You get what you pay for: Danish & Australian employment programs neglect long-term unemployment

Denmark is an international leader in employment programs. It spends far more than most countries on employment assistance for unemployed people. Since its activation (‘activering’) reforms in the 1990s, it has a proud record of keeping unemployment low and especially reducing long-term unemployment. Contrary to neoliberal orthodoxy, the Danes achieved this despite having one of the highest unemployment benefits in the OECD.

A new report finds that changes announced in 2016 shifted the efforts of Municipal employment services away from the more disadvantaged and longer-term unemployed people towards others who were ‘easier to place’ in jobs.

I was there completing by PhD fieldwork in 2013 when those programs were reviewed, and warned that the proposed changes could weaken employment services for people unemployed long-term. That warning was based on experience of the transition from the Job Network to Job Services Australia in Australia.

So this is a story of two countries, performance-based funding of employment services, and how the best paid plans can go astray.

The success of Danish labour market policy (1994-2008)

Denmark is a labour market success story. For many years its ‘flexicurity’ model has kept both unemployment and labour market income inequality low. This was achieved by a combination of high minimum wages and unemployment payments (paid at up to 90% of previous wages), limited regulation of hiring and firing (though no less than in the Anglosphere), higher turnover of jobs as a result, and investment in education and training to help people adapt to changing labour market conditions. Crucially, this included the largest investments in employment and training programs for unemployed people in the OECD, who were required to take up those opportunities. These programs were targeted to people unemployed long-term.

From the introduction of its activation (‘activering’) policy in the mid 1990s to the recession of 2008, unemployment fell from 9% to 4%. Unlike in Australia, as unemployment fell the share of people reliant on unemployments long-term (over 12 months) also declined. The long-term share of recipients of municipal social assistance (generally regarded as ‘harder to place’ than unemployment insurance recipients) fell from 44% in 1994 to 39% in 2005.  So as the employment tide rose, all boats were lifted. Among all people classified by the statistical authority as unemployed, in 2008 just 12% were unemployed for over 12 months, among the lowest in Europe.  This was important for social reasons (prolonged unemployment is bad for health and entrenches poverty), and also economic reasons (to keep structural unemployment in check).

% of unemployment that was long-term (<12 months) in 2011

ltu in europe14

Since the recession in 2008, the flexicurity model has come under pressure from a  number of directions. A Liberal-led government cut the maximum duration of unemployment insurance from 4 years to 2 years, and more recently access to benefits for migrants was severely restricted by a new rule imposing stringent requirements social assistance payments (including a requirement that people must have at least 225 hours paid work a year to receive them) for people who have not resided for at least seven of the last eight years in Denmark.

A shift from ‘regular activation’ of people unemployed long-term (2016)

The subject of this blog is a change to the funding of municipal employment services in 2016 associated with the Koch Review of employment services in 2014. Instead of guaranteeing regular ‘activation’ through employment programs for people unemployed long-term, the new system reduced funding for municipal employment assistance as the duration of unemployment increased. This turned on its head the previous emphasis on reducing prolonged unemployment.

“The Expert Committee also recommends abolishing the obligation to participate in repeated active measures in order to achieve a greater focus on contents and results. For this reason unemployed persons shall only be obliged to participate in one activating course after six months of unemployment….
“It is recommended that the same rate of reimbursement of municipal expenditure for all benefits and regardless of participation in active measures be introduced and that the rate is reduced over time.” Danish expert group (Koch) report on active employment policy(2014) p9.

I was in Denmark in 2013 concluding fieldwork for my PhD thesis on activation policies in four countries, and spoke with academics and officials there.

My message was this: if funding of employment programs diminishes with the duration of unemployment, there is a risk that people unemployed for a long time would be ‘parked’ and offered limited help.

The logic was straightforward: while there is a strong incentive for the national government to invest in people unemployed long-term, municipalities would have much less incentive to invest in this group unless the government supported and rewarded them to do so.

As the Dutch experience showed, if national funding of municipalities depends on quick reductions in the number of people on Social Assistance payments, they are likely to concentrate on ‘easy wins’ and not those who face greater labour market disadvantage.

I relied mainly on Australian experience with the transition from the Job Network to Job Services Australia in 2009 (as explained later).

Some background: unemployment payments and employment services in Denmark

Unlike Australia, the Danish system of unemployment benefits is divided into two parts: unemployment insurance for workers who have sufficient employment experience in Denmark to qualify (which runs for up to 2 years and is paid largely by union-run social insurance funds), and social assistance for people with low private incomes who are not eligible for unemployment insurance or disability payments (including insured workers whose unemployment insurance expires, new migrants, and many young people and sole parents who have not accumulated entitlements to unemployment insurance). Social assistance is paid by municipalities.

Employment services for unemployed people – both insured and uninsured – were run by municipal employment offices, though the help received by the two groups differs sharply. Insured unemployed people were more likely to receive job search assistance, vocational training or traineeships, while people on social assistance, especially those considered furthest from the labour market, were more likely to be engaged in municipal ‘projects’: team-based work experience programs that didn’t pay wages or offer vocational training.

Why was the funding model for Danish employment services changed?

The pre 2016 arrangements for funding municipal employment services were far from perfect.

Traditionally, municipal welfare services paid little attention to the labour market as employment services were the responsibility of the national employment service (AMS). Further, municipal social workers viewed most of their social assistance clients as too disadvantaged to join the paid labour force.

This changed progressively over the 1990s and 2000s in response to national ‘activation’ policies for unemployed people first introduced by the Social Democrats in 1994, which progressively extended job search and training requirements for unemployed people.

In the mid 2000s, a Liberal-led government pushed for greater uniformity of employment assistance for insured and uninsured workers. Those on Social Assistance, along with Unemployment Insurance recipients, were required to participate in regular, intensive activation programs once they were unemployed for 6 months. Those assessed with ‘problems other than unemployment’ (social barriers to employment) were required to join municipal projects or employment preparation programs.

Controversially, the national employment service (run by the national labour market authority with employers and unions on its governing board) was progressively replaced by a network of local services (run by municipalities with oversight from a restructured labour market authority). Also, the maximum duration of unemployment benefits for insured workers was halved from 4 to 2 years (after which those still out of work and lacking private income would need to apply for municipal social assistance).

In response to reports of wide variations in the quality and quantity of activation of municipal social assistance recipients, the Liberal-led government progressively reduced municipal discretion over the frequency and character of ‘activation’ for social assistance recipients. To do so, it use the ‘power of the purse’, as most municipal social assistance benefits and employment services were funded by the national government. Following the American ‘welfare to work’ model, municipalities were financially penalised if recipients were not ‘activated’ on time, and employment assistance funding was more generous for the government’s favoured programs (work experience in private employment) than municipal projects or vocational training, which were frowned upon.

The result was a complex and prescriptive model of employment assistance financing that was widely criticised (including by the Employment Ministry itself) by the time the recession came in 2008.

Also during the 2000s, access to benefits for new migrants was severely restricted – a policy pursued by both sides of politics in Denmark in response to a backlash against non-European migration, especially among older working class voters. This revealed a major fault-line in the previously inclusive Danish welfare model.

The Koch review was instigated by a Social Democratic-led government in 2013, five years after the recession. That government was under pressure from the unions to restore four-year unemployment insurance payments and invest more in vocational training for insured unemployed workers. It declined to restore the previous duration of unemployment insurance, but accepted the Koch review’s recommendation of a right to vocational training for insured workers after 6 months’ unemployment, and took a number of steps (including a guarantee of work or training for unemployed young people) to ease the employment impacts of the recession on those most affected.

When the Koch review’s first report on employment services for insured unemployed workers turned its attention to funding arrangements for employment assistance, it voiced the widespread criticism of the complex, prescriptive model and turned to the New Public Management rulebook for answers.

One possible response within the NPM framework, which Danish policy makers had to date resisted, was to fund employment services according to outcomes as well as service inputs, as their Dutch and Australian cousins had done. The idea here is that the purchaser should not concern themselves too much with how employment outcomes are achieved; it should reward outcomes achieved.

A well known problem with performance-based financing of employment services is the difficulty in measuring the net impact (value added) by a service. Outcomes based funding biased services towards low-risk, low-cost interventions such as job search assistance and motivational strategies, and opened up opportunities for service providers to cherry-pick those unemployed people they considered most likely to secure employment. The losers in this process were those most disadvantaged in the labour market, who need employment services willing to invest substantially and patiently in employment preparation, and work intensively with employers willing to give them an opportunity.

These problems can be overcome by investing in better measurement of net impacts (as Australia’s Star Ratings system attempts to do) and by hybrid funding models that combine funding for investment and activity (for example, wage subsidies) with outcome payments for those assessed as more disadvantaged. A variant of this hybrid approach was advocated by Australia’s Expert Panel on Employment Services (of which I was a member) in 2019.

The Koch review recommended a funding model for employment services that sits somewhere between prescriptive funding of service inputs (such as activation courses and placements) and an outcomes-based funding model. It proposed a simpler, uniform  schedule of payments to municipalities for employment assistance (regardless of the type of help provided) that diminished as people were unemployed for longer. The idea was to encourage municipalities to find people jobs as quickly as possible. In the funding model that was adopted, the reimbursement rate from national government was reduced over time from 80% in the first four weeks, 40% in weeks 5-26, 30% in weeks 27-52 and 20% from week 53 onwards (see graph below).


In its second report (on employment services for social assistance recipients), the Koch commission warned that:

”The restructuring will thus give the municipalities a greater part of the financial responsibility for public benefits and thus increase the incentive to invest purposefully in getting citizens on the margins of the labor market in permanent employment. Conversely, there is also some risk that municipalities are more reluctant to provide company-focused offers to the most vulnerable citizens if they are uncertain whether the action will actually bring the individual citizens into work [emphasis added].”

Expert group on the assessment of active employment efforts (2015), New avenues for jobs – for citizens on the outskirts of the labor market, p176

What was the outcome of this change in targeting away from the most disadvantaged?

A recently-released report of the Danish National Social Research Institute  found that the new reimbursement system for municipal employment services did alter incentives for municipalities, but not always in the way intended.

Two-thirds of the municipalities have re-organized the job centers, and at the same time they often set up business cases for how resources are best allocated…
The results indicate that reimbursement has had a positive effect on the strongest unemployed (unemployment benefit recipients and social assistance clients who are closest to the labor market), but on the other hand a negative effect for more vulnerable groups, especially social assistance beneficiaries furthest from the labour market. So the change in the state reimbursement scheme to the municipalities means that the municipalities reap the low-hanging fruits, while those more disadvantaged are neglected (at least in the first year of the reform that we have examined). [emphasis added]

Pedersen N et al (2020),The reimbursement stairs and the low-hanging fruits in the job center. Danish National Social Research Institute (VIVE), 27/1/20

The Institute’s findings were nuanced: in municipalities with a large share of people unemployed long-term, that group was more likely to be targeted for investment in employment assistance, while those with lower rates of long-term unemployment were more likely to neglect people unemployed long-term.

The Australian case: from Job Network to Job Services Australia

This is covered in more detail in previous blog.

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.

In the Job Network model, employment services were tendered by the Employment Department every three years. Providers were paid a commencement fee per jobseeker and outcome fees for a range of specified employment and training outcomes, and had access to a Jobseeker Account to invest in work experience, training and other supports. 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 paid to providers.

The highest fees were paid in the ‘Customised Assistance’ stage which lasted for six months following the 12th or 24th month of unemployment (earlier if the jobseeker was assessed as disadvantaged). In addition, unemployed people were required to undertake  ‘Mutual Obligation’ for six months of each year of unemployment (alternating with Customised Assistance), This involved work related activity such as training, part time employment, or work for benefits.

In 2009 Job Services Australia (JSA) replaced the Job Network, retaining its basic structure. A key change was the shift from duration-based targeting to risk-based targeting. People were streamed into more intensive services according to their Job Seeker Classification Instrument scores rather than unemployment duration. The higher streams (3 and 4) attracted higher fees for providers.

After 12 months’ unemployment, a jobseeker entered a Work Experience phase which attracted lower payments for the provider than those in Streams 3 or 4. Duration of unemployment was but one of many factors considered when assigning people to streams.

Resources were thus shifted from ‘curing’ long term unemployment towards ‘preventing’ it.

Typical administrative and job-seeker account payments for participants in the Job Network and participants in Stream 3 of Job Services Australia (the second-most disadvantaged stream) are shown in the graph below.

The JSA payments for those at high risk of long-term unemployment ($2,200) were much higher in the first year of unemployment than Job Network payments ($853). Conversely, in the second and third years, the Job Network payments were much higher. While payments for employment outcomes under JSA were higher for people unemployed long-term, this did not compensate for the lower payments for service inputs as only a minority achieved a payable outcome.

jsa fees13

The graph below compares employment outcomes for JSA participants in 2013 with those for Job Network participants in 2008. JSA outcomes were consistently lower (the green bars show the difference between the two), due to a weaker labour market at the time.

employment outcomes ltu13

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%. For those unemployed over 3 years, outcomes were 41 percentage points less under JSA.

It was these graphs I used in meetings in Denmark to argue for sustained investment in employment assistance for people unemployed long-term.


The latest evidence from Denmark points in the same direction as that from Australia:  financial incentives for investment in employment services work, but they don’t always work well for people with major barriers to employment.

Since it’s hard for national governments to accurately assess the impact of employment services on their job prospects (especially the degree to which they would have secured a job without assistance), they should directly fund and then monitor the help that people unemployed long-term receive from services on the ground. This inevitably introduces a degree of inflexibility and complexity into the purchasing of employment services.

Otherwise, non-government employment service providers (in Australia) or municipal employment services (in Denmark) will lean towards investment in the lowest-cost, fastest-acting help for those in the target group they assess as closest to employment (even if the target group is people unemployed long-term or otherwise disadvantaged). That usually means job search assistance combined with motivational strategies  (including frequent  compulsory interviews and courses and penalties for non-attendance). They are unlikely to risk investing in more costly, slower-but-longer-lasting interventions such as wage subsidies and vocational training.

One way to overcome reluctance to invest in the help that many people disadvantaged in the labour market need is to remove local discretion, by running national employment programs (work experience, training and so on). National programs never completely  disappeared from the employment services landscape in countries like Australia and the United Kingdom that followed NPM principles and funded employment services to outcomes. Yet they are not always the most cost effective approach.

An alternative, adopted in Australia, is to finance the more intensive and expensive (but necessary) assistance for people unemployed long-term through a fund which providers can use for a range of interventions (not including ‘core’ employment services such as job matching and job search assistance). The fund would be used to pay for assistance agreed between the provider and each unemployed person. That’s the logic of the ‘Employment Fund’ in Australia’s jobactive program, though in practice it’s been under-resourced and under-utilised by providers (these problems are related to other elements of the performance-based purchasing model, and a long-standing reluctance of Australian governments to adequately invest in employment services).

In the Danish context, consistent reimbursement of a percentage of municipal expenditures on employment services (preferably calibrated so that a higher share of investment in the most disadvantaged is reimbursed), together with monitoring of the intensity and effectiveness of assistance, would have similar effect.

In order to avoid the ‘parking’ of those who are the most disadvantaged in the labour market, the fund should be topped up in each year of employment. Failure to do so in the Job Services Australia scheme (and in the ‘jobactive’ program) meant that people unemployed for a long time did not get the help they needed, and the share of people unemployed long-term rose (at great cost to governments and those affected).

At the same time, the system should reward employment outcomes, in order to avoid the opposite problem: repeated investment in ineffective and meaningless activation.

In Australia this is done directly, via outcomes payments to employment service providers. In Denmark, it is done indirectly by not reimbursing 100% of municipal benefit payments. This raises the risk that, rather than investing in assistance to improve employment outcomes, municipalities may ‘get tough’ with social assistance payments, leaving some people without jobs and benefits. For this reason, it is generally better to reward employment outcomes directly.

Governments should work with municipalities/providers to evaluate ‘what works’ and inform and support best practice. They need to exercise informed stewardship of the system, rather than deposit money in a ‘black box’ then stand back and hope that if the financial incentives are right for municipalities, providers and unemployed people, it will all work out in the end. It often doesn’t, and the system loses credibility with governments and the community.

Finally, government and providers need to pay more attention to local labour markets  and the needs of, (and incentives facing) employers. Employment programs are likely to fail if they ‘activate’, prepare and train unemployed people for paid work without connecting with employers throughout the process. There are two sides to the labour market.












Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s