In testimony before the Royal Commission into the Robodebt Scheme, in a session that lasted an entire day, former Department of Social Services executive Serena Wilson finally broke.
“ I could have acted and I – and I didn’t,” she said on Wednesday afternoon. “I took no step to stop it.”
She was “ashamed” now to think of how she knew the robo-debt program was illegal at least by 2017 – although the inquiry has shown evidence this was known at the beginning of 2015, before it began – but insisted she “felt quite removed” as a senior leader in DSS with other priorities.
Commissioner Catherine Holmes was not persuaded. “There were hundreds of thousands of recipients affected by this. It was a massive program. It was supposed to recoup 1.7 billion [dollars],” she said.
“How removed could you be when you’re the deputy secretary of the department responsible for policy and the way recipients are ultimately treated?”
While Wilson is the first senior public servant to admit she failed to stop the monstrosity that was robo-debt, her evidence fits with that given by the Australian Taxation Office deputy commissioner Jeremy Hirschhorn last Friday, and others, that this project was being railroaded by the Department of Human Services. Figures there were being deceptive.
Although ministerial accountability is a key feature of Australian politics – and that will come into play later – the truth about the origins of robo-debt is that it came from within the “business integrity division” of the DHS.
This is a tale of two departments, linked by policy: the Department of Social Services developed it, while the Department of Human Services delivered it. The thematic similarities ended there. In witness testimony given so far across the first two weeks of this royal commission, the two were unalike in behaviour or dignity. Under what a colleague called the “authoritarian” rule of then secretary Kathryn Campbell, the DHS and ministers had developed a close working relationship. A sense of pre-empting government demands around budget savings was a useful adaptation for an increasingly corporate agency where welfare recipients were now “customers”.
A DHS minute prepared for and signed by the then national manager of the business integrity division Scott Britton, on June 30, 2014, is the earliest reference so far discovered to the origins of robo-debt. This document asks “that you approve the further exploration and trialling of the four concepts” for compliance activity listed. Among the four concepts is a “trusted data assessment” that would use information provided by third parties – for example, income data from the tax office – to “enable automatic updates to customer records”. The second was labelled “intervention through self-service” where welfare recipients would be told to interact or supply information. Recognisably, these are two core features of what would become robo-debt.
In his handwriting on this document, Britton notes: “There are considerable policy issues associated with this intent. We would obviously look to leverage from dereg[ulation] welfare reform. It will be important to commence the policy and legislative review to determine the scope and boundaries of possibility.”
We now know that, by December 18, 2014, DSS had internal legal and policy advice that provided such a boundary: it was not lawful under the Social Security Act to average or smooth income in this way, given benefits are assessed and paid fortnightly, and debts calculated in this way would likely be wrong, especially for people who earned money erratically over a given year via casual work or the gig economy.
In fact, the old DHS process guides for Centrelink compliance officers said exactly that. It was hardly new information.
Despite all of this, robo-debt would make it out of conception and into a policy proposal in early 2015. It would be briefed to the then Social Services minister, Scott Morrison, and become a budget measure and then a “pilot” from July 2015.
While more issues were identified in this manual phase, where staff still checked other sources of information, robo-debt would be turned on full bore by July 2016, attracting horror story after horror story in the press and an “own motion” review by the Commonwealth ombudsman by early 2017. This review led to a new set of legal advice by Anne Pulford, the same DSS lawyer who wrote the 2014 opinion against robo-debt. It came heavily qualified on a narrow technical question but was used anyway to “justify” the entire scheme to the ombudsman, who months later found robo-debt to be lawful. This ombudsman’s report was then relied on by DSS in public to prove – against a mountain of its own internal logic – that the scheme had a legal basis.
External legal advice provided for the first time by law firm Clayton Utz in August 2018 reiterated the point of the 2014 advice and was deliberately left in draft. Robo-debt continued for another year and was suspended only after a class action lawsuit was launched and the government finally sought an opinion from the solicitor-general, who confirmed in eerily similar language the entirety of the position put by the 2014 legal and policy advice: the program was illegal.
The problem, as ever, was money. By October 14, 2014, Britton emailed a director who reported to him, Jason Ryman, requesting “detail about what’s in the pool for each year back to 2010-11”.
Data-matching with the tax office had happened in its current form at least since 1991, when the Data-matching Program (Assistance and Tax) Act took effect, but depending on the level of detail required was also done under a voluntary 2004 protocol developed by the two agencies.
This data was a lot like a potential iron ore deposit: rough, unhewn and the extremely valuable bits were difficult to extract because it required manpower. There were 900,000-odd potential “discrepancies” hinted at by these matches.
“With the PAYG data, there was more cases or possible discrepancies than we were able to actually do any level of investigation,” Britton said.
But what if they could automate that process? What if those hundreds of thousands of potential “discrepancies” hinted at by the matching could be “treated”? What if they just sent letters to customers telling them there was “an anomaly” and get them to do the rest?
That would be a significant payday for the department, would win political brownie points and, happily, attract investment to “modernise compliance” infrastructure that would allow them to do even more in the future. That, of course, would all but guarantee happy political masters down the road, as now.
It is not news, perhaps, that the poor have been used as an extractive resource by the state for some time. But this is an important consideration.
“Am I understanding your evidence,” Britton was asked, “that it was not uncommon when the budget was under consideration for your branch to identify means of increasing compliance activities, to increase savings for the Commonwealth for the purposes of the budget?”
“Yes,” he replied.
Tony Abbott’s “debt and deficit disaster” catchcry had become, in Coalition government, a directive to find savings to assist “budget recovery” or repair. Britton remembers conversations like this with his boss, Mark Withnell, as permeating the general work and culture of the DHS around that time.
By early 2015, this imperative had become such that, against the legal advice received by the DSS and apparently communicated to DHS deputy secretary Malisa Golightly (who died in December last year), an executive minute was drafted to be sent to then Social Services minister Scott Morrison. This in itself was odd, given the usual process was for such matters to go to the minister for Human Services (at the time, Marise Payne) and then on to the senior ministry.
An earlier version of this brief, since “watered down”, stated: “Some of the proposals may have fundamental impacts on security policy and legislation.” Someone in the DSS edited this brief to cross out “fundamental impacts” and replace the phrase with “significant implications for”.
DSS deputy secretary Matt Flavel, who was not in the department at the time and was still selected as the “proper” person to answer questions for this inquiry, was grilled about this change.
“It appears as though the change in language had the effect or was designed to have the effect that the minister would be more inclined to give the proposal his approval,” senior counsel assisting the royal commission Justin Greggery, KC, said. “That is, it reduced a potential hurdle of significance.”
Flavel did not agree.
“It still says ‘significant implications for social security policy and legislation’ and then further goes on to say, ‘will need to be assessed’,” Flavel said.
“I mean, this is a proposal. This is, you know, at the stage of sort of saying, in lay terms, ‘Do you want us to work further on this to work up a proposal for the budget?’ ”
That was exactly what Scott Morrison wanted, apparently expressing a particular interest in the PAYG proposal as it was then known. It had a hefty savings tag attached: a so-called fiscal balance contribution of $1.64 billion.
Serena Wilson, then deputy secretary of the DSS with responsibility for its social security elements, was on leave for a few weeks during late January 2015. By the time she returned, “possibly 8th or 9th February”, Morrison’s interest had been piqued and a devastating series of events had been set in train.
Greggery spent some time establishing a sequence of logic that puts Wilson firmly in the sphere of knowledge about the robo-debt idea before November 12, 2014, when a branch manager, Murray Kimber, emailed her about a potential new budget measure that “sounded OK but as we discussed it did have a number of issues”.
“Following further investigation we have confirmed as much,” Kimber writes.
This email noted that the then Social Services minister, Kevin Andrews, who Morrison would replace by the end of December, had written to his prime minister, Tony Abbott, regarding potential 2015-16 budget measures and was going to investigate the “feasibility” of chasing debts from overpayments at greater volume.
Kimber realised, with the legal and policy advice due any day, that they hadn’t given Andrews any precise detail on how they were going to do this – partly so they didn’t have to lock in anything illegal. A “plan B” was to go after high-value debts instead.
When Scott Morrison became Social Services minister before Christmas that year, then DHS secretary Kathryn Campbell broke her annual leave and flew to Sydney on December 30 to brief him.
He asked how welfare compliance worked and whether there were any proposals to “strengthen” this and Campbell went away to debrief her executives on January 5 the following year. She told the inquiry on Thursday she could not recall whether deputy secretary Malisa Golightly mentioned the embryonic robo-debt at this meeting.
But the idea was still bouncing around.
Back at DSS, on January 16, 2015, Wilson was sent a detailed summary of both the policy and the legal advice regarding the DHS proposal, which again made it explicit that the concept was a non-starter. Wilson responded on January 19: “Thanks, I was concerned when Malusa [Malisa Golightly] described it to me. I will go back to Malusa and let Finn know.”
Three days later, however, both DSS secretary Finn Pratt and Wilson met with Scott Morrison. Wilson went on leave the next day.
“We have been provided with information which indicates that you and Mr Pratt met minister Morrison on 22 January, 2015,” Greggery said. “You had your clear policy answer to the proposal that was being discussed and you had your clear legal answer.
“The uncertainty about [robo-debt] was allowed to develop under your watch, wasn’t it? Because instead of saying ‘No’ to the proposal going to the minister, you engaged with DHS about how it might be framed.”
Wilson said they were “trying to be collegial”. When pressed, she added: “I thought we had killed it.”
Greggery asked if she said that in the meeting with Morrison – she did not – before elaborating on his own point.
“The clearest way to kill it,” Greggery said, “was to commit the advice, both legal and policy, to writing and send it to your own minister.”
Wilson said she wished now that she had.
“In hindsight,” Greggery said, “there was nothing stopping you, was there?”
She conceded there wasn’t.
Wilson retired from the public service in mid-2017 and, on the way out the door, destroyed a series of “personal” notebooks and diaries she says contained nothing of business importance, just “personal jottings that were mainly lists”. Later, under questioning about whether she told her boss, Finn Pratt, about the early robo-debt concepts, she said she would usually meet with him in person and “I would use my personal notebook, I guess, as a way of making sure that I had raised something and ticking off that I... that I had”.
Wilson cannot recall whether she raised robo-debt with Pratt. He gave evidence on Thursday morning that he didn’t recall being briefed by her on the legalities of the scheme at any later date.
“I am sure Ms Wilson would have talked to me about the issues in relation to the scheme,” he said. “I don’t recall any discussion about legalities or issues of that sort.”
When Wilson returned from leave in February 2015, she realised the robo-debt proposal was not dead but very much alive and had a meeting in her office with DHS’s Golightly. Again, there is no record of this meeting and another official from Human Services who was apparently there has never been identified. Greggery says he believes the meeting did happen but questions why it wasn’t in Wilson’s first statement to the royal commission.
“I’m not suggesting it’s a false memory,” he said. “But I am suggesting you always had it and you just didn’t put it in your first statement.”
Wilson denied this.
The significance of this meeting, she says, is that the “bottom-line” assurance she sought from Golightly at its end was that the Department of Human Services would not use income averaging as a means of calculating a debt and that there would be “no change” to the underlying approach that had been used for years. Wilson says she was given this assurance but also noted a curious exchange.
As Greggery explained: “According to your recollection, Ms Golightly tells you that Minister Morrison expressed his interest in the PAYG proposal but communicated to her that it was difficult to pass legislation through the senate at that time.”
Wilson said Golightly had “reflected on the difficulty of legislative passage”. And, and this is key, she said: “… as a consequence, I was told that Minister Morrison wished to proceed in a way that did not require legislative change”.
What didn’t make sense to counsel assisting, nor Commissioner Holmes, was the projected savings of $1.7 billion – a figure Wilson had seen – and how they might possibly be achieved without the key components of robo-debt that she had apparently ruled out.
According to Wilson, it was about “scaling up in a way that selected more records quickly, that used risk-profiling to identify where there was a greater risk of the possibility of an overpayment and providing a more digital or e-service delivery system that supported the process of correcting information”.
Greggery asked why welfare recipients would need to correct information if an appropriate debt calculation method was being considered, as Wilson claims she was told it would be.
“To correct… I mean… recipients being presented with data… and I… which had been the previous process,” she said, struggling to find words.
The previous process, Greggery reminded her, was not about raising the debt but about checking whether assumptions were correct by deploying compliance officers to ring employers or find other sources of information.
“What I undertook to do was, having pressed back that income smoothing or income averaging could not be part of a design and that there will always be sufficient opportunities for the record to be corrected,” she said before being interrupted again.
What record? Greggery asked.
Wilson paused for long periods during her subsequent answer.
“The PAYG data to be explained,” she said.
“Between the customer and Human Services, potentially with the assistance of other information that might be available at that time,” she said.
This suggested, Greggery said, “that the PAYG data has been applied to income fortnights, which requires explanation”.
Here Greggery was methodically building a case that he suggests is like so: Serena Wilson knew or should have known that even after apparent “assurances” were given that income averaging was not being used, it was, in fact, a key feature of the proposal that was now being developed into a budget measure. Similarly, she should have known this in late 2015 when her branch manager, Murray Kimber, asked whether they should “be raising any concerns for the minister about the MYEFO [mid-year economic and fiscal outlook] proposals”.
“Irrespective of your belief, you have a relatively senior officer in your department asking you for guidance about something they are concerned about,” Greggery said.
“And your guidance as deputy secretary was to say, ‘We are not raising it.’ ” Is “that because you understood from your earlier conversation with Ms Golightly that it would be a pointless exercise because it was already under way – legislation couldn’t be changed but the departments had committed to the government that it would happen.”
No, Wilson replied. “It was because I genuinely believed that the measure was not going to proceed on the basis of income averaging.”
Kathryn Campbell confirmed on Thursday she worked with Golightly on the brief to Morrison. It promised to continue work with DSS on proposed legislative changes. So what did she remember of this in the lead-up to the budget three months later?
Her response was characteristic of others she would give on the matter.
“At that time? I don’t recall any further details,” she said.
While all of this was unfolding in DSS, public servants in DHS knew exactly what was happening. The “pilot” stage of robo-debt from 2015 tested the approach on 1000 welfare recipients but, crucially, with staff oversight. A June report on this trial noted there are “58 per cent of customers who have not contacted within the prescribed period of time, 21 days” and that “the matched data has/will be applied to the customer’s record”.
Not only was this noted, the inquiry heard, but it was likely intentional.
Scott Britton, who is now with the National Disability Insurance Agency, and his direct report, Jason Ryman, both spoke on the stand about the use of “behavioural insights” or “influencing” of customers by choosing very specific language and systems to shunt people into “verifying their information”.
By the time the online compliance intervention – essentially the first version of robo-debt following the manual pilot – began in July 2016, DHS didn’t add any design nudges to help people complete their details and clarify potential debts. Instead, officials subtracted information and used framing that maximised the number of people who never got in contact.
That was a clean sweep for DHS; it meant they would simply raise the debt they had guessed without any other information, no checking and no pushback from a welfare recipient. It has emerged throughout this inquiry that the data used to do this was likely not being used legally, at least for the first phase of the program.
Last Friday, Greggery asked Australian Taxation Office deputy commissioner Jeremy Hirschhorn about the status of the data-sharing.
“How does the ATO satisfy itself, in the context of a bulk data release which would otherwise involve the commission of offences on a large scale?” he asked.
Hirschhorn said the ATO had trust in the DHS and believed they were treating the data accordingly. “Certainly,” he said, “if we were provided information that it was not being lawfully used, that would be something that we would consider very deeply, because we would be very worried that we would be breaching the Tax Administration Act.”
Later briefings within the ATO revealed alarm that senior executives were “exposed” to the robo-debt scheme and on July 7, 2017, Tyson Fawcett from the tax office wrote to DHS representatives with alarm bells ringing.
“I’m seeking an urgent discussion (next week at the latest) on this. Otherwise, I ask that you cease and desist the usage of the data until we have your assurance around the data use.”
It is apparent, in evidence, that senior people within the Department of Human Services had no intention of changing the way the department used the data because to do so would jeopardise a significant budget measure.
On November 18, 2016, months after robo-debt officially began, a departmental liaison officer from DHS embedded in the office of the then minister for Human Services, Alan Tudge, wrote to senior staff with a request from Tudge’s senior adviser Mark Woods “asking for an update budget measures/compliance targets for Business Integrity”. Woods was asking when the department would achieve its $500 million target, the email said.
Asked about this, Britton said it was linked to “compliance measures”.
“I’m not sure where the 500 came from, whether it was an amount set by the minister’s office as a target or whether it’s part of a forecast,” he said.
“But I do remember the pressure at the time around the achievement of savings, and then what this meant was more regular reports to government around the achievement of those savings.”
The email alarmed Malisa Golightly, who emailed Mark Withnell and Scott Britton asking for a report, noting she believed there was no target but a “milestone”. Days earlier, she had written on a report for the online compliance intervention in her own hand notes for Britton and his boss, Karen Harfield.
“I know that Oct was the first few weeks of our increased numbers of interventions, but these savings figures (on the surface of it) appear very low,” she wrote. “Are we happy that the savings are commensurate with interventions?”
In January 2017, when the ombudsman was investigating both DHS and DSS regarding the “authority” by which income averaging was being used in calculating debts, Wilson is alleged to have commissioned Anne Pulford to provide new legal advice to “justify” the scheme to the ombudsman. Wilson denies this, but concedes she briefed Finn Pratt about an apparent “incongruity” between the 2014 advice and the 2017 advice, so that he was not “blindsided”.
“You had a choice,” Greggery said this week. “You could act consistently with the code of conduct which applies to you as an Australian public servant and give full and frank advice, or you could go down the road of doing things, taking steps which ensured the continuation of what you then believed to be unlawful.”
Wilson said she lacked the “courage” to speak up and was asked to elaborate on what she meant. It was Kathryn Campbell’s arrival as secretary of DSS after Pratt’s retirement in 2017 that changed things.
“It was a less warm and... mutually respectful relationship. It was... a different cultural experience,” she said, “a more authoritarian environment.”
It was fear, too, that was baked into the assumptions about robo-debt: that people would be afraid of the government and pay up.
They had every reason to be afraid. The brief that went to Morrison’s office from Campbell’s DHS in early 2015 was emphatic that a welfare crackdown “will send an important message around the country that ‘you’re next’ ”.
As the debt trap chugged to life in late 2016 – an algorithm and inattention that unleashed untold trauma on people who had never done anything wrong – Scott Britton nominated the project lead, Jason Ryman, for a public service pin known as an Australia Day Achievement Award. The entire Department of Human Services nominated itself for an Institute of Public Administration Award for its “modernising compliance intervention” program. Neither Ryman nor the department won.
The hearings continue.
This article was first published in the print edition of The Saturday Paper on November 12, 2022 as "Up to their ears in robo-debt".
For almost a decade, The Saturday Paper has published Australia’s leading writers and thinkers. We have pursued stories that are ignored elsewhere, covering them with sensitivity and depth. We have done this on refugee policy, on government integrity, on robo-debt, on aged care, on climate change, on the pandemic.
All our journalism is fiercely independent. It relies on the support of readers. By subscribing to The Saturday Paper, you are ensuring that we can continue to produce essential, issue-defining coverage, to dig out stories that take time, to doggedly hold to account politicians and the political class.
There are very few titles that have the freedom and the space to produce journalism like this. In a country with a concentration of media ownership unlike anything else in the world, it is vitally important. Your subscription helps make it possible.
Select your digital subscription