How to improve welfare

The Australian welfare system in its current state is a complex system of payments, providing a social safety net for people who, through unemployment, disability or other factors, require financial assistance to meet basic living expenses.

Many current welfare payments were designed decades ago, with levels that may not be appropriate for the needs of modern Australian households nor represent the best value for money for Australian taxpayers.

Consider the Newstart payment for those who are unemployed and seeking work. Twenty years ago the unemployment benefit was about 90 per cent of the age pension payment. Today it has declined to 60 per cent. Driving this growing divergence is the pension’s more generous indexation arrangements and a substantial one-time increase in 2009.

The current level of the Newstart payment is about $550 per fortnight, which is about 40 per cent of the minimum wage and considered by most groups to be inadequate.

The adequacy is particularly concerning in regions with high unemployment rates and poor employment prospects, or for special groups such as older workers, persons with mental health issues or single parents where finding and maintaining employment is challenging and the risk of longer term unemployment is real.

The current Newstart rate is about $340 per fortnight below the single person poverty line. This implies that, without additional income sources, these people are estimated to be in poverty and have a large ‘poverty gap’ which measures the depth of poverty. From the perspective of poverty, our modelling shows that, dollar for dollar, an increase in the Newstart payment reduces poverty more than any other change to welfare payments.

Alleviating poverty

The welfare system is designed around a number of core policy objectives. One of these objectives is to alleviate poverty. With one payment falling behind other payments and the living standards of most other Australians, the welfare system can and should allocate scarce resources better, from the perspective of poverty alleviation.

The ANU Centre for Social Research and Methods has focused on developing a new methodology to calculate payment levels that optimally allocate payments to ensure policy objectives are best achieved. Our initial work focuses on poverty in Australia.

We have done this using Policymod, our recently developed microsimulation model of the Australian tax and transfer system. PolicyMod is a statistical model based on 19,000 actual households in a 2015–16 Australian Bureau of Statistics survey.

Using this model, we simulated about 1,000 different potential payment levels for our major payments including pensions, allowances (such as Newstart), family payments, parenting payment and rent assistance. These payments make up over 80 per cent of all welfare cash payments in Australia.

Using these simulations, we established a statistical relationship between each payment and poverty in Australia. Factoring in some further restrictions around the budget and payment changes, we were able to optimise the payment levels that would minimise poverty in Australia.

Our initial results found that poverty could be reduced by up to 11 per cent without any change in the social services funding level. A 15 per cent increase to the current budget would reduce poverty by up to nearly 27 per cent, while it is possible to maintain the current poverty level even with an eight per cent reduction in the social services budget.

The modelling suggests that current payment rates for our major welfare payments are often a long way from ‘optimal’ levels where we wish to minimise poverty. Dramatic changes to current payment levels are required to minimise poverty under a budget neutral scenario. For example, the existing Newstart payment would be increased from its current level of $550 per fortnight to a much more generous $800 per fortnight.

The age pension and parenting payment single payments would be increased modestly from current levels. To offset the cost of these increased payments, rent assistance and family tax benefits would be lower.

These ‘optimal’ policy settings relate to the objective of minimising the household poverty gap – which considers household income after taxes and benefits. Where we deduct the cost of housing from income and minimise the ‘after-housing’ poverty gap, we find similar results except rent assistance is increased, family payments are kept at existing levels, the age pension is modestly reduced and the Newstart payment and parenting payment single are increased.

These estimates include some important restrictions that ensure the Newstart payment must be at least 10 per cent lower than the age pension and that each welfare payment cannot deviate from current levels by more than a factor of 0.6 or 1.6. We do this in recognising the political and financial realities of changing existing payments would cause problems for current recipients. Without these restrictions, it is likely that poverty rates would be even lower.

The modelling ensures the age pension is higher than the unemployment payment – even though the unemployment payment may better target poverty than the age pension. This restriction follows the logic that pensions are usually longer term payments whereas Newstart is, at least ideally, a shorter term payment.

Given the strength of the statistical relationship and the speed of modern optimisation routines for complex mathematical problems, we found our ‘optimal solution’ to payment levels in less than a second! This provides the modeller with great flexibility in testing the model with a range of restrictions or different policy objectives.

Our research is not designed to provide a single answer for policymakers. However, it does provide a useful new modelling tool to policymakers and researchers to better understand how well existing or alternative welfare measures achieve their policy objectives.

As our research continues we expect to develop similar modelling tools for a range of different policy objectives. Policymakers and researchers can then use these tools independently or combine the results to better guide their own intuition and understanding of the impacts of policy.