Automation in the Targeted Compliance Framework: when the law is changed but the system isn’t
An investigation to understand why a change in legislation, which had been developed by Department of Employment and Workplace Relations (DEWR), had not been implemented in its policies, processes and systems, and why it took DEWR a significant amount of time to discover this.
To receive income support, many job seekers are required to complete tasks called 'mutual obligations'. DEWR and Services Australia manage job seekers’ requirements to complete these tasks under a framework called the Targeted Compliance Framework (TCF).
This first report shines a light on whether DEWR and Services Australia (the agencies) cancelled job seekers’ income support in a lawful manner by using automated processes and the safeguards they established to ensure future cancellation decisions are lawful, fair and reasonable.
The recommendations provide lessons for all agencies to ensure automated decision-making systems are aligned with law and policy and are subject to ongoing testing and assurance to ensure decisions are lawful, fair and reasonable.
Key Findings
- Agencies’ cancellation of income support under s42AF(2) was not lawful.
- Agencies did not ensure their respective processes and computer systems were complying with the amended legislation.
- Agencies’ quality assurance activities did not identify the unlawful cancellations.
- There was significant delay between identifying the unlawful cancellations and pausing all cancellations.
Key recommendations
- The Secretary of DEWR not resume cancellations until satisfied that the identified errors have been rectified and that policies, processes and systems are in place that will ensure cancellations comply with the law.
- Ensure systems in place that provide ongoing assurance that the administration of the TCF complies with the law and relevant policies.
- Prepare a plan for identifying and assessing the scale and impact of legal, policy and administrative errors in the TCF, and for their timely remediation.
- Proactively and quickly rectify identified issues with automated decision-making that have the potential to have adverse impacts on people in vulnerable circumstances.
