Lessons in Lawfulness

An analysis of the Commonwealth Ombudsman report finding more than 100,000 Centrelink files are affected by 'income apportionment' -- an unlawful method of calculating debts -- from 2003 to 2020.

Last week, on 2 August 2023, the Commonwealth Ombudsman published a short report, titled Lessons in Lawfulness, outlining its findings after having investigated the lawfulness of 'income apportioning' — a method of debt calculation used by Australian welfare agencies DHS and DSS since at least 2003 (ie, for some 20 years). As the Report states, this process, as it turns out, is unlawful.

This is an astonishing finding, and all the more so in the wake of the Robodebt Royal Commission. On the day following the publication of the Report, I recorded a long video (1hr and 40 mins) that provides my detailed commentary and speculation on the meaning and remarkable implications of this unambiguous document. The video is available below:

The video details many of the aspects of the situation that are important, including the four recommendations that the Ombudsman makes in respect of what the agencies — DHS and DSS — should do now.

Those recommendations, in a nutshell, are as follows.

Recommendation 1

Firstly, in the context of the legal advices the agencies have received, and in view of the fact that the Ombudsman states the practice is unlawful, the Report recommends that Services Australia (SA) and the Department of Social Security (DSS) should obtain a declaration of the law. It makes this recommendation in two parts, as recommendation 1A and 1B.

Notably, this first recommendation refers not only to the way in which the agencies have historically calculated the debts, but also to the way they have calculated the debts since 2021, when the Ombudsman first brought the matter to the agencies’ attention. Since that time, the agencies have developed a position or form of guidance contained in the so-called General Instructions that directs decision-makers about how to putatively lawfully calculated social security entitlements prior to December 2020. However, because there has not been any legal declaration on this guidance, nor on the original, pre-Guidance calculation method, the first recommendation asks the agencies to get that clear advice on both methods.

1A: SA and DSS should seek the opinion of the Solicitor-General (SG) through a referral under Law Officers Act 1964 (Cth).

1B: SA and DSS should seek a Federal Court of Australia ‘opinion’ through referral by the Administrative Appeals Tribunal (AAT) under the Administrative Appeals Tribunal Act 1975 (Cth).

Before we move on, it is worth looking at the legal provisions underlying these recommendations.

Recommendation 1A

The referral to the SG is said to be authorised by the AG under the Law Officers Act 1964 (Cth) (‘LO Act’) but, as is common in brief Reports of this nature, the Report does not name the section authorising this action.

Section 12 of the LO Act sets out the functions of the SG as follows:

12  Functions of Solicitor‑General

                   The functions of the Solicitor‑General are:

                     (a)  to act as counsel for:

                              (i)  the Crown in right of the Commonwealth;

…

                     (b)  to furnish his or her opinion to the Attorney‑General on questions of law referred to him or her by the Attorney‑General; and

                     (c)  to carry out such other functions ordinarily performed by counsel as the Attorney‑General requests.

Thus, s 12(b) and (c) both indicate that the SG will do as is required of them when the AG refers a question of law to them (s 12(b)) or makes a request of them (s 12(c)).

However, as the Report notes (p 12), the SA and DSS have already approached the SG and the SG has

declined to provide advice, stating [that] SA and DSS should first share the different legal advices between the legal professionals involved to see whether [they] can agree. The agencies have [since] shared all legal advices with the legal professionals [of the other agency] and are in discussions to see whether the different views can be bridged by discussion.

It seems to me that the SG is entitled to decline to give advice to a department head or Minister in circumstances where it is not clear that they are acting as counsel for the Department or Minister. As s 12 of the LO Act states, the SG’s functions include acting as counsel for the following entities, presumably when they are defendants or plaintiffs:

(i)  the Crown in right of the Commonwealth; (ii) the Commonwealth; (iii) a person suing or being sued on behalf of the Commonwealth; (iv) a Minister; (v)  an officer of the Commonwealth; (vi) a person holding office under an Act or a law of a Territory; (vii) a body established by an Act or a law of a Territory; or (viii) any other person or body for whom the Attorney‑General requests him or her to act.

However, it is unsurprising that, where two agencies of the Commonwealth are not in agreement as to whether something is lawful or unlawful, it might be prejudicial (giving rise to a conflict of interest) if the SG were to expressly or implicitly act only for one to the other — even without intending to — by preferring the legal position of one agency over the other, thus rendering one or both of the agencies vulnerable to liability.

It may also be so that the SG has a more general implied right to refuse to write a legal opinion on request from the agencies (or give any other advice) since it appears to be outside their functions — except, perhaps, where it can be said that the SG is acting for one of the agencies, or both. And it would not appear that the request is that the SG act for the agencies. Thus, there is no statutory function that required the SG to give advice or to write an opinion for the agencies without being quasi-retained by them in circumstances where it can be said that the SG is acting as counsel for them.

It is for this reason that it appears the original recommendation takes the form that it takes, whereby the Ombudsman recommends that the AG refers the matter to the SG, and presumably this would occur under s 12(b) or s 12(c) of the LO Act, and it would require the SG to answer the question of law. Notably, as I say in my video summary, the Report also clearly states that there are two issues on which the SG is to give an opinion, thus indicating that the questions of law (QOLs) may already have been formulated, and that the Ombudsman is merely waiting for the AG to refer the QOLs to the SG. In this context, it should be asked: Why has the AG not referred the QOLs to the SG already?

Recommendation 1B

The ‘fallback’ recommendation proposed by the Ombudsman is that the agencies refer the matter to the Federal Court of Australia (FCA) in the course of hearing an AAT proceedings. For this to happen, someone from the AAT, presumably a member hearing a matter and having informed the President of the Tribunal (the statute uses the expression ‘the Tribunal’ in reference to the referring entity), would refer that matter as one that poses a question of law (QOL) that should be resolved by the FCA under what the Report simply indicates is some innominate power under the Administrative Appeals Tribunal Act 1975 (Cth). Again, the Report names no section or method for its recommendation.

(As an aside, I should note that it’s quite striking to consider the way in which these recommendations are made in such a skeletal way, the subtext being that they have already been researched and discussed with the agencies to which the recommendations are being ‘officially’ now made. Indeed, it is obvious that this recommendations have already been subject to some degree of consultation and — alas, when one considers the problem of the Ombudsman’s ‘inadvertent’ collusion with the agencies (primarily DHS) during the robodebt incvestigation — a degree of ‘negotiation.’ However, it could also be said that the Ombudsman report on this occasion, even if it has been subject to such discussion with the target agencies, is also a much more stoic and hard-nosed document that the Ombudsman’s robodebt investigation.

One might hope, for example, that no one at the agency has been able to edit or delete parts of this Report . It certainly does not read as though that has occurred. But, then again, as I pointed out before, the document is skeletal; and so much of its substantive findings have to be inferred. (For instance, why does the Ombudsman feel so comfortable stating that method I straightforwardly unlawful — and to teach ‘lessons in lawfulness’ to the agencies — if there agencies are still in dispute about whether the legal advice says so? It seems clear that the Ombudsman is not prepared to accept the legal advice of whichever agency is trying to spin the s 1073B problem into a smaller one — our perhaps even into a nullity).)

Although the Report names no section for the FCA referral, I have conveniently addressed this issue before. The process by which the AAT refers a matter to the FCA is via s 45 of the AAT Act 1975 (Cth). My post on that process is viewable here. But to quickly recap, the section provides as follows:

45  Reference of questions of law to Federal Court of Australia

(1)  The Tribunal may, with the agreement of the President, refer a question of law arising in a proceeding before the Tribunal to the Federal Court of Australia for decision. The Tribunal may do so on its own initiative or at the request of a party to the proceeding.

…

(2)  The Federal Court of Australia has jurisdiction to hear and determine a question of law referred to it under this section.

(2A)  If, after consulting the President [of the AAT], the Chief Justice of the Court considers it appropriate, that jurisdiction is to be exercised by the Court constituted as a Full Court.

(3)  Where a question of law arising in any proceeding has been referred to the Federal Court of Australia under this section, the Tribunal shall not, in that proceeding:

(a)  give a decision to which the question is relevant while the reference is pending; or

(b)  proceed in a manner, or make a decision, that is inconsistent with the opinion of the Federal Court of Australia on the question.

As we can see in s 45(1), the AAT does not even need the parties (Services Australia or the customer) to ask the AAT to refer the matter to the FCA (and presumably the AAT would not need their consent, assent or approval either). Thus, if the AAT heard a case in which the income apportionment issue arose tomorrow, it could, now with the Ombudsman’s Report in hand, refer the matter to the FCA for determination—just as the Ombudsman is now recommending. Even if one of the parties objected, it seems there would be no bar to such a referral. And while the Ombo’s recommendation is directed to the agencies themselves, they need not ask the AAT to refer the QOL to the FCA — and indeed may not even be relevant to the process by which it is referred.

All that seems to be required is that the the reference be made. Interestingly, the full court of the FCA (a three-judge bench) may be constituted to hear the matter if the President of the AAT and the Chief Justice of the FCA agree. This means that two personalities would be involved.

Notably, the President of the AAT resigned last year in unusual circumstances, including in the context of allegations of bullying made by users of the Tribunal against some 17 Tribunal members. The former President was appointed on 1 April 2022 for the statutory period of seven years; however, her Honour, who is also concurrently a Justice of the FCA, wrote to the AG and the Governor General (separately) on 30 November 2022 to resign following a meeting with the AG on 25 November.

Prior to her Honour’s appointment, a review of the amalgamation of the AAT authored by the Hon Ian Callinan KC, former Justice of the High Court, was published (in 2019). That review recommended a number of measures to improve the AAT, which had previously not included the separate migration and refugee tribunals but, with the amalgamation in 2015, had enlarged to include tribunal divisions for these matters. Notably, in response to criticisms that had been made in the period after 2015 about political appointments, the review recommended that some political experience may be appropriate and ‘useful experience’ for tribunal members ‘undertaking merits review’ (p 170-71). Additionally, however, the review observed that

overwhelmingly submitters favour, as I do, the appointment of people of good character, even temperament, diligence, possessing legal qualifications, and selected as a result of a transparent process.

At a time when many members of the AAT had been criticised for not holding legal qualifications, this recommendation appears to have been a small affirmation of the importance of legal training to the work of the AAT. In later passages, the review also noted, while characterising the interaction between political decisions, policy and tribunal decision-making, that AAT and other merits review decisions are inevitably open to debate and contestation, they being a part of the application and implementation of policy decisions expressed as it may be in legal enactments. In this context, the AAT review notes that

Members should not be thin skinned. Persons seeking to work as Members of the AAT need to understand that they are undertaking merits review of executive decisions in matters upon which fair minded people can and do disagree, often strongly so. Competent Members of the AAT themselves will from time to time disagree about the merits of a particular executive decision. (para 10.32, p 171)

Given what is described, above, as the inevitably and even desirably culture of policy contestation at the AAT, it may be that different views among members may exist and materialise about whether a matter that comes before a first or second tier member will be one that raises an appropriate QOL for reference to the President of the AAT and, by extension, to the FCA. Of course, in the wake of the Ombudsman’s Report, any political or policy-based objections are likely to have diminished force.

In addition to the former President of the AAT, who sits on the Federal Court of Australia and may well be involved in hearing and determining a QOL referred to the FCA by the AAT, other judicial officers in the FCA, as would be expected, have significant experience in welfare law. For instance, another Justice of the FCA was director of the CDPP during the important welfare criminal law South Australian Supreme Court (full bench) and High Court decisions of DPP (Cth) v Poniatowska. Similarly, the Chief Justice of the FCA was lead counsel for the applicant (instructed by Victoria Legal Aid) in DPP (Cth) v Keating — a fundamentally important case in welfare law and criminal prosecutions.

The stories of the Poniatowska and Keating matters are far too remarkable in their own right to do them any kind of justice through repeating them tersely here. However, it should suffice to say that the fact that there is a vast amount of experience in the Federal Court about welfare law, and welfare crime matters, will mean that any referral is bound to be met with not just a learned but an historically informed and, ex hypothesi, an historically coherent approach, appropriately flowing from precedent in like (albeit different) matters.

Recommendation 2

The second recommendation in the Ombudsman’s report relates to the guidance given to decision-makers and said to be contained in the General Instructions described earlier. As noted, the agencies have developed guidance to avoid committing what the Ombudsman has said are unlawful calculations by reference to s 1073B of the Social Security Act 1991 (Cth) — a provision that was repealed and replaced with a new version in December 2020 through the Social Services and Other Legislation Amendment (Simplifying Income Reporting and Other Measures) Bill 2020, the Bills Digest for which is here.

(The timing of this amendment Bill is notable; the Digest for the Bill was published on 24 February 2020, which means that it precedes the settlement of the class action (Prygodicz v Commonwealth of Australia (No 2) [2021] FCA 634), which was settled by means of Deed of Settlement in November 2020 and finally approved by approval judgment by Murphy J of the FCA on 11 June 2021. If the robodebt class action may be said to have brought the issues relating to income apportionment to the government’s attention (or may be said to have prompted a response), the government would have been working on that response for some months prior to the class action being settled.)

However, even though the General Instructions have been created, they are viewed by the Ombudsman to be inadequate. And, notably, this assessment of these debts as ‘inadequate’ has been made in the Report even before the SG or FCA comments on the unlawfulness of the debts. These are those debts that are raised pursuant to s 1073B — and the surrounding provisions (in their pre-amendment forms) — and which are said to be unlawful because they are not calculated or coded consistently with that section.

In short, the Ombudsman suggests that the General Instructions are inhibiting reviews of historical debts. The first part of the second recommendation is therefore that the General Instructions be amended so that

delegates are not inhibited from exercising discretion to consider whether to review historical debts. [My emphasis.]

Delegates, in this recommendation, would mean SROs and AROs. Dealing with a Centrelink ARO can be quite an impersonal and blunt episode. Many people report feeling unheard and that their matter was not properly considered. This may be a problem; however, it is certainly not the end of the line, and many people go on to appeal their debts to the AAT, where the outcomes and nature of the review is fart superior, with the analysis of all arguments being generally very thorough and at the end of which detailed reasons are proffered.

However, it is not surprising to learn from this report that the SME and ARO reviews, which occur on the phone, are said to be potentially ‘inhibited’ and law an ability to permit those delegates to exercise discretion. Indeed, it has been my experience (vicariously) to understand that AROs have in fact been asked about apportioned income during the Robodebt period and have been met with denials that apportioned or averaged income across multiple fortnights are a reason too review the debt. While I have seen several copies of ARO reasons, the scant detail often provided in those reasons and the very rudimentary nature of the analysis means that it is difficult to tell whether the AROs even understand what the customer is identifying when they bring it to their attention, or whether the ARO’s General Instructions are so restrictive as to mean that they cannot review a matter unless very specific evidence is adduced.

For example, below is an excerpt of an ARO decision from 22 January 2021 — in the period during which the Ombudsman had begun its investigation but before the agencies had responded. As can be seen, the prospect of ‘averaging’ and its potential unlawfulness is raised by the customer; however, the ARO gives no actual consideration to the problem, stating that because the income amounts had been provided by the employer, there was nothing that could be done. In the context of the Robodebt Royal Commission, the notion that ‘employer verified’ income reports could be inadequate evidence — or evidence that could be rendered invalid if it was apportioned across multiple fortnights — was completely misunderstood or underrated.

As can be seen, it appears the ARO has no conception of the notion that the ‘coding’ of the employer amounts is the problem — not the amounts themselves. Even if the employer provided correct information, if the income data is averaged or apportioned across multiple fortnights, it will be affected by the income apportionment problem described in the Ombudsman’s Report.

On the basis of the above and other ARO decisions, I have no doubt that many people who have had their incomes apportioned and who even brought this fact to the attention of AROs (I saw several instances through being asked to look at files as a volunteer with Not My Debt) will have received the same response as the customer above. The more concerning aspect of this, however, is that these unsuccessful reviews would have occurred during the so-called ‘freeze’ period of 2020 to 2023. This suggests that the 13,000 frozen affected review requests identified by the agencies as being affected and the 87,000 files being affected in total is probably an unreliable figure.

Another question arising from this second recommendation is: What do the General Instructions look like? Helpfully, we have access to some documents within what is often called the ‘Operational Blueprint’ — a system of internal rules and guidances (some might describe it as ‘soft law’) that now is given the title ‘Operational information’ (OI) The front page of this library of documents is here.

In the section dealing with the Earnings Apportionment Tool (EAT) (a document with code name 107-03020010), there are multiple sections or tabs: Background; Process; Resources; and Training & Support. Each contains questions and answers. Each answer is an instruction about how to use the EAT (proverbially known as the ‘EAT sheet’) which suggests that this documentation, the OI, may well be the same documentation that is referred to as the ‘General Instructions’ in the Ombudsman’s Report.

But there are more convincing arguments for concluding that the OI and ‘General Instructions’ are the same. For instance, the Resources tab includes a section devoted to ‘Example questions and answers to support Service Officers.’ One of the questions instructs officers how they should assess the calculation of ‘daily payslips’ prior to 7 December 2020. That, of course, is the date when the amending legislation was passed, in the form of the Social Services and Other Legislation Amendment (Simplifying Income Reporting and Other Measures) Bill 2020, that rendered the income apportionment problem putatively resolved (from that point onwards).

If it is correct that what the Ombudsman’s report calls the General instructions are the same as the OI notes, then it is interesting to note that these instructions are fairly elliptical or terse as to what is not permitted. As the note above makes clear,

Fortnightly or monthly payslips that show a day without a date cannot be coded in this way [ie, in the way that fortnightly payslips with income dates can be]. Over a fortnight or month, there is no way to identify which Tuesday or Saturday the customer worked.

Service Officers must decide the best method for assessing each customer's employment income. The same method may not be suitable in every case.

Explain all coding decisions on a DOC.

In reality, this is no guidance at all. It merely delegates or leaves to the decision-maker the task of determining how to raise a debt in circumstances where one method is said to be unavailable (they ‘cannot be coded in this way’). Even the instruction that it ‘cannot be coded in this way’ does not quite emphasise that this may be an unlawful operation or, at least, an operation not authorised by statute. If, as I am suggesting, the OI text above is the same as the general Instructions, it is no surprise that the Ombudsman’s report might find this text inadequate. Then again, it seems improbable that this is all that the agencies have have produced regarding this question for decision-makers, precisely because the guidance is so uninstructive and does not present a solution at all. In any case, we do not have access to more information and can only speculate as to what internal instructions have been given.

Recommendation 3

The third recommendation indicates the broader issues at play here — issues that speak to welfare injustice and even suggest miscarriages of criminal justice. Even so, it is a very mild recommendation — a benign intervention — in the circumstances. The ombudsman recommends that SA (and it is SA that is singled out here) should consult with DSS and the Commonwealth Director of Public Prosecutions (CDPP) to work out how to review and treat past, present and future criminal prosecutions that involve unlawful apportionment under s 1073B. The recommendation, for clarity, is written in these terms:

We recommend Services Australia, in consultation with DSS and the CDPP, develops a strategy for how agencies will approach historic, current and future criminal prosecutions associated with administrative debts involving income apportionment under section 1073B of the Social Security Act 1991. The strategy should include an agreed policy position on evidence required to prove administrative debt amounts and the impact this will have on prosecutions and convictions.

The clear inference to be drawn from this recommendation is that people have been, are being, or may continue to be prosecuted for offences where the use and unlawful misuse of s 1073B is a part of the evidence-gathering process. We would be talking about offences here under the relevant four crime sections of the Criminal Code that are normatively used to carry out welfare prosecutions: namely:

• s.134.1(1) Criminal Code: Obtaining property by deception;

• s.134.2(1) Criminal Code: Obtaining a financial advantage by deception;

• s.135.1(5) Criminal Code: General dishonesty—causing a loss; and

• s.135.2(1) Criminal Code: Obtaining a financial advantage.

Anecdotally, the most common welfare prosecution offence is the last of these: s 135.2(1). Notably, this section does not require a very serious 'mental ‘fault’ element; the fault element is not intention, recklessness, or negligence, but merely knowledge. In essence, if a person receives an amount from Centrelink to which they are not technically entitled, and they know that this amount has been received without that entitlement, the offence would be very likely made out, the facts being otherwise unexceptional.

My general and rough sense is that many welfare prosecutions and convictions under s 135.2(1) would be affected by unlawful income apportionment, depending on how the advices wash out and how the meaning of ‘unlawfulness’ ends up being particularised in any determinative legal holding. However, if the practice of using insufficiently particularised (ie, non-fortnightly or disaligned payslips, or payslips that exceed, in terms of the number of days, the relevant instalment period under s 1073B) payslips or customer reports is as common as it instinctively would seem (for example, the problem of misaligned fortnights, for one, has long been identified), then it would make sense that many prosecutions would involve the apportionment of income contrary to s 1073B because the problem of disaligned fortnights (instalment period versus payslip period) has been remedied by apportionment of income that extends beyond one instalment period.

I will not go too far into the technical basis of the 1073B problem and its various causes right here. Three likely causes are identified above: namely, (a) non-fortnightly or (b) disaligned payslips, or (c) payslips that exceed, in terms of the number of days, the relevant instalment period under s 1073B. I have also given my preliminary analysis in the video at the base of this post (not the top).

But the nature of the problem is quite technical and I am still working it out myself, as it appears the section interacts with the sections of the Social Security (Administration) Act 1999 (Cth) (eg, s 43, which concerns daily rates) and also overlaps with other matters arising from common law statutory interpretation, such as whether the person is receiving ‘regular payments’. (There is some common law authority on this point, which I have not examined in detail as of yet.) However, the basic lineaments of the problem look to be most significant. I will come back to some other unexamined source on the problem below.

Regardless of the technical basis of the 1073B problem, the reality is that it is very likely many people were prosecuted for incomes affected by apportionment. The third recommendation mercifully envisages SA taking the initiative of designing an approach to this fundamental problem of criminal justice with DSS and CDPP. Unlike what happened in response to the problem of an estimated 15,000 welfare convictions following the Keating v DPP (Cth) case.

On Twitter I once offered commentary about why I felt there was a connection between the Poniatwoska and Keating matters and the origins of robodebt. I will not repeat those thoughts here, but provide a link to the thread below:

https://twitter.com/chrisrudge/status/1655465107315621890?s=20

One of the most regrettable aspects of criminal justice history, to my mind and knowledge, is the fact that up to an estimated 15,000 people (on the government’s own estimates) are likely to have been innocent of any crime but nevertheless were prosecuted and convicted under s 135.2 of the Criminal Code between 2001 and 2011 for Centrelink overpayments (welfare debts). This is not to say that these debts did not exist; however, as this 1073B issue now makes clear, many of them may well have not existed or be much less than they were. Thus, there may have been a double injustice for some of these 15,000 who were prosecuted and convicted for omitting to report their income to Centrelink and faced conviction. Not only would they have been wrongly convicted for a non-criminal omission to report their income (it is accepted that reporting is required, but it was not certain, so said the full High Court in Keating, that in all circumstances, such an omission would have been a criminal offence. (Indeed, in those cases, based on the way the evidence was adduced and briefed to the CDPP, it likely that most would not have been capable of being convicted on a proper application of the law to the evidence — that is, prior to the operation of s 66A of the Social Security (Administration) Act 1999 (Cth).)

But setting aside the detailed facts of the Poniatowska and Keating cases, the point is that the CDPP cannot, once more — as was the case following the Keating decision of the High Court — simply be expected to review files on request. In other words, the CDPP should not merely act in a passive capacity and wait for any request from any individual who suspects that may have been affected, or their debts may have been affected, by putatively unlawful income apportionment under s 1073B. No. Instead, the CDPP should play an active role and, together with the agencies, identify the number of prosecuted and convicted people affected (or whose files are affected), and follow up on a case by case basis.

It was alarming to discover during the robodebt RC that the CDPP had created a checkbox to organise any Keating-affected matters into one bucket so that they could keep track of appeal requests but that they had not informed anyone proactively that their matter might be Keating-affected or, in respect of those who already had identified it themselves, that they were one of a group. It was also concerning for other reasons, as previous FOI requests to the CDPP about Keating-affected matters had indicated that there was no record of how many there were. I have raised this before (see thread below):

https://twitter.com/chrisrudge/status/1636728632826552321

In any event, the point I am at pains to make here is that the CDPP should take a proactive role in identifying how many people’s convictions are affected by the s 1073B problem and set up a remediation process for them. It should resemble the Gordon class action process but be run entirely by the welfare agencies, the CDPP, and an indepndent oversight body, who can ensure that the integrity of the remediation process is unimpeachable.

Recommendation 4

The fourth and final recommendation flows from the third. In one sense, it is entierly banal. The Ombudsman simply recommends that SA and DSS provide the CDPP with:

• a copy of any legal opinion the SG may proffer in the future (which would presumably addressing the s 1073 rule problem);

• copies of any and all draft and finalised legal advices that have been obtained by the agencies in relation to apportionment under s 1073B (the daily attribution rule); and

• a copy of the Ombudsman’s finalised report, which would presumably encompass both the short Lessons in Lawfulness report, any version of the report that is more detailed, and any copy of the report that arises as a result of the second investigation (this is the ‘investigation 2’ mentioned on p 1 — the investigation that is not just into the lawfulness of certain apparent misuses of s 1073B, but into what has been done to remediate this problem as it has affected different parties, including customers).

Clearly, the provision of these documents to the CDPP, in circumstances where the matters impact on the past work of the CDPP and the lives of those who were prosecuted and convicted, should not really need recommending. Nevertheless, given that the CDPP are probably the only agency among the three who can identify the extent to which the legal problem identified by the Ombudsman in the Social Security Act 1991 (Cth) ‘translates’ across to the offence provisions of the Criminal Code — and most commonly s 135.2 — it will be crucial for the CDPP to have all the legal advices it can get to determine the best reading of the law and then, presumably, to undertake its own legal researchers into the way in which the defect in administration impacts on the prosecution processes and the evidentiary bases of those prosecutions. But it cannot all be about the elements of the offences in the Code. The recalculation of any debt will have ramifications for and raise questions about past convictions and prosecutions.

If, for instance, a debt is reduced by thousands of dollars, could it be said that the same sentence (good behaviour bond, community service, gaol) would have been imposed? Would the same term of imprisonment have been imposed, if one was imposed? Would the person have been prosecuted at all? At various times, the ‘threshold’ for criminally prosecuting overpaid welfare recipients has changed. When prosecutions were at their peak, between 2006 and 2011 (roughly), it was often said that amounts as low as $500 would be prosecuted. But what if anyone who was unlucky enough to have been prosecuted for an overpayment of $500 was, following a review in light of the apparent s 1073B defect, discovered to have been not overpaid at all, or even underpaid? The implications are extraordinary to consider.

Section 1073B income apportionment and robodebt: Interwoven phenomena — not different things

As I research the 1073B problem, I am discovering much in the Robodebt Royal Commission (RRC)’s Document Library that already identified something like this problem, even in the earliest stages of the robodebt scheme’s development. For instance, ‘s 1073’ is mentioned even before the NPP was finalised in the context of DSS and DHS disagreeing about whether legislative changes would be or ‘might' be’ needed — and this appears in the evidence of the Royal Commission’s report itself. Section 1073, of course, is an entirely different provision to s 1073B; however, even the reference to a section only a few provisions removed from the infringed section indicates a general awareness of the problems that may be associated with the relevant sections related to ‘ordinary income.’

After all, s 1073 specifies when and how certain amounts may be taken to be received weekly over 12 months (and this would have instinctively been of interest to the deeming of ATO data, which comes in the form of a single figure, and was used to spread income across a whole year evenly as having been ‘received’ over 12 months). However, s 1073, on its terms, seems to be an obviously inapplicable and wrong provision altogether for the use for the apportionment of an annual figure (as in the ATO figure used in robodebt) to divide income into fortnights (the provision only permits weekly apportionment), as, for instance

1. it does not allow ‘ordinary income from remunerative work undertaken by the person’ to be subject to its operations; and

2. it does not allow apportionment of income from a lump sum into fortnighlty earnings, as was done through the EAT sheet in robodebt but, rather, only allows a lump sum to be attributed to weekly receipts.

There are other reasons the provision does not work either; however, in short, it is a red herring. For this reason, it seems odd that s 1073 should be referred to in the Royal Commission at the early stage of discussing the legality of the scheme, and it is possible that, in the below text, the DHS official used s 1073 in place of s 1073B. As is quoted on page 75 of the RC’s report:

Ms Golightly forwarded Ms Halbert’s email to Mr Withnell for comment, describing some of the changes to the brief as “a little frustrating.”

Mr Withnell replied, stating:

In relation to the legislative change – I got hold of Murray but he had left for the day so couldn’t give me specifics on which provision in the legislation – his only comment is that it relates to attribution of income. This suggests maybe Section 1073 but he had no idea really. I left a message for Andrew Whitecross which he didn’t return but I have just tried him again and he has just got back to me – his advice is that they believe that several provisions in Chapter 3, General Provisions Relating to Payability and Rates would need to be changed and possibly others. I think if just change their ‘would’ to ‘may’ and go with that? Or we indicate that DSS have advised that they believe that legislative change would be required. We will be unable to get a clear resolution tonight and we have flagged potential legislative or policy impacts elsewhere. The other key thing is that the measure could still be done using the existing arrangements – just at a higher cost…

Ms Golightly replied to Mr Withnell’s email and agreed with his suggestions.144 With respect to his statement that the process could be done under existing arrangements, but at a higher cost, Ms Golightly wrote that “…we might keep that up our sleeve for the moment in case we need it as a contingency later on.”

In addition to the many mentioned of s 1073B we see in materials quoted in or otherwise collected in the RC, the Report also contains many (seemingly something like 28) AAT1 decisions — extracted as a part of the RRC Report — that refer to improper or invalid income apportionment specifically under s 1073B. And so this cannot be said to have been a new problem, now. The AAT1 decisions begin identifying the oddness of robodebts and the way they seem to grate against s 1073B from as early as 2016. However, as I hope is clear, it is not just robodebts that would be contrary to s 1073B, but any debts affected by apportionment that extnds beyond the statutory ‘instalment period.’

In fact, while I am beginning to describe the technicalities of s 1073B, I will provide a link to my video explanation of the operation of the provision here. Please note the disclaimers about this being just an academic gloss on the section.

Section 1073B was identified as being unlawfully used in the Amato submissions in 2019

The other very important note to make is that the very defect now recognised by the Ombudsman seems to have been already recognised by the applicant in Amato, and was identified in the submission of the applicant in 2019. See paragraph 56.4 below:

As the submissions clearly state, ‘s 1073B do not authorise apportionment of the kind done under the EIC program.’ And so it makes no sense now, some four years later and following a Royal Commission into the very issue that was engendered by the Federal Court’s holding (albeit a settled matter in which the Court merely made a statement) in the Amato matter — namely, robodebt — to say that s 1073B is only just now being identified as a problem. Indeed, it was picked up by the AAT1 in 2016, as we’ll see below.

The remediation of robodebt may be all for nothing

This discovery potentially destabilises all of the remediation that has already occurred via the Robodebt class action process and may require reassessement of those debts already reassessed.

Speaking of robodebt, one wonders whether the category 3 and 4 robodebts — which required no remediation as the person had later cooperated with the agencies and provided additional materials (bank statements, employment records, and so on) — may yet turn out to be invalid debts due to the operation of s 1073B? Indeed, this problem would seem to necessitate the recalculation of all of the robodebts already refunded (after all, these debts could result in ‘underpayments’!) as well as the recalculation of all of the debts that were ‘validated’ through mechanisms such as additional ‘proof points’ being adduced by customers, and then subject to such tools as the Net to Gross Calculator and the ordinary EAT sheet (as a tab in the ADEX schedule).

Survey of AAT1 decisions making reference to s 1073B

To conclude the blog post, I now quote from the AAT1 decisions that make reference to the operation of s 1073B in the context of assessing robodebts:

• In November 2016, member Harvey wrote:

  As I noted above, the documents do not set out how the figure of $20,046.09 was reached. However, the ADEX documents suggest that the fortnightly income amounts that were taken into account were worked out by apportioning an annual amount evenly into each fortnight. The relevant Rate Calculator, in section 1067G, requires regard to be had to the person’s fortnightly amount of income and section 1073B requires that a person’s income in a particular instalment fortnight is taken to be earned evenly on each day in the fortnight [18].

• On 27 April, 2017, member Schuster wrote:

  The Tribunal found the Applicant incorrectly reported his income (at [[18]] ). It but also, however, found that: [the Applicant’s] correct entitlement to NSA must be worked out under the rate calculator in section 1068 of the Act. The income test requires the department to work out [the Applicant’s] ordinary income on a fortnightly basis. Section 1073B of the Act also requires that employment income earned over all or part of an instalment period is to be apportioned over each day in the instalment period (at [19]).

• In May, 2017, member Smith wrote:

  … the tribunal is not necessarily satisfied as to the amount of the debt said to be owed. This is because, in the Department’s calculations, it has averaged [the Applicant’s] income over longer periods than any single fortnightly instalment period. As the applicant’s income/hours of work fluctuated each fortnight, the approach taken by the Department in calculating the debts is not the correct application of section 1073B of the Act [14].

  While it would certainly have been easier for the Department to obtain total income information from the ATO and average it, rather than obtaining specific weekly/fortnightly pay information from [the Applicant’s] employers, administrative expedience is no reason to misapply the law [15].

• On 26 June 2017, member Horsburg found

  that, although the ARO did not note it, income averaging was used to identify and raise the debt; this was contrary to the requirements of the Social Security Act 1991 (Cth) s 1073B. In particular, the Tribunal found that apportioning income in this case ‘produced artificial and arbitrary results’ [9].

• On 14 July, 2017, member Ducrou wrote:

  … The Applicant believed that his income should not be apportioned over 14 days because he didn’t work weekends and public holidays (application of s 1073B).

• On 3 August, 2017, member Halstead found

  It is evident that in calculating the debt, the Department applied averaging of income received by [the Applicant] from the various employers through the 2012/13 and 2013/14 financial years. The effect has been that income was incorrectly attributed throughout the relevant period because [the Applicant’s] actual income from employment was not apportioned as required by section 1073B of the Act [12].

• and, again, member Halstead found in August 2017:

  The Tribunal found that the amount of debt determined for the Applicant for the relevant period was based upon a rate of carer payment that had been incorrectly calculated for the purposes of s 1064 of the Act because averaging of income was used rather than the application of s 1073B to actual periodic income. In such circumstances, the Tribunal was satisfied the debt under review was not correct and needed to be recalculated [9].

• and, again, member Halstead found in 20 August, 2017:

  The Tribunal noted that the authorised review officer did not refer to s 1073B of the Act in coming to the conclusion that a debt could arise from this method and stated that ‘[a]pportioning] earnings in that way is contrary to the provisions of that section and the Act supplies no alternative method such as that assumed by the authorised review officer’ [9]. The Tribunal stated the apportionment had ‘produced artificial and arbitrary results’ [9].

• and, again, member Halstead found on 3 Jan 2018:

  [it was] clear and undisputed that in calculating the debts, the Department applied an averaging of the income received by [the Applicant]. The effect has been that income was incorrectly attributed throughout the relevant period because [the Applicant’s] actual income from employment was not apportioned as required by section 1073B of the Act.

• and, again, member Halstead found on 7 Feb 2018:

  The Tribunal stated it was clear from the earnings apportionment record and ARO notes that in calculating the debts under review the Department had applied ‘an averaging of the [Employer] income for the 2005/06 year’ with the result being that ‘income was incorrectly attributed throughout the relevant period because [the Applicant’s] actual income from employment for each fortnight was not apportioned as required by section 1073B of the Act’ [10].

• and, again, member Halstead found on 14 Feb 2018:

  It is evident that in calculating the debts, the Department applied an averaging of the annual income received by [the Applicant]. The effect has been that income was incorrectly attributed throughout the relevant period because [the Applicant’s] actual income from employment each fortnight was not apportioned as required by section 1073B of the Act [11].

• by member Manetta on 12 Sep 2018:

  DSP is treated differently by the legislation. Whereas Step 1 in the relevant module containing the income test for youth allowance and newstart allowance calculations is: Work out the amount of the person’s ordinary income on a fortnightly basis. The same step in the DSP calculation, found in section 1066A, Module F1 of the Act (for persons under 21) is:

  Work out the amount of the person’s ordinary income on a yearly basis. The stipulation is the same for persons over 21 (section 1064). Sections 1073A and 1073B then mandate the averaging of that income.

• by member Halstead on 6 Feb 2019:

  While averaging of that income over the relevant period may approximate his earnings, that method is not an accurate measure for the purposes of section 1064 (the carer payment rate calculator) because the income was not apportioned as required by section 1073B for each fortnightly social security payment period [10].

• by member Horsburg on 7 March 2019:

  The Tribunal found Centrelink calculated part of the Applicant’s debt by apportioning the Applicant’s income as notified by the ATO, evenly across all the relevant weeks [9]. It stated: ‘This is contrary to the provisions of 1073B of the Act and I cannot affirm a debt calculated on that basis. Other parts of her debt have been calculated properly using her payslips’ [9].

• by member Horsburg on 8 March 2019:

  This process is contrary to the provisions in sections 1073B and 1073C of the Act. This casts into doubt the calculation for the 2010/2011 year. It may be that [the Applicant] has a debt for this period, but it will not be the debt calculated by this faulty method [9].

• by member Halstead on 8 May 2019:

  The Tribunal noted that the Department obtained taxation records for the Applicant from the ATO and then used that information to determine the debts now under review. The Tribunal stated it was clear that in calculating the debts, the Department applied an averaging of the income received by the Applicant, with the effect being that income was incorrectly attributed throughout the relevant period because the Applicant’s actual income from employment was not apportioned as required by s 1073B of the Act [10].

• by member Halstead on 1 July 2019:

  Averaging income throughout the relevant period does not accurately reflect his periodic income in the circumstances [9].\

  It is not correct to apply averaged fortnightly income in the calculation of youth allowance payments for the period [the Applicant] was employed by [employer]. The effect of using averaging is that income was incorrectly attributed throughout the relevant period. As a result, the income was not apportioned as required by section 1073B of the Act [10].

• by member Halstead on 7 August 2019

  The application of averaged income does not accurately reflect [the Applicant’s] actual fortnightly income in the circumstances. It is not correct to apply averaged fortnightly income in the calculation of austudy or newstart allowance payments for the periods [the Applicant] was employed. The effect of using averaging is that income will be incorrectly attributed, which means [the Applicant’s] actual income is not apportioned as required by section 1073B of the Act [9].

• by member Halstead on 30 Oct 2019

  The effect of section 1073B of the Act is that the recipient of a social security payment is taken to have earned one fourteenth of the total amount they receive from employment income during a fortnightly instalment period on each day of that period. This is referred to as the daily apportionment of earnings and allows the rate of austudy to be calculated based on a daily earnings amount [7].

• by member Halstead on 8 January 2020:

  …the Tribunal cannot be satisfied the use of averaged income accurately reflects the income received by [the Applicant] from employment on a fortnightly basis. It is not accepted that averaging income for the period he received newstart allowance, for which bank statement records are not available, has been correctly apportioned as required by section 1073B of the Act [12].

• by member Halstead on 19 February 2020:

[the Applicant] said her income from employment varied significantly from one week to the next during the second period because she worked on a casual basis. Averaging income does not accurately reflect her periodic income in the circumstances. It is not correct to apply averaged fortnightly income in the calculation of youth allowance payments for the period [the Applicant] was employed. The effect of using averaging is that income was incorrectly attributed throughout the relevant period. As a result, the income was not apportioned as required by section 1073B of the Act [15].

• by member Halstead on 25 March 2020:

Averaging income does not accurately reflect [the Applicant’s] periodic income in circumstances where her wages regularly varied. It is not correct to apply averaged fortnightly income in the calculation of youth allowance payments for the period [the Applicant] was employed and specific information about her earnings is not currently known. The effect of using averaging is that income was incorrectly attributed throughout the relevant period. As a result, the income was not apportioned as required by section 1073B of the Act [10].

• by member S De Bono on 16 April 2020 (notably, this holding is different to the rest; it appears to be erroneous with respect to the Ombudsman’s view of the provision):

income averaging was different to income apportionment and that income averaging is permitted in accordance with subsection 1073B(2) of the Act. Income apportionment is when income is evenly apportioned over a period of time in the absence of other supporting information other than the matched income data from the ATO; this is known as a ‘robo debt’.

• by member Horsburgh 18 May 2020

…Centrelink discovered a discrepancy of $12,155.64 between the gross income and the information from the ATO. Centrelink made no attempt to discover why this discrepancy existed. Instead, they apportoned the amount of $12,155.64 over the remaining period of the financial year from 15 April 2014 to 26 June 2014 and raised the debt under review [7].

During the period of the debt, [the Applicant] was not working. She had no earnings in that period. It follows from this that the debt under review is both arbitrary and unlawful. It is arbitrary because it was imposed without any investigation of the discrepancy. It is unlawful because it is contrary to the provisions of section 1073B of the Act, which does not allow such apportionments [8].

• by member Horsburgh on 9 June 2020

According to the Centrelink documents, the debts under review were calculated by using information provided by the Australian Taxation Office. This information was apportioned evenly across the relevant date periods. This method of calculation is contrary to the provisions of section 1073B of the Act. That section requires income from employment to be apportioned to the fortnight in which they were earned. That is to say, the method of calculating these debts was unlawful. It follows from this, that [the Applicant] does not have these debts. If [the Applicant] does not have the debts, no penalties can be imposed upon her. I set aside the decision under review. Any monies recovered from [the Applicant], must be repaid to her [8].

• by member AJ Halstead on 24 June 2020

  The amount of the debts determined for [the Applicant] for the periods where averaged income has been applied has not been correctly calculated in accordance with section 1068 of the Act. This is because the evidence does not establish that [the Applicant’s] income was the same from one period to the next and so an averaging does not reflect actual income for each social security payment period. In such circumstances the debt for those periods needs to be recalculated with reference to actual periodic income [9].

  The Department is to obtain detailed payroll records from each of [the Applicant’s] employers for which averaged income has been applied... In the event that information is not available, grossed-up amounts assessed from bank statement records obtained from the relevant financial institution for [the Applicant’s] bank account may be used. The amount of income received for each period specified in section 1073B of the Act is to be applied according to that provision. [the Applicant’s] actual entitlement to newstart allowance is to be calculated thereafter according to section 1068 of the Act [10].

I have not yet analysed all of the above quotations, but do get the sense that most of them, with at least one exception, identified that s 1073B is not lawfully applied for these robodebts, from 2016 through to 2020.

Thanks for reading this post and I will obviously await more news as this extraordinary legal story develops.