Robodebt, Refunds and Redress – The Questions Left Unanswered

adminSocial security rights review

Written by Anastasia Radievska with Terry Carney, Charley Brumby-Rendell and Miles Browne, based on their presentation ‘What next for Robodebts’ delivered on October 27, 2020 in partnership with EJA and Community Legal Centres Queensland

The November 2019 Amato decision concerning the illegality of robodebt continues to have flow-on impacts for recipients and for government policy. Despite the refund process initiated in June 2020 and the $1.2 billion class action settlement in November 2020, questions remain about redress for victims of robodebt and lessons learnt to avoid a repeat of the program.

A total of $721 million in refunds was announced in May 2020, with 373,000 people receiving refunds averaging $1900.  The rollout of the refunds has been characterized by a string of challenges, including unclear timelines and administrative difficulties.  Centrelink did not have the contact details of almost 200,000 people impacted by the program, and the onus was placed on those who did not receive refunds to contact Centrelink and provide their bank details.  Despite these difficulties, by October 2020 $697.1 million had been paid back to affected recipients and a further $398 million in debts were to be dropped.  The class action settled in November 2020 meant in addition  $112 million will be paid in compensation, although around $16 million of this has been requested for legal costs.  This compensation is expected to be paid out this year, although the timelines are likely to be determined as part of the settlement.

Despite the clear decision in Amato that sole reliance on income averaging was an unlawful way to raise debts, the status of other robodebt elements remains unclear.  A key concern surrounding robobebt was the application of a 10% penalty fee under section 1228B(1)(c) of the Social Security Act where the debt arose wholly or partially because the person refused or failed to provide information in relation to the person’s income from personal exertion, or knowingly or recklessly provided false or misleading information in relation to this income.  The conditions for determining that a person has failed to do so has not been clarified by the Commonwealth.  In Amato, the Commonwealth never articulated a rationale for the 10% penalty fee and conceded that in Deanna Amato’s case there was not sufficient material before the decision-maker capable of supporting the conclusion that s1228B(1)(c) was satisfied.  It was announced by Centrelink in May 2020 that any penalty fees (and interested) added to a robodebt would also be subject to a refund.

The implementation of the Single-Touch Payroll (STP) from December 2020 onwards addresses some of the system inadequacies that contributed to robodebt.  Due to changes in the way income is assessed, recipients report the employment income paid to them in a reporting period, rather than income earned, derived or received in that period.  Through STP, employers will report payroll information to the ATO on a regular basis. Centrelink recipients will then be given income data from the ATO and asked to confirm whether this data is accurate.  This will avoid some of the difficulties recipients encountered trying to work out their income before tax..

However, the changes to the law implemented as part of the move to STP include inputs that leave room for uncertainty and subjective decision-making by Centrelink when earnings are erratic or do not relate to an easily identifiable employment period such as weekly or monthly pay cycles.  When an amount of employment income is paid “in respect of a particular period” (an “employment period”), the amount will be averaged over an “assessment period”.   The assessment period will have the same number of days as the employment period, but it will begin on the first day of the (two-week) instalment period during which the amount is paid.  It is not entirely clear what is meant by the “particular period to which the employment income relates”.  It could be the pay cycle period, the number of days of work or the period during which the person worked.  For example, if a person works only on Tuesday and Thursday in the first week of a month and is paid monthly, the person may be required to report the period as a month, two days or three days.  Each different approach will have a different effect.  Past authority might suggest that the correct answer hinges on locating the period(s) when the money was ‘earned’ but the new system is designed to focus on ‘receipt’; clarity is needed.

It is possible that Centrelink may seek to put the onus on the recipient to provide information about a given period, including in situations where the recipient does not or is unable to provide this information. Further, where no period is able to be determined, there is no express limitation on the Secretary’s discretion to determine the period over which the averaging power in clause 1073BA(2) will operate, although there are some factors the Secretary may take into account in exercising this discretion. These include the nature and period of the person’s work, whether they were receiving a Centrelink payment at the time, and any financial hardship caused As with the previous provision about spreading lump sums, this involves some subjectivity and thus potential unfairness if not done carefully. Several other key elements of robodebt remained unaddressed by the Amato decision and income reporting reforms.

The practice of garnishing tax returns to repay a debt under s1230C(2) was not considered in detail in Amato, which only confirmed that the garnishing in Deanna Amato’s case was illegal because she did not have a lawful debt.  The satisfaction of the requirements for tax garnishing was not separately considered by the Federal Court in Amato.  It is therefore unknown whether Centrelink will make changes to ensure that it has made contact with a person and attempted to enter into a payment arrangement with them before flagging their debt for garnishing. Further it is unclear whether any change will be made to the past practice of simply advising the Australian Tax Office (ATO) of a list of debts as determined by Centrelink, leaving garnishing to then automatically be applied by the ATO without any independent investigation of its own.

The accessibility and adequacy of merits review for debt decisions also remains a key concern for future enforcement of Centrelink debts.  The appeals process for robodebts did not provide sufficient accountability for decisions, as recipient requests for review of their debt were able inappropriately to be diverted into ‘reassessments’ which could not themselves be reviewed until the reassessment was finalised.  Requests for review by an authorised review officer (ARO) were also refused by Centrelink on the basis that there was no new information and insisted on provision of information for ‘reassessment’.  It is unknown whether reassessments will continue to form part of the debt review system.  The broader effectiveness of merits review also requires re-examination in the wake of robodebt.  The inability of ARO and AAT reviews, as well as Commonwealth Ombudsman and the Australian National Audit Office, to address the illegality of income averaging on a systematic level raises serious questions about how well these mechanisms ensure accountability.

The robodebt refund process and recent income reporting reforms confirm that the income reporting system must be better adapted to serve the needs of recipients in an accountable way.  Remaining uncertainties concerning central planks of robodebt and their continuation in the new STP process indicate that further adaptation is required to create a fair income reporting system.  Accountability for answering the difficult questions posed by robodebt is key to providing redress and avoiding a repeat of this episode in Australia’s social security history.