Crawford School of Public Policy

Michael Di Francesco presents a two part series on the costing of public policy using the case of the COVID-19 JobKeeper program. In this first piece, he sets the scene by recounting the program’s genesis, the challenges of developing cost estimates in times of uncertainty, and the hostile reception that the program’s ‘reporting errors’ received. What can the JobKeeper miscalculation tell us about the status of costing as an input to public policy?

At a time of declining trust in public institutions, one of the more unremarked aspects of the COVID-19 policy response in Australia has been the steeling of government. A common threat compelled uncommon political bipartisanship and resuscitated the legitimacy of expert advice. At the same time, conventional tools of evidence-based policy have been under great stress, raising questions about their fitness for purpose. One such tool is the costing of public policy. Is the politics of ‘precision’ in costing itself a cost to good policy-making? Can we better frame expectations about how costing informs policy in times of uncertainty? Using the contentious JobKeeper program as a case in point, this analysis argues that the ‘value add’ of policy costing should reside less in numerical precision and more in the policy design content embedded in its methods.

Time is of the essence

In late March 2020 the COVID-19 crisis reached an initial peak in Australia, demanding extraordinary public health and economic interventions. Flattening the pandemic curve required shuttering much of the non-essential economy. The impacts on unemployment were immediate and massive. Over a matter of weeks the Australian Government announced a sequence of increasingly large economic support packages. The last of these was the JobKeeper initiative, a colossal program of time-limited wage subsidies paid directly through employers who had incurred steep falls in turnover owing to lockdown. Its key aims were to avoid further separations and to retain the employer-employee relationship as a ‘bridge’ for post-COVID-19 recovery.

Oversighted by Treasury and delivered by the Australian Tax Office (ATO) through existing tax system infrastructure, the program was, at the time of its announcement, estimated to apply to 6.5 million employees and to cost $130 billion, the largest single Budget measure in Australian public financial history. While the policy was criticised – most notably for quixotic eligibility rules and harsh cash flow imposts on shocked businesses – the appropriation bills passed in a bipartisan fashion at a special parliamentary sitting in early April.

A ‘reporting error’

However, in a late May 2020 joint media release, Treasury and ATO revealed a ‘reporting error’ in the estimates of employees expected to access JobKeeper, ostensibly as a result of incorrect business responses through the ATO online application portal. As the ATO focused its program delivery on payment integrity, and because the employer estimates of eligible employees were consistent with original Treasury demand forecasts, the error was only detected after reconciling actual payments made with the estimates submitted by employers. Expected demand for the payment was adjusted down to 3.5 million people, and the cost estimate revised to $70 billion. Embarrassingly, the disclosure was made the day after the Treasury Secretary, Steven Kennedy, had given testimony to the Senate Select Committee on COVID-19 with nary a mention of possible variations to the original estimates.

Revisions to budgeted cost estimates are not unusual, but the magnitude and direction of the JobKeeper adjustments were. They were met by media incredulity and, as cracks in political bipartisanship started to appear, political ridicule. Some outlets duly reported the revision as an ‘error’, others characterised it as a ‘blunder’ or ‘stuff-up’. Leader of the Opposition Anthony Albanese channelled these sentiments, chorusing that the revision was a ‘mistake that you could see from space’. In the policy parrying that ensued, the anticipated underspend was either an opportunity to ameliorate debt commitments, or a chance to reassign fiscal windfall by widening eligibility to industries (such as the university sector) omitted from the original policy.

What happened: Not the ‘most severe scenario’

Soon after the media release, both Prime Minister Scott Morrison and Treasurer Josh Frydenberg accepted political responsibility for the estimates revision, with Morrison likening the situation to a home building contractor favourably revising a quote. The backstory to the costing process was revealed at a follow-up session of the COVID-19 Senate Select Committee, where both the Finance Minister Mathias Cormann and the Treasury Secretary appeared, with the latter issuing his own mea culpa. JobKeeper was, they said, a textbook example of a ‘demand driven’ program, where estimates of take-up must be based on the available information and assumptions about behaviour. At the time of its conception, there was no formal epidemiological modelling of how COVID-19 would evolve in Australia, and no clear sense of what an economic ‘lock-down’ would entail. The extent of any shock was unknown, and what guidance there was came principally from comparative experience in the worst hit countries of Europe. The program was, therefore, costed with the Department of Finance against the ‘most severe scenario’, with the Treasury supplying speedily updated unemployment forecasts. When the Australian lock-down proved to be narrower in scope and shorter in duration, demand was much lower than estimated. As Minister Cormann noted caustically, cost estimates ‘are not 100% accurate predictions of what will happen’.