The Parliamentary Budget Office (PBO) published a paper earlier this year entitled ‘Behavioural assumptions and PBO costings’ to support public understanding and transparency around how behavioural responses are incorporated into our costings.
In the course of parliamentary debate, new policy ideas are often met with the question: ‘how much will it cost?’. Answering questions such as this is one of the core functions of the PBO.
Any parliamentarian in the Commonwealth parliament can ask the PBO to assess the estimated financial implications, or ‘cost’, of policy proposals they are considering. A policy costing provides an estimate of the impact that a proposed policy would have on the budget if it were implemented. Such costings can be requested on a public or confidential basis, and are prepared to the same standards and using the same concepts as those published by the Government in budget papers.
Our advice is typically sought by members of the Opposition and minor parties as they cannot ask government departments for such advice. We also publish a post-election report that costs all election commitments for the Coalition, the Australian Labor Party and the Australian Greens (smaller parties are also able to opt in).
A key element of a costing relates to the details of the policy as set out by the requestor. These are referred to as ‘specifications’. The specifications for a cash payment, for example, would generally include: the start date, the cohort of the population eligible for the payment, the criteria that must be met to receive it, and how the payment would be delivered. As the PBO does not provide policy advice, these specifications must be provided by the parliamentarian in order for the policy to be able to be costed.
A costing critically depends on two things – the specifications of the policy and the assumptions about how people are likely to change their behaviour in response. Although parliamentarians provide the specifications, it is the PBO that makes the assessment of the nature and quantum of behavioural responses.
The behavioural component – how affected individuals, businesses or organisations would respond – can be contentious with a range of differing views around potential policy responses. The paper released by the PBO earlier this year explained the principles we use to come to a robust view of behavioural assumptions, making sure our assumptions are consistent across policies, draw upon empirical evidence, and are informed by expert judgements. The paper also included a list of behavioural assumptions made in costings that were publicly released during 2019, including costings of the election commitments made by the major political parties. Where available, this list includes links to any research or data we used in determining the assumptions.
Some policy proposals are intended to induce a particular behavioural response – for instance, the intention of subsidising dental care may be to increase the frequency of dental visits for particular groups, which means that demand for dental care is expected to increase in response to the policy.
Conversely, some policy proposals may induce behavioural responses that are unintended or responses which act against the policy’s purpose. A proposal to increase the tax rate for higher income earners, for example, may lead some high-income earners to find ways to reduce their taxable income so as to avoid paying additional tax, resulting in the proposal raising less revenue than it would have otherwise.
An important consideration then in preparing costings is both the intended and unintended responses associated with a policy.
In determining the likely behavioural response, the PBO is guided by three principles, namely that the assumptions are:
- consistent across similar policies;
- backed by empirical evidence; and
- informed by expert advice.
Consistent across policies
In order to achieve consistency, we draw upon behavioural assumptions for policy proposals we have previously costed, where appropriate. Before an election, it is not uncommon for different parties to have the same or very similar policy commitments, and it is essential that we do not make inconsistent assumptions across similar election commitments.
A number of our costing models incorporate default elasticities – a measure of the degree to which individuals or organisations respond to a given change in wages, tax rates, or prices for particular goods or services – that are drawn from empirical literature, which helps ensure that assumed behavioural responses are proportionate to the policy impact.
We also draw upon behavioural assumptions that have been made by other government agencies when costing budget measures, in the interest of preparing costings consistent with those prepared for the Government.
Before accepting any assumptions that we have or another agency has made, we carefully scrutinise these. For instance, we examine whether there have been changes to government policy or economic conditions that suggest a different behavioural response may occur. We also regularly review literature and undertake new analysis as data become available to keep behavioural assumptions up to date.
Evidence-based
Peer-reviewed academic research is usually considered the ‘gold standard’ for quantifying behavioural responses.
For some policy proposals there is a wealth of reliable empirical literature to draw upon that estimates how those affected by similar proposals have responded in the past, although this is not always the case. For instance, elasticities estimated in the literature tend to be more general than specific, or they may apply to an international, rather than Australian, context.
As an organisation we have access to high-quality historical data that may be used to estimate behavioural responses to past policy changes, or to help inform a behavioural assumption. Our own research can be used to complement academic research, or to inform cases where reliable academic or other empirical research is not available.
There are, however, times where there is little empirical research to draw on or data may not be available. On these occasions, the best we can do is apply our professional judgement or seek advice from experts in the field.
Informed by expert advice
Expert advice is particularly useful for new and novel proposals where there is little empirical evidence to draw upon. We ensure that any consultations with experts are conducted in a way that maintains the confidentiality of requests from parliamentarians and is with the permission of the requestor.
The experts we reach out to include people in government departments, business or academia that specialise in particular policy areas. We have a panel of advisors who have expertise across a broad range of areas and are available for consultation when the need arises.
Ultimately, all behavioural assumptions require a degree of our own expert judgement. Even when we have collected a strong evidence base to inform a particular assumption, such evidence rarely relates to exactly the same policy in the same context.
There is a degree of uncertainty associated with all behavioural assumptions and this is noted in our costing advice to parliamentarians. We provide details of these assumptions to help parliamentarians understand how individuals and businesses affected by their policy proposals are likely to respond, and to provide assurance to the wider public that we think carefully about the assumptions we make.
We welcome your feedback on our assumptions
We welcome scrutiny and comments on the assumptions we make in order to ensure that they are robust and based on the best evidence. Supporting empirical evidence for these or alternative behavioural assumptions would be particularly welcome (via [email protected]).
For those empirical researchers considering topics for future work, the list of behavioural assumptions published in our information paper could help to identify areas where there is, at present, little empirical evidence around behavioural responses.
For instance, more detailed research on how different groups (for instance, by income, age, household type or gender) respond to policy changes can be more valuable in informing policy decisions than using a simple ‘one size fits all’ aggregate response.
Examples of areas for further research include:
- the impact of changes to out-of-pocket costs on the uptake of private health insurance;
- the impact of subsidies for dental care on attendance rates;
- estimates of taxable income elasticities for different income levels;
- the investment response of companies to tax incentives; and
- the impact of superannuation tax concessions on contributions and on total savings.
In conclusion, behavioural responses are an important component of policy costings. The PBO takes a thorough approach when exercising its professional judgement on appropriate assumptions to ensure consistency across costings and the inclusion of the best research and expert advice. However, we are always willing to receive feedback on our assumptions or to access new, emerging research or expert opinion in order to produce the best possible cost estimates. Ultimately, this helps to enhance the policy debate, which is of value to us all.
Recent Comments