History of the Conservative Bias Allowance
In the 1987-88 Budget, the Commonwealth Government introduced the Conservative Bias Allowance (CBA) to improve the accuracy of expenses forecast over the forward estimates, by providing an allowance for potential variations in program estimates due to 'program specific' factors.41 The CBA is not intended to capture variations in economic parameters (for example, changes in the Consumer Price Index) or policy decisions (for example, a measure to change the rate of the Age Pension). Program specific parameter variations may include, for example, an increase in the expected proportion of Age Pension recipients who are single, leading to higher average costs per recipient than estimated (see Box 3 of this explainer).
The CBA percentage is based on historical program-specific parameter variations that occur as a given year progresses from being the third forward estimates year (FE3) to being the budget year. As these variations in expenses tend to be positive (upwards revisions) in the forward years, the CBA has been set at a positive rate throughout its history (Table B1 and Figure B1). The CBA amount is calculated as a percentage of total general government expenses, less GST payments to states and territories. The CBA percentage was initially set at 0.75 per cent of outlays in FE3. It was gradually increased to 4 per cent by the 2005-06 Budget, before being revised down to 2 per cent by 2009-10. The CBA percentage has remained stable from 2009-10 to the time of publishing this explainer, in 2021-22.
Table B1: History of changes to the CBA
Figure B1: CBA percentage, by budget
Source: Commonwealth budgets.
Note: FE stands for forward estimates – FE1 is the first year of the forward estimates period, FE2 the second year and FE3 the third year.
The effectiveness of the Conservative Bias Allowance
To investigate whether the CBA improves the accuracy of budget estimates, the CBA allocated in FE3 can be compared with the total reported program specific parameter variations for a given year over the period it is included in the budget, from when that year is FE3 to its final outcome.42 While these might not align for a given year, if the long-run average of program specific parameter variations is similar to the CBA percentage, then the CBA is improving the accuracy of budget estimates.43
For example, in the 2014-15 Budget, the year 2017-18 had a CBA of 2 per cent of total expenses ($8.2 billion). For the CBA to exactly account for reported variations in program estimates, the sum of all the 'program specific parameter variations' for 2017-18, through each budget update from the 2014‑15 MYEFO to the 2018-19 Budget, should equal $8.2 billion. However, the sum of these was $5.1 billion, or 1.2 per cent of the original expenses estimate for 2017-18, suggesting that the CBA slightly overestimated expenses for this year.
Figure B2 compares the relationship between the CBA reported in FE3 and the total reported program specific parameter variations between FE3 and the budget year. Over the last three decades, there has been a tendency to revise expenses upwards, except between 2005-06 and 2008‑09 and more recently from 2018-19 onwards although the recent revisions are likely due to the COVID-19 pandemic. The long‑run average of 2 per cent is the same as the level factored into the CBA, confirming that the CBA accounts for the tendency of budgets to underestimate expenses.
Figure B2: Program specific parameter variations compared to the CBA percentage
Source: Commonwealth budgets and PBO analysis.
Note: Amounts are expressed as a percentage of total government expenses minus GST payments to the states and territories. This includes all reconciliations in budget updates from 1993-94 onwards, except where reconciliations were not published such as at the 1999-00 Budget and the 2010 Pre-election fiscal outlook. The long-run average excludes variations from 2019-20 and onwards as they may have been impacted by the COVID-19 pandemic.