Late last month saw the release of the Administrative Appeals Tribunal ruling in FTXB; Secretary, Department of Social Services (‘FTXB’). This long-awaited ‘lead’ case (or test case) is a three-tribunal-member decision with Justice Kyrou sitting as President. It provides us with guidance on when and how unlawfully apportioned debts can be recalculated. The stakes were high. As we have covered elsewhere, the Department’s internal sampling exercises have confirmed an enormous number of files may be affected.
In this important case, the Department of Social Services (‘Department’) sought a convenient pathway to recalculate these legacy files. What it got was an unsteady goat track — something that can only be trod by the well trained, and at great cost. In the end, the Tribunal found it is possible to recalculate the hundreds of thousands of debts that have been unlawfully calculated between 2003 and 2020. But the Federal Government now faces several urgent ethical questions:
Will it proactively review files to ensure they are lawfully calculated? Or will it repeat its past failures — such as the robodebt ‘communications campaign,’ where over 150,000 communications yielded only a ‘few hundred bites’ from victims?
How could any reprocessing of aged debt files reflect the Robodebt Royal Commission’s strong warning about the forensic inequity of raising historic debts?
Will it identify and waive unlawful debts where the cost of recalculating them exceeds the value of the actual debt?
Will it target reviews, using algorithmic or machine-learning processes, towards the files at highest risk of substantial miscalculation?
Could it make an ‘adjustment response’ offer, whereby those who have been subject to an unlawful debt are offered an adjusted debt based on the average variance within their risk-profiled category of debt? The pragmatic use of partial waivers for administrative errors is not uncommon.
Will it find, waive or partially waive any cases where recalculation is functionally impossible — due to missing information, lost files or residual averaging?
Let’s first deal with what the Tribunal found; analysis of the questions of remediation outlined above will be deferred to a later post.
The AAT Hears a ‘Lead Case’ (or Test Case)
The matter was precipitated by an odd course of events. Both the Ombudsman and the Department have been clear that the past practice of income apportionment is unlawful. For at least seventeen years, Services Australia mishandled payslips when it did not have daily breakdowns of earnings or dates exactly matching Centrelink fortnights. The agency adopted an unlawful practice of spreading the total fortnightly amount across two Centrelink fortnights. It was unlawful because there was no legal basis for that practice.
The respondent in this case was one of the people affected by this process. He has now successfully had the unlawfully apportioned debt reduced from $911.98 to $806.16. When he appeared before AAT1, however, the Tribunal endorsed the income apportionment. The Department accepted it could not simply follow that ruling. It holds, and has already said it intends to follow, internal legal advice conflicting with the AAT1 decision. It therefore sought to ensure a settled position by concession, as occurred in the recent federal court matter of Brooks v Operator, National Redress Scheme for Institutional Child Sexual Abuse [2024] FCA 725.
Equality of Arms Breaks a Cycle of Silence
Government public relations professionals have been at pains to frame income apportionment as ‘different from robodebt’ and ‘super-duper complex.’ But this giant pile of debts carries some deeper lessons — not least, that social security law is prone to silences that ensure that unlawful Centrelink practices can go unquestioned, often for years.
This was not the first time the Department had put its arguments to the AAT2. In the 2022 case of Judd, the Commonwealth government’s preferred interpretation of s 1073B — the ‘income received’ method — was affirmed. The government appearances for those hearings included a senior counsel, four solicitors and an EL Director. The Tribunal recorded that:
As the Applicant was self-represented, the Tribunal varied the usual order of proceedings, so that the Applicant had the benefit of hearing the Respondent’s opening submissions first, after which she was provided with an opportunity to provide her opening submissions to the Tribunal.
Let us hope such scenes become less common with the launch of the new Administrative Review Tribunal (ART) next month. Reforms such as the ART Guidance Panel will not, however, be a panacea for the everyday and frontline challenges of dealing with Centrelink systems. Consider the fact that the timeliness standard for simply getting a proper explanation of a decision is currently met in only 21% of cases. We must remember that only a tiny faction of experiences within the social security system ever receive ventilation in the public Tribunal (AAT2). It is vital to capitalise on opportunities for oversight. The monstrous scale of income apportionment itself underlines this.
Delivering on the rule of law and access to justice in this opaque system has never been more imperative. In FXTB, those principles were promoted by the efforts of three barristers: Kateena O’Gorman SC, Laura Hilly and Tim Farhall. Their appearances were facilitated by the Victorian Bar Pro Bono Scheme. They were instructed by the firm Arnold Bloch Leibler, which also represented the respondent on a pro bono basis. Pro bono representation for the respondent was facilitated by Justice Connect. Their work has ensured a well-developed decision informed by, and responsive to, the best versions of the legal arguments on both sides.
Of course, it is only in test cases like FXTB that pro bono schemes of this nature are available. And it is worth emphasising that Economic Justice Australia this week called on Government to adequately fund social security lawyering in this country to avoid future ‘total system failures.’
The Department’s ‘Right to Elect’ Approach Goes Down
The case boiled down to how the Department of Social Services was lawfully permitted to interpret and apply a core statutory phrase in the social security legislation: namely, the phrase ‘ordinary income is to be taken into account in the fortnight in which it is first earned, derived or received.’ This phrase was contained in s 1073B of the Social Security Act 1991 (Cth) (although it was removed in 2020).
Where a debt has been unlawfully income apportioned, the original problem was that some aspect of the employment evidence (eg, a daily breakdown of earnings) was not available to the Department to use in their calculations. The Department argued that, in such circumstances, it was permitted to simply change its calculation method in a way that calculated the ‘ordinary income’ of the employee based on the date the income was received. After all, the Department did not have relevant or precise details of when the income was earned, but it could work out when it was received.
The Department argued that the key phrase, which said ordinary income could be ‘taken into account in the fortnight in which it is first earned, derived or received,’ could be read as if it created a discretion allowing the Department to choose the method in situations of uncertainty. The Department thus had a right to elect how it calculated the debts. The law, the Department contended, did ‘not mandate that a single method of accounting be used and does not require the Secretary to select only one of three accounting methods’ [125].
In other words, the Department opposed any approach that involved requiring Services Australia (also known as Centrelink) to take extra steps to confirm the date upon which income was earned. As it argued:
a construction which required… the Secretary to obtain better evidence of facts that were within a social security recipient’s knowledge ought to be rejected as an impractical construction that does not accord with the express terms of [the legal test] [116].
The Tribunal rejected this approach. It found there was no right to elect which form of income could be used to calculate a debt (ie, earned income, derived income, or received income).
Instead, the provision required that precedence be given to one of those categories, and that the first-in-time had to be given precedence:
Where more than one of the three verbs applies to the income amount, point H23 requires determination of which of those verbs is first to be engaged. [149]
The key question Department will now have to explore is the number of cases when the date earned preceded the date received.
The Tribunal’s construction depended on interpreting the word ‘first’ as a kind of instruction to the Department about when, in a temporal sense, ordinary income was to be ‘taken into account’ in circumstances when it is ‘earned, derived or received’ in different fortnights. As the Tribunal said,
For the purposes of point H23, it is clear that ‘first’ has a temporal connotation in the sense that, if ordinary income is ‘earned’ ‘derived’ or ‘received’ in different fortnights, it must be taken into account in the fortnight in which the first of those events occurs. [130]
The Tribunal then warned that Department’s nominated approach:
carries a risk that recipients of social security benefits will receive less than the amount to which they are entitled or will be pursued for debts that are either not owed or are for amounts that exceed what is owed [178].
This is concerning given that the Department’s preferred method — which has now been rejected by the Tribunal in this case — is likely to have been applied to some ‘in-train’ debt reviews that were unpaused only a few months ago. Of course, this is not to mention the many thousands of debts likely to have been calculated in this way since 2003. In any case, when the decision of FTXB was handed down last month, Services Australia amended its apportionment webpage to ‘note’ the ruling in FTXB and to advise it is seeking to ‘understand the impacts of the decision.’ In short, the Department is trying to work out what to do next.
The Tribunal did not just reject the government’s arguments, however. It also rejected some of the core elements of the respondent’s arguments (ie, the arguments of the social security recipient). One of the facts in FTXB was that the respondent’s income had been paid in accordance with an hourly rate. The Tribunal was not satisfied that, just because income had been determined by an hourly rate, it could be understood as having been ‘earned’ upon the completion of each hour’s work. That is, the Tribunal rejected the idea that ‘the income would be ‘earned’ when the employee’s side of the bargain was completed’ (even if the employee did not have an entitlement to sue for payment at that time) — a second argument advanced by the respondent.
Instead, the Tribunal adopted its own approach to determining when income is earned for the purposes of s 1073B. This approach focuses on identifying the contractual arrangements between the employer and employee as to when a full pay period is completed such that a person becomes entitled to a payment. Arrangements could easily stipulate that payment is to occur that day or at a separate, later date.
Discerning the Recipient’s Employment Arrangements is Essential
Ultimately, the Tribunal adopted the following interpretation of ‘earned’ income:
[T]he time at which a person earns ordinary income… is when that person becomes legally entitled to the income. That time may not necessarily coincide with the time at which the income is payable. That is because a person may become legally entitled to income on day one (and thus earns the income on that day), with that income to be paid on a future day. [136]
This holding immediately brings to attention the fact that evidence of the contractual and payment arrangements between employee and employer will be essential to calculating when income is earned.
In its reasons, the Tribunal also warns against making assumptions or conducting a superficial analysis that emphasises individual days in the calculation process. It warns that in many cases potential remuneration can be accrued daily or hourly but only ‘crystallises’ into an entitlement (and therefore only becomes ‘earned’ ordinary income) at the end of the person’s pay period:
… ‘earned’ cannot be equated with ‘accrued.’ A benefit may be accruing on a daily basis without giving rise to a legal entitlement to be paid that benefit at the end of each day. For example, if a worker is entitled to be paid wages for each day of a working week at the end of Friday, the worker may be accruing a right to be paid for each day worked at the end of each work day but the entitlement to be paid does not arise until the end of Friday.
One does wonder whether this statement raises fresh issues for debts which have been calculated from detailed timesheets submitted by recipients. Were some of those debts impermissibly calculated from mere accruals of hours or days and not ordinary income ‘earnings’ as now been defined by the Tribunal?
The overarching message is that care must be exercised to ascertain the specific arrangement of employment so the decision-maker can pinpoint the point at which hours worked or accrued crystallise into a legal entitlement to be paid. According to the Tribunal, it is at the time (and only this time) that ordinary income may be ‘earned’ under s 1073B.
The Department did not support the Tribunal’s approach. It was concerned that it would permit people to benefit in situations where they had negotiated arrangements that let ‘accrued hours worked’ build up before a later, larger payout would be received by them. The Department submitted that a person could defer their legal entitlement to be paid; the deferrer could thus receive social security benefits in the period during which they accrued unpaid hours of work. When they were eventually paid, the deferrer’s reduction in social security entitlements would be confined to those fortnights in which they ‘earned’ their income. The reductions in social security benefits could thus be minimised.
This was not the first time the Department raised this issue; it has long emphasised this problem. Indeed, it seems that this hypothetical prospect of ‘fraud’ by artificial deferral that is at the heart of the entire saga of s 1073B.
The Department argued against any interpretation of s 1073B that required Services Australia to obtain evidence of a particular contractual or employment arrangement between recipient and employer. However, as the Tribunal found, s 1073B appears to demand this very thing — that decision-makers obtain knowledge about this legal aspects of this relationship:
The time at which an employee becomes legally entitled to wages will depend upon the legal arrangements that apply to the employment relationship. Those legal arrangements may arise from a contract of employment (whether oral or in writing, express or implied or inferred from a course of conduct) or may be governed by a legal instrument such as a statute or an industrial award. [139]
It appears the Tribunal holding also requires the Department to confirm the date on which recipients were paid by their employer and to ask whether this date is the same date on which a recipient ‘earned’ the income. That is because the date (or time) on which a pay period ends may differ from the day or time on which wages are actually paid. Problems might arise where no document — no contract, no statutory rule, or no award — can specify when an employee earns their income. In that case, inferences may be necessary to draw from what evidence is ascertainable. As the Tribunal writes at [157] (emphasis added):
Where an employer complies with the law (including tax law), a precise date upon which an employee receives a wage will usually be capable of being ascertained from the employer’s business records and documents in the possession of an employee, such as payslips and bank statements. A precise date upon which an employee earns a wage (in the sense of becoming legally entitled to the wage) will be capable of being ascertained from the legal arrangements governing the relationship between the employee and their employer. The date will be readily identifiable if a written contract of employment, a statute or an award specifies it. If no document exists which expressly sets out the applicable arrangements, in many cases, those arrangements will be capable of being inferred from other documents.
The Tribunal states — with striking confidence and certainty — that ‘it is unlikely that there will be any cases where a conclusion could be drawn as to when an income amount was received but not as to when it was earned or derived.’ It then provides six case examples, which we will discuss in a later blog post. That later post will be focused on the remediation pathway and the complex, knotty issues in recalculating debts subject to unlawful calculation inconsistent with the methods discussed above.
In the final lines of the paragraph already quoted above, the Tribunal identifies the method to be adopted in circumstances where no documentation ‘sets out the applicable arrangements’ of employment and when ‘those arrangements’ must be inferred from other documents. As the Tribunal states (again at [157]),
Such other documents include payslips and bank statements. Further, oral evidence from an employee (or another person with whom they worked) about a workplace’s practices and arrangements concerning rostering, timesheets and pay cycles may corroborate documentary records or indeed be sufficient on its own to identify when an income amount was earned.
One of the difficulties with this method is its generality. What method should be adopted, for instance, to ‘gross up’ net income records that appear on a bank statement to ascertain the gross income that was earned by the recipient? Problems with existing calculators have been discussed on this blog before. Moreover, to what extent can oral evidence from a recipient’s co-workers be relied on in determining when income was earned — especially in circumstances where such investigations may need to be undertaken many years subsequent to the employment time?
A Fair Call? Any Unresolved Issues?
Reading the decision, the most unsteady part is how the Tribunal treats the concept of ‘derived’ income and the meaning of ‘legal entitlement.’ We agree with the Department that the core reason the three forms of income — ‘earned, derived and received’ — were created was to ensure no one gamed the system by structuring arrangements in their favour. The concept of ‘derived’ employment income could have important uses in cases where (for example) a seasonal worker in a family business deliberately delays their pay or entitlement until the end of a season or a contract. Would such an act involve ‘gaming’ the system or merely be arranged for convenience’s sake? Or consider a worker who works only short hourly shifts of a few hours at a time in a small business. That employee might only request their accrued pay to be tallied and paid when it reaches a reasonable threshold (ie, more than $100) for reasons of convenience and reduced administrative paperwork. This would surely not be an example of ‘gaming’ the system?
In its reasoning on the meaning of ‘derived’ income in s 1073B, the Tribunal (at [147]) first raised the prospect that the term could have been used to apply to forms of income other than ordinary pay, such as gifts:
… it may well be the case that the word ‘derived’ is included… to cater for particular types of ordinary income which are not of a kind that can be described as being earned. It is not necessary to form a final view on this issue.
One difficulty not seemingly contemplated by the Tribunal is that s 8 of the Social Security Act 1991 (Cth) contains a definition of ‘income from personal exertion’ that itself includes the verb ‘derived’:
‘income from personal exertion’ means an income amount that is earned, derived or received by a person by way of payment for personal exertion by the person but does not include an income amount received as compensation for the person's inability to earn, derive or receive income through personal exertion.
This definition seems to contemplate that someone can derive employment income by effort for the purposes of the definition of ‘derived.’ As gifts and other ‘unearned’ forms of payment are not understood to have been ‘derived’ from ‘personal exertion,’ it would seem the Tribunal’s conjecture as to the possible meaning of derived may be questioned.
Ultimately, the Tribunal concludes that there is one core, common element to the verbs ‘derived’ and ‘earned.’ It finds at [147] that ‘what gives rise to the earning or derivation of the income amount is that the person has a legal entitlement to receive that amount.’ (Our emphasis.)
The respondents’ submissions gave strong emphasis to employment law cases where courts have already explored the possibility that a person may have a legal entitlement to recover money (or to demand payment) for work undertaken but before the end of a pay period has been reached. The Tribunal, however, put these cases to one side, distinguishing them from the questions raised on this administrative appeal. This elision, however, is arguably a source of untidiness. The respondents relied on the approach taken to this question by the full bench of the Fair Work Commission in a case concerning ‘modern award’ legislation under the Fair Work Act 2009 (Cth). There, the Commission seemed to recognise the possibility that accruing working hours may create an actionable entitlement to pay that occurs prior to the completion of a full pay period:
The period over which wages accrue is of significance because the employee only earns wages for completing each such period. Absent express provision for accrual, there may be a question as to whether an employee whose employment ends part way through a pay period is entitled to wages for the incomplete period. For example, if an employee is paid weekly and the employee’s employment ends after three and a half days of a weekly pay period, there may be a question as to whether the employee is entitled to any wages for those three and a half days. If the employee’s wages accrue for working a week, the employee will not be entitled to wages for the three and a half days (although the employee may have a basis to claim some other form of compensation). If the employee’s wages accrue on a daily basis, the employee will be entitled to wages for the three complete days that were worked in the pay period, but not for the half day. If the employee’s wages accrue on an hourly basis, the employee will also be entitled to wages for complete hours worked on the half day. [Emphasis added]
Despite the apparent relevance of this analysis, the Tribunal found the statements of the Commission (and related statements from similar cases) of limited relevance. That was because these cases
involved determination of unresolved issues about the legal rights of employees as against their employers in circumstances where employers were refusing to pay them for work that they had performed. [163]
Nevertheless, one could argue that these sources are relevant because they consider whether the act of working (creating a ‘unit’ of accrued earning) creates the earliest legal entitlement to payment. Why should the fact that such an inchoate or hypothetical entitlement may go unclaimed (because the employer later fulfils a broader entitlement) make a difference? Given that the word ‘first’ is used in the statutory provision — and given the importance of that word in the Tribunal’s analysis — shouldn’t the core task of the decision-maker be to identify the earliest (‘first’) moment at which the legal entitlement to a payment of any type arises? Despite the force of logic in this analysis, the Tribunal instead took the view that, where the sum was clearly earned and received on the evidence, an entitlement might only ‘crystallise’ where the pay period comes to an end and the legal entitlement thus arises.
The decision ultimately boils down to a constructional choice between a number of possible interpretations. As law teachers, we teach our students the methods of statutory interpretation; we recognise that, sometimes, the way statutory provisions are interpreted may be contested (and overruled by superior courts), including where the construction process can involve ascertaining the underlying purpose of the statute and parliament’s intention. The submissions in this case duelled over the standard of accuracy required by the social security law in calculation of payments (and debts). There is considerable reliance on the ‘practical consequences’ of each interpretation in this decision (and others). Indeed, the process of statutory interpretation often depends on ‘a balance of considerations’ and may often consider whether ‘absurd results’ follow from a given interpretation. Given that s 1073B contains multiple undefined verbs, and is ambiguous as to when or if any of them will ‘bite’ (apply), it was probably always unlikely that a neat and clear interpretation of the section, free of questions or uneven remainders, could be reached.
Lessons for the Road Ahead
For us, the statute was always poorly drafted and overly general. Arguably, its design was based on a perceived risk that recipients might take advantage of the social security system by artificially deferring their income payments; it was probably not drafted with the Department’s everyday business of calculating figures and making decisions in mind. It may have expected the internal processes of the Department to have filled the gap. Indeed, this is precisely what the Guides to Social Policy Law: Social Security Guide does. However, in creating a law with these three broad categories of income, and with an emphasis on a first-in-time event, the legislature has confused matters. In the drafting, the section appears to have placed undue focus on catching (or counteracting) those hypothetical recipients who might seek to avoid reductions to their benefits.
But we should not miss the bigger picture here. This decision in FTXB confirms that the core income test used to assess entitlement to social security benefits in Australian law was unclear for decades. In turn, this lack of clarity resulted in the misapplication of the test by officials and Tribunal members for at least seventeen years. Few things more clearly embody punitive conditionality than demanding people report income according to a method so unclear that it has confounded legal experts for so long.
The decision is also an important reminder to all government decision-makers: never treat decisions like widgets or easy currency. We need to value frontline decision-makers who have both the experience and ability to make inferences from evidence to ascertain facts — even when those inferences lead to conclusions that favour the recipient. This raises questions about how efficient or practical any process of recalculating these debts will be. It would take people of the calibre of Colleen Taylor to recalculate these files in a way that was fair and reliable — an army of such people. After years of not supporting staff, it may be assumed there is no such army.
In our next blog post, we will examine what lies ahead in terms of recalculation and what choices the Government might have to make around reviewing and rectifying it enormous pile of unlawfully calculated debts.
An interesting analysis. If I could make some points:
(1) The AAT was able to sidestep the whole issue of whether apportionment is lawful. The Ombudsman's view that it was unlawful because it used s 1073B is now widely discounted. The apportionment methodology has nothing to do with s 1073B. I haven't seen an argument that convinces me apportionment is either unlawful or factually inaccurate. The idea that apportionment does not accurately reflect a person's fortnightly entitlement misses the point that apportionment is used to calculate debts, not fortnightly entitlement. In the AAT2 case of Lyall, the Department argued its 'receipt view' was correct. And yet in debts totalling $18,000 the difference between its calculations and apportionment was about $120. Hardly a basis for inferring that no decision-maker in their right mind could think apportionment was ok.
(2) At paragraph 130 the AAT seems to think its interpretation of the fortnight 'first earned, derived or received', i.e. earned, derived or received, whichever event comes first, is consistent with the view of the parties that it means first earned, first derived or first received. They are plainly not logically consistent as the AAT's interpretation logically identifies one fortnight, whereas the interpretation of the parties logically identifies three fortnights (which may or may not coincide). I think the parties got it right but the Applicant's attempt to identify the relevant fortnight according to available evidence is a conflation of construction with evidence need to make findings w.r.t a provision already construed. The AAT's interpretation also misses an opportunity to provide an analysis that would be useful in dealing with pension cases where the word 'first' does not appear.
(3) As you've pointed out, the radical notion that earnings are to be understood as crystallising when there is a legal entitlement certainly raises issues. In one sense, the idea that you can go to a work contract to work out when there is a legal entitlement to earnings begs the question. Work contracts typically aren't about earnings. They typically concern work conditions, work hours, rates of pay and frequency of payment. Nothing about earnings per se.