Run-Away Robo-debt Train Derailed: Pity About Inadequate Damage Clean-Up?

adminSocial security rights review

Terry Carney

On 19 November 2019, the morning after conceding its inability to defend the Federal Court test case challenge to robo-debt in Amato,[1] radical changes were announced by internal Centrelink email.[2]

As previously explained,[3] robo-debt was an illegal, immoral and ill-constructed scheme for recovering mainly non-existent or seriously inflated debt amounts,[4] as now confirmed by the Federal Court orders.  Those affected by robo-debt were often vulnerable and overwhelmed by the process.[5]  A revenue-raising measure to balance the Budget,[6] robo-debt generated around 900,000 debts over its life.  Centrelink itself estimates that around 600,000 debts may be based solely on ATO averaging, with at least 220,000 of those entitled to refunds.[7]

More than two years after it was publicly called out for failing all three of mathematics,[8] law,[9] and IT design[10] — the 19 November changes to robo-debt were the first substantive changes to be made.  Previous adjustments were minor cosmetic ones (such as slight improvements of correspondence or identification and handling of some vulnerable clients).

The November 2019 changes were that:

  1. Debts would no longer be raised solely on the basis of an ATO average but would require additional proof;
  2. Previous OCI identified debts will be ‘methodically’ reviewed, with a focus on debts where the person did not reply to the original OCI correspondence;
  3. These will be reviewed to assess whether ‘further information is available to clarify the amount of any debt owed’; and
  4. ‘Affected’ persons will have recovery ‘frozen’ but otherwise there is no change to collection arrangements.

So what does this mean for people flagged as potential debtors under future ATO data matches, and what is the position for the speculated 220,000 or more people where a debt has previously been raised (if not yet always recouped)?

No future robo-debts solely based on tax office data-matches.

It is clear that Centrelink now accepts that it cannot reverse the onus of proof by requiring alleged debtors to locate fortnightly payslips or bank records to contradict the ATO averaging assumption.  But how much more proof is Centrelink proposing to collect?  At the time of writing the answer is worryingly unclear.

The Minister simply spoke of obtaining ‘additional proof points’, as too does point #1 in the internal email above, and as also was implied by the ‘script’ provided to Centrelink call centre staff immediately after the backdown.[11]  This falls well short of what the law requires.  To discharge its onus of proof, Centrelink must obtain adequate proof of  actual income for each and every fortnight in the debt period.  If the alleged debtor does not volunteer payslips or bank records when asked (or fails to reply to a request to do so) then Centrelink must obtain such documents by using its statutory powers to demand them of employers or banks.  Otherwise there is no debt in those fortnights.  It is unclear whether this is yet fully understood by Centrelink.

This may yet prove to be a short to medium term worry for welfare advocates.  This is because Government may yet concentrate its efforts on bringing on-line the new scheme offering almost ‘real time’ matching of fortnightly income data reported to Centrelink against equivalent information from the ATO’s ‘one-touch’ payroll system.  Such virtually contemporaneous ‘matching’ not only would provide a realistic ability promptly to explain discrepancies, but also prevent large debts from accumulating.[12]

Either way, future robo-debts thankfully now seem unlikely.

Past victims of robo-debt

The most worrying questions concern the very large numbers of past victims of robo-debt.  ‘Victims’ is advisedly applied to describe such people here, because so many were understandably ignorant of whether they truly had a debt (not understanding how large can be the difference between the correct fortnight by fortnight calculation and one done on the basis of an average) or they were intimidated into or lacked the confidence to contest Centrelink’s assertion of a debt.[13]

Past victims of the unlawful and inaccurate robo-debt methodology fall into a number of different categories.  They differ in the process for raising the debt (did they engage or not, for how long, and in what ways), and in whether the debt remains unrecovered, partially recovered or fully collected (such as through garnishee of a tax refund to offset it).  Irrespective, serious doubts must be held about how extensive and how adequate is the redress to be made available for those past wrongs, as now explained.

1. Debts raised or recovered against those who did not respond initially.

Centrelink’s decision to commence its reviews of past debts with those who did not reply was sensible.  They are among the most vulnerable groups impacted:  those least likely to know how the system works, or to be intimidated by the might of government.  They apparently are to receive a letter.  But what will the letter say?  If past robo-debt contact letters are any guide, it probably will not even clearly identify why there may not be any debt at all, or at best an inflated one.

  • What is the response going to be if there again is no reply from the person? At law it is clear that the whole debt amount remains unproven and uncollectable.  Is that what Centrelink understands?
  • What is the response going to be if the person replies but is able to supply no records at all, or only a few of them? Again at law the so-called debt (or the unproven parts of it) remains no debt at all until Centrelink uses its compulsory powers to obtain the information (something it did on less than 1,000 instances out of some 600,000 robo-debts originally[14]).

2. Debts raised or recovered from people who engaged but could locate no or only some records.

Prior to the November 2019 backdown, the principal justification of robo-debt went along the lines that it was perfectly reasonable to ‘ask’ people to explain why ATO data match averaged income suggested a certain size of overpayment debt, and to put the onus on that person to disprove (or ‘clarify’) the position by producing their payslips or bank records from up to seven years previously.  To the objection that there had been no advice to keep these documents for more than the six months stipulated on the Centrelink website or that employers might no longer be in business or prepared to produce such records, it was said that Centrelink would ‘assist’ in obtaining them.

It is a matter of record that fewer than 1,000 alleged debtors ever received any such assistance.  Some simply gave up altogether, but an unknown number of others had debts adjusted on the basis of a few, usually their most ‘recent’ such records.  The questions that still remain to be answered are:

  • Will everyone with a past robo-debt receive a ‘methodical’ review of their debt to identify the fortnights where proof of debt is lacking? If not, on what basis can it be said that a past debt is being maintained on any basis other than the now abandoned ‘solely on an ATO average’?  If no future debt is to be raised solely on an ATO average (and certainly it is unlawful to do so) past debts cannot be treated any differently.  They too must all be validated by ‘proofs’ for each and every fortnight of the overpayment being maintained, irrespective of how the person has responded to the debt (an unlawful debt does not magically grow legality because ‘unprotested’).
  • Will partially proven debts already partly or fully recovered (whether by repayments or garnishee powers) be treated in the same way as debts that remain outstanding? Under the law (as now apparently also accepted by Centrelink) there can be no such difference in treatment.  A lack of proof of a debt is a lack of proof of a debt, whether it be still outstanding or one paid involuntarily or under a misapprehension that it was owed.

These are the burning questions the answers to which will determine how well or otherwise the Government cleans up the justice mess left by the unfortunate robo-debt saga.


Robo-debt remained a legal, political and ethical disaster for over three years, during two of which it appears Government continued to rely on ‘internal’ advice of the scheme’s legality before belatedly obtaining an independent opinion which agreed with critics that it was not lawful to raise a debt on the basis of an ATO average.[15]  It is beyond welcome that this has finally been accepted.  However the devil is in the detail of the belated reforms, which still may not fully align with the legal obligation to obtain proven fortnightly income information for each and every fortnight of any debt raised in the future, and may yet short-change those with pending or repaid debts falsely raised in the past.

This is why the legal actions which precipitated the Government’s belated change of heart have been so important.  The class action formally launched 24 hours after the Government backdown,[16] remains on foot.  The finding of illegality of the robo-debt at the heart of the Amato case settlement is both unequivocal and clear and ought to ensure that past debtors receive full justice.  If not, success in the class action is surely capable of ensuring that robo-debt is rendered fully compatible with the rule of law.[17]  For there remains a serious risk that the so-called ‘reformed’ scheme is another ‘smoke and mirrors’ play.  Past stratagems for avoiding legal accountability have included waiving debts once test cases were launched, and not ever appealing to the publicly visible second tier of the AAT (its General Division) the several hundred rulings of first tier AAT hearings invalidating robo-debts on the basis of lack of legal foundations (hearings for which are held ‘behind closed doors’ and for which no reasons are made public).[18]

That is why welfare rights advocates are right to remain on guard.  It is why people affected should promptly lodge Authorised Review Officer and AAT review requests for any future or past victims of robo-debt’s averaging algorithm where inadequate redress is provided.  And it is also why justice for all may yet ultimately hinge on obtaining further judicial orders.

[1]              Deanna Amato v The Commonwealth of Australia VID/2019/611-0  The initial test case was effectively spiked by waiver of all of the debt in question: Madeleine Masterton v Secretary, Department of Human Services of the Commonwealth VID73/2019.

[2]              Terry Carney, ‘Robodebt failed its day in court, what now?’ on The Conversation (28 November)<>

[3]              Terry Carney, ‘Robo-debt: Challenges and Opportunities for Administration and Accountability’ (2019) 5(1) Social Security Rights Review  <>.

[4]              For details: Terry Carney, ‘The New Digital Future for Welfare:  Debts without legal proofs or moral authority?’ (2018) (March) UNSW Law Journal Forum 1-16 <>.

[5]              National Legal Aid Secretariate, ‘Rethink Robo-debt: Building a fair and accurate system people can trust’ (Canberra: Submission 26 to the Senate Community Affairs References Committee Inquiry into Centrelink’s Compliance System 27 September 2019) <>.

[6]              Laura Tingle, ‘The robodebt scheme was a political disaster — but that’s not necessarily why the Government is ditching it’, ABC News Saturday 23 November 2019 <>.

[7]              Paul Karp, ”Pay the money back’: robodebt, the Coalition’s backflip and how it ‘hounded’ welfare recipients’, Guardian Australia Saturday 23 November 2019 <>.

[8]              In that a half-yearly or other ATO average of earnings cannot supply any relevant evidence of the value of the constituent elements—here earnings for each and every fortnight: Terry  Carney, Submission 22, Senate Legal and Constitutional Affairs References Committee reference on ‘The impact of changes to service delivery models on the administration and running of Government programs’

[9]              In that it unlawfully required clients to disprove ‘debts’ raised on basis of an average, when the law places that onus on Centrelink, and calls for substantial proofs: Carney (n 8).

[10]            In that it was neither designed nor subsequently adjusted in conformity with the ‘smart design’ consultation with affected parties or the guidelines set by the Administrative Review Council: Carney (n 8).

[11]            Paul Karp, ‘Robodebt class action to go ahead despite overhaul of Centrelink debt recovery’, Guardian Australia Wednesday 20 November 2019 <>.

[12]            See the early iteration of this:  Budget 2019-20, Part 2, Table 2 Expense Measures, p 158.

[13]            National Legal Aid Secretariate, ‘Rethink Robo-debt: Building a fair and accurate system people can trust’ (n 5); generally: Hansard, Senate Community References Committee 9 October 2019;fileType=application/pdf#search=%22community%20affairs%20references%20committee%22.

[14]            Hansard, Senate Committees Community Affairs References Committee hearings 9 October 2019,;query=Id%3A%22committees%2Fcommsen%2F9d96e564-4f94-4261-9e90-989189421105%2F0000%22.

[15]            Karp (n 7), quoting the Attorney-General.

[16]            Katherine Prygodicz & Ors v Commonwealth of Australia VID1252/2019 at  Terry Carney, ‘Robo-debt class action could deliver justice for tens of thousands of Australians instead of mere hundreds’ on The Conversation (18 September)<>

[17]            Terry Carney, ‘Bringing Robo-debts Before the Law: Why it’s time to right a legal wrong’ (2019) 58(August) Law Society Journal of NSW 68.

[18]            Terry Carney, ‘Robo-debt Illegality: The seven veils of failed guarantees of the rule of law?’ (2019) 44(1) Alternative Law Journal 4.