Cross-posted from Independent Australia:
By substituting manual checking of data for a computerised algorithm in a typical Coalition cost-cutting exercise, Centrelink now faces thousands of man hours dealing with the debacle including a frenzy of class action suits. John Haly reports.
CENTRELINK HAS been fraudulently issuing debt notices to people who owe no money. Persons so identified are then harassed and threatened to the point that they pay this un-owed debt rather than being penalised by a system, which they already know actively disparages them.
Some 20,000 people a week receive notices of debts — allegedly to recoup incorrect welfare payments. Said notices are triggered by a controversial automated debt recovery system, now under intense criticism because of what is essentially (or intentionally?) the flawed logic of its computer algorithm.
An inherent incompatibility exists between Centrelink’s financial information and the Australian Tax Office (ATO) records. It’s a matter of timing. Centrelink has information about its payments made fortnightly, and possibly data relevant to jobs that clients were offered and accepted. Centrelink is unlikely to be aware of the continuing circumstances of that job or subsequent ones found independently in the course of any given financial year. The ATO has only an annual summary of income. There is no breakdown into weeks, fortnights or months. There is also no breakdown of pay rates, when it was specifically known they earned it, or what changes to income streams occurred in the course of the year.
The ATO data is, therefore, incompatible with Centrelink’s data. The Government is comparing apples with oranges.
Despite this, Centrelink’s algorithm takes your yearly income as reported to the ATO and averages it over each fortnight of the year. To assume absolute consistency for all fortnights is absurd on a number of levels. The only group that may get close to this pattern are the fully employed and, even then, there are allowances, overtime, uneven hours, holidays, sick leave, RDOs, wage rises, wage falls, changes of roles and any manner of occurrences that will alter the payroll for any individual over any given week/fortnight.
Certainly, the most unstable employment group – and the most likely to have variants – are the unemployed. This is common sense that if you are dealing with people who move in and out of employment in any given year, where they may move from poverty one fortnight to sufficiency (or, if lucky, excess) the next. It is also common sense that averaging clients’ yearly income will most likely produce inaccurate results by which to measure any given actual fortnight.
So what does Centrelink do? They take the actual fortnightly records held by Centrelink along with any limited volunteered data and try to cross-reference it against a fortnightly averaging of annual taxation income data. The normal presumption of statistical probability would tell you the likelihood of such figures matching for this demographic is extremely unlikely. You would have to presume the mismatches will be the most common occurrence. In the absence of specific information in Centrelink’s internal records for discrepancies, inquiries should be made tentatively as to why there might be aprima facie case for a mismatch in numbers.
The onus of proof should also be on Centrelink (and not the client), as the process is so obviously flawed. This is something fully recognised internally within Centrelink, if not by the political policy makers. In the face of the inherently flawed logic of this approach, innocence until proven guilty would be the legally and ethically prudent course of action.
Instead, the Turnbull Government implements a process that presumes people to be guilty (of debt) until proven innocent. As a result, some 20,000 welfare recipients a week have been receiving notices that they have 21 days to prove their “innocence”, or be hit with penalties. These include a 10 per cent debt recovery fee, gaol time, a restriction on travel. The event for which they are being investigated may be anywhere up to six years in the past. Some recipients are paying up, not because they accept that they actually owe the debt, but simply because they can’t locate evidence from past years, or because they fear the repercussions of a punishing government bureaucracy.
If you have ever had to deal with Centrelink or any of its private job network partners, you will be well aware of how punitive they are. Surprisingly to the Government – apparently – this is producing a backlash.
Minister for Human Services Alan Tudge insists the automated process is not flawed and, despite protests to discontinue the letters, he is forging ahead with gusto. For Tudge to claim to the ABC that he “wasn’t aware of anyone who was completely convinced they don’t owe money but have been given a debt notice”, is either a grotesque case of wilful ignorance or a lie in the face of a growing body of evidence otherwise.
The one aspect of this is the sheer workload this must be creating for Centrelink. Let’s assume Alan Tudge is correct that the error rate is only 20 per cent, which is contrary to what centrelink whistleblowers reveal is the case. Giving him the full benefit of the doubt, 20,000 letters a week represents 4,000 fraudulent claims a week. This amounts to 16,000 a month and 192,000 a year.
After 1.04 million data matching discrepancy letters in a year, they will not even cover the number of unemployed in this country (1.199 million), let alone all the other welfare recipients for other reasons. Alan Tudge expects the system to “generate 1.7 million compliance notices”, which by his own estimates means at least 340,000 letters in error. Of course, the Centrelink compliance officer whistleblower that spoke to The Guardian suggests the percentages of errors are vastly larger. Given that all of this was not only easily identifiable but unavoidably self-evident prior to the system being switched on, how is any of this not fraudulent?
At the current letter-writing rate (if they can maintain it) this will take over a year and a half to complete, although Mr Tudge thinks it will take three years. By then, the Australian legal fraternity will be in a feeding frenzy of class action suits with minimally, I calculate, 340,000 clients with legitimate grievances with the Government.
To keep on top of the “erroneous” case load – if Tudge is correct – requires the equivalent of 105 Centrelink officers processing each claim within an hour in a 38 hour week. This presumes an ability for each officer to address, research, confirm and redress an error on each letter in just one hour. But there appears to be mounting evidence it takes much more time. Plus it doesn’t factor in the equivalent of the 421 Centrelink officers devoting a single hour in a 38 hour week needed to process other claims.
But these figures are very conservative. The backlog of work is going to be extraordinary. No wonder it is so difficult to get through to Centrelink on the phone. It was nearly impossible to get Centrelink on the phone when there was only 20,000 debt recovery letters sent in a year, but now that they are doing it every week. It’s absurd. As for other means of communication, even compliance officers are complaining they cannot access the Centrelink online system efficiently, let alone customers.
In truth, even if Alan Tudge did call a halt, Centrelink will probably still be spending thousands of man-hours dealing with the consequences of this flawed and fraudulent system. The same would be true if the Commonwealth Ombudsman began investigating Centrelink’s debt recovery system and put a stop to it, disregarding the costs in legal redress, which are sure to follow.
Nothing about this course of action makes any logical sense, except to see this as class warfare against our vulnerable and easily disparaged citizens.
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Australia License