JobKeeper was formulated in the midst of an economic fog of war, with few concrete predictions available for what the true impact of COVID-19 would be. Scenarios from the Doherty Institute suggested that, at the low end, Australia could see millions of infections and up to 50,000 deaths from COVID-19, while GDP could fall by as much as 25 per cent.
“In this uncertain context, it was prudent to design the policy to be robust to whatever circumstances unfolded, to be demand-driven, but to cost the JobKeeper policy under the assumption that very significant constraint measures would be required more akin to a lockdown,” Treasury secretary Dr Steven Kennedy told the Senate COVID-19 committee.
But despite the economic impact being substantially smaller than initially expected, JobKeeper uptake still appeared to be trending towards the worst-case scenarios.
“Treasury considered that this could reflect variations in distributional impacts of the shock across firms, a higher proportion of firms than we had expected being able to demonstrate their eligibility, or the shock being more significant than other real-time indicators were implying,” Dr Kennedy said.
Treasury engaged with the ATO to try and resolve the numbers, with discrepancies discovered as early as two weeks into the scheme. It was later discovered that a number of businesses had incorrectly filled out the application form, leading to massive overestimation of how much the scheme would cost.
“Once employers began to confirm the actual employees covered as part of stage two, we identified within a fortnight that the numbers were less than expected,” ATO commissioner Chris Jordan told the committee. “By May 20 when numbers continued to stay below estimates, we did a deep dive into the 900,000 applicants on a [line-by-line] basis.
“It became evident that a number of mistakes were made in the stage one box that asked for the number of employees.”
Treasury was informed and analysed the impact on the afternoon of 21 May. Dr Kennedy took full responsibility for the bungle, but noted the mistake only occurred in 0.1 per cent of applications.
