Ombudsman report shows Centrelink's data-matching tool was mostly a waste of time

A report from the Commonwealth Ombudsman has made a number of recommendations and a handful of comments on the handling of Centrelink's online compliance intervention (OCI) program -- which became colloquially known as robo-debt -- by Services Australia.

After a pilot of 2,600 individuals in 2015, the OCI program went live in 2016.

The data-matching elements were meant to automate a lot of the manual work performed by staff.

The system automatically compared the income declared to the ATO against income declared to Centrelink, which resulted in debt notices, along with a 10% recovery fee, being issued whenever a disparity in government data was detected.

The system, however, didn't account for how the ATO handled income -- as an annual figure. As such, Services Australia would "average" the ATO amount over the fortnights in the relevant period in order to compare it against the individual's reported income.

The federal government paused the automated data-matching element of the initiative in November 2019 and in May 2020 confirmed "it was legally insufficient to use ATO income averaging either fully or partially" to raise debts, estimating refunds to the tune of AU$721 million would be paid to individuals caught up in robo-debt.

This would see around 470,000 debts refunded. As revealed during Senate Estimates last month, that figure is actually 507,000 debts raised using income averaging.

Once the November pause was announced, Services Australia began reviewing debts manually.

When a debt was categorised as being wholly or partially raised on the basis of income averaging, Services Australia froze recovery action on that debt. Recovery action continued on all income compliance debts, including those ultimately determined to be raised wholly or partially using income averaging, until they were categorised.

People could also make contributions towards a debt during this time, too.

"We acknowledge that not all debts raised under the IC Program relied wholly or partially on income averaging," the Ombudsman said in its report [PDF]. "However, in the absence of knowing which were and were not, we are concerned that Services Australia nevertheless continued to recover income averaged debts despite a high risk that many of those debts were 'legally insufficient'.

"Despite this, there was a period of some months during which debt recovery action was continuing in relation to many debts which Services Australia knew had a high likelihood of being raised on 'legally insufficient' grounds. We consider this created a high degree of risk for Services Australia and unnecessary hardship for individuals affected."

The Ombudsman said the better approach would have been to pause debt recovery action on all debts raised under the income compliance program.

"During this investigation, our Office raised concerns with Services Australia about its decision to continue recovery action on income compliance debts while it undertook its identification process, rather than freezing all income compliance debts before undertaking the process," the report says.

"Services Australia advised it was not technically feasible to do this, as the design of its debt management system did not enable it to easily freeze only those debts raised under the IC Program.

"In particular, it advised it was unable to isolate an individual's income compliance debt from other debts they may owe."

Making a total of nine recommendations and seven comments, the Ombudsman said Services Australia should have recorded appropriate indicators on individuals' records to capture the method of calculating debts, including whether it relied wholly or partially on income averaging or whether other "proof points" were relied on, and if so, which ones. This also includes the flagging of the debt being raised using income averaging or not.

"In our view, if Services Australia had recorded reliable system indicators to track what information was relied on to raise the debt, including whether a debt had been raised relying wholly or partially using averaged data, this would have reduced the intensive and manual nature of this process and reduced the adverse impact upon affected individuals," the Ombudsman declared.

"This would have enabled greater understanding by Services Australia about the proportion of debts raised under the IC Program that relied wholly or partially on income averaging.

"In addition, it would have facilitated a clearer picture within the agency, and greater ability to respond to external requests for information, about the use of income averaging as part of the IC Program."

Services Australia engaged KPMG to provide external assurance on the debt identification process. The Ombudsman said the assurance report noted that the process "was reasonable given the inherent limitations of the underlying technology and customer data, and the manual approach that was adopted in the identification of the customer population with income averaging applied".

KPMG also noted in its report that Services Australia acknowledged and accepted a high residual risk that not all income compliance debts would be correctly identified for refund.

The government accepted seven of the recommendations, responding to the remaining two with "noted".

HERE'S MORE

• AU$723.1 million repaid to Centrelink customers for incorrect robo-debts
• Canberra agrees to AU$112m compensation in Centrelink robo-debt class action settlement
• Australian government says no to shuttering Centrelink compliance after robo-debt inquiry
• Centrelink's new data-matching approach to avoid robo-debt 2.0