An in-depth look at our robo-debt test case

An in-depth look at our robo-debt test case

Tuesday, 14 April 2020

In this article lawyers Miles Browne and Charley Brumby-Rendell shared their experience of running our successful test case against robo-debt, with the Law Institute Journal.

L–R: Client Madeleine Masterton, Managing Lawyer Miles Browne, Lawyer Charley Brumby-Rendell and client Deanna Amato
L–R: Client Madeleine Masterton, Managing Lawyer Miles Browne, Lawyer Charley Brumby-Rendell and client Deanna Amato


In November 2019, Centrelink’s Robodebt system came to an abrupt halt after years of public criticism. The Commonwealth conceded in the Federal Court matter of Amato v Commonwealth of Australia (Amato) that the averaging method at the heart of the system was unlawful. Victoria Legal Aid (VLA) clients led the successful legal challenge. However, the outcome is also the result of the hard work of individuals, advocates and community organisations who drew attention to a system that was inaccurate, unfair and unlawful.

The rise of robo-debt

Centrelink introduced the robo-debt system in late 2016 with little publicity. The system allowed the agency to increase the number of potential debts it could raise more than 50-fold – from around 20,000 over the course of a year to 20,000 every week. During this time, VLA saw a 300 per cent spike in calls for legal help from people who had received robo-debt letters in the mail. There was a more than 500 per cent increase in visits to Centrelink information on the VLA website. Our team, which specialises in administrative law and Commonwealth entitlements, spoke with distressed and weary people who told us about the confusion created by impenetrable Centrelink systems and the practical barriers to locating historical evidence demanded by Centrelink to disprove debts. People experiencing mental health issues, or who spoke English as a second language, were hit particularly hard. One of our clients told us that after trying to deal with the robo-debt system, 'I didn’t want to get up and face the day'.

Understanding the problems with the robo-debt system – and its averaging method – requires going back to before robo-debt was introduced. For social security payments such as Newstart, Austudy and Youth Allowance, the amount a person is paid by Centrelink in a fortnight depends on any income they earned for that fortnight. They are required to declare their fortnightly income to enable Centrelink to calculate their payments. The complex technical framework for these calculations is set out in the Social Security Act 1991 (Cth), and did not change with the introduction of robo-debt.

Since 2001, Centrelink has data-matched with the Australian Taxation Office (ATO) to identify potential overpayments caused by recipients incorrectly declaring their income. The data used is the annual PAYG payment summary information reported by employers to the ATO. Where a potential discrepancy was identified, Centrelink would contact the person and ask them to provide more information.

Pre-robo-debt, if they did not respond, or were unable to provide the information, Centrelink would contact their employer or bank directly to obtain evidence of that person’s earnings in each fortnight they received a Centrelink payment. That breakdown of income in each individual fortnight allowed Centrelink to determine whether there had been an overpayment, in which case a debt was raised.

What changed?

Robo-debt introduced two radical developments where a person did not respond, or did not provide the requested information – income averaging and reverse onus.

Income averaging – the use of averaged ATO income data to calculate a debt

Centrelink averaged-out the annual ATO income data over each of the fortnights between the employment start and end date recorded in the ATO income data. That averaged amount became the input for fortnightly income for the purposes of determining the amount of any debt.

Reverse onus – the requirement that people disprove this speculative debt to avoid enforcement action

Centrelink created a reverse onus by requiring people to disprove the resulting robo-debt. Unless a person could prove the robo-debt was incorrect – for example, by providing old employment payslips or bank statements which triggered a recalculation – Centrelink would enforce the debt against them.

A key concession of the Commonwealth in Amato was that raising a debt solely based on 'averaging' of ATO income data was unlawful. Robo-debts would also almost never be correct as the calculation would only work if a person’s income was exactly the same every fortnight. Seventy-nine per cent of people caught up in robo-debt were on Newstart, Youth Allowance or Austudy (1).  People who receive these payments tend to be in casual and unstable employment and, by definition, do not consistently work full time.

Adding penalty fees and debt collectors

Centrelink often added a 10 percent penalty to the robo-debt on the basis that a person had not engaged with the system (for example, by not providing the additional requested information). In the first iteration of robo-debt, this occurred in approximately 72 per cent of debts raised (2). A person’s failure to engage with the robo-debt system is not relevant to the criteria for adding a penalty in s. 1228B of the Social Security Act.

Where a person was not currently receiving payments, the robo-debt would be referred to a debt collection agency, and a notification would be sent to the ATO requiring the ATO to garnish that person’s tax refund when their tax return was lodged. Many of our clients only became aware of a robo-debt when their tax refund was taken or they were contacted by debt collectors.

A key concession by the Commonwealth in the Amato decision was that it was unlawful to add a 10 per cent penalty fee to the robodebt based on the information they had, and to seize a tax refund when there was no lawful basis for the debt.

Requesting a review

If a person challenged the robo-debt, they were often told that the debt would not be reconsidered until they provided payslips or bank statements. A request for statutory internal review was usually met with the same response or diverted to an 'informal' reassessment. Only 1.1 per cent of debts received a statutory internal review (3). Many people who obtained a statutory internal review found that the review officer continued to use averaging to determine the robo-debt.

Challenging robo-debt

From 2017, there was sustained public criticism of robo-debt from people affected, community networks and legal centres, academics and other advocates. This advocacy triggered a Commonwealth Ombudsman investigation and reports, Senate inquiries, growing public awareness and media reporting. There were some tweaks to robo-debt as a result of this scrutiny, but the use of averaging remained consistent. Based on what we were hearing from our clients, we had serious concerns about the lawfulness of this process. This view was confirmed by administrative law experts who spoke publicly, including Peter Hanks QC who eventually appeared on behalf of our clients (4). Given the number of clients seeking advice and assistance with this issue, VLA recognised the need for authoritative guidance from a superior court about whether key features of robo-debt were unlawful.

Our first test case – Madeleine

Our first client to challenge the lawfulness of robo-debt in the Federal Court was Madeleine Masterton. In June 2018, Centrelink raised a robodebt of $3725.04 against Madeleine and added a 10 per cent penalty. Madeleine’s situation was typical of many robo-debt recipients. She was a young woman whose robo-debt was raised for a period several years earlier, from when she was studying and working part time.

In February 2019, on Madeleine’s behalf, VLA filed an application for judicial review seeking orders that the debt demand and the decision to add the penalty were unlawful. Within a week of Madeleine’s proceeding being filed, Centrelink reduced her robo-debt to $602.87. At the first case management hearing, the Commonwealth sought time to recalculate Madeleine’s robo-debt again. The second recalculation removed the use of averaging, accepted Madeleine’s original declared income and resulted in there being no debt. The Commonwealth then argued that in the absence of a debt, there was no longer a justiciable matter. Ultimately, this question was not determined as VLA filed a second application for judicial review in June 2019 on behalf of Deanna Amato, which was subsequently listed for final hearing. Madeleine’s matter was deferred until the resolution of Deanna’s case. Following the outcome in Deanna’s case, Madeleine’s matter was discontinued with the Commonwealth to pay her costs.

Our second test case – Deanna

Deanna Amato’s case was a clear demonstration of the most controversial aspects of robo-debt – in particular, the inaccuracy caused by using averaged ATO income data. Deanna worked for the same employer for the entire financial year in question, before switching from full time to casual employment when she began studying in February. Her robo-debt related to six fortnights from April to June when she received Austudy. By averaging ATO income data, Centrelink overestimated Deanna’s income by almost fourfold for those fortnights.

Deanna first became aware of the alleged $3215.38 robo-debt almost a year after it was raised, when her tax refund of more than $1700 was taken as a repayment. Centrelink had sent correspondence to her old address which she had moved out of almost three years prior (and after she had ceased receiving payments). Centrelink then added a 10 per cent penalty as Deanna had not 'engaged' with Centrelink. Deanna instructed us that she wanted to bring the robo-debt test case 'Because I think the process isn’t lawful and I feel bad for the tens of thousands of other people who have had to go through this without the help that I have had.'

After Deanna commenced the litigation, Centrelink decided to check whether her robo-debt was correctly calculated. Centrelink used its information-gathering powers to obtain payslips from Deanna’s employers and determined that Deanna was actually only overpaid $1.48 in total, which was waived (5). Deanna received a refund for the amount Centrelink had already taken. A subsequent Freedom of Information request revealed that the calculations showed that Deanna was actually underpaid just over $480. Deanna’s case was listed for hearing in December 2019.

In the period leading up to the hearing, another Senate inquiry into robo-debt was underway with hearings across the country, with more people affected by Centrelink and advocates calling for the robo-debt system to be re-considered. Pressure continued to mount as law firm Gordon Legal announced its intention to initiate a class action against the Commonwealth seeking compensation for those with unlawful robo-debts.

Deanna’s robo-debt win

The central argument was the application of relevant Social Security Act provisions to key elements of the robo-debt scheme. Deanna argued that it was unlawful for Centrelink to demand a debt where the decision that a debt had arisen was:

  • based only on a predictive and speculative judgment by reason of her income over an irrelevant period (being her annual income) that there was some chance that she was overpaid in the relevant periods (being each of the six fortnights)
  • not based on any real evaluation of the material before the decision-maker nor on the information that could have been obtained by Centrelink from Deanna’s employer.

One week after we filed our submissions and just two weeks before Deanna’s final hearing, Centrelink staff were directed to stop raising debts based solely on averaging ATO income information. The surprising announcement removed the key element that made robo-debts inaccurate and unlawful, and brought an enormous sense of relief for those who had been agitating for change for three years.

The following week, Deanna was successful in her test case, obtaining orders made by consent (6). The Federal Court declared that the demand for payment of the alleged debt was not validly made because the information before the decision-maker was not capable of satisfying them that a debt was owed; a critical requirement under the Social Security Act.

The court annexed as notes to the Orders a statement justifying the making of the orders (the contents of which was provided by the parties in a joint letter to the court) (7). In summary the conclusion that Deanna owed a debt was not open on the material before the decision-maker because:

  • the assumption underlying the use of the averaged ATO income data was that Deanna’s annual income was earned in equal fortnightly amounts
  • there was no probative material before the decision-maker to support that assumption
  • there was probative material, being Deanna’s declared income, which indicated that she had not earned her income in equal fortnightly amounts.

In the circumstances, there was no material before the decision-maker capable of supporting the conclusion that a debt had arisen under the Act and, therefore, the conclusion was irrational.

In addition to there being no debt, there was also not sufficient material before the decision-maker capable of supporting the conclusion that the criteria for imposing the 10 per cent penalty were satisfied. As there was no debt, the decision to garnish Deanna’s tax refund was invalid and as a result, Deanna was entitled to interest on that sum. After the outcome, Deanna summed it up when she told us, 'My robo-debt should never have occurred in the first place ... I hope others who have paid dodgy debts will also have a way to get their money back'.

Where to from here?

Gordon Legal filed proceedings in the Federal Court in November. As of 4 February 2020, almost 10,000 people have registered to be part of the class action.

Centrelink is currently looking at all debts raised under the robo-debt system and freezing repayments for up to five years on those debts where it has identified that averaging of ATO income data has been relied on. The Amato test case has already settled that any debt raised solely using averaged ATO income data is unlawful. Thousands of people are still waiting for potential refunds. However, at the time of writing, Centrelink is yet to announce what steps it proposes to take in relation to the potentially hundreds of thousands of unlawful debts.

Our view is that the effect of Amato is that these unlawful debts must be set aside and speedily refunded.

Miles Browne is the Managing Lawyer of the Economic and Social Rights Program. Charley Brumby-Rendell is a senior lawyer in the Economic and Social Rights Program .


  1. Services Australia, Submission No. 20 to Community Affairs References Committee, Senate, Inquiry into Centrelink’s Compliance Program, p 21, download from the Parliament of Australia website.
  2. Based on statistics in the Services Australia’s Submission, note 3, p 17.
  3. Services Australia Submission, Note 3, p 23.
  4. Along with Frances Gordon, Kateena O’Gorman and Glyn Ayres.
  5. See s. 1237AAA of the Social Security Act
  6. Order of Justice Davies in Amato v The Commonwealth of Australia (Federal Court of Australia, VID611/2019, 27 November 2019).
  7. Services Australia Submission, Note 9.

Was this helpful?