skills, putting people in contact with employers who might take them on, or by improving the efficiency of job search.

The overall impact of an employment program equals the referral effect + the program effect – attachment effects. Quantitative studies should be supplemented by qualitative research that examines what actually happens to unemployed people and prospective employers when they participate in a program. Here I focus on net impact studies of work for benefits programs.

Previous studies of employment impacts

Economist Jeff Borland  conducted the first and only independent net impact study of Work for the Dole. He found that participation in the original pilot scheme for young unemployed people in the late 1990s actually reduced the probability that unemployed people had left income support 12 months after referral to the program. He suggested this was due to attachment effects (they were too busy with Work for the Dole to search for jobs effectively). There appeared to be no significant program effects.

Subsequent official net impact studies by the Employment Department found reductions in reliance on unemployment payments after referral to Work for the Dole. A study in the mid 2000s found that referral to the program reduced reliance on income support 12 months later by 6.1 percentage points. (Dept of Education Employment and Workplace Relations 2010, ‘Labour market assistance, a net impact study.’)

That’s not a bad outcome for a labour market program but I’d be more confident of the results if they were backed by independent studies. International studies of the effects of work for benefits schemes give little cause for optimism. An early literature review by MDRC Research, a prominent independent employment program evaluator in the US, found that work for benefits schemes did not improve people’s job prospects. A later report by the New Zealand Department of Labour concluded that neither New Zealand’s ‘community program’ nor similar programs in other countries improved job prospects.

Evaluating Work for benefits schemes in the UK

The most thorough recent evaluations of work for benefit schemes were undertaken in the UK: a country with a rare record of careful independent evaluation of its labour market programs. Here we examine recent evaluations of two UK work for benefit programs: the Mandatory Work Activity (MWA) program introduced in 2011 and the Community Activity Program (CAP) introduced in 2013.

Mandatory Work Activity

Mandatory Work Activity comprised 4 weeks of compulsory work for benefits for 30 hours a week. It was targeted towards people at risk of long term unemployment. It attracted much controversy and a successful British High Court challenge by an unemployed woman required to work for her benefit for a private employer.

The official evaluation of MWA found significant referral or ‘threat’ effects on the likelihood of benefit receipt:

“in the first 3 months MWA decreased the likelihood of claiming benefit by up to 5 percentage points: the proportion of referrals in receipt of benefit was 77% at week 10 compared to 82% for non-referrals at week 10. This strongly suggests that MWA had caused a deterrent effect of claiming benefit – a very high proportion (64%) of those who did not start signed off their JSA claim.

But, as shown in the graph below, these effects were reversed within another three months, as many of those who left benefits returned:

“However, in the subsequent period the impact decreased, returning to zero by 21 weeks following the referral (benefit receipt for both referrals and nonreferrals was 74%). This means that in this period a higher proportion of individuals returned to benefit amongst referrals than non-referrals. Overall, the benefit impact during the first 21 weeks equated to referrals being off benefit for an average of about 4 days more than had they not been referred.” p39.                      

                      Effects of Mandatory Work Activity on benefit reliance

Note: green line represents MWA participants

Crucially, the program had no impact on people’s employment prospects (so those who left benefits weren’t for the most part getting jobs and some were diverted to disability pensions). This is shown in the graph below:

“unlike for benefit, a MWA referral had no impact on the likelihood of being employed compared to non-referrals. This suggests that although the benefit impacts suggested a deterrent effect from claiming benefit for the first 10 weeks, individuals were not going into employment.” p40                            

                        Effects of Mandatory Work Activity on employment

Note: green line represents MWA participants

These data suggest that this short work for benefits program had significant ‘threat’ effects that were later reversed, but no impact on the future job prospects of unemployed people. What did the Government do when it received this evaluation? It expanded the program! As labour market researcher Jonathan Portes pointed out in a blog:

“Briefly, what the analysis shows is that the programme as currently structured is not working. It has no impact on employment; it leads to a small and transitory reduction in benefit receipt; and worst of all, it may even lead to those on the programme moving from Jobseekers’ Allowance to Employment and Support Allowance.  Sadly, the DWP’s press release doesn’t mention any of these key points. Nevertheless, one would naturally assume that as a consequence the programme would be redesigned to achieve better outcomes.

However, in a Written Statement, the Minister for Employment said:    ‘I am also pleased to announce the Government has decided to expand the Mandatory Work Activity scheme. The expansion will enable Jobcentre Plus to make between 60,000 and 70,000 referrals to Mandatory Work Activity each year, based on the current experience of the scheme, at a cost of an additional £5 million per annum. This decision has been taken as the result of careful consideration of the positive impacts demonstrated within the Impact Assessment.’ Unfortunately (says Portes) it is very difficult indeed to reconcile this statement with the impact assessment itself.”

Community Action Program

In 2013 a new work for benefits program called Community Action Program (CAP) was trialled for very long term unemployed people (2 years out of work). This was carefully evaluated as part of a ‘very long term unemployed trailblazer’ scheme.

The scheme had these features:

• “A randomised controlled trial to give a high degree of confidence that any observed differences in outcomes are attributable to the support options,

• A 13 week pre treatment period to test deterrence, informing and maintaining awareness of forthcoming support throughout the period, and

• A 26 week period of treatment consisting of either:

–     Standard Jobcentre Plus support – this is the control group within the trailblazer and consists of flexible and personalised adviser based support.

• Community Action Programme (CAP) – delivered by contracted providers and comprising 26 weeks of full time work experience or provider-led job search support.

• Ongoing Case Management (OCM) – an intensive offer of flexible and personalised adviser based support, delivered by Jobcentre Plus (JCP) through increased interventions over 26 weeks.” p6

A randomised trial is the ‘gold standard’ of net impact studies. Unemployed people were randomly assigned to a treatment group referred to the CAP,  an alternative scheme called ‘ongoing case management’ (an intensive form of job search assistance), or a control group who received ‘standard JobCentre Plus support’ (mainly job search assistance) as shown in this flow chart:

The likelihood that unemployed people involved in the trial were still receiving income support up to 91 weeks after referral to each of these three options was compared. The results are shown in the graph below: 68% of benefit recipients in  ‘standard JobCentre Plus support’ still received benefits 91 weeks later, compared with 66% of those referred to CAP and 64%  referred to ‘case management’. So referral to CAP reduced benefit reliance by 2%, which was 2% less than the impact of case management.

Effects of CAP, Case Management and standard job search assistance on benefit relianceThe graph below compares employment outcomes for those referred to CAP and the other two options in the trailblazer. It turns out that CAP had no significantly greater impact on employment than the standard ‘Jobcentre Plus’ option. Ongoing case management did increase the likelihood of employment compared with the standard service, by just 1%.

So the recent British evaluations confirm what others have found: referral to work-for-benefits programs increases the likelihood that people will leave benefits, but many soon return. Participation these programs  has no significant impact on the likelihood of paid employment. A fair summary of the pros and cons of these programs was offered by Tony Wilson from CESI, the British employment services advocacy body. His conclusion: it’s reasonable for unemployed people to be required to join programs that help them secure a job and Governments should fund them, but work for benefit schemes generally aren’t worth the effort.

Evaluation of the 2014 Work for the Dole scheme

When the Abbott Government restored direct funding for Work for the Dole and also its previous ‘default program’ status in 2014, it authorised an evaluation of the impact of the new scheme in a number of ‘trial’ areas.

This found that in those areas, the new scheme increased participation in Work for the Dole by 28 percentage points for unemployed people with similar characteristics. So it was successful in reviving the program. However, 27% of participants did not complete the program and 60% of unemployed people responding to a survey agreed that the program’s aim was to ‘put people off claiming benefits’. Consistent with this, the report found that:

‘There is clearly stigma associated with the programme and those who participate in it.’

The new program was less successful in helping people secure jobs.

While the report estimated that the program:

increased the rate of part-time/casual paid employment reported to JSA providers by 7 percentage points more than in the comparison areas controlling for other characteristics (from a baseline of 13%).

it warned that this was due to an unknown extent to people declaring existing employment on referral  (i.e. a ‘threat effect’).

The report estimated the impact of the program on placement in new jobs as follows:

‘an additional 2 percentage point increase in the probability of job seekers having a job placement controlling for other characteristics (from a low baseline of 14%).’

It’s not clear from the report exactly how the net employment impact of the 2014 scheme was measured: the report refers to a ‘difference in differences’ method using administrative data, but warns that it was too early in the program’s development to properly measure its employment impact. Unlike the British evaluations, there was a lack of longer-term data to assess how many people returned to benefits after leaving them  (The report refers to ‘short term’ employment impacts but it’s not clear how soon after referral to the program these were measured).

A key finding of the 2014 evaluation  on the employment impact of Work for the Dole appears to be that participation in the scheme increased the probability of future employment by 2 percentage points (interestingly, the same as the British CAP program), which may well be within the net impact study’s margin of error. The evaluation suggests that ‘threat effects’ (such as declaration of existing employment) are significant but does not appear to measure them directly. Yet, similar ‘threat effects’ would probably be achieved by any program of compulsory, intensive ‘activation’ of unemployed people.

Participation in an employment program should make a difference to people’s job prospects and work-for-benefits schemes do not do this. To refer people to a scheme that has no impact merely to discourage them from staying on benefits is a cynical exercise.

Conclusion

To summarise: six months of compulsory work-for-benefits has none or very marginal impact on the future job prospects of long term unemployed people and the $300 million spent on the scheme would be better invested in other programs. Even the standard job search assistance offered by Jobactive providers (limited as this is for most people) probably yields better employment outcomes.

The Australian and British Governments are among the biggest investors in work for benefits schemes. Despite public commitments to ‘evidence based policy’ and ‘value for money’, these Governments are slow learners.

Activation policies in the 1990s and 2000s: Denmark and the United Kingdom compared.

I haven’t blogged for a while. My excuse is that I’m writing up my PhD thesis comparing the emergence of activation policies in four countries: Australia, United Kingdom, Denmark, and the Netherlands. As you can imagine, this is taking a while!

In case this is a topic that interests you, here’s a sample of my work: a paper I presented at the FISS conference in Sigtuna Sweden in 2014 comparing activation in the UK and Denmark:

https://www.researchgate.net/publication/272482524_Comparing_activation_policies_in_Denmark_and_the_United_Kingdom_testing_the_convergence_thesis

The conclusion? These two pioneers of activation policy took the same ideas – structural unemployment, activation, and New Public Management, and implemented  them differently. They put the activation policy jigsaw together in different ways. Path dependency was at work here: the two countries had very different sets of labour market and social security institutions, and still do.

The ‘Qualification Program’: Norway’s answer to entrenched long term unemployment?

With an unemployment rate of 4.1%, Norway must be getting a few things right. In 2007 they did what few countries do: seriously invest in an employment program to reduce entrenched, long term unemployment.  The two-year ‘Qualification Program’ aims to overcome social barriers to work as well as low skills. Participants must undertake full time activities and caseloads are 1:18. In a new twist on work incentives, they receive higher income support than similar unemployed people. It’s not cheap, but 4 years after starting the program, long term unemployed people are 18% more likely to be employed.

Seven years after the global recession, long-term unemployment is worrying experts and policy makers across the world, even in the US where time-limited benefits and a ready supply of low skilled low paid jobs previously seemed to hold it at bay. Yet few countries are getting to grips with the core of the problem: by investing in programs that bring the most disadvantaged long term unemployed people into jobs. One exception is Norway, which with an unemployment rate of 4.1% can boast success in labour market policy.

In 2007, Norway introduced its ‘Qualification Program’ (or ‘QP’) for people at risk of prolonged unemployment, including those who lack unemployment insurance entitlements and rely on municipal social assistance payments: a group often neglected European labour market programs.  The name is a bit misleading: its aim is to ‘qualify’ people for jobs by overcoming social barriers to work as much as weak vocational skills.The program combines secure and generous income support (higher than most social assistance payments) intensive employment and social support over a two year period, and a requirement to engage full time in those activities.

It is being  evaluated with an emphasis on its long term impacts. One evaluation found that it boosted the probability of employment for a group of unemployed people whose prospects were otherwise weak by an average of 10% after 2 years and 18%  after 4 years. This comes at a price: the average annual cost of additional employment assistance through the program is EUR$6,000 spread over 2-3 years. The higher unemployment benefits cost a similar amount. Much or this is eventually retrieved as more program participants move off benefits: a key evaluation question is how much.

Context

Norway has two systems of unemployment payment: unemployment insurance paid by industry based social insurance funds which requires a contribution history, and social assistance which is run by  municipalities.  Unemployment insurance is relatively generous but social assistance payments are generally lower and are not legislated. Municipalities have  discretion in setting rates of payments and other conditions of payment.

In 2007 when the QP was introduced the Norwegians faced a common problem in European unemployment benefit systems: employment assistance was fragmented and there was poor cooperation between the public employment services and municipalities to assist people on social assistance payments to secure work. A common assumption was that social assistance clients were not ’employable’.

One part of the solution was a merger of the benefits agency and public employment service, and formal collaboration arrangements between this new agency and municipal welfare offices to meet the employment needs of both insured and uninsured unemployed people. Offices of the new ‘NAV ‘agency continue to serve insured unemployed people but are co-located with municipal social services to make labour market assistance more accessible to people on social assistance payments.

The other part was the QP which has been rolled out region-by-region in conjunction with the establishment of NAV offices. This facilitates evaluation since similar people are assisted through the ‘old’ and ‘new’ systems in different localities.

Target group

The Programme is targeted at people:

•  with ‘significantly reduced earning ability’ and no or limited national insurance rights;
• 􀂃 who are ‘trapped’, or in danger of getting trapped, in a passive situation characterised by income poverty; and
• 􀂃 who are considered to have a chance of getting a job through individual follow-up, even if this implies a lengthy process.

This assessment is conducted by the NAV, which administers the program. As the above criteria suggest, there is much room for discretion when making referrals. Assessors do not use a statistical assessment tool such as the Australian Job Seeker Classification Instrument. Most participants previously received municipal social assistance but unemployment insurance recipients may also meet the criteria.

Participation is voluntary but once enrolled, participation in employment related activities is a condition of receipt of the ‘Qualification Benefit’ (discussed below).

The table below compares unemployed people referred to the QP and those not referred. QP participants are much more likely to be women, migrants, and to lack complete high school qualifications. They are much less likely to have labour earnings over the previous year. This all suggests that QP participants are significantly more disadvantaged than the average unemployed person.

Income support

One innovative feature of the QP is that participants receive higher income support payments (in most cases) than they would if on social assistance. This is based on the view that people who are severely disadvantaged in the labour market need stability in their incomes and lives in order to fully benefit from employment assistance. This is a  departure from the conventional view (informed by neo-classical economics and research on transitions to employment among less disadvantaged unemployed people) that lower benefits and time limits spur people to search for jobs more intensively.

I suspect it’s also an attempt to patch a few holes in the social assistance system, which (unlike most countries) is not based on uniform legislated entitlements. Instead each municipality determines rates of payment and this has led to inequity and inconsistency.

A second innovation is that the payment is not income tested. If an unemployed person secures full time work within the two year period of the program they keep the full benefit as a wage supplement.

The qualification benefit for a single adult is NOK145,762 (approx. EUR17,000) per year for participants 25 years or older, while those below 25 years receive 2/3 of this.

By contrast, unemployment insurance for a single person is 62% of the previous wage and the average wage is around EUR50,000 (so an former average wage-earner might receive EUR30,000) which can be claimed for up to two years. Those ineligible for unemployment insurance may transfer to the much lower social assistance payments.

So the ‘qualification benefit’ is high by OECD standards and higher than what social assistance clients would otherwise receive, but lower than unemployment insurance.

Employment assistance

In return for the Qualification Benefit, participants must engage in 37 hours a week of work preparation activity negotiated with the NAV. Examples of these activities are provided below.

As these examples show, a wide variety of activities are included on a spectrum from immediate referral to an employer or work experience placement to therapeutic programs and sheltered employment for those assessed as most disadvantaged.

Participation in the activities is compulsory, and continues until the individual secures paid employment (in which case they continue to receive the benefit) or abandons the benefit.

A key feature of the program is the low caseloads. Each case worker has only 18 clients at a time.

Impact

A quantitative evaluation of the impact of the program on employment found that the probability of being employed fell by an average of 9% in the first year (due to the ‘lock in effect’ of program activities), then rose by 10% after 2 years and 18% after 4 years. The graph shows the average proportion of each year spent in employment before and after entry to the program.  The decline in employment in the years leading up to program entry was attributed to the labour market disadvantage of participants, and the improvement was attributed to the program.

These are strong results for a program targeting disadvantaged, long term unemployed people. There are a few qualifications: most of the jobs obtained were part time and low paid, and after four years, the average cost of the program was still not recouped through lower benefit payments. The latter result is due in part to the  design features of the qualification benefit: for the two years of the program the benefit is paid regardless of labour earnings at a level higher than most social assistance payments.

How well does it work and could it work better?

Few programs are effective in assisting the most disadvantaged long term unemployed people into jobs, on a large scale. Activity requirements and job search assistance alone are not often effective for that client group. They need more intensive (therefore costly) programs to build work capacity, link them with a suitable employer (not just the first job available), and ensure that it lasts.

‘Lock in effects’ from intensive work experience training, or therapeutic programs are common because they delay job search. What matters here is the strength and sustainability of the subsequent ‘treatment effect’. If the QP does boost average employment impacts by 18% after four years, the delayed impact is probably worth it. A 10 percentage point improvement in job prospects after two years is also pretty good, especially if their chances were very low to begin with.

If these employment outcomes were full time, sustained jobs it’s very likely the program would eventually pay for itself, since most participants were otherwise likely to remain on income support (and not pay income tax) for years to come. However, the evaluation finds that the program is not yet ‘cost neutral’ four years after program entry (on average). The extra costs (the higher qualification benefit, more intensive employment assistance, and loss of potential earnings in the first two years) still exceed the gains at that stage though the gap is closing. The evaluation suggests that the employment gains after four years would have to be sustained for longer before the financial benefits of the program to overtake its costs.

The evaluation may have exaggerated the costs to government. To assess this accurately it would have to subtract the unemployment payments clients would have received on average over the four years in the absence of the program. Also, no account is taken of the effects of the program on income tax receipts.

There are three uncertainties about the estimated impacts of the program.

First, did the evaluation effectively control for ‘unobserved characteristics’ and the possibility of selection bias into the program? While it targeted disadvantaged job-seekers, voluntary programs usually involve an element of ‘creaming’ since the more motivated and confident you are, the more likely you are to join. The risk of ‘creaming’ is heightened by the discretionary nature of assessment of eligibility for the program.

Second, to what extent are the strong long-term results due to the the fact that the program is new and run on a (relatively) modest scale?  It’s common for a new program assisting a group who previously received little or not help (and were not expected to actively search for jobs) to have good outcomes. Will its effectiveness decline once it’s brought up to scale?

The third question is the most interesting: to what extent can the results be attributed to the qualification benefit or the more intensive employment services? Financial incentives to take up employment (such as the lack of an income test) often don’t work well for disadvantaged job-seekers because their main problem is not a lack of incentive to take up job offers: it’s that they receive too few of them.

The main advantage of the higher and more stable qualification benefit is likely to more subtle: that it supports and rewards people who commit themselves to the program. An interesting feature of the benefit is that it is paid like a wage (by a payroll office) rather than a social assistance benefit.

I suspect it is the more intensive employment services (and the requirement to stick with them to keep the benefit) that make the most difference. A caseload of 18 people is well below the 100 or more in most employment service systems (including Australia’s). If the case workers are well trained and focus on employment outcomes, that alone should make a big difference.

Here there appears to be a weakness in the implementation of the program: most of the caseworkers apparently come from a municipal social service background.  Their focus seems to be mainly therapeutic (removing barriers such as homelessness or addictions before turning to employment goals). In other countries where social assistance recipients were first ‘activated’ into the labour market, including Denmark and the Netherlands, considerable effort was devoted to retraining and re-orienting  caseworkers from social work to employment assistance.

The nature of the employment assistance received is also critical. In the examples above, the client referred to work experience in a regular workplace was much more likely to succeed than the one referred to ‘social activities’ and sheltered employment. Of course, the latter was likely to face steeper barriers to employment (and that’s why she was referred to those programs). However, the evaluation evidence strongly suggests that even in such cases a program is more likely to succeed if participants are placed as early as possible to a regular work environment where they are treated (as far as possible) as a regular paid employee. This was the experience, for example with Individual Placement and Support (IPS) services for people with severe mental illness in Australia which are also being trialled in Norway.

On the other hand,  programs that emphasise social support or work outside mainstream employment, and neglect the employment objective, such as the ‘Special Municipal Projects’ in Denmark  and Personal Support Program in Australia are relatively ineffective in improving job prospects due to their  ‘lock-in effects’.

Conclusions and suggestions

The QP appears to have achieved what few programs assisting the most disadvantaged unemployed people do: a significant and sustained  improvement in their job prospects (of more than 10 percentage points) over the long term. Norway deserves to be congratulated for investing in long term unemployed people when other countries that once were once pacesetters in employment assistance for that group (including Australia, the Netherlands and the United Kingdom) seem to have retreated from earlier commitments  in pursuit of short term budget savings (by either failing to subsidise intensive assistance for people unemployed long term or sharply reducing labour market assistance spending across the board).

It remains to be seen whether the positive early results of the QP will hold as the program matures and is brought up to scale.

It’s important to target the program to those most likely to benefit, and understand which elements of the program contribute most to these outcomes, because the program is loosely targeted and relatively costly and it may not need all of its elements to sustain its outcomes as it is scaled up.

Targeting would be improved by using a statistical assessment tool such as Australia’s Job Seeker Classification Instrument, in addition to long term unemployment status. This would also facilitate evaluation.

Previous evaluation evidence suggests that low caseloads and intensive support with a strong and direct employment orientation make the greatest difference for this target group. It’s important that this element is maintained and upgraded.

The benefits of this intensive support would be maximised if the program had a consistently strong emphasis on employment in the regular labour market, rather than sheltered jobs or social support in its own right. Experience with similar programs suggests that this requires considerable investment in recruitment and support for prospective employers.  Caseworkers  need a combination of labour market knowledge and nous, and the skills and social awareness to deliver the social support component of the program.

On the other hand, employment incentives are probably less important for this target group than for less disadvantaged jobseekers. The program may be equally effective without a non-means tested benefit payment. For example, a 50% reduction in benefits for every Krone earned may still encourage people to take on part time jobs (and this is more liberal than most social assistance income tests).

Adequate and stable income support is nevertheless important – both to prevent poverty and support engagement with the program. If this is lacking in Norway’s social assistance system, maybe it’s time to give it a firm legislative basis as other countries have rather than leave vital matters such as rates of benefit to the discretion of each Municipality. If social assistance clients need financial encouragement and support to participate in programs like QP, consideration could also be given to a modest ‘program participation supplement’.

Finally, robust evaluation is essential: both to establish which elements of the program work best and in what combination; and to win the inevitable arguments over secure and permanent financing. Finance Ministries care about long term fiscal impacts so its worth evaluating them carefully and thoroughly (even though the social impacts are equally important). The road to effective activation of long term unemployed people is littered with too many small scale, temporary, pilot schemes. The QP is a rarity that’s worth preserving and improving.

Scarcity: The use and abuse of psychology to reduce unemployment.

The idea that psychology (rather than shortages of jobs or skills) plays a role in unemployment is controversial. Yet employment service providers report that motivation makes a difference. A decade ago, an Australian Government study argued that the attitudes of unemployed people influenced their job prospects. But it left many questions unanswered: what lay behind those attitudes? Where they permanent or a fleeting response to current circumstances? A new book by American economist and psychologist team Mullainathan and Shafir gives us new insights into how people respond to poverty. It reveals how ‘scarcity’ shapes people’s lives and behaviour. This has implications for the way governments treat unemployed people.

Along with employment opportunities, skills and financial incentives, the motivation of unemployed people to search for work has an impact on their employment prospects. This can be viewed in three different ways.

The ‘homo economicus’ view holds that this is due to financial incentives. If people don’t increase their incomes much by working then they might ‘rationally’ decide to remain on income support. Financial incentives are always important, but don’t seem to explain much of our 6% unemployment rate. Most unemployed people are single. A single unemployed adult would more than double their disposable income by taking a full time job at the minimum wage. The rate of unemployment in Denmark is similar yet unemployment insurance payments there are typically equal to 90% of a modest fulltime wage.

A second view is that people’s attitudes – shaped by their family and social environment – hold them back from improving their lives. This is the so-called ‘culture of poverty’ theory. Advocates of this view such as Laurence Mead , argue that labour market reform and better financial incentives won’t do much to reduce unemployment. Paternalistic polices (such as Australia’s ‘Work for the Dole’) are the answer. While obligations to search for jobs do reduce unemployment, evidence is lacking that most social security recipients suffer from the kind of social ‘dysfunction’ hypothesised by Mead.

A third view, based on new research informed by behavioral economics, holds that the experience of poverty and unemployment itself holds people back. In addition to a lack of access to jobs, skills and opportunities, people living in poverty experience scarcity, and scarcity messes with the mind.

(1) It’s about attitude

The view that people’s attitudes towards unemployment affect their job prospects was presented in a study commissioned by the Howard Government in 2001. This ‘attitudinal segmentation’  study commissioned by the Australian Employment Department classified unemployed people according to their attitudes towards unemployment as (among other categories) ‘drivers’, ‘cruisers’ or ‘withdrawn’:

“In 2001 the Department of Employment and Workplace Relations (DEWR) commissioned a study to develop a model of job seeker attitudinal segmentation, in which job seekers were grouped based on their attitudes towards finding work. The model was developed in several stages. In-depth interviews with a range of job seekers were first undertaken to identify what underlies job seekers’ attitudes and drives their behaviour. This research resulted in the development of a framework of eight job seeker segments.

Job seekers were asked to agree or disagree with each of the attitudinal statements using a 10-point scale for their responses. Extensive statistical analysis of responses to the attitudinal statements was undertaken to develop an algorithm to allocate job seekers to segments, as well as to quantify the segments.” p3

The research used the same tools applied by market researchers to segment their potential customer base. The results of this research are summarised in the figure below:

How might such research be used? The Department argued that:

“The characteristics associated with each of the segments can help identify the types of
assistance required to most effectively generate the desired behaviours of active job search, economic participation and reduction of income support reliance. The survey also identified that more than half (57%) of job seekers were in the less motivated segments. This indicates a strong focus for assistance on increasing the “benefits” of job search and paid employment and decreasing the “costs” of moving from unemployed to employed.” p20

There’s nothing novel about this prescription. It comes from the ‘homo-economicus’ textbook with its emphasis on work incentives. It’s a short step from here to punitive ‘solutions’. In  a presentation in 2003 which referred to the attitudinal segmentation research as evidence for welfare ‘dependency’, Peter Saunders, then with the Centre for Independent Studies (CIS), cited Mead in support of time limits on unemployment payments:

“As Lawrence Mead puts it: “Disadvantaged people without jobs find no end to reasons why working is impossible for them… They avoid personal responsibility and blame circumstances beyond their control…a mentality is at work that refuses to believe that opportunity exists, even when it does” (Lawrence Mead, ‘The New Politics of Poverty’ Basic Books, New York, 1992, pp.143-5)” p13

The political reaction to this research was predictable. The then Employment Minister turned his attention on the 16% of job-seekers identified as ‘cruisers’, argued publicly that ‘cruising’ was the reason many failed to show up at compulsory interviews, and promised to ‘get tough’.

Missing from the ideological reactions to this research was any serious attempt to work out where these ‘attitudinal’ responses to unemployment came from. Were they due to a ‘mentality’ among poor people as Mead suggests, were people reacting to financial incentives (the payment of benefits), or was something else going on? Until we work his out, classifying unemployed people according to their ‘attitudes’ will only reinforce stereotypes.

There were plans to use the attitudinal ‘segments’ identified in surveys such as this one to inform the provision of employment services, but they came to little. The idea that the kind of assistance (or punishment) unemployed people receive should depend on their responses to a psychological survey carried serious risks for unemployed people and governments. ‘Orwellian’ is a word that springs to mind.

Some providers do use their own assessments of people’s psychological reactions to unemployment to inform service provision. One approach used in Australia takes account of the impact of fear and self esteem on motivation among job-seekers. Providers using this model place less immediate emphasis on active job search where they identify these factors as barriers to employment.

Other providers use their own assessment of people’s motivation as a signal of whether it it is worth investing in employment assistance. Since motivation is not measured by the assessment tool used by Centrelink to determine the financial rewards for providers if they find someone a job (and it would probably not be possible to measure it accurately), there are financial benefits for providers who can identify which of their most ‘disadvantaged’ job-seekers are more ‘motivated’ (and therefore more likely to find a job with a bit more support). The problem with this strategy is that people whose motivation is impaired by severe labour market disadvantage will be ‘parked’, that is denied help beyond the minimum contractual requirements. The most effective providers don’t ‘park’ people who seem to lack motivation right now. They make an effort to find out why, and work out a way to engage them by holding out the prospect of employment.

(2) It’s about how we respond to our circumstances

An alternative view is that poor people’s attitudes and responses to their circumstances are shaped by their experiences at the time rather than a fixed outlook on life (e.g. ‘ambitious’ or ‘lazy’), or economic incentives.

A new body of research in psychology and behavioural economics offers insights into the way people adapt to poverty. To my mind (as someone who has worked with unemployed people), these insights are more grounded in reality than either ‘homo-economicus’ or the view that unemployment and poverty are signs of deeper social dysfunction. This is not to say that economic incentives and social background are unimportant: in some cases they make all the difference. It’s just that they don’t provide a firm evidence base for policies to reduce poverty or unemployment in the majority of cases.

A new book by American economist and psychologist team Mullainathan and Shafir neatly summarises a raft of recent research on how people respond to ‘scarcity’ . As the authors point out, it’s not just a scarcity of money that affects behaviour. Scarcity of time can produce the same results.

“Scarcity captures the mind. When we experience scarcity of any kind we become absorbed by it. …For the hungry, that need is food. For the busy it might be a project that needs to be finalised. For the cash strapped it might be this months’s rent payment. ….Scarcity changes how we think.”

The authors note that this has a positive side: people experiencing poverty are often better at managing their weekly budget because their minds are focussed on financial survival. But there’s a downside:

“Because we are preoccupied by scarcity, we have less mind to give to the rest of life. We can directly measure mental capacity or, as we call it, ‘bandwidth’. We can measure fluid intelligence (a key resource that affects how we process information and make decisions). We can measure executive control, a key resource that affects how impulsively we behave. And we find that scarcity reduces all these components of bandwidth. Being poor reduces a person’s cognitive capacity more than going without one full night without sleep. It’s not that the poor have less bandwidth as individuals. Rather the experience of poverty reduces anyone’s bandwidth.”

The authors offer a few practical examples. Firefighters in the US are more likely to die from motor vehicle accidents than other people. In their rush to respond to a call, with their minds fully occupied with dozens of urgent tasks, they often forget small things like buckling their seat belts. Similarly, the greatest risk from mobile phone use while driving does not come from holding the phone: it’s the loss of attention to the road while we speak and listen.The authors call this ‘tunnelling’:

“The narrowing of the visual field in which objects inside the tunnel come into sharper focus while rendering us blind to everything peripheral, outside the tunnel.”

They illustrate this with psychological tests. Before undergoing a standard test of cognitive ability, people on different income levels were asked to imagine how they would deal with a situation where their car required a $3,000 in mechanical repairs. To those on higher incomes, this wasn’t too much of a distraction. It didn’t have much affect on their scores on the test. For those on low incomes it was too close to the bone. Their scores plummeted. The distraction of working through a financial problem had a similar impact on their test scores as the loss of a night of sleep.

At a practical level, ‘tunnelling’ means that people who are short of money are more likely to focus on their immediate needs than future plans. People in poverty are often too busy juggling their finances, and responding to the minor and major crises related to poverty (the threat of electricity disconnection or eviction, finding the money for a school excursion, or chronic ill health) to raise their horizons beyond short term survival.

Once again, we should careful not to fall back on stereotypes. Do all people in poverty fall into this trap of short-term thinking? Clearly not. But the psychological researched cited by Mullainathan and Shafir suggests that it is harder for people whose minds are focussed on immediate survival and crisis management to make plans to improve their circumstances over the longer term, including through effective job search.

How many unemployed people face these circumstances is an open question, but with Newstart Allowance set at $37 a day financial worries are rarely far from people’s minds. A recent Senate Inquiry into proposed changes to benefit waiting periods revealed that 70% of people who apply for Newstart Allowance seek a waiver of the standard one week waiting period on financial hardship grounds. Most enter unemployment with few resources behind them. For example, most people on  Newstart Allowance rent their housing and about half don’t own a car.

Policy applications

The insights from ‘scarcity research’ have implications for the way we provide services for unemployed people, for example, the treatment of people when they first apply for unemployment payments.

Much of the recent debate over solutions to unemployment has focussed on how new claims for unemployment payments should be treated. This is often based on the simplistic view that ‘welfare creates its own dependency’, so that once people enter the system they will become stuck there.

In the 2014 Federal Budget, the government proposed that young unemployed people be made to wait for 6 months for payment in the hope that this would spur them to find a job. Like proposals to put a time limit on benefits, this is based on a ‘homo economicus’ view of unemployment: change incentives and people will respond. Yet few labour market economists think this is a good idea. It ignores the demand side the labour market (a lack of jobs) and leaves people exposed to poverty. This is the ‘starve them into submission’ approach to unemployment. The 2014 budget proposal (and the 1 month wait proposed in 2015 after the 6 month wait was rejected by the Senate) tried to reduce the risk of impoverishment by exempting people assessed as having a higher risk of long term unemployment, but such assessments are far from 100% accurate. If there aren’t jobs available where people live that are suited to their skill level, then even less ‘disadvantaged’ unemployed people will remain out of work. This harsh approach shifts the financial risk of unemployment from government to unemployed people.

Paternalists argue for a different approach: impose stringent requirements (such as intensive job search) backed by close supervision of unemployed people. The idea is that those who can easily find employment will leave benefits more quickly, while those facing greater difficulty can be identified and offered more help. The advantage of this approach over long waiting periods is that on the face of it income support is not denied to people and it at least makes a serious attempt to treat these two groups differently. And there is evidence to suggest that supervised job search helps reduce unemployment by keeping people engaged with the job market and improving the effectiveness of their efforts to find work.

It’s not so clear when and how employment requirements should be applied. One view holds that job search and other requirements (such as working for the dole) should be imposed intensively as soon as people claim benefits, in the hope that people with good job prospects don’t persist with their claim. This idea is called ‘diversion’.

When this strategy was applied to sole parents claiming Temporary Assistance for Needy Families (TANF) benefits in the United States as part of a welfare ‘reform’ in the mid 1990s, it did turn many people away from income support, but not just those who didn’t need it. Many financially disadvantaged people who had difficulty dealing with public bureaucracies or meeting work requirements (for example due to a disability) were ‘turned away’ from benefits and experienced severe hardship. A study by Acs found that successful claims for TANF declined sharply yet the financial need for income support did not. Blank found that a decade after the US welfare ‘reform’, one in five low-income mothers at risk of poverty received neither TANF nor wages for a year at a time. Most of this group had virtually no source of income and they and their children were impoverished.

In 2005, the Australian Government trialled an experiment called ‘rapid connect’ in which unemployed people claiming benefits were referred more quickly to their first Job Network (employment service) interview. Instead of being given 14 days to register with the Job Network, they were required to do so within two days of their first appointment at Centrelink. When this was evaluated it was found to increase the probability of employment within the first 14 days, though from a low base – from 1% of applicants to 4%. Unfortunately the evaluation did not examine what happened to people who did not pursue a claim, including the extent to which they reapplied later. ‘Rapid connect’ had a downside. Unemployed people had too little time and too little information to choose the Job Network provider best suited to their needs. The majority did not make a choice and were assigned to a service by Centrelink, or simply opted for the nearest provider. This meant that people were often poorly matched with providers. If they didn’t find employment quickly, this was likely to reduce their future job prospects.

It also meant that from the outset, their relationship with employment services was passive. People were following requirements rather than taking the most effective action to find a job.

‘Scarcity’ theorists would advocate a different approach. Rather than force people to undertake intensive activities (including making their first contact with unfamiliar services) in a rush when their minds are focussed on how they will pay their rent and feed themselves, assure them that financial support will be provided if they register with an employment service and give them the time and information they need to choose one.

The first interview with an employment service provider would also be structured differently. Instead of gathering information in a mechanical way (following screen prompts) to populate a standard list of activity requirements and pushing people to sign immediately, it would better to start with an open ended discussion of how they see their career goals and job prospects, so that the focus turns to their own plans to secure employment and how the provider can help. This shifts the responsibility and initiative to where it should lie, with the job-seeker. Benefit requirements should be explained, but that could come at the end of the interview, once a positive relationship has been established with the consultant. It would help if providers were paid to spend more than half an hour with people in the first interview.

Scarcity theory also offers insights into the reasons people miss appointments with Centrelink and employment services, and what might be done about this. For at least a decade, around one third of these interviews were missed. There are often good reasons for this, including that the person has found employment or that they are ill. Yet a small minority of job-seekers account for a high proportion  of missed appointments.The main response of successive governments to this problem has been to tighten penalties for non attendance. Some of these changes were sensible – including the idea of suspending payments until the person attends an appointment, and then restoring them. Most people who still need income support quickly contact Centrelink and the appointment can be re-scheduled.

Less attention has been paid to the reasons people miss appointments. The Employment Department surveys unemployed people from time to time about their experience of employment services. Why not ask whether they missed appointments, why this happened, and what might be done to make it easier to meet their requirements? Innovations such as reminders by text message have made a difference, but more that could be done, especially for people who are likely to face the greatest difficulty such as those with limited literacy or poor mental health. Sanctions could then be limited to the minority of people who deliberately fail to comply.

When unemployed people are ‘tunnelling’ in response their immediate problems, we need to find better ways to bring their attention back to requirements such as employment service appointments. Simply threatening to cut payments may be counterproductive: some people will avoid contact out of fear. People sometimes avoid opening letters from Centrelink altogether for this reason.

Finally, there is something governments can do to reduce financial scarcity for people who are unemployed: improve the Newstart Allowance. Making people live and search for work on $37 a day doesn’t improve people’s job prospects; it degrades them.

Does Work for the Dole work?

The Australian Government is determined to expand its flagship Work for the Dole program. From July 2015, working for benefits for 15 hours a week will be the default activity for 6 months of every year for unemployed people on Newstart or Youth Allowance. One fifth of employment services spending is devoted to the program – around $300 million a year. Aboriginal and Torres Strait Islander peoples in remote communities face continuous Work for the Dole as long as they receive unemployment payments. The scheme has always been controversial, but does it work? This blog looks at recent evidence of the impact of ‘work for benefits’ schemes on transitions to paid employment in the UK.The answer: not much at all.

The birth, decline and re-birth of Work for the Dole

Prime Minister Tony Abbott’s first Ministerial gig in the Howard Coalition Government was in employment services. One of the first programs he expanded was Work for the Dole.

‘If they’re not working for wages they’ll be working for the Dole’ (Tony Abbott, Minister for Employment Services)

Work for the Dole was the Howard Government’s flagship employment program. By the mid 2000s, it was the default compulsory activity for long term unemployed people for 6 months of every year. ‘Job avoiders’ had to work for their benefits for 11 months a year.

The Rudd Labor Government kept Work for the Dole as one of a series of compulsory activities organised by Job Services Australia providers for long term unemployed people. Since it was no longer the ‘default’ activity for people who did not choose another program, and was no longer separately funded (instead JSA providers had budgets to draw upon to finance a range of options), participation in Work for the Dole diminished in favour of vocational training programs.

Now it’s back with a vengeance!

The program is being expanded across the country from July 2015. People on Newstart Allowance for 12 months will undertake 15 hours a week’s unpaid work for 6 months, unless they’ve already enrolled in an alternative employment or training program approved by their Jobactive (formerly Job Services Australia) provider. People under 30 years will work unpaid for 25 hours a week after 6 months’ unemployment.

In remote Indigenous communities, people on unemployment payments will face continuous ‘Work for the Dole’ for 25 hours a week. Unlike the previous Community Development Employment Program in which they worked in community-controlled services and received ‘top up’ payments (so that it was more like waged employment), they will have to work for benefits in community jobs and projects organised by Work for the Dole Coordinators contracted by the Government, many of whom are private for-profit companies.

Finding quality placements on such a large scale will be a challenge. The Government has even suggested that picking up and dropping off their children from school will count as participation in Work for the Dole in remote Indigenous communities.

Controversy

From its beginning in 1997 Work for the Dole attracted controversy. Some argued that unemployed workers were being exploited in work for which they were paid less than the hourly minimum wage. While 15 hours a week of work for benefits was close to the minimum hourly wage at that time, the ‘full time’ variant of Work for the Dole introduced later required 25 hour’s work each week (about $10 per hour in current values). Others criticised the nature of the work on offer (usually ‘make work’ activity in low-productivity jobs) and the lack of vocational training.

Supporters argued the program improved work motivation and ‘soft skills’ like punctuality and communication and improved self esteem, though an early official evaluation of its impact on self esteem, by Dr Winefield found that it had no impact and that participants were already strongly motivated to work:

“Although, WFD had no impact on participants’ self-esteem, it had clear ongoing benefits on other measures of psychological well-being, although whether these benefits were lasting or short-lived remains to be seen. Research has shown that self-esteem is sensitive to changes in employment status. The current finding implies that participants did not perceive WFD as a change in their employment status.

 Similarly, although WDF had no impact on work involvement, the level of work involvement was so high that there was little room for improvement.” (Winefield, A 1999, “Measuring the impact of the Work for the Dole pilot projects on work ethic and self esteem.”)

The views of participants varied: many objected to being forced to undertake work they felt wasn’t helping find paid work while others enjoyed the work or companionship with fellow workers.

Work for the Dole was not a labour market program. It did not have an explicit goal of helping people secure paid employment. Its official goals were mixed. They ranged from fulfilling a ‘moral’ requirement that unemployed people should ‘give something back’ for their benefits, to discouraging people from staying on benefits, to building soft skills and raising self esteem, and undertaking work of benefit to local communities. Depending on the audience and political mood, it was portrayed either as punishment for bludgers or a balm for the psychological damage of prolonged unemployment.

The program’s chameleon-like character made it hard to evaluate objectively. To my way of thinking, the litmus test for any employment program is whether it improves people’s job prospects. If not, then there’s no good reason to force people to undertake it. If it does stand a good chance of improving an unemployed person’s job prospects (if it’s relevant to their particular circumstances and there’s no ethical objection to the program itself) then it’s reasonable to require people to participate.

Although Work for the Dole was not formally a labour market program, the Howard Government nonetheless claimed that it improved people’s job prospects. Whether this was the case has been debated among experts from the outset.

Evaluating employment programs

The best way to assess this is a ‘net impact evaluation’, in which employment and off-benefit outcomes are compared over time with outcomes achieved by a similar ‘control group’ of unemployed people not referred to Work for the Dole. Employment programs have three different effects:

1. Referral or ‘threat’ effects: Some people leave benefits rather than participate in the program
2. Attachment or ‘lock-in’ effects: Participation in some programs reduces job search activity
3. Program effects: Ideally, participation in a program improves employment prospects, for example by improving skills, putting people in contact with employers who might take them on, or by improving the efficiency of job search.

The overall impact of an employment program equals the referral effect + the program effect – attachment effects.

Quantitative studies should be supplemented by qualitative research that examines what actually happens to unemployed people and prospective employers when they participate in a program. Here I focus on net impact studies of work for benefits programs.

Previous studies of employment impacts

Economist Jeff Borland  conducted the first and only independent net impact study of Work for the Dole. He found that participation in the original pilot scheme for young unemployed people in the late 1990s actually reduced the probability that unemployed people had left income support 12 months after referral to the program. He suggested this was due to attachment effects (they were too busy with Work for the Dole to search for jobs effectively). There appeared to be no significant program effects.

Subsequent official net impact studies found reductions in reliance on unemployment payments after referral to Work for the Dole. A study in the mid 2000s found that referral to the program reduced reliance on income support 12 months later by 6.1 percentage points. (Dept of Education Employment and Workplace Relations 2010, ‘Labour market assistance, a net impact study.’) That’s not a bad outcome for a labour market program but I’d be more confident of the results if they were backed by independent studies.

International studies of the effects of work for benefits schemes give little cause for optimism. An early literature review by MDRC Research, a prominent independent employment program evaluator in the US, found that work for benefits schemes did not improve people’s job prospects. A later report by the New Zealand Department of Labour concluded that neither New Zealand’s ‘community program’ nor similar programs in other countries improved job prospects.

Evaluating Work for benefits schemes in the UK

The best recent evaluations of work for benefit schemes were undertaken in the UK: a country with a rare record of careful independent evaluation of its labour market programs.

Here we examine recent evaluations of two UK work for benefit programs: the Mandatory Work Activity (MWA) program introduced in 2011 and the Community Activity Program (CAP) introduced in 2013.

Mandatory Work Activity

Mandatory Work Activity comprised 4 weeks of compulsory work for benefits for 30 hours a week. It was targeted towards people at risk of long term unemployment. It attracted much controversy and a successful British High Court challenge by an unemployed woman required to work for her benefit for a private employer.

The official evaluation of MWA found significant referral or ‘threat’ effects on the likelihood of benefit receipt:

“in the first 3 months MWA decreased the likelihood of claiming benefit by up to 5 percentage points: the proportion of referrals in receipt of benefit was 77% at week 10 compared to 82% for non-referrals at week 10. This strongly suggests that MWA had caused a deterrent effect of claiming benefit – a very high proportion (64%) of those who did not start signed off their JSA claim.

But, as shown in the graph below, these effects were reversed within another three months, as many of those who left benefits returned:

“However, in the subsequent period the impact decreased, returning to zero by
21 weeks following the referral (benefit receipt for both referrals and nonreferrals
was 74%). This means that in this period a higher proportion of
individuals returned to benefit amongst referrals than non-referrals. Overall,
the benefit impact during the first 21 weeks equated to referrals being off
benefit for an average of about 4 days more than had they not been referred.” p39.

                      Effects of Mandatory Work Activity on benefit reliance

Note: green line represents MWA participants

Crucially, the program had no impact on people’s employment prospects (so those who left benefits weren’t for the most part getting jobs and some were diverted to disability pensions). This is shown in the graph below:

“unlike for benefit, a MWA referral had no impact on the likelihood of being employed
compared to non-referrals. This suggests that although the benefit impacts suggested a deterrent effect from claiming benefit for the first 10 weeks, individuals were not going into employment.” p40

                           Effects of Mandatory Work Activity on employment

Note: green line represents MWA participants

These data suggest that this short work for benefits program had significant ‘threat’ effects that were later reversed, but no impact on the future job prospects of unemployed people.

What did the Government do when it received this evaluation? It expanded the program! As labour market researcher Jonathan Portes pointed out in a blog:

“Briefly, what the analysis shows is that the programme as currently structured is not working. It has no impact on employment; it leads to a small and transitory reduction in benefit receipt; and worst of all, it may even lead to those on the programme moving from Jobseekers’ Allowance to Employment and Support Allowance.  Sadly, the DWP’s press release doesn’t mention any of these key points. Nevertheless, one would naturally assume that as a consequence the programme would be redesigned to achieve better outcomes. However, in a Written Statement, the Minister for Employment said:  
 ‘I am also pleased to announce the Government has decided to expand the Mandatory Work Activity scheme. The expansion will enable Jobcentre Plus to make between 60,000 and 70,000 referrals to Mandatory Work Activity each year, based on the current experience of the scheme, at a cost of an additional £5 million per annum. This decision has been taken as the result of careful consideration of the positive impacts demonstrated within the Impact Assessment.’

Unfortunately (says Portes) it is very difficult indeed to reconcile this statement with the impact assessment itself.”

Community Action Program

In 2013 a new work for benefits program called Community Action Program (CAP) was trialled for very long term unemployed people (2 years out of work). This was carefully evaluated as part of a ‘very long term unemployed trailblazer’ scheme.

“The scheme had these features:

• A randomised controlled trial to give a high degree of confidence that any observed differences in outcomes are attributable to the support options,

• A 13 week pre treatment period to test deterrence, informing and maintaining awareness of forthcoming support throughout the period, and

• A 26 week period of treatment consisting of either:

– Standard Jobcentre Plus support – this is the control group within the trailblazer and consists of flexible and personalised adviser based support.

• Community Action Programme (CAP) – delivered by contracted providers and comprising 26 weeks of full time work experience or provider-led job search support.

• Ongoing Case Management (OCM) – an intensive offer of flexible and personalised adviser based support, delivered by Jobcentre Plus (JCP) through increased interventions over 26 weeks.” p15

A randomised trial is the ‘gold standard’ of net impact studies. Unemployed people were randomly assigned to a treatment group referred to the CAP,  an alternative scheme called ‘ongoing case management’ (an intensive form of job search assistance), or a control group who received ‘standard JobCentre Plus support’ (mainly job search assistance) as shown in this flow chart:

The likelihood that unemployed people involved in the trial were still receiving income support up to 91 weeks after referral to each of these three options was compared. The results are shown in the graph below: 68% of benefit recipients in  ‘standard JobCentre Plus support’ still received benefits 91 weeks later, compared with 66% of those referred to CAP and 64%  referred to ‘case management’. So referral to CAP reduced benefit reliance by 2%, which was 2% less than the impact of case management.

Effects of CAP, Case Management and standard job search assistance on benefit reliance

The graph below compares employment outcomes for those referred to CAP and the other two options in the trailblazer. It turns out that CAP had no significantly greater impact on employment than the standard ‘Jobcentre Plus’ option. Ongoing case management did increase the likelihood of employment compared with the standard service, by just 1%.

To summarise: six months of compulsory work for benefits had no significant impact on the future job prospects of very long term unemployed people and increased the likelihood of leaving benefits by just 2% compared with standard job search assistance (similar to that received by unemployed people assisted by Job Services Australia). More intensive job search help had better outcomes than work for benefits, even though work for benefits was a more expensive program.

Like Australian Governments, UK Governments just don’t seem to learn when it comes to work for benefit schemes – or don’t want to learn.

A fair summary of the pros and cons of these programs was offered by Tony Wilson from CESI, the British employment services advocacy body. His conclusion: it’s reasonable for unemployed people to be required to join programs that help them secure a job and Governments should fund them, but work for benefit schemes generally aren’t worth the effort.

Long term unemployment: what can be done about it?

The following is a paper I presented at the long-term unemployment conference in 2014. It argues that the fact that two thirds of people on Newstart Allowances have received it for over a year and half for over two years signals policy failure. Governments have failed to invest in the regular work experience, training and capacity-building, and connections with employers needed by most people who have been out of paid work a long time.

JSA (and before that Job Network) rewards providers for low-level job search assistance. Average caseloads are over 100. This might work when people are close to employment already, but it’s not good enough for those with low skills, weak (or no) employment experience, or a disability. In theory, paying providers according to job outcomes is a good idea, but as in other countries where this has been tried, it hasn’t worked out as the policy makers planned. To begin with, Governments only get the quality of employment services they are prepared to pay for. In Australia they have not been prepared to pay for it.

The previous Government built a program based on short-term low quality training (average spend $300 per course, % employed after course 30%). The present Government is building a new one based on Work for the Dole (average spend $2,000, % employed after program 23%). Neither approach was successful (except in the sense that it kept unemployed people busy).

Long term unemployment: the ‘achilles heel’ of the Job Services Australia model

Paper presented by Peter Davidson at Long Term Unemployment conference, Gold Coast 18-19 August 2014

Two-thirds of recipients of Newstart Allowance have been unemployed for more than a year. Long term unemployment is associated with poverty, poor health, and higher levels of structural unemployment. As the population ages and gaps emerge in the paid workforce, Australia will have an opportunity to solve one of our worst social problems and meet one of our most pressing economic needs at the same time.

While sustained economic growth is essential to reduce long term unemployment, it is not sufficient. Evidence from Australian and international program evaluations suggest that employment services can improve the job prospects of people unemployed long-term.

Australia’s largest public employment program, Job Services Australia (JSA), is under review and a revised program is expected to be announced this year.

 This paper attempts to answer the following three questions:

• What forms of employment assistance are most effective in reducing long term unemployment, based on recent international evidence?

‘Activation’ of unemployed people can reduce long term unemployment by requiring and assisting people with reasonable good job prospects to search for employment more effectively. Activation on its own is not sufficient. People who are unemployed long term or at risk of it usually face specific hurdles to employment such as low skills, a lack of work experience, or disabilities. Broadly speaking, paid work experience (using wage subsidies) in regular jobs and substantial vocational training (preferably linked to a job) are relatively effective in overcoming these barriers to employment while very short training courses (for example, less than three months) and ‘make work’ schemes, whether paid or unpaid (work for benefits) are relatively ineffective. That said, the international evidence suggests that successful interventions are generally those which are tailored to individual needs (of jobseekers and employers), rather than standardised.

•  Are resources in the JSA system effectively targeted to assist long term unemployed people?

Australian and international evidence suggests that it is generally cost-effective to target the most intensive help towards people who are unemployed for one to three years, since fewer of this target group would find employment without assistance. The JSA system shifted resources from people unemployed long term towards those people unemployed for less than a year who were assessed as at-risk of long term unemployment. As a result, providers are typically funded to interview a person unemployed for one to two years every two months and purchase only $500 worth of work experience or training, on average. While not conclusive, the evidence suggests that this shift of resources reduced the effectiveness of the program in assisting those who were already unemployed long-term. 

• Does the JSA system encourage efficient investment in work experience, training and other supports needed by long term unemployed people?

Overall Australian public investment in labour market assistance has been low – about half average OECD levels – since the Job Network was introduced in 1998. Within this pool of funds, Australia spends relatively more on job search assistance and less on work experience and training programs for unemployed people. While in theory outcomes-based funding of non-Government employment services should encourage cost-efficient investment in disadvantaged jobseekers, in practice it has rewarded providers who concentrate on low cost job search assistance rather than patient investment in work experience and training. Since the late 1990s, Governments have directly funded their preferred forms of work experience and training to fill some of the resulting gap in employment assistance.

In the Job Network period, a ‘work for benefits’ scheme (Work for the Dole) was favoured over vocational training, and this was reversed with the introduction of JSA. If funds invested in Work for the Dole (a relatively ineffective program) were replaced by substantial vocational training linked to employment opportunities, this would likely have improved the effectiveness of employment assistance. However, providers were only resourced to purchase relatively ineffective short courses and faced pressure to place people in readily-available State Government-funded courses to meet jobseeker activity requirements at low cost. Average employment consultant caseloads were over 100 which left little room for individualised assistance.

Under these conditions, the JSA program was likely to have a similar net employment impact to the Job Network, since the basic design of the two programs was otherwise similar. Average ‘gross’ employment outcomes following participation in JSA were at first similar to those of the Job Network, with just under 50% of participants in employment three months after leaving the program. Average employment outcomes fell after the GFC and declined further during 2012, but this is likely to be due to adverse labour market condition rather than changes in employment assistance.

The conference presentation will conclude with a brief assessment of the Government’s restructure of Job Services Australia, which was not available at the time of writing.

 

Keywords

Unemployment, long term unemployment, unemployment benefits, labour market programs, welfare reform.

The problem: growth in long term unemployment and its effects

There are many good reasons to reduce long term unemployment, including its health and social effects, and the link between long-term and structural unemployment[1].

An individual’s employment prospects decline sharply the longer they are out of work. This is due to long term unemployment itself and other factors such as poor health or low skills that predispose people to long term unemployment. Almost half (47%) of new recipients for Newstart Allowance (60%) leave that payment within the first 6 months. However, among those already on unemployment payments for 12 months, approximately half still receive those payments a year later[2]. Among unemployed people receiving Job Services Australia services (JSA) in 2012, 58% of those unemployed for one year, and 91% of those unemployed for two years, were in the JSA system a year later[3].

While the unemployment rate has fallen over the past two decades, a growing proportion of individuals reliant on unemployment payments are long-term recipients.

Figure 1 shows trends in unemployment over the last two decades, a period during which the unemployment rate fell from 11% in 1993 to 4% in 2007, and then rose again to 6% in 2013 following the international economic downturn (GFC) in 2008[4].

Figure 1: Unemployment rate, Australia

Source: ABS, Labour Force survey.

Figure 2 shows changes in the duration profile of spells on unemployment payments over the same period[5]. In 2013, approximately two thirds (65%) of the 753,000 recipients of Australian unemployment payments (Newstart Allowance and Youth Allowance – Other) had received income support for at least one year. Half (49%) had received it for over two years[6].

Figure 2

Source: Department of Social Services (2014); Department of Employment (2014d).

As unemployment fell between 1993 and 2007, the proportion of recipients out of work for less than one year declined in line with the fall in the unemployment rate (figure 1) as they picked up opportunities in an improving labour market. The proportion unemployed for one to two years remained fairly constant. However the proportion unemployed for over two years (‘very long term unemployed’) rose steadily, suggesting that this group was more disadvantaged in the labour market and benefited less from the more favourable labour market conditions. Following the Global Financial Crisis in 2008 more people entered unemployment and the proportion of recipients unemployed long-term declined. After 2009 this pattern was reversed as the inflow to unemployment payments subsided and a growing proportion of unemployed people received unemployment payments for two years or more. Their number was boosted from 2012 by successive Budget decisions to transfer sole parents on the Parenting Payment Single payment to the lower Newstart Allowance, since they generally rely on income support for longer periods of time[7] .

What forms of employment assistance reduce long term unemployment?

To answer our first question, we need to measure the effectiveness of different forms of employment assistance. A key challenge for effectiveness evaluations is to distinguish between outcomes that would have been attained without assistance (the ‘deadweight cost’ of a program) and the difference it makes to employment and other outcomes (its ‘net impact’).

‘Net impact’ evaluations aim to measure the value added by employment programs by comparing outcomes achieved by a sample of program participants and those of a comparable control group. This can be done through experiments in which individuals are randomly assigned to control and treatment groups (rarely been used in Australia), by using regression techniques to ‘match’ program participants and non-participants according to observed characteristics such as age and qualifications (the method most commonly used), or using ‘natural experiments’ which take advantage of differences in the timing or targeting of programs when they are introduced or expanded (for example where a new program is restricted to people within a certain age group)[8].

No net impact study of the effects of the JSA system has yet been published, but we know about the effects of its predecessor program the Job Network on employment outcomes and exits from income support from a number of evaluation studies of that program, including net impact studies. Specific interventions such as work experience programs, training, and job search assistance have also been evaluated using these methods, both here and overseas. Relevant conclusions are summarised below.

• On average, employment impacts are greater for long term unemployed people

Figure 3 compares the net impact (in percentage points) of the highest level of Job Network assistance, Customised Assistance, on the probability of exits from income support for clients with different unemployment durations, 12 months after commencing each ‘stage’ of Job Network assistance[9]. 

Figure 3

Source: Department of Education, Employment and Workplace Relations (2010).
Note: Estimated impact of the program on exits from income support 12 months after commencement in the program. Estimates for those unemployed over 24 months are for the ‘second stage’ of Customised Assistance.

The estimated distribution of net impacts forms an inverted ‘U’ shape, with impacts rising with unemployment duration up to around two years, then falling. This is consistent with evaluation findings in a number of other OECD countries[10]. This suggests that it is cost efficient to target more expensive forms of employment assistance towards long term unemployed clients. The reason for this is that their employment prospects are lower without assistance and deadweight costs are minimised. It is noteworthy that those who received Customised Assistance before 12 months of unemployment were assessed as facing a high risk of long-term unemployment using a profiling tool. The relatively low average net impact for this group illustrates the limitations of profiling tools in predicting prolonged unemployment and targeting employment assistance. It is worthwhile to ‘intervene early’ where people are at risk of long term unemployment but deadweight costs are greater because we cannot predict with complete accuracy which individuals will become long-term unemployed without further help.

An average ‘net impact’ of up to a 10 percentage point increase in employment in the short to medium term is fairly typical of the results of these evaluation studies[11]. While these results are modest (and imply a deadweight cost of 90%), if they are sustained for a number of years across the population of long term unemployment benefit recipients then long term unemployment and reliance on income support would be greatly reduced[12].

• Effects of activation and job search assistance

The average net impacts of a generic program such as Job Network mask a diversity of impacts from different interventions on different target groups. We need to shine a light into the ‘black box’ of employment assistance to understand which interventions make the most difference under what circumstances.

In assessing the impact of different of different interventions, it is useful to separate them into the combination of ‘activation’ (keeping people engaged with the labour market through activity requirements, penalties and regular interviews) and job search assistance (practical help with job search) which together form the generic core of most labour market assistance; and specific work experience and training programs for more disadvantaged jobseekers.

‘Activation’ has been found to significantly increase transitions to employment at relatively low cost, especially for less disadvantaged jobseekers and in the early stages of establishment of a rigorous activation regime[13]. This appears to be mainly due to its incentive or ‘threat’ effects: when they are interviewed or new activity requirements are imposed, many unemployed people intensify their job search and find employment. Examples of significant employment effects from Australian activation policies include the introduction of ‘Newstart interviews’ for long term unemployed people introduced in 1990 which were estimated to increase exists from income support 6 months later by 8%, and the new activity requirements for sole parents with school age children introduced in 2006 which were estimated to increase exits from income support 12 months after a benefit claim by an average of 12%[14].

However, there is also evidence to suggest that returns to activation diminish when its intensity is increased in a mature system of activity requirements, interviews, and compulsory referral to employment programs. For example, when an additional ‘9 month intensive review’ interview was added to the schedule of regular meetings with Australian employment services in the late 1990s, the evaluation found that it had no measureable impact[15].

Further, the intensification of activity requirements and compliance monitoring appears to be less effective for the most disadvantaged jobseekers, presumably because (without additional help) their employment prospects remain limited even if they search more intensively. A ‘Job Seeker Diary’ introduced for unemployed people to record their job search efforts for Centrelink had much weaker employment impacts for long term than short term unemployed people[16]. The introduction of ‘review interviews’ for people unemployed for five years or more was also found to be ineffective.[17] 

• Effects of work experience and training

For people with significant barriers to employment (which includes most long term unemployed people), activation and job search assistance is more effective if backed up with work experience training and other programs that improve work capacity. This was the logic behind the successful New Deal for Young People (NDYP) in the United Kingdom, which offered a sequence of intensive job search assistance for up to three months followed by compulsory referral of those still unemployed to one of four ‘options’ – a temporary subsidised private sector job, education and training, voluntary work, or unpaid work on environmental projects[18]. When the corresponding ‘New Deal’ program for long-term unemployed adults (NDLTU) was first introduced in 1989 it offered activation and job search assistance but not the ‘options’. When it was found that the program was much less effective than the NDYP, at first job search requirements and contacts with advisers were increased. This made little difference to employment outcomes. When ‘options’ such as work experience and training were later added to the program, employment outcomes doubled^[19].

Policy makers have sought to combine the best of both worlds – consistent activation and flexible investment in work capacity building – by incorporating individually tailored employment and training programs into regular spells of intensive activation. International examples include the New Deals in the United Kingdom and regular intensive activation of long term unemployed people in Denmark since the mid 1990s[20]. Critical to the effectiveness of these approaches is the intensity and quality of engagement between a jobseeker and their caseworker[21].

Net impact evaluations of work experience and training programs suggest that[22]:

• Temporary wage subsidies in private sector employment are relatively effective because employers often keep the jobseeker on after the subsidy expires, but positive impacts are likely to be partly due to selection effects (‘creaming’ through provider or employer selection of the ‘best candidates’)[23]
• Temporary employment in ‘additional jobs’ in the public sector (‘job creation’) is relatively ineffective due to ‘lock-in’ effects and the poor quality of the work experience usually provided, except for a minority of people with severe disadvantages and very limited recent work experience^[24]
• Work for benefit schemes have substantial ‘referral’ or ‘compliance’ effects (since some people leave benefits when referred) but little impact on employment prospects among those who commence the program (since the work experience is usually far removed from paid employment in the open labour market and there is little upgrading of skills)[25]
• Vocational training is effective, especially for people with latent skills such as sole parents who left education early, but positive employment impacts can take years to occur, and as with the other programs above, ‘lock in effects’ (withdrawal from active job search during program participation) are significant during program participation. On the other hand, very short courses are ineffective. Training is generally more effective if linked to on the job work experience or preparation for a particular job[26].

• Emerging approaches

In recent years, attention has turned to new interventions to assist long term unemployed people into employment.

One of these is called the ‘demand led’ or ‘employer partnerships’ approach in which employment service providers enter into a formal agreement with an employer to train and place disadvantaged jobseekers with them. The Australian Council of Social Service, together with the Business Council of Australia and Australian Council of Trade Unions last month released a statement calling for the broader adoption of these approaches[27]. The insight of this approach is that to achieve a match between a disadvantaged jobseeker and an employer, employment services providers should devote as much effort to working with the employer as improving the skills and motivation of the jobseeker. Government-backed programs that adopted the demand led approach include the ‘Ambition’ program in the United Kingdom and the Australian Employment Covenant[28].

A second emerging approach, referred to as ‘wrap-around servicing’, is to integrate traditional employment assistance with other support services such as mental health, social housing and family support services. The insight behind this approach is that these services are likely to be more effective when offered together than separately or in sequence. The integration of employment and mental health services in Individual Placement and Support (IPS) schemes has shown some promise in Australia[29]. Rather than treat mental illness first and then refer to employment, the IPS approach combines job referral and support in employment, either coordinated between mental health and employment service providers, or through employment counsellors attached to mental health clinics.

A suite of ‘wrap around service’ schemes for disadvantaged jobseekers was funded by the Australian Government in its ‘Building Australia’s Future Workforce’ initiative announced in the 2011 Budget[30]. Unfortunately detailed information on participation in these schemes and their effectiveness is not yet available so these are not discussed further in this paper.

Key features of the Job Services Australia program

The principal employment services program for unemployed people is Job Services Australia (JSA), under which the Employment Department contracts out all employment assistance to for profit and not for profit providers who are paid according to a combination of service inputs, investment in work experience and training and other supports, and employment outcomes. This replaced the similarly designed Job Network program in 2009.

Australia was among the pioneers of contracting out of public employment services to private providers and remains the only OECD country where the entire national network of employment services is contracted out[31]. The Job Network replaced the Commonwealth Employment Service and a range of national work experience and training programs for disadvantaged jobseekers from 1998. Employment services were tendered by the Employment Department every three years, offering providers a fixed share of ‘business’ (jobseeker referrals) within each ‘Employment Services Area’. Providers were paid a commencement fee per jobseeker and outcome fees for a range of specified employment and training outcomes, especially jobs lasting at least three months. On commencement of unemployment, each individual was assessed using a ‘Job Seeker Classification Instrument’ that estimated their risk of long-term unemployment. Their score, together with their duration of unemployment, determined the level of fees (especially outcome fees) paid to providers.

The idea behind this new system of outcomes based funding was that by giving jobseekers a choice of provider and by leaving the provider to decide the form of assistance offered to each jobseeker, assistance would be tailored to individual needs. There are similarities between this ‘black box’ contracting model and the new Work Program for long term unemployed people in the United Kingdom[32]. A significant innovation was the introduction of a five-tier ‘Star Ratings’ system to provide information for jobseekers and (more importantly, as it turned out) the Employment Department on the relative performance of providers. Unlike outcome payments, which were to a large extent a deadweight cost to the Government (since most employment outcomes would have been achieved without assistance) the Star Ratings were based on estimates of the net impact of employment services on people’s employment prospects – or the value-added by the service. The Star Ratings were used to award ‘business shares’ to better-performing providers[33].

In 2003, the Active Participation Model of the Job Network was introduced. The new model moved away from the ‘black box’ approach in response to concerns about ‘parking’ of jobseekers (provision of minimal or no service) and excessive profit-taking. Providers were contracted to offer jobseekers a fixed ‘continuum’ of service, with the highest payments reserved for the Customised Assistance stage which lasted for six months following the 12^th or 24^th month of unemployment (earlier if the jobseeker was assessed as disadvantaged). Customised Assistance combined a period of intensive job search assistance with activities such as training or work experience. To reduce parking and profit-taking, around one third of total funding was quarantined in a ‘Jobseeker Account’ for providers to invest in these and similar interventions to overcome barriers to employment.

Another key element of the Job Network service continuum was ‘Mutual Obligation’. For six months of each year of unemployment (alternating with periods of Customised Assistance), a jobseeker was expected to undertake six months of work related activity such as training, part time employment, or work for benefits. The default activity was ‘Work for the Dole’, a ‘work for benefits’ scheme offering six months’ work experience on projects of community benefit.

The Job Services Australia (JSA) program that commenced in 2009 retained the basic structure of the Job Network model, including

• Service payments to ensure regular interviews and supervision of activity requirements;
• An Employment Pathway Fund (EPF) to replace the Jobseeker Account;
• Outcome payments focussing on employment that is sustained for 3 to 6 months, with higher payments for long-term and more disadvantaged jobseekers;
• The performance management system based on Star ratings and regular (though less frequent) ‘business share re-allocations’ among providers.

The key changes included:

• Risk-based rather than duration-based targeting:
  The ‘Job Seeker Classification Instrument’ and a disability assessment tool were used to divide participants into four ‘Streams’ of labour market disadvantage. People with substantial health and social barriers were shifted from a previous program (Personal Support Program) into mainstream employment services as ’Stream 4’ of JSA.
  Higher service fees and EPF credits were targeted towards those in the higher Streams rather than long-term unemployment. After 12 months’ unemployment, a jobseeker entered a ‘Work Experience Phase’ which attracted lower payments for the provider than ‘Stream 3’ or ‘Stream 4’ jobseekers in the first year of unemployment. Resources were thus shifted from ‘curing’ long term unemployment towards ‘preventing’ it.
• More flexible sequences of activation:
  Providers were given more discretion to determine the form and timing of regular periods of intensive activation, in lieu of a fixed ‘service continuum’.
• Greater encouragement of training:
  Incentives for participation in vocational training were increased, both for jobseekers and providers, and new training programs were funded including for vocational education and training under a ‘Productivity Places Program’ (PPP)[34]. Resources for some other programs including ‘Work for the Dole’ were reduced.

Table 1 shows levels of participation in each Stream and the Work Experience phase. It indicates that 27% of participants at September 2013 were in the Work Experience Phase for long-term unemployed people, and that 39% were assessed as highly disadvantaged in the labour market (Streams 3 and 4). These categories are not mutually exclusive as participants usually remain in their ‘streams’ when they enter the Work Experience Phase.

Table 1: Participation in Job Services Australia, at September 2013

Stream	Number	Proportion
Stream 1	267,000	35%
Stream 2	193,000	25%
Stream 3	150,000	19%
Stream 4	156,000	20%
Work experience phase (also included in Streams)	207,000	27%
Total	770,000	100%

Sources: Department of Employment (2014c).                       * In September 2013

What employment outcomes have been achieved by Job Services Australia participants?

Figure 4 shows trends in average employment outcomes three months after program participation for Job Network ‘Intensive Support’ participants (those unemployed for over 3 months) and all Job Services Australia participants, from the last three years of measured Job Network outcomes ( up to 2010) through to 2013.

Figure 4

Source: Department of Employment (2014b); ABS, Labour Force survey; Department of Employment (various years), Labour Market Assistance Outcomes.
Note: Employment outcomes three months after leaving the program. The last outcomes from Job Network were reported 12 months later in 2010.

Figure 4 indicates that average employment outcomes declined from 54% to 48% following the Global Financial Crisis (GFC), continued to decline after the introduction of JSA (the results of which were measured from 2010), and fell more sharply in 2012. In the absence of a net impact evaluation of JSA, it is hard to draw clear conclusions from these data on the relative effectiveness of the two programs[35]. The decline in 2012 corresponds with the start of a second three-year JSA contract period, a time when services are disrupted by agency closures and the transfer of jobseekers to new agencies. More significantly, both of these declines in program outcomes corresponded with sharp reductions in employment growth in 2009 and 2012, as shown on Figure 5.

Figure 5

Source: ABS, Labour Force survey.

Is assistance for disadvantaged jobseekers appropriately targeted in Job Services Australia?

The JSA fee structure shifted resources from long-term unemployed people towards shorter-term unemployed people at risk of prolonged unemployment. Table 2 shows that the highest levels of service fees and EPF credits are paid in respect of ‘Stream 3’ and ‘Stream 4’ clients in their first year of unemployment. Once these jobseekers reach 12 months of unemployment they usually move into the Work Experience Phase, during which service fees are approximately halved and Employment Pathway Fund credits fall to around $500. In contrast, Customised Assistance, the highest level of Job Network assistance, was targeted towards long term unemployed people, though a minority of shorter-term unemployed people facing a high risk of long term unemployment were also targeted.

Table 2:
Comparison of the highest level of assistance in Job Network and Job Services Australia

	Job Network: (Duration-based targeting)	Job Services Australia: (Risk based targeting)
Highest level of assistance	Customised Assistance	Streams 3 & 4
Duration	6 months	12 months
Payments to providers in typical cases	Service fees of $600-$900 for fortnightly to monthly interviews Plus $500-$900 in Jobseeker Account	Service fees of $1,100-$1,900 for fortnightly to monthly interviews Plus $1,100-$1,900 in Employment Pathway Fund
Targeted towards	Unemployed 12-18 months, Unemployed 24-30 months, ‘Highly disadvantaged’ (high risk of LTU) in first 12 months)	Those with disability/social barriers (Stream 4), Others at high risk of LTU (Stream 3), in the first 12 months of unemployment

Source: Davidson P & Whiteford P (2012).
Note: ‘LTU’ refers to long-term unemployment.

Figure 6 compares typical funding levels for providers in respect of Stream 3 jobseekers and recipients of Job Network Customised Assistance[36]. In the case of a typical Stream 3 jobseeker (one ‘at risk’ of long term unemployment), service fees and EPF credits are $600 lower for people unemployed for 1 to2 years than in the Job Network, reflecting the targeting of Job Network Customised Assistance towards long term unemployed people. They are also $1,000 less than funding levels for the same jobseekers in their first year of JSA assistance. For those unemployed for 2 to 3 years, JSA payments are higher due to an additional $1,100 credit to the EPF at that stage. However this is expected to fund a full 11 months of intensive employment-related activity for every remaining year of unemployment[37]. 

Figure 6:

Note: Based on typical cases, includes service fees and Jobseeker Account or Employment Pathway Fund credits.

Outcome payments and Star Ratings calculations are also heavily weighted towards Stream 3 and 4 clients. One fifth of the weights used for Star Ratings are for ‘full’ employment outcomes for Stream 3 and 4 clients respectively, compared with 6% for each of Streams 1 and 2 respectively. Outcome Payments for ‘full’ outcomes are also much higher for Stream 3 and 4 clients (typically $2,200 in the first year of unemployment for a ‘full’ employment outcome compared to $1,400 for Stream 2 and $400 for Stream 1).

Unlike service fees and EPF credits, outcome payments increase with unemployment duration (to around $3,100 after the second year of unemployment and $3,900 after the fifth year for Streams 3 and 4). However, outcome payments do not offer much reward for provider investment, even for the most disadvantaged jobseekers, as investments in labour market assistance have significant deadweight costs[38].

In the absence of a net impact study for JSA, one way to assess the efficacy of the shift from duration to a risk based targeting is to compare average ‘gross’ employment outcomes of Job Network and JSA clients by duration of unemployment.

Figure 7 compares employment outcomes for Job Network Intensive Support clients (those unemployed for at least 3 months) and JSA clients, by unemployment duration. 

Figure 7

Source: Department of Education, Employment and Workplace Relations (2008 & 2011), Labour market assistance outcomes, years ending September 2008 and September 2011.
Note; ‘JN IS’ refers to the Intensive Support phase of Job Network, which commenced after 3 months’ unemployment.
‘% reduction’ is the difference between the 2008 and 2013 values, divided by the 2008 values.

Figure 7 shows that in 2013, employment outcomes following participation in JSA were 19% lower than in the Job Network in 2008[39]. Conclusions cannot be drawn from this about the relative effectiveness of the two programs since as noted previously, the fall in outcomes is probably mainly due to business cycle effects.

More salient are the relative employment outcomes at different stages of the unemployment spell. While average outcomes were 19% lower for JSA as a whole, the reduction in outcomes for short term unemployed people was 13% while the reduction for people unemployed for 1 to 2 years was 26%. Interestingly, the reduction in employment outcomes was slightly less for those unemployed for two to three years (25%), the stage of the unemployment spell where funding of JSA services increases (figure 6).

These changes in relative employment outcomes could be due to the changes in program design and targeting described above, or to changes in labour market conditions which disproportionately affected long term unemployed people. Regardless of the reason, these results raise serious concerns about the reduction in investment in long term unemployed people. There is a growing risk that people will become ‘stuck’ in long term unemployment, especially since over half of participants in the Work Experience Phase of the JSA program (that is, over 100,000 people) have been on income support for over three years and the graph indicates that outcomes for that group are 41% less than they were five years ago[40].

The answer to our second question: Are resources in the JSA system effectively targeted to assist long term unemployed people? is ‘no’.

One option to improve target efficiency would be to raise the level of service payments and EPF credits for those unemployed over 12 months to the same level as applies to ‘Stream 3’ jobseekers in their first year of unemployment. This would broadly equalise support for those at risk of long term unemployment and those already unemployed long term. Since a majority of long term unemployed people come from Streams 3 and 4, this would not create an incentive for providers to delay employment outcomes until people are unemployed for 12 months in order to receive higher payments.

Does the JSA system encourage efficient investment in work experience and training?

Since the introduction of the Job Network, there have been few national work experience and training programs targeted to unemployed people apart from the ‘Work for the Dole’ program which provides up to three months of unpaid work experience on projects of community benefit.

The reason for this is that it was considered more efficient for employment service providers to decide which forms of work experience or training are best suited for each jobseeker[41]. In the Job Network and subsequent employment service purchasing models, the main levers available to policy makers to encourage and support efficient investment in work experience and training are outcome payments, Star Ratings, and notional ‘investment funds’ such as the Jobseeker Account or Employment Pathway Fund.

Investment incentives in the employment services system

While on the face of it providers are rewarded for employment outcomes regardless of how they are achieved, Australia’s employment services system has always encouraged some interventions and discouraged others. Broadly speaking, funding to short term outcomes encourages providers to focus on relatively inexpensive forms of job search assistance and to avoid investment in longer term and more costly activities such as paid work experience and substantial vocational training courses that might delay employment outcomes[42]. The introduction of the Jobseeker Account (now Employment Pathway Fund) was partly intended to compensate for this bias against patient investment in work experience and training[43].

The employment services system during the Job Network period (1998 to 2008) favoured ‘work for benefits’ schemes since Work for the Dole was the default program for regular Mutual Obligation activities and it was one of few work experience and training programs that was separately funded[44]. There was also a bias against vocational training since jobseekers were not generally permitted to undertake courses of more than 12 months’ duration while on unemployment payments, and under the prevailing ‘work first’ policies, providers were discouraged from approving part-time courses of more than a few months’ duration[45].

The JSA system reversed the previous bias in favour of ‘work for benefits’ and against training. Work for the Dole was no longer funded as a separate program, so JSA providers had to use EPF credits or their ‘own’ funds to finance such placements. This was intended to encourage cost-efficient investment in unpaid work experience by bringing it within the scope of the outcomes-based system of employment services purchasing^[46].

The removal of specific funding for ‘work for benefits’ programs encouraged greater use of training to fulfil requirements for providers to organise compulsory periods of intensive work-related activity during the Work Experience Phase. Vocational training courses were generally a cheaper option for providers, to the extent that they could refer people to free or low cost training in the State Government- run Vocational Education and Training (VET) system. Commonwealth funding for the VET system was also boosted through a ‘Productivity Places Program’ in which half the places were reserved for unemployed people. Significantly, under an ‘Earn or Learn’ policy announced in 2009, young people under 21 years who had not completed 12 years of schooling (‘early school leavers’) were required to participate in education or training instead of actively seeking employment until they achieved an equivalent qualification to 12 years of schooling^[47]. Further, parents and carers were no longer required to abandon part time training courses if they prevented them from taking up an employment offer^[48].

Data on investment in wage subsidies and training by employment service providers

Table 3 compares average provider investments in training and wage subsidies (paid work experience) activities undertaken by Job Network and JSA providers using the Jobseeker Account and EPF, respectively. These data do not include places in programs that were funded separately, such as Work for the Dole which is discussed separately below.

 

Table 3: Comparison of Job Seeker Account and Employment Pathway Fund expenditures for long-term unemployed people (2003-2011)

	Number or % of jobseekers that received assistance	Average cost of assistance	% of all Jobseeker Account/EPF expenditure2.
Job Network Customised Assistance (2003-06)
Training.1.	25%	$350	23%
Wage subsidies	12%	$2,590	23%
Job Services Australia (2009-11)
Training .	44,000 3.	$303	33%
Wage subsidies	13,000 3.	$2,050	21%

Sources: Department of Employment and Workplace Relations (2006); Department of Education, Employment and Workplace Relations (2012); Department of Education, Employment and Workplace Relations (2011).
^1. Includes training in job search skills but not the Job Search Training courses offered earlier in the APM continuum.
^2.Other expenditures (each of which attracted lower shares of total credits) included clothes and equipment, fares and petrol, interpreters, additional provider contacts, professional services, and relocation assistance.
^3. Number of exits from these activities in the JSA ‘Work Experience Phase’ for Year ending September 2011. This is not expressed in the LMAO report as a percentage of all Work Experience Phase participants.
A comparison of the unit and total costs of training and wage subsidy places purchased by providers suggests that the number of training places purchased exceeded the number of wage subsidies by a factor of seven to one under the Job Network and a factor of ten to one under JSA[49]. However, the training was purchased at a very low average cost ($300 to $350) indicating that it was either very short-term nature or that providers referred jobseekers to State Government-funded vocational education and training places.

These data do not include separately-funded places under national wage subsidy schemes (paid work experience) or Work for the Dole (unpaid work experience). Table 4 shows that these wage subsidy scheme places (which are separate to those funded by providers in table 3) rose from up to 3,000 per year under the ‘Wage Assist’ program in 2007 to 10,000 a year from 2011 under the more recent ‘Wage Connect’ scheme[50]. Nevertheless, the overall use of wage subsidies remained low (and much lower than the use of training) under both the Job Network and JSA regimes. Work for the Dole places were more widely available in the Job Network era, with 65,000 participants overall in 2007-08. Places in that program fell considerably in the JSA era – to 28,000 in 2010-11. 

Table 4: Separately-funded paid and unpaid work experience places

	Job Network period (2007):	Job Services Australia period (2011)
Wage subsidy schemes	3,000	10,000
Work for the Dole	65,000	28,000

Sources: Department of Education Employment and Workplace Relations (2008b) Australian Government (2011) Department of Education, Employment and Workplace Relations (2007 and 2011), Labour market assistance outcomes, Canberra.
Another source of data on investment in work experience and training is the OECD’s social expenditure data base, though this only extends to 2011, the first two years of JSA (figure 7). In 2009, overall investment in labour market assistance for unemployed people in Australia was relatively low, at about half the OECD average (0.3% of GDP compared with 0.6%)[51]. Of these funds, most were devoted to ‘placement and related services’ (comprising assessment, job matching, and employment counselling but also includes expenditure by JSA providers in work experience and training using the EPF). Since the beginning of the Job Network, direct Government outlays on work experience and training programs for unemployed people have been very low by OECD standards. Expenditure on the Work for the Dole program (the main component of the ‘direct job creation’ category since 1998) fell, and direct Government spending on training for unemployed people rose, with the introduction of JSA in 2009.

Figure 7

To summarise, since the introduction of the Job Network, Australian employment assistance programs have been heavily weighted towards low-cost job search assistance. The introduction of JSA saw a shift within our relatively modest investment in work experience and training from work for benefits schemes towards vocational training, which was mainly of a short-term nature.Source: Organisation for Economic Cooperation and Development (2014).

Note: Prior to 2001, expenditure on placement and related services was not recorded. After 2001, this head of expenditure includes training and work experience purchased using the Jobseeker Account/EPF. ‘Direct job creation’ refers mainly to and the Work for the Dole program and its equivalent for Indigenous people, the Community Development Employment Program (CDEP).

The effects of the changed profile of investment in work experience and training in the JSA program

Table 5 compares employment outcomes from various activities undertaken by long term unemployed people in the Work Experience Phase of JSA in 2013.

Table 5: Outcomes of employment assistance (year ending September 2013)

Activity	% employed 3 months after completion
Wage subsidy	47 1
Training	30
Work for the Dole	23
Total	28

Source: Department of Education, Employment and Workplace Relations (2013), Labour market assistance outcomes, year ended September 2013 – Job Services Australia; Department of Education, Employment and Workplace Relations (2013).

1. Year to February 2013, % employed 6 months after completion.

On the face of it these data suggest that wage subsidies are much more effective than either training or ‘work for benefits’ programs. However, they show ‘gross’ not ‘net’ employment impacts and therefore do not take account of differences in the characteristics of participants. For example, only one in three of those receiving wage subsidies in 2013 were long-term unemployed. Only one net impact study of labour market assistance was undertaken since the introduction of JSA in 2009. This assessed the impact of expenditure of EPF funds on wage subsidies on the probability of being off income support 12 months after commencement of the placement, which was estimated to be 14 percentage points[52].

To assess the relative effectiveness of different interventions in the absence of a more comprehensive net impact study, we rely on the international (and previous Australian) evidence presented earlier.

Well targeted wage subsidy schemes are generally considered effective. The introduction of a separate wage subsidy scheme in 2012 (Wage Connect) suggests that policy makers believed that the JSA payment structure did not of itself encourage an optimal level of investment in wage subsidies (for example, only 11,000 in the year ending September 2011 – see table 3). Wage Connect was consistently over-subscribed, which suggests that there was more than enough employer demand for places. However, the scheme was capped at only 10,000 places per year so would have made little impact on the overall unemployment level.

The data cited in the previous section show that participation levels in vocational training rose substantially following the introduction of JSA, but that most of the training was either in very short courses, or it was funded from sources outside the JSA program[53]. Little is known publicly about the nature of the courses offered and qualifications obtained, or employment outcomes following participation[54]. Given the evidence presented previously on the poor results from short vocational training courses, and the low level of investment in training by providers, the increased use of training was not likely to significantly improve employment outcomes.

In regard to work for benefit schemes, it is clear that participation in Work for the Dole fell substantially with the introduction of the JSA regime. If those resources were redirected towards more effective interventions, such as wage subsidies or more substantial training, then the reduction in investment in work for benefits schemes was likely to improve employment outcomes. However this does not seem to have happened.

Overall public funding for employment services declined with the introduction of JSA[55], and funding levels have not increased substantially (through indexation or otherwise) over the past six years.

As noted previously, evaluations point to the importance of individualised help for the most disadvantaged jobseekers. This suggests that average caseloads, and the degree of discretion given to Employment Consultants to offer long term unemployed people the most appropriate help for their needs and local labour market conditions, are key predictors of the likely effectiveness of JSA. In 2012, the average caseload for a JSA consultant was 114, which was similar to caseloads in the Job Network era[56]. This makes it very difficult for providers to offer individualised help to more than a small minority of jobseekers assisted[57]. In a major survey of employment consultants, 72% indicated that: ‘the IT system dictates how I do my job’[58].

In the absence of up to date evaluation studies, the answer to the third research question: Does the JSA system encourage efficient investment in work experience, training and other supports needed by long term unemployed people? is less conclusive than the other two questions.

While the JSA model is more flexible than the Job Network (especially with the removal of the rigid ‘service continuum’), employment consultants lack the resources and the time to offer an individualised service to their (average of) 114 clients. It is likely that cost considerations and the availability of externally funded program places (especially in the VET system) had a large influence on their decisions to refer jobseekers to programs. The survey cited previously indicates that 31% of consultants considered that the ‘availability of program places’ was a key influence in this decision making process[59].

The limited use of relatively effective programs, such as wage subsidies and more substantial or individually-tailored training programs, suggests that provider investment in long term unemployed people in the JSA era was no more effective on average than in the Job Network era, and may be less so given the shift of program resources from long term unemployed people to those at risk of long term unemployment.

Conclusions

As unemployment rates have fallen over the last 20 years, the profile of unemployed people receiving income support has become more disadvantaged. Two thirds of Newstart Allowance recipients have received income support for over a year, and half for over two years. High levels of long term unemployment entrench poverty and structural unemployment.

The labour market ‘activation’ of unemployed people can reduce long term unemployment by requiring people with reasonable good job prospects to search for employment more intensively and effectively. Activation on its own is not sufficient. People who are unemployed long term or at risk of it usually face hurdles to employment such as low skills, a lack of work experience, or disabilities. Employment programs such as vocational training and paid and unpaid work experience are designed to help people overcome these and other barriers to employment.

Broadly speaking, among these programs paid work experience (using wage subsidies) in regular jobs and substantial vocational training (preferably linked to a job) are relatively effective while very short training courses (less than three months) and ‘make work’ schemes, whether paid or unpaid are relatively ineffective. That said, the international evidence suggests that successful interventions are generally those which are tailored to the individual needs of jobseekers and employers, rather than standardised. This suggests that the caseloads and degree of discretion exercised by Employment Consultants are critical factors.

Over the last two decades, some Governments including Australia’s have moved away from funding work experience and training programs directly and towards paying contracted employment service providers according to the outcomes they achieve. This principle underpinned the Job Network and its successor program, JSA.

Yet competition among providers and funding to outcomes are not ‘silver bullets’ to reduce long term unemployment. If providers are rewarded for short term results and funding levels are low, they have little incentive to undertake patient investment in programs to help overcome people’s barriers to work and tend to focus on low-cost job search assistance methods. To overcome this problem, and the associated problem of ‘parking’, around one third of the resources in both the Job Network and JSA programs were quarantined for investment in such activities as work experience and training, generally during periods of compulsory ‘intensive activation’.

The main changes introduced when the JSA model replaced the Job Network included a shift in resources from intensive support for long term unemployed people towards assistance for those at risk of it. Evaluation data suggests that this is not an optimal use of resources, since intensive employment assistance has its greatest impact, on average, once a jobseeker is unemployed for one to three years. Since the introduction of JSA, employment outcomes for long term unemployed people have declined sharply relative to those unemployed for less than a year. The diversion of resources away from long term unemployed people (who typically receive an interview every two months plus $500 worth of further assistance) has probably contributed to this result. There is a strong case for targeting more resources towards people unemployed long term, by funding providers to offer at least the same level of support to people who are long term unemployed as those at risk of it.

Overall levels of public funding for employment assistance for unemployed people in Australia are low by OECD standards and concentrated on low-cost job search assistance and activation. Within the envelope of our very modest additional expenditure on work experience and training, the JSA model shifted investment away from work for benefits (Work for the Dole) towards vocational training. Shifting resources away from work for benefits is probably cost effective as the unit cost of Work for the Dole was over $2,000 and the impact of ‘work for benefits’ programs on the job prospects of participants is very limited. However, the same can be said of the short training courses to which these resources were diverted in the JSA model. The modest boost to wage subsidies for long term unemployed people is likely to be cost effective, but only 10,000 places have been allocated annually despite strong employer demand. There is a strong case for higher investment in wage subsidies for people unemployed long-term.

In response to provider concerns about ‘red tape’ and excessive control of their work by the Employment Department, the JSA model loosened those controls to a degree. It gave providers more latitude to determine the nature and timing of periods of intensive activation and assistance for disadvantaged jobseekers. However, this was unlikely to lead to genuine ‘individualisation’ of employment services, or improved employment outcomes, while average employment consultant caseloads remained at over 100 jobseekers. A substantial increase in provider service fees for people unemployed long term as suggested above is needed to reduce caseloads to make room for individualised assistance.

In the absence of net impact evaluations, we do not know whether the JSA model is more effective in assisting disadvantaged jobseekers. The limited evidence available from Australian and international evaluations suggests that it is unlikely to be substantially more or less effective than the Job Network for the average participant, but that outcomes for people at risk of long term unemployment have probably improved at the expense of those already in that situation.

When it comes to employment services, as in other fields of public policy, Governments achieve the outcomes they are prepared to pay for. 

References

Anderson T et al (2004), Work-Based Learning for Adults: an evaluation of labour market effects commissioned by the Department for Work and Pensions, London: National Centre for Social Research.

Australian Bureau of Statistics (various years), Labour Force survey, Canberra

Australian Council of Social Service (2006), The role of education and training in Welfare to Work policies, Sydney Australian Council of Social Service.

Australian Council of Social Service (2012), Submission to Senate Inquiry into the Adequacy of Allowance payments, Sydney:Australian Council of Social Service.

Australian Council of Social Service, Business Council of Australia, and Australian Council of Trade Unions (2014), Partnerships to assist disadvantaged jobseekers into secure jobs, Melbourne.

Australian Government (2008), Budget Paper No 2, Budget Papers, Canberra.

Australian Government (2009), Budget measures, Budget Papers, Canberra.

Australian Government (2011), Building Australia’s Future Workforce, Budget Papers, Canberra.

Baker A (2013), Not Looking for Work: The Rise of Non-Jobseekers on Unemployment Benefits, Sydney:Centre for Independent Studies.

Bivand, P (2006). Evaluation of StepUp pilot, final report. London:Department for Work and Pensions.

Borland J, & Wilkins R (2003), Effects of activity tests on exit from payments – the 9 months intensive interview, , Melbourne: Melbourne Institute Working Papers 25/03.

Bowman D & Lawlor J (2010), Evidence and experience: Report on the Individual Placement and Support project for disadvantaged job seekers with mental health issues, Melbourne:Brotherhood of St Laurence.

Bredgaard T & Larsen F (2005). Contracting out the public employment service, a quasi market analysis. Transitions and risk, new directions in social policy. Melbourne:Centre for Public Policy, University of Melbourne.

Breunig R et al (2003), Assisting the long-term unemployed: Results from a randomized trial, Economic Record, Vol 79.

Brock T et al (1994), Unpaid work experience, for welfare recipients, Findings and Lessons from MDRC Research, Washington:Manpower Development Research Corporation.

Card D, et al (2009), Active labour market policy evaluations, a meta analysis, IZA Discussion Paper No 4002, Bonn: Institute for the Study of Labor.

Considine M (2001), Enterprising states, Cambridge:Cambridge University Press.

Considine M et al (2012), Increasing Innovation and Flexibility in Social Service Delivery Australian Report back to Industry Partners, Melbourne: School of Social and Political Sciences University of Melbourne.

Danish Ministry of Employment (2000), More People at Work, Analysis Paper 6, Effects of activation measures, Copenhagen, at: http://www.bm.dk/Aktuelt/Politiske%20aftaler/~/media/BEM/Files/Dokumenter/Politiske%20aftaler/Flere%20i%20arbejde/Effekter_af_aktiveringsindsatsen.ashx

Davidson P and Whiteford P (2012), Overview of Australia’s system of income and employment assistance for the unemployed, Paris:OECD Social Employment and Migration Working Paper No 129

Department for Work and Pensions (2012), Early impacts of Mandatory Work Activity, London: Department for Work and Pensions.

Department for Work and Pensions (2013), Support for the very long term unemployed trailblazer longer term analysis of benefit impacts, London:Department for Work and Pensions.

Department of Education, Employment and Workplace Relations (2008a), Portfolio Budget Statement, Budget Papers, Canberra.

Department of Education, Employment and Workplace Relations (2008b), Welfare to Work Evaluation Report, Canberra.

Department of Education, Employment and Workplace Relations (2010), Labour market assistance, a net impact study, Canberra.

Department of Education, Employment and Workplace Relations (2011), Labour Market Assistance Outcomes, Year ending June 2011, Canberra.

Department of Education, Employment and Workplace Relations (2012a), Employment Pathway Fund evaluation: wage subsidies, Canberra.

Department of Education, Employment and Workplace Relations (2012b), Labour Market Assistance Outcomes, Year ending June 2011, Canberra.

Department of Education, Employment and Workplace Relations (2012c), Employment Pathway Fund evaluation, Canberra.

Department of Education, Employment and Workplace Relations (2012d), Submission to Senate Inquiry into the Adequacy of Allowance payment, Canberra.

Department of Education, Employment and Workplace Relations (2013), Response to Senate Education Employment and Workplace Relations Committee, Question on Notice EW0868_13, Canberra.

Department of Education, Employment and Workplace Relations (2014), Response to Senate Standing Committee on Education and Employment, Question on Notice EM0094_14, Canberra.

Department of Employment (2014a), Response to Senate Employment and Workplace Relations Committee Question No EM0094_14

Department of Employment (2014b), Response to Senate Estimates Question NoEM0172_14.

Department of Employment (2014c), Response to Senate Standing Committee on Education and Employment, Question on Notice EM0948_11, Canberra.

Department of Employment (2014d), Response to Senate Standing Committee on Education and Employment, Question on Notice EM0093_14, Canberra.

Department of Employment (2014e), Response to Senate Standing Committee on Education and Employment, Question on Notice EM0098_14, Canberra.

Department of Employment and Workplace Relations (2002a), Active Participation Model, Budget Papers, Canberra.

Department of Employment and Workplace Relations (2002b), Evaluation of the Job Network, stage 3 effectiveness report, Canberra

Department of Employment and Workplace Relations (2006), Jobseeker account evaluation, Canberra.

Department of Social Services (2014), Statistical summary of social security payments, Canberra.

Finn D and Schulte B (2008), Employment first, activating the British Welfare State, in Eichhorst W, et al, Bringing the jobless into work? , Berlin:Springer.

Fowkes L (2011), Rethinking Australia’s employment services, Sydney:Whitlam Institute

Freud D (2007), Reducing dependency, increasing opportunity, London: Department for Work and Pensions,

Goul Andersen J and Pedersen J (2006), Continuity and change in Danish active labour market policy 1990-2007, Centre for Comparative Welfare Studies Working Paper no. 2007-54, Aalborg: Department of Economics, Politics and Public Administration, Aalborg University

Graversen B and Jensen P (2006), A reappraisal of the virtues of private sector employment programmes, IZA Discussion Papers No. 2230, Bonn: Institute for the Study of Labor.

Graversen B and Van Ours J (2008), How to Help Unemployed Find Jobs Quickly: Experimental Evidence from a Mandatory Activation Programme, Journal of Public Economics, 92:2020-2035.

Hasluck C and Green A 2007, What works for whom? Leeds:Department for Work and Pensions Research Report No 407.

Heckman J, et al (1999), The Economics and Econometrics of Active Labor Market Programs, in Ashenfelter O and Card D, Handbook of Labor Economics 3:1865–2097, Amsterdam:Elsevier.

Hotz V et al (2004), The long term gains from GAIN, NBER Working Papers No 807, Cambridge MA:National Bureau of Economic Research.

Jespersen S, et al (2004). Costs and benefits of Danish active labour market programs. Working paper No1. Copenhagen: Danish Economic Council.

Johri R, et al (2004), Evidence to date on the working and effectiveness of ALMPs in New Zealand’ Wellington:New Zealand Department of Labour and Ministry for Social Development.

Kalisch D (1991), Long term UB recipients, the impact of Newstart and economic conditions, National Social Policy Conference, Sydney: July 1991.

Killackey E, et al (2008), Vocational intervention in first-episode psychosis: individual placement and support v. treatment as usual, The British Journal of Psychiatry 193:114–120.

Machin S and Manning A (1999), The causes and consequences of longterm unemployment in Europe. in Ashenfelter O and Card D, Handbook of Labor Economics, Paris: Elsevier.

Martin J and Grubb D (2000), What works among active labour market policies, OECD Economic Studies 30:1, pp79-113, Paris:Organisation for Economic Cooperation and Development.

Meagher N (2009), The Role of Training and Skills Development in Active Labour Market Policies, International Journal of Training and Development 13:1, Malden MA:Wiley‐Blackwell.

O’Connor B (2009), Improved participation requirements for parents and carers, Media Statement, Canberra: Minister for Employment Participation.

Organisation for Economic Cooperation and Development (2001), Innovation in labour market policies, the Australian way, Paris:Organisation for Economic Cooperation and Development.

Organisation for Economic Cooperation and Development (2005), Employment Outlook, Paris:Organisation for Economic Cooperation and Development

Organisation for Economic Cooperation and Development (2012), Activating jobseekers, how Australia does it. Paris:Organisation for Economic Cooperation and Development

Organisation for Economic Cooperation and Development (2014), Social expenditure data base

Productivity Commission (2002), Report of the independent review of the Job Network, Canberra:Productivity Commission.

Saunders P. and Taylor R (2002), The price of prosperity : the economic and social costs of unemployment. Sydney: University of New South Wales Press.

Tempelman C, et al (2010), Cost benefit analysis of reintegration, Report commissioned by the Ministry of Employment and Social Affairs, Amsterdam:SEO.

Van Reenen J (2003), .Active labour market policies and the British New Deal for Young people in context. Cambridge MA :National Bureau of Economic Research Working Paper No. 9576.

Vanstone A (1996), Reforming employment assistance, Canberra:Department of Employment, Education Training and Youth Affairs.

Wells W (2001), From Restart to the New Deal in the UK, in Organisation for Economic Cooperation and Development (2001), Labour market policies and the public employment service, Paris.

Wren T (2012), Demand-led workforce strategies, presentation to Australian Council of Social Service conference, Melbourne: March 2012.

Endnotes

[1] Machin S and Manning A (1999); Saunders P. and Taylor R (2002).

[2] Department of Education, Employment and Workplace Relations (2012d); Australian Council of Social Service (2012).

[3] Department of Employment (2014e).

[4] Australian Bureau of Statistics (various years), Labour Force survey, Canberra.

[5] This is a different statistic to long term unemployment as measured in labour force surveys. Some unemployment payment recipients are not classified as unemployed (e.g. because they are undertaking training) while some long term unemployed people are not eligible for income support (e.g. due to their partners earnings).

[6] Department of Employment (2014d).

[7] Department of Education, Employment and Workplace Relations (2012), op cit.

[8] Heckman J, et al (1999).

[9] Details of the Job Network and Customised Assistance are provided in a later section of this paper. Customised Assistance was offered to short term unemployed people assessed as ‘at risk’ of long term unemployment, and to people already unemployed for one to three years.

[10] For example, a Danish evaluation of the net impact of labour market programs for unemployment insurance clients found that impacts rose for the first 3 years of unemployment and then fell. See Danish Ministry of Employment (2000).

Similarly, a net impact evaluation indicated that programs were more effective for Dutch unemployment insurance clients when implemented after one year of unemployment. See Tempelman C, et al (2010).

[11]Card D, et al (2009).

[12]For an influential argument for such investment on narrower cost-benefit grounds, see Freud D (2007).

[13] Graversen B & Van Ours J (2008); Wells W (2001).

[14] Kalisch D (1991); Department of Education, Employment and Workplace Relations (2008b).

[15] Borland J, & Wilkins R (2003).

[16] Borland J & Tseng T (2003).

[17] Breunig R et al (2003).

[18] Van Reenen J (2003).

[19] Hasluck C and Green A (2007).

[20] Goul Andersen J and Pedersen J (2006); Finn D and Schulte B (2008).

[21] Hasluck C and Green A (2007).

[22] For good summaries of international evaluation studies see: Card D,et al (2009); Martin J and Grubb D (2000) and Hasluck C and Green A (2007).

[23] Jespersen S, et al (2004); Graversen B and Jensen P (2006).

[24]; OECD (2005); Bivand, P (2006).

[25] Johri R, et al (2004); Department for Work and Pensions (2012); Department for Work and Pensions (2013); Brock T et al (1994).

[26] Meagher N (2009); Hotz V et al (2004); Anderson T et al (2004).

[27] Australian Council of Social Service, Business Council of Australia, and Australian Council of Trade Unions (2014).

[28]Wren T (2012).

[29] Bowman D & Lawlor J (2010); Killackey E, et al (2008).

 ^[30]Australian Government (2011).

[31] For a more detailed history of the employment services system since the 1990s, see Davidson P & Whiteford P (2012); OECD (2001); OECD (2012).

[32] Vanstone A (1996).

[33] Productivity Commission (2002). Star Ratings were developed using regressions to compare actual employment outcomes with predicted outcomes based on individual characteristics.

[34] OECD (2012), op cit

[35] Their client populations are different. Figure 4 excludes the least disadvantaged Job Network participants (those unemployed for less than 3 months) but the results for JSA include a relatively disadvantaged group (Stream 4 participants) who were excluded from the Job Network as most were referred instead to the PSP program.

[36] Stream 3 clients are assessed as facing a high risk of long term unemployment. Outcome payments are not included in these calculations due to a lack of publicly available data on typical employment outcomes achieved.

[37] Australian Government (2011).

[38]If employment assistance increases the probability of payable outcomes by 10%, the reward for provider investment that results in a ‘full outcome’ for a person unemployed for 5 years would average $3,900 X 10% or about $400 per head.

[39] As a proportion of outcomes achieved in 2008.

[40] In September 2013. See Department of Employment (2014a).

[41] Vanstone A (1996).

[42] Bredgaard T and Larsen F (2005); Department of Employment and Workplace Relations (2002b); Productivity Commission (2002).

[43] Department of Employment and Workplace Relations (2002a).

[44] The Government funded Community Work Coordinators to establish work experience placements and projects, and connect Job Network participants with them, at an average unit cost of around $2,000.

[45] Australian Council of Social Service (2006).

[46] The OECD reported that providers did not consider they can fund Work for the Dole places at their previous level using EPF credits. See OECD 2012, op cit.

[47] Australian Government (2009); Baker A (2013).

[48] O’Connor B 2009.

[49] Table 5 indicates that the overall share of Jobseeker Account/EPF credits invested in training rose substantially from 23% to 33% while the share devoted to wage subsidies remained stable at around 22%.

[50] Department of Education Employment and Workplace Relations (2010b); Australian Government (2011).

[51] Organisation for Economic Cooperation and Development (2014).

[52] Department of Education, Employment and Workplace Relations (2012a).

[53] Short courses costing less than $1,000 might include ‘responsible service of alcohol’ courses necessary for people to work as bartenders. The Government funded an additional 180,000 vocational education and training (VET) places in 2010-11 under its ‘Productivity Places Program’, of which about half were earmarked for unemployed people. However, the LMAO data indicate that of 181,000 JSA activities in 2010-11 that comprised training courses, only 20,000 were financed under the Productivity Places Program. OECD (2013); Department of Education, Employment and Workplace Relations (2012), Labour Market Assistance Outcomes, Year ending June 2011, Canberra.

[54] For a detailed discussion of the poor state of data on participation in and outcomes from training by jobseekers, see OECD (2012), op cit.

[55] Australian Government (2008).

[56] Considine M (2001).

[57] Fowkes L (2011).

[58] Considine M et al (2013).

[59] Considine M et al (2012).