Casebook

Robodebt

A scheme that sent about 20,000 debt notices a week, where the manual process had sent about 20,000 a year. It rested on a reading of the law the department had been told in 2014 was wrong. When a tribunal kept finding its debts unlawful, the department treated each ruling as one person's outcome, not as evidence against the scheme.

System
Automated welfare-debt raising: annual tax-office income averaged across fortnights to assert overpayments, with the burden of disproof placed on the recipient.
Where
Australia · Services Australia (Centrelink)
When
Jul 2016 to Nov 2019
Scale
About 470,000 debts raised unlawfully. The class-action settlement the Federal Court approved in 2021 covered roughly 381,000 people, with repayments, wiped debts, and interest valued at about A$1.8 billion.
Time to halt
Three years and four months
Errors carried by
Recipients, who had to produce payslips, often for years past, to disprove debts raised by averaging. The scheme had been booked as a budget saving of more than a billion dollars before it raised a debt.
Halted by
The Federal Court, on a consent order the Commonwealth agreed to hours before a hearing it would have lost.
After the failure
Process unchanged

The record

What happened

From the primary sources listed at the end. Where a finding is quoted, it is the inquiry's, not ours.

Timeline
  1. 2014 The department is advised in writing that income averaging cannot prove a debt.
  2. Jul 2016 The automated scheme launches. Debt notices go from about 20,000 a year to 20,000 a week.
  3. Apr 2017 The Ombudsman reports that notices do not explain how a debt was calculated.
  4. 2017–19 A tribunal rules individual debts unlawful dozens of times. The department appeals none of the rulings, so none becomes a precedent that binds the scheme. Each ruling fixes one case. The scheme keeps running.
  5. Nov 2019 The government concedes a Federal Court case it was about to lose. The scheme is halted that month.
  6. Jul 2023 The Royal Commission reports: “Robodebt was a crude and cruel mechanism, neither fair nor legal, and it made many people feel like criminals.”

From July 2016 the Online Compliance Intervention replaced a manual process in which a compliance officer reconciled a welfare recipient's reported fortnightly income against employer records. The automated process took the annual income the tax office held, divided it evenly across the year's fortnights, compared each fortnight to what the recipient had declared, and raised a debt for the difference. The recipient was then asked to produce payslips, often for years past, to disprove it. Income averaging cannot establish what a person earned in any given fortnight, and the Department of Social Services had received legal advice to that effect in 2014.

The Commonwealth Ombudsman reported in April 2017 that the scheme's notices did not explain how a debt had been calculated and that the reversal of the onus of proof was causing serious distress. The Administrative Appeals Tribunal found individual debts unlawful in dozens of matters from early 2017. The department did not appeal those decisions, which would have created a binding precedent, and did not treat them as evidence about the scheme. In November 2019, in Amato v Commonwealth, the Commonwealth consented to a Federal Court declaration that a debt raised by averaging was not lawfully made. The scheme was halted the same month.

The Royal Commission into the Robodebt Scheme reported on July 7, 2023. It found the scheme unlawful from the outset, that the government had been told so, and that the department's handling of the legal advice, the Ombudsman, and the tribunal decisions amounted to a sustained refusal to look at the evidence the scheme was generating about itself.

Scoring

The six safeguards

One finding per safeguard, with the clauses it falls under and the law it applies.

  1. drifted

    Capability

    What could the assembled system do?

    The automated process could raise about 20,000 debt notices a week where the manual process had raised about 20,000 a year. Nothing about the authority to raise debts was re-examined when the capacity to raise them grew by that factor, and the review capacity on the other side of the notice, the recipient's ability to challenge a debt, did not grow at all.

    STD-08 §4.2 STD-08 §4.3 Law I Law XI

  2. failed

    Authority

    Which actions was it permitted to take, for whom, until when?

    The authority to raise a debt by averaging did not exist under the Social Security Act, and the department held written advice saying so before the scheme launched. The scheme ran for three years on an authorization whose legal basis had been examined and found absent. Authority that rests on a reading of the law its holder knows to be wrong is not authority.

    STD-07 §2.1 STD-08 §2.1 STD-08 §2.5 Law II Law III

  3. failed

    Evidence

    What evidence justified that authority?

    An averaged annual figure is not evidence of fortnightly income. The scheme asserted a debt, then asked the person to supply the evidence that would have been needed to assert it. Notices did not state how the figure was computed. The Royal Commission found the scheme unlawful from the outset.

    STD-02 §1.1 STD-02 §1.2 STD-06 §2.4 Law III

  4. drifted

    Dependency

    How hard had withdrawal or substitution become?

    The scheme was booked as a budget saving of more than a billion dollars before it had raised a debt. Once the savings were counted in the budget, withdrawing the scheme had a cost in money and in politics that no one inside the department was in a position to accept. The department had come to depend on the scheme, and that is why the tribunal's findings were handled one case at a time instead of acted on.

    STD-06 §5.1 STD-06 §5.3 Law V

  5. failed

    Standing

    Who could challenge a decision, and did the challenge have to be answered?

    Recipients could object, and many did, and the tribunal found for them repeatedly. But an objection that succeeds only for the objector, and changes nothing in the scheme itself, has no force against the scheme. The department chose not to appeal, which kept each finding from becoming precedent, and did not record the pattern as a finding about the scheme.

    STD-02 §8.3 STD-02 §8.5 STD-02 §9.3 Law VII

  6. failed

    Correction

    Which interventions were still possible, and how fast?

    Switching the scheme off was technically easy, and it was halted within the month once a Federal Court case forced the question. For three years before that, nobody halted it. No one whose job it was to halt it had been given the authority to, and the people who had the authority had a reason not to. A correction that only a court can make is not a correction the operator holds.

    STD-06 §1.3 STD-06 §4.2 STD-02 §3.2 Law IV Law VI Law XII

Remediation

Avoid this failure in your system

The diagnostic that tests for this defect, the standard that would have required a fix, and the essay that explains why it recurs.

1. Run the diagnostic

Delegation audit

Asks whether the policy a system applies has an expiry date and a named event that forces a review, and whether the people who bear its errors can challenge a decision and get an answer by a deadline.

2. Apply the standard

STD-02: Contestability and recourse

Requires every notice to give, when it is sent and in plain language, the reasons for the decision and the evidence it rests on. The person can ask for a human review without supplying new evidence.

3. Read why it recurs

Challenge density

When challenges arrive faster than anyone can answer them, why is the fix more reviewers or a smaller scheme, and not fewer people allowed to challenge?

After the failure

Did the institution change?

Whether the institution changed the rule, category, or workflow that produced the errors, or corrected the errors and left that process as it was.

Process unchanged

The tribunal's repeated findings of unlawfulness were handled as individual outcomes and never changed the scheme. The department did not appeal, which kept each finding from becoming precedent, and did not record the pattern as a finding about the scheme. For three years each challenge ended in a settled case; the institution that ran the scheme never revised the rule. The scheme was halted only by a court, and the institutional change came after that, from a Royal Commission, not from the operator's own machinery.

Read the theory: exception learning · Corrective learning eval suite

The record that was not there

Every case reduces to a record the standards require and the operator did not hold. This is that record.

A written record of the income-averaging policy: where it came from, what it assumed, and what would force a review of it. The 2014 advice was enough to force one. So was the first tribunal decision against a debt. Either would have put the policy, and the scheme's authority to raise debts under it, under review. STD-08 §2.3 sets a deadline for that review.

Further reading

An essay that uses this case

Published elsewhere, as part of a wider argument. It is not one of the primary sources.

Stop the Machine, Not the Person · The Crumple Zone

Uses Robodebt as its central case. Income averaging produced debts that people's own records contradicted, objectors were asked for years of payslips, and a budget counting on $1.5 billion in recoveries kept the scheme running. Its argument: only an outside body with binding authority could have halted it.

The primary record

Courts, statutory inquiries, regulators, and parliaments, in that order of weight. Operator and NGO documents appear only where they record what the primary record established. No press coverage.

  1. Inquiry Royal Commission into the Robodebt Scheme, Report · Jul 7, 2023
  2. Regulator Commonwealth Ombudsman, Centrelink's automated debt raising and recovery system · Apr 10, 2017
  3. Court Amato v Commonwealth of Australia, Federal Court consent orders · Nov 27, 2019
  4. Court Prygodicz v Commonwealth of Australia (No 2) [2021] FCA 634, settlement approval · Jun 11, 2021