Quantifying welfare law (part 1)
Is quantative analysis of social security decisions in the AAT possible — or useful?
It is important for any researcher who is using any method of quantitative analysis to identify the quality of the dataset or database with which they are working. Clearly, selecting the best database, with the most complete and searchable number of datapoints, will be critical to the depth and breadth of any knowledge generated from the analysis. If the database is limited, the results will be skewed — and potentially misleading.
Using quantative methods to analyse AAT decisions realting to welfare will only be useful if the dataset is reasonably deep and broad. So how does the AAT publish its decisions, and where?
As the AAT website indicates, ‘AAT decisions with written reasons are published on the AustLII website and may also be available from other legal publishers.’ These other publishers seem to include non-open-access databases, such as Westlaw, as well as some partially open-access online publishers, such as Jade. A cursory search on Jade suggests that it publishes AAT decisions at the same rate as Austlii.
Because, for reasons I will explain, most published AAT decisions relating to welfare will always be marked with the words ‘(Social services second review),’ it is relatively easy to seach for that phrase (making them a neccessary item) and then, by sorting the results by date, to compare the decisions published recently by Austlii and Jade. Using this method indicated that both databases are publishing the same decisions at the same rate.
But before I write more about accessing data through databases, we should examine some of the endogenous limits on the published data themselves.
The AAT does not publish all its decisions. In fact, I’m quite sure the AAT is not required to publish any of its decisions. In its Publication of Decisions Policy (Policy), the AAT confirms that it is ‘authorised’ to publish its decisions under s 66B of the AAT Act (although it may not publish decisions when it is not lawful to do so). This does not seem to guarantee that it will always publish its decisions, however, because the AAT Act does not appear to require it to.
Neverthless, the Policy — a form of soft law, perhaps — directs the AAT to publish decisions in the public interest. As the AAT website states, the reasons AAT members’ reasons are published is because doing so ‘promotes public trust and confidence in our decision-making and the transparency of our operations.’ Since being updated in 2017, the Policy directs the AAT to publish decisions with written reasons in most circumstances. Generally the AAT will publish decisions in its nine main divisions. They are the Freedom of Information Division; the General Division; the National Disability Insurance Scheme Division; the Security Division; the Small Business Taxation Division; the Taxation & Commercial Division; and the Veterans’ Appeals Division.
Under cl 2.2(a) of the Policy, the AAT will not publish a decision if there is some legal constraint or prohibition on publication—such as a legislative provision prohibiting publication—then the decision will probably not be published.
In welfare law, appeals are made at two levels or two tiers.
The lower level is called the Social Services & Child Support Division. Decisions at the lower level are less expensive and usually occur in one hour, with no legal represntative from Centrelink. It is an informal process.1 The lower level decisions are usually called AAT1 decisions.
The higher level is called the General Division of the AAT. These are final appeal decisions. This means that they have already been heard, and presumably dismissed, at the lower level. These higher (highest) level decisions are called AAT2 decisions.
In terms of citations for social security cases, AAT1 decisions relating to child support look like this:
Name and Child Support Registrar (Child support)  AATA 111 (1 January 2021)
AAT2 decisions, by contrast, look like this:
Name and Secretary, Department of Social Services (Social services second review)  AATA 111 (1 January 2021)
You can tell that it is in AAT2 decision because the citation includes the words “second review” in parentheses.
Notably, AAT1 decisions (first reviews) relating to Centrelink payments are very difficult to find publicly. In fact, I have not found any. The only published AAT1 decisions I can find in the Austlii or Jade databases are those relating to child support. These are decisions in which the Child Support Registrar is the respondent. It appears that decisions of the AAT1 in which the Secretary of the Department of Social Services is the respondent are simply not published.
Curiously, however, the AAT states in its Policy-related webpage that it does publish these decisions:
We publish a selection of decisions with written reasons made in our other divisions:
Migration & Refugee Division – all decisions made in some categories of cases and a randomly selected percentage of decisions in other categories, and
Social Services & Child Support Division – a randomly selected percentage of child support decisions.
Decisions with written reasons are not published if a law or a Tribunal order prohibits their publication.
Notably, the Policy-related webpage states that only a ‘randomly selected percentage of child support decisions’ will be published. It appears, then, that no social services decisions are published.
This immediately has consequences for any quantitative analysis of welfare law in Australia. How many decisions of the AAT1 are overturned by the AAT2? What percentage of AAT1 decisions are successful? These and many other questions cannot be answered without a database of published decisions. They could possibly be answered if those AAT1 decisions were published in partial form, anonymised form, or even in a way that simply showed that they occurred.
Nevertheless, it is possible to deduce some quantitative data from the publication of the AAT2 decisions. For instance, it is almost certain that every decision of the AAT2 is the result of an unsuccessful decision in the AAT1. However, we still cannot really determine much about the AAT1 decisions.
Consulting the AAT’s annual reviews is not neccessarily helpful. Althought the 2019-20 Annual Review offers caseload data, it does not disclose the number of Centrelink AAT1 decisions; only the total number of AAT2 decisions. Reporting on the caseload, the Review notes
The number of applications for second review [AAT2] of Centrelink decisions increased in 2019–20, reflecting to some degree the higher number of finalisations of first reviews of Centrelink decisions in the Social Services and Child Support Division.2
Unless I am missing something, this statement simply means that the increased number of AAT2 decisisons indicates an increased number of AAT1 decisions. This is the point I made earlier — namely, that all we can read into the AAT2 data about the AAT1 decisions is that there have been at least an equal number of AAT1 decisions as there have been AAT2 decisions.
Professor Terry Carney has rightly bemoaned the fact that AAT1 decisions are not published. In an article analysing why the AAT failed to remedy robodebt before a settlement was reached in the FCA (although that settlement remains, at the moment, unapproved by the FCA), Carney argues that the secrecy of the AAT1 — facilitated by the private hearings and the non-publication of its decisions — was a critical reason why robodebt could go on for so long as it did, without oversight beinging its injustices to the attention of the public or mainstream communities. Carney’s analysis is as follows:
First, why did the AAT fail? Of course generally it did not fail those individuals with the stamina to press their case through up to two internal review (original decision-maker review and then authorised review officer) and AAT1 hearings. But the closed AAT1 hearings and non-publication of its decisions (partly a failure of AAT leadership in not publishing any de-identified precedent cases), coupled with what appears to be a deliberate policy by Centrelink not to appeal to AAT2, meant that the AAT failed to exercise any normative influence, and led to justice being administered in secret (and thus its ruling not being open to public debate). This latter is not of small moment.3
Carney’s observations represent strong, important and legitimate criticisms of the two-tier AAT system. However, to play devil’s advocate, one might propose that there are also some privacy protections afforded by the AAT1’s secrecy that might benefit an applicant enmeshed in this system.
For example, if a person was being prosecuted for offences realting to social security receipts (such as offences under s 135.2(1) of the Criminal Code), they might seek a review of the decision to raise the debt. If that person, after going through the ARO process, sought review in the AAT1, the Tribunal member in the administrative appeal would almost certainly be required to consider certain factual aspects of the applicant’s case that may have a bearing on the applicant’s criminal trial.
For instance, the Tribunal member might consider whether there were any reasons to waive or write-off the debt, and may consider whether any special circumstances exist that might fall within s 1237AAD of the Social Security Act 1991 (Cth). That section provides that the respondent may waive a debt where satisfied that
(a) the debt did not result wholly or partly from the debtor or another person knowingly:
(i) making a false statement or a false representation; or
(ii) failing or omitting to comply with a provision of this Act, the Administration Act or the 1947 Act; and
(b) there are special circumstances (other than financial hardship alone) that make it desirable to waive; and
(c) it is more appropriate to waive than to write off the debt or part of the debt.
A consideration of this section might touch on matters relevant to the criminal proceedings being brought against the applicant in the criminal system. Thus, if this consideration were published, it may be possible that it could be adduced as evidence in a court and used against the defendant, provided no bar to admissibility existed.
In these circumstances, it may be preferable that the AAT1 decisions remain unpublished, lest they be used against an applicant in any future or contemporaneous criminal proceedings. of course, on the other hand, it may well be possible for the Department to share the reasons handed down by the AAT1 in respect of any application—even if they were oral — to the Commonwealth Director of Public Prosecutions anyway. In that case, the CDPP may well be able to use the information and the reasons just the same.
Clearly, the non-publication of AAT1 decisions means that the possibility of undertaking quantitative analysis of welfare law is partially limited. Without knowing how many of these first reviews are successful or unsuccessful, we will only have an incomplete picture of the nature of the administrative process.
In part 2 of this post, I will turn to the published decisions of the AAT2 and consider the extent to which a quantitative analysis of those decisions can be usefully undertaken to shed light on the workings of the AAT system in welfare matters.