Social Security and Robo-Debt

Raising debts without legal authority is wrong – and illegal. Centrelink’s Online Compliance Intervention System, colloquially known as robo-debt, raises social security debts without adequate legal proof or authority.

There has been a lot of media attention, as well as the Ombudsman report, Parliamentary inquiry and many Tribunal decisions, about robo-debt since it was introduced in 2016. The government denies this, but fails to put any counter argument. Administrative Appeals Tribunal (AAT) decisions have found such lack of legal authority. The government has not appealed these decisions, but continues the Centrelink robo-debt system for people of working age (while apparently delaying its extension to pensioners until perhaps as late as 2021).

What is robo-debt?

Robo-debt is the colloquial name for the Online Compliance Intervention System introduced by Centrelink in mid-2016. This system matches Australian Taxation Office data about reported earnings of social security recipients over the whole employment period, against fortnightly earnings declared by the recipient each and every fortnight during which she or he received Newstart (unemployment) or Youth allowance.

Under social security law, payments such as Newstart or Youth allowance must be reassessed in each and every fortnight as income changes. There is an earnings bank ‘offset’ if income in a previous fortnight fell below or within a disregard range, designed to encourage casual or part-time work. This adjusts for any overpayment, or underpayment, in each fortnight.

Robo-debt uses a different method to determine if there is an overpayment of Newstart or Youth Allowance. It calculates and applies a fortnightly average of gross earnings in the whole employment period as the basis for raising a debt because of overpayment. The data used goes as far back as 2010. However, the average calculated does not tell anything about each of the fortnights which comprise it. As a result, the so-called ‘debts’ raised in this fashion are nearly always entirely false (in fact, there is zero debt) or greatly inflated in size. To rely on the average where casual employment is the norm is as risible as saying that the same amount of rain falls every fortnight based on yearly averages.

The Commonwealth Ombudsman’s report confirmed that this is the consequence of the robo-debt calculation method. It found that in the sample assessed, the alleged ‘debts’ dropped from original amounts of a few thousands of dollars to nothing or to a few hundred dollars once fortnightly pay was verified.

Prior to robo-debt, Centrelink obtained payslips from clients who had kept them, or used its statutory powers to obtain fortnightly pay records from employers. It did this in seven percent of all data matches—unlike robo-debt, which seeks to capture all matches. That original practice was consistent with the obligation of Centrelink to correctly find and prove any debt.

Why must government prove a social security debt?

Robo-debt lacks legal authority. It is Centrelink’s legal obligation to prove up debts for three reasons.

1. Social security legislation states that there is a debt ‘if and only if’ a provision makes it such.
2. The Full Federal Court in McDonald (1984) 1 FCR 354) explained that Centrelink bears the practical onus of proving a debt; and
3. An allegation of ‘debt’ raises the bar in terms of that proof, under the long-standing High Court ‘Briginshaw principle’ which dates back to 1938.

The robo-debt process fails to discharge this obligation of Centrelink to prove the debt. The method of calculating an average can only ever be correct if the person concerned had only one job and their rate of pay never varied over time. This is extremely rare. In many robo-debt cases, the recipient of Newstart or Youth Allowance may have up to a dozen different jobs, all of which were casual or episodic. That fluctuation and diversity of jobs and earnings is why debts disappear or are drastically reduced on proper investigation and calculation.

Shifting the onus: You must disprove your debt

Robo-debt shifts the onus to alleged debtors to ‘disprove’ what Centrelink itself is legally obliged to establish. Until late 2016, Centrelink’s own website only advised keeping payslips for six months. Few citizens will have kept records from seven or eight years ago, yet the Online Compliance Intervention puts the onus on debtors to load such information into the Centrelink system, or obtain it afresh from employers.

This makes sense if recipients have kept payslips or can readily obtain them. Yet many employers are likely to have ceased business after a period of years, or may be uncooperative. Casual work is notorious for under-payment and other scams. Centrelink rarely provides assistance to obtain the correct information, meanwhile it maintains that these debts exist before the Administrative Appeals Tribunal.

What does government say about robo-debt?

Centrelink states that nothing about the Online Compliance Intervention scheme is ‘inconsistent with the legislation’ (quoted in Karp & Knaus 2018). While it is true that there is no provision outlawing robo-debts, as explained in this article, there is a provision stating that only legislated debts may be raised. It is the common law that brings a robo-debt undone due to lack of any relevant proof of it.

More recently, the government has stated that robo-debt ‘did not automate debt recovery, nor change how income was assessed or how debts were calculated’.

It is true that there remains (or should remain) a human act by a person in Centrelink, of endorsing the result of the data match and any new information provided by the alleged debtor. However, in practice it is the averaging that applies in the vast majority of cases.

And while robo-debt does not ‘change’ the law about how income is assessed and debts calculated— no amendments having been made to the legislation —it is simply not accurate to imply that the current law is being correctly applied, when an average is substituted for the legally required precise and variable specific fortnightly amounts.

Is the government legally correct about robo-debt?

The government claim to have a legal foundation for the robo-debt approach has never been tested. Neither the Ombudsman inquiry nor the Parliamentary inquiry into robo-debt properly addressed the ‘law question’. Nor has Centrelink ever taken the many invalidations of robo-debts in the first level of AAT review (AAT1) for testing at the second level of AAT review (AAT2).

The only case decided at the second review (AAT2) level to date was taken by an applicant after Centrelink accepted the AAT1 ruling that it must obtain actual fortnightly figures, which the applicant still thought too high (Re: Gosse 2018). An AAT1 pre-hearing direction in early 2017, which sought written argument on the precise elements of the argument for lawfulness of robo-debt also failed to elicit any such basis for sustaining the government claim. Since AAT1 decisions remain unpublished, the lack of any proper legal testing of the government claim for a legal basis for its scheme remains most worrying.

Social security clients are among the most vulnerable members of the community. That is why we have a system of social security. But the rate calculations for payments such as Newstart are very complex. Faced with a ‘claim’ of a debt, based on the wrong information, most recipients of payments especially from past years may simply assume that Centrelink must be correct (‘it was so long ago’). They will be frightened and embarrassed into paying money they do not owe.

A better way forward

How has this situation come about? Was robo-debt simply a rushed savings measure, given MYEFO estimates it will yield $2.1 billion over the forward estimates? Ombudsman inquiry data finding a high proportion of sample debts to be totally false, and others grossly inflated, suggests that a large proportion of the MYEFO estimate of $2.1 billion of ‘savings’ is actually an unlawful, and immoral (or even ‘extortionate’), ‘tax’ on the vulnerable.

This is a shamefully regressive outcome. It also tarnishes the image of new approaches such as ‘machine learning’, which otherwise hold great promise of sounder, fairer and more efficient benefit administration. That could be achieved if such technological innovations were rolled out in a sensible and fair way, as the government’s own Administrative Review Council proposed in 2004.

 

Further reading

This article is based on Carney, T 2018, ‘The new digital future for welfare: Debts without legal proofs or moral authority’, UNSW Law Journal Forum, March, pp. 1-16.