Andrew Leigh

Randomised Trials, Employment Services and Work For All
Connect Up 2023, Surfers Paradise
Tuesday, 25 July 2023

Introduction

I acknowledge the traditional owners of the Yugambeh region, and all First Nations people present here today. I commit myself, as a member of the Albanese Government, to the implementation in full of the Uluru Statement from the Heart, including a constitutionally enshrined Voice to Parliament. Thanks to CoAct CEO Simon Brown and chair Lynn Smart for the invitation to speak at ConnectUp 2023.

In 1944 and 1945, as war raged in the Pacific, HC (Nugget) Coombs and a team of fellow economists in the Ministry of Post-War Reconstruction drafted a White Paper on Full Employment. Commissioned by Prime Minister John Curtin, the result was punchy and bold. It began with an excoriating denunciation of the way that the economy in the interwar years had served Australians:

‘Despite the need for more houses, food, equipment and every other type of product, before the war not all those available for work were able to find employment or to feel a sense of security in their future. On the average during the 20 years between 1919 and 1939 more than one-tenth of the men and women desiring work were unemployed.’

The White Paper on Full Employment didn’t mince words. Traditional economists might have called unemployment ‘inefficient’. The White Paper said unemployment was ‘evil’. Yet it also recognised that work needed to be meaningful. The White Paper praised workplace flexibility. It celebrated technological advances. It rejected ‘schemes designed to make work for work’s sake’. It might be true that jobs could be created by paying one group of people to dig holes, and another group of people to fill them up again. But the authors of the 1945 White Paper recognised that for many people, work is integral to their sense of dignity. It was a bold, visionary document, which set out to remake the labour market so that it worked for everyone.

Inspired by John Curtin, Anthony Albanese addressed the National Press Club two years ago, and announced that Labor would deliver our own White Paper. As he said of Curtin’s White Paper, ‘The courage to imagine greater opportunity for all in peace, the leadership to begin that work even in the midst of war… It is an energy that I am proud to say drives me.’ 

Nearly 80 years after Curtin, the economy has changed a great deal. Careers are longer. More women are in paid employment. More Australians work in services, and fewer in agriculture and manufacturing. Workers have more education and use more advanced technologies. We produce several times more per hour than did workers in the 1940s. 

Yet some things have not changed. Work is more than a pay packet. It provides a sense of dignity and connection. It integrates people into the social fabric. As writers such as Jon Cruddas have noted, human beings do not see ourselves primarily as consumers, but as producers. When we meet a stranger at a barbecue, we’re unlikely to ask ‘what do you buy?’. Instead, the more common question is ‘what do you do?’.  

The importance of work is borne out in wellbeing data. Research by economist Nick Carroll findsthat to retain the same level of emotional wellbeing, a person who is unemployed would need to have an income that is $40,000 to $80,000 higher than they had when employed. Studies from the United States and Germany have reached similar conclusions.

At a macroeconomic level, mass unemployment can be seen as a fundamental market failure. When people want to work but cannot find suitable jobs, the productive capacity of the economy is impeded. 

But this impact also plays out at a personal level. When society fails to equip people with the skills to succeed in the labour market, we don’t just hurt their earnings potential, but their potential to experience the meaning and fulfillment that comes with a good job.

Later this year, the government will hand down our employment white paper. We do so in an environment of low unemployment, though not as low as when Curtin’s White Paper was being written. In 1945, the unemployment rate was 2½ percent, even lower than today’s rate of 3.5 percent.

From the Commonwealth Employment Service to Workforce Australia

Central to our goal of sustaining full employment is a revamp of the employment services system. Australia’s employment system was transitioned from the insourced Commonwealth Employment Service to the Job Network in 1998, marking a fundamental shift in Australia’s employment service. The primary role of government in this sector shifted at that time from being a provider to a purchaser of services. A network of providers was established to help Australians move from welfare into work. 

That model was used for the next few decades, in various iterations. In 2009, it became Job Services Australia. In 2015, it became jobactive. In 2022, we saw the introduction of Workforce Australia . The approach shifted to a partially insourced model: services are now delivered to more job-ready participants on the caseload through Workforce Australia Online.  

The current system has been strengthened by changes made in response to an independent expert panel. For many jobseekers, online servicing better reflects today's work and job search environment. I grew up in an era of classified newspaper advertisements – a far cry from the world of online job finding platforms such as Seek, LinkedIn, CareerOne, as well as specialised platforms such as APS Jobs, JobAccess, Ethical Jobs and ArtsHub. 

However, the current system also has a number of weaknesses. The current system is based on an ideology employed over the past 20 years, to reduce the number of people claiming benefits and, ultimately, reduce the cost of running the system. This ideology has shaped the statistical data held in the system. The Department of Employment and Workplace Relations has an excellent knowledge of whether a person is claiming benefits, but the current data assets do not provide a sufficiently clear picture of their earnings once they find a job . Because what gets measured gets managed, this data limitation ends up pushing policy towards ‘any job’ rather than ‘the right job’. It can lead to an underinvestment in education and training, which might have long-term productivity costs.

Another factor is that the caseload has changed. Looking back to other moments at which the system was reformed, Australia’s unemployment rate averaged 8 percent in 1998, 6 percent in 2009, 6 percent in 2015, and 4 percent in 2022. Today, it’s just 3.5 percent. Low unemployment is welcome news. But in an era of low unemployment, a larger share of job seekers are people with significant barriers to employment, or whose skills don’t match the available jobs. 

The job context has changed, and it’s important that employment services are designed to work in the landscape of the day. And, while the current compliance system is considered less punitive relative to other OECD countries both in the value of penalties and when they are applied, our system should ensure we don’t cause unnecessary hardship or make people less employable on the way.

Indeed, CoAct has identified several concerns with the current system. You have argued that the current system provides limited opportunities for social enterprises to work with the most disadvantaged jobseekers. You note that it provides insufficient flexibility to innovate. And you have raised the concern that the Job Seeker Snapshot may not provide accurate results, since jobseekers may want to present themselves in the best possible light and believe that disclosing barriers to employment is detrimental because it may impact their continued receipt of social security payments. 

The Australian Government recognises the challenges of reform. Recent years have seen many changes to the employment services system, including the creation of the digital caseload. Deeds for the latest round of service provision were only just signed last year, immediately before the caretaker period began.  We are not after change for change’s sake, and we are sensitive to the risk of disrupting the sector. But equally, we will not just adopt a ‘set and forget’ approach to a set of deeds for programs which have forecast expenditure of $7.3 billion over five years (from 2022-23), making it the Commonwealth’s largest non-defence procurement.

The Workforce Australia Employment Services Inquiry

That is why we have taken a methodical approach to reform. In August 2022, Minister Tony Burke established an inquiry into Workforce Australia Employment Services. The inquiry is headed by my colleague Julian Hill.

I cannot imagine a better person to run this inquiry than Julian Hill, who I have known for 25 years, and am proud to call a friend. One of the smartest and most experienced Members of Parliament, Julian has worked as a public servant at the state and federal levels, and as an elected official in local and federal government. Julian represents a community where the unemployment rate has generally been around double the metropolitan average. 

Julian has taken on the inquiry with his usual forensic focus, engaging broadly across the sector, reading everything, and holding 11 public hearings in five different cities to-date. If you have not done so already, I would encourage you to read Julian’s speech to the National Employment Services Association CEO Forum on 4 October 2022, along with the published Submission Guide, which lays out the Committee’s rigorous approach to the inquiry. 

Also serving on the committee are Coalition MPs Russell Broadbent and Aaron Violi, independent Rebekha Sharkie, and Labor members Andrew Charlton, Daniel Mulino and Louise Miller-Frost. I know the three Labor members very well and can attest that they come to the inquiry with many years of thinking about how to boost employment among the most disadvantaged. They are deeply engaged with the big questions that animate this important inquiry. 

The Workforce Australia Employment Services inquiry will report to the House of Representatives by the end of this year. It will be one of the most important parliamentary reports this term, and we will take time to study it deeply.

My job will then begin. As Assistant Minister for Employment, my task is to work with Minister Tony Burke on the government’s response to this inquiry.  

The Value of Randomised Trials

In implementing the response, a major focus will be on improving the quality of evidence. 

In this year’s Budget, the Australian Government established the Australian Centre for Evaluation, with a remit to conduct rigorous evaluations, with a focus on randomised trials. 

In some contexts, government relies heavily on randomised trials. When assessing new pharmaceuticals, we want to know how they perform against a randomly selected control group who have been giving sugar pills, saline injections or some other form of placebo. 

Medical researchers have long recognised the fact that most patients eventually get better. Suppose I gave you an ineffective medicine when you were sick, and then came back a fortnight later to measure your vital signs. I might erroneously conclude that my snake oil helped you heal, even although it had no impact at all. 

Silly as this sounds, it is actually not far off from the way in which many evaluations are currently conducted. Studies comparing people before and after they receive an intervention are commonplace. Every year, governments pay millions of dollars for studies like these. Like snake oil, they make money for the purveyors, but don’t do much good for the world.

To see the problem in a familiar context, consider this chart, produced by US randomista Jon Baron, who runs the Coalition for Evidence Based Policy. It shows the results from the US Department of Health and Human Services’ Comprehensive Child Development Program, which provided intensive case management services to low-income families with young children. As the graph shows, employment rates for mothers in the program more than doubled over the five years in the program.

This might sound impressive, until you compare it with the results from the randomised control group who did not participate in the program. They saw almost identical gains.

As Jon Baron points out, ‘If the Comprehensive Child Development Program had been evaluated in the usual non-rigorous way (examining employment outcomes without reference to a control group), it would've been deemed highly effective.’

The same pattern holds true for children in the program. Over time, the share who were ‘at risk’ fell substantially. A naïve researcher might conclude that the program benefited children.

But it turns out that almost exactly the same improvement can be seen among children in the control group. 

Without a randomised trial, researchers could reach precisely the wrong conclusions about the effectiveness of such a program.

Bringing rigorous evaluation to government programs is in the interests of all citizens. I’m yet to meet a taxpayer who would like to see more of their taxes spent on programs that don’t deliver. 

But improving the quality of evaluation is especially important for the most disadvantaged. People who rely on government services suffer most when those services do not work. Rigorous evaluation isn’t just about improving the efficiency of government; it’s also vital for reducing inequality.

Randomised Trials of Employment Services

Those of you in the room, will, of course, be asking how evaluation can make a difference for your clients, who include First Nations Australians, people with a disability, older Australians and people from culturally and linguistically diverse backgrounds.

The answer is that by applying randomised trials we will increasingly build a robust evidence base around what does and doesn’t work. We will also be able to know with confidence who benefits most from programs. As a wise randomista once observed to me, no program has the same impact on all participants. They might be more effective for older participants, for regional participants, or for First Nations participants. Rigorous evaluation can help tailor programs for particular cohorts.

We will work carefully to ensure that randomised evaluations are conducted in a robust ethical framework, with appropriate oversight. In saying this, it is also worth recognising that perpetuating ineffective programs is itself an ethical failure. Government has an ethical obligation to measure the impact of programs, and continually improve them.

By improving the evidence base, Australians who are out of work will have access to programs and supports that we can be confident will work for them and their situation. Focus group discussions, in-depth interviews, and surveys with program participants are important complements to this work – and ensure that evaluations can inform what works – and how best to deliver services that are informed by the latest evidence.

This represents a philosophical difference from the way in which the employment services system has historically operated. Large, long, inflexible contracts have not been nimble enough to adapt to sudden shocks, such as COVID. The system has been set up as though we already know everything about how to help people find work. A refreshed system must be driven by the approach of test-learn-adapt, not the approach of set-and-forget. For the sake of the clients that it serves, it must have a stronger approach on learning and on incorporating that learning into implementation. If the challenge was simple, prolonged joblessness would be non-existent. The fact that long-term unemployment persists in the current labour market suggests that we would be better served by a system that is grounded in modesty, evidence, and continuous improvement.

Conclusion

Curtin’s Full Employment White Paper was no mere marketing document. As I have noted, double-digit unemployment was the norm in the interwar period. Yet in the three decades after World War II, unemployment dropped to around 2 per cent. From Federation until the outbreak of the Second World War, Australia’s unemployment rate only once dipped below 3 per cent. From the end of the Second World War until the early 1970s, unemployment only once went above 3 per cent. French economists refer to this 30-year postwar period as Les Trente Glorieuses (‘The Glorious Thirty’). The same could be said of the Australian labour market over these three decades.

Successive Australian Governments have built a complicated employment services system. In implementing the response, a major focus for the Albanese Government will be on improving the quality of evidence. How we evaluate it is critical. The evaluation of government programs must be based on rigorous methods, particularly through the use of randomised trials. This work will build a better feedback loop within government and produce tangible outcomes, evidence-based policy and a culture of continuous improvement.

We will work with the sector, employing rigour, passion and focus to deliver the goal of work for all.