NATSEM Modelling of Federal Budget Should not Surprise
NATSEM was initially established at the University of Canberra in 1993 to develop microsimulation models for the Federal Government and to undertake broad social and economic modelling and research. A major modelling task was to develop STINMOD - a model of the personal income taxation and government benefits system.
The Federal Government heavily relies upon this model through Treasury, Social Services and Employment to understand how policy impacts on families – both example families (cameos) and the broad impacts on different socioeconomic groups across the country.
The model is based on actual families in ABS survey data and the data is updated using the inflation, wages and population data from the ABS and government benefit and taxation numbers are aligned with government administration numbers to improve accuracy. The model can also project over the forward estimates using assumptions for these variables. These assumptions are generally based on the most recent budget assumptions and are usually not controversial since they rarely change much from one budget to the next.
The NATSEM budget modelling shows a total of $18 billion (over the forward estimates) in savings (and tax increases) when the 2015-16 Budget is combined with old measures yet to pass the senate. The bottom 20 per cent of income households account for 33 per cent of these savings while the top 20 per cent only account for 7 per cent of the total savings. As a share of income the bottom 20 per cent lose around 3 per cent while the loss for the top 20 per cent is negligible (-0.1 per cent). The largest impact is for low income single parents (-5.5 per cent) and couples with children (-3.9 per cent).
When NATSEM released the results of our budget analysis the results were met with criticism and caution by the Government. Interestingly, there was no such concern from other economic modelling outfits or academic experts. Standard responses by Government politicians included concerns over ‘assumptions’ and general trust issues of modelling. The reality is STINMOD is little more than a sophisticated calculator with only minimal use of assumptions and those few assumptions line up with budget assumptions in any case.
STINMOD applies the rules of the taxation and government benefit system as outlined in Government legislation. If a family has a given income level and the family has children of a given age STINMOD calculates the income tax liability and government benefits in the same way a calculator on the Centrelink or ATO website would. STINMOD applies these rules to each of the 44,000 families in the ABS surveys in its underlying database. By adding up the impacts on all these families STINMOD can estimate the impact across all households in Australia, socioeconomic groups and the aggregate budget impact.
The NATSEM STINMOD model does not calculate ‘second-round’ effects of policy. While such modelling is possible it is by its nature imprecise. NATSEM’s STINMOD model provides a simple ‘day-after’ impact of policy. This is the same impact the Budget standards demand.
An example of STINMOD modelling is our modelling of the new child care policy as part of the broad suite of policies in the Federal Budget. The STINMOD model calculates who would win and lose and by how much if that policy were enacted tomorrow and nobody had a chance to alter their child care or work arrangements. The new package has mostly winners (around 70 per cent or 780,000 families) while the rest are either no worse off or worse off. NATSEM expects around 270,000 families (one in four) to be worse off due to the tougher new work test and the cap on the child care subsidy price.
While NATSEM does expect the lowest income families to benefit modestly from the new child care package the vast share of the benefit will go to middle and upper middle income families – since it is these families that predominantly use formal child care.
‘Second-round’ effects modelling would attempt to estimate the change in hours worked by parents in response to the new, more generous package. This is not a simple calculation like ‘day-after’ modelling, rather it involves quite complicated econometric modelling that attempts to estimate behavioural change. Such modelling could conceivably attempt to incorporate impacts such labour supply responses, substitution between informal and formal care, price and supply responses from child care providers. The modelling may suggest some behavioural change such as an increase in workforce participation but there is no guarantee that in the current labour market that would translate into actual jobs and greater hours worked.
The Productivity Commission’s modelling did attempt to model some of these factors in their model for child care which takes on many of the features of the Government’s proposed model. They suggest the behavioural impacts are not likely to be large. Certainly not significant enough to compensate the 1.3 million low and middle income families impacted by the budget by an average of over $2,000 per year by 2017-18.
Ultimately, one doesn’t need to do any sophisticated modelling or make any assumptions to understand that $18 billion in net savings by the government (including an increased child care spend) will impact low income families more heavily than high income families. The savings are largely welfare payments that are directed mostly to low income families so it stands to reason if you cut those payments the impact will be felt by low income families. Since child care payments predominantly go to middle and high income families it is unlikely that a small increase in these payments will make much difference to the overall balance of the impact.
 The 2015-16 Budget introduced a number of new childcare programs - Inclusion Support Programme, Community Child Care Fund and Additional Child Care Subsidy that replace a number of existing programs such as Community Support Programme, Special Child Care Benefit, and Jobs Education and Training Child Care Fee Assistance. NATSEM expects the new programs to be more generous than the ones they replace but the underlying survey data and a lack of policy detail does not permit the modelling of such schemes. Given the small share of the total childcare funding envelope NATSEM does not expect the modelling of such schemes would greatly alter the results.