Local Audit and Accountability Bill

Written evidence from Karen Heath (LAA 06)

1 I respectfully request that the committee should consider carefully the NFI data mining provisions contained in the Local Audit and Accountability Bill.

2 As the NFI currently operates, thousands of innocent, entitled people are being inappropriately suspected of fraud, when no inconsistency has been found in their case. This causes distress and harm and is unjust. I note that the research note provided for Parliament repeats a position which the NFI has repudiated, namely that where a match is found this is a case where fraud is suspected [1] . The most recent NFI position on the definition of a match is: "where a match is found it may indicate that there is an inconsistency that requires further investigation". Such a definition opens the door wide to the use of data mining and other data analytics methods of highlighting people whose situation is perfectly regular for potentially intrusive and damaging investigations. These are fishing trips within fishing trips.

3 This represents an erosion of civil liberties; engages human rights and Data Protection Law, including Article 3 of the 1st Protocol to the ECHR; and it breaches principles of natural justice - all in a context where there is no right of appeal against the administrative decision to assume a person is a fraud. The decision itself may be secret. Suspects often can’t get the evidence simply because there is none.

4 Where councils are involved, especially if they provide taxpayers with misleading information about data mining, the unnecessary distress and harm arising constitutes a failure in the duty of care of councils regarding vulnerable adults, who feature largely on some NFI hit lists. Disseminating misinformation is also bad practice in terms of governance, public law, and the explicit requirements of some legal frameworks.

5 Prestigious organisations, including the Joseph Rowntree Reform Trust and Liberty, have expressed concerns about the NFI. For example, see the Liberty’s submission on the NFI provisions within the Audit Commission Act.

6 I agree broadly with the Information Commissioner’s response to the consultation on the draft bill. I commend this to the committee, especially the comments on clause 90.

7 Of special interest ate the two reports numbered 801/851 (the primary CT exercise) and 802/852 (rising 18s) in the 2010 NFI roll out. Some councils were advised that the NFI pilots probably involved electoral law offences, and refused to supply their registers. Some still refuse, I believe. I refer you to the Department’s own submission to the consultation on the present Code of Practice on this specific issue [2] . Electoral law is not, however, the only controversial aspect of these exercises.

8 Regarding interpretation of the output from the ‘primary’ CT data-mining exercise, I have a copy of a letter from Bob Neill while Under Secretary of State. He explains that he has told the Audit Commission that all interpretations should make it clear that this output does not show lack of entitlement, nor that the taxpayer had failed to provide information required by law, nor even maladministration by the council, as residents validly entered on the electoral register may fall to be disregarded. I agree with Bob Neill. What relevance, then, can this output have for auditors? I regret that the A.C. appears not to have replied or to have heeded his advice. Rather, when the matter was raised originally, it simply changed its FPN and brought forward proposals to revise the code’s definition of data matching.

9 More recently, Brandon Lewis has written to my MP, having apparently been advised that the primary SDP match excludes households receiving or claiming the discount on the basis of disregards, and includes only those claiming or receiving the discount on the basis that they live alone. As Bob Neill made plain, this is incorrect in fact and in law. The NFI has stated that it is ‘merely pedantic’ to point out that the people identified as ‘hits’ on this exercise are not ‘claiming’ to live alone [3] . Given that making a false claim would provide evidence of fraud, this is very far from being a pedantic point. Mr Lewis has also stated that the NFI is trying to improve the quality of the data sent by councils. It should be made plain that the thousands of abortive investigations do not result from councils sending in poor quality data, but from the fact that misunderstanding of the data has been built into the exercise from the outset. I have seen a print out of my own ICT records. These show me as having receiving a ‘disregard discount allowance’. There is no such thing as a ‘disregard allowance’. My legal position at the time was that I was receiving a discount ‘of the appropriate amount’ via a demand notice issued on the compulsory assumption that the same rate would apply on every day of the coming tax year. At the very time that this committee is debating this legislation the NFI is preparing to receive data relating to thousands and thousands of households which it misinterprets, and to use the electoral register to compile lists of people it wants to be investigated even though there is nothing at all discrepant or inconsistent in their situation. This includes my data, relating to my financial position, and I object.

10 This is of public interest: ‘scandals’ about entitled people being falsely suspected of CT fraud or threatened with cancelled discounts have hit the press. My council had to apologise publicly for its pilot, such was the outcry.

11 I will accept, for the sake of argument, that it might be necessary and proportionate to use data matching to identify actual inconsistencies giving rise to a reasonable suspicion of fraud. But I oppose data mining to indicate on statistical or similar grounds that a family’s situation ‘may’ involve an inconsistency. Nobody should be subjected to a fraud investigation on that basis. It appears that the majority of CT investigations are abortive i.e. no actual inconsistency, leave alone dishonest intention, is discovered, but prejudicial PR materials and inflated statistics threaten to overwhelm common sense and are adversely affecting local government administration. A letter from Bob Neill to the Guardian highlighted the latter issue.

12 Whichever approach the committee favours (evidence-based or evidence-free investigations in search of fraud) the decision should be clear within the statute, and fully debated in Parliament, and not left to whoever drafts the new code. Experience in Scotland shows that the following pattern is predictable: the NFI gets powers through on the basis that it matches data to identify suspicious inconsistencies, gets the power to write its own code, subject to limited requirements to consult, and then produces a document from which all limiting requirements which would prevent the use of data mining and other statistical techniques to identify people as ‘risks’ of being frauds for investigation have been systematically removed.

13 I point out to the Committee that it is the explicit policy of the Cabinet Office to use data mining to fight fraud: therefore, one can expect a similar process to take place here. One might reasonably suspect that knowledge of this is why the NFI was so happy to find a new home there.

14 In my evidence I have had to balance the need for detailed technical information to support my points against the need to be brief. What follows develops my point in some detail.

The inadequacies of the current code of data matching practice and current definitions of data matching

15 The present code states that the aim of data matching is to identify inconsistencies, that where a match is found it indicates that there is an inconsistency. The model fair processing notification (FPN) provided within the code contains the same definition.

16 However, this is not how the NFI works in practice. When this was pointed out to the NFI in the context of the CT reports, its response was a) to come up with a new definition of data matching b) to modify its FPN, asking participants to alter theirs [4] c) to draw up proposals apparently designed to eliminate any requirement for inconsistencies from the code [5] and d) to generally deny that any such limiting definition applied.

17 At this point the activity becomes not data matching, but data mining, an activity about which Parliament expressed concern [6] . The aim of stipulating that the code should be laid before Parliament was to prevent data mining. But as the ICO points out, the code lacks legal force.

18 Even code provisions intended to ensure data security have been waived. The NFI had regard to the code, but decided against complying.

What definition of data matching will apply when the Act is law?

19 The research notes to the Bill state that it includes a definition: it does not. It is not good enough to say that data matching ‘involves’ comparing data sets to see how far they ‘match’. Data sets contain different information for legitimate reasons. Differences are inevitable given the different frameworks governing our lives. The view of the NFI is that what makes an exercise legal is whether the exercise as a whole, including the subsequent investigations help to prevent and detect fraud. They deny in principle any requirement for a case by case justification. These ‘investigations’ are fishing trips, often masquerading as ‘reviews’ to avoid letting people know the truth about what is going on.

20 I ask the committee to protect the rights and privacy of innocent, entitled people by ensuring that a clear definition, such as the one in the present code, goes onto the face of the bill so that there is no room for manoeuvre or function creep.

21 Staff at the Department have said that the new code will contain a definition of data matching as identifying actual inconsistencies. However, the present provisions are also said to transfer the NFI to another owner. But if the Government intends to continue with the NFI as it is, it will need a different definition. If the proposals previously on the table within the Audit Commission are produced when the new code is designed, there will be opposition to any limiting definition within the code.

22 We may continue to see thousands of abortive investigations into people whose situation is perfectly proper, with adverse consequences in terms of loss of public trust and injustice.

23 Therefore, I ask the committee to accept the recommendations of the ICO in respect of the code.

Accountability and Transparency

24 We have a lack of transparency to the extent that in order to obtain the Audit Guide relating to the NFI reports in question I had to mount an F of I Tribunal Appeal, whereupon the AC provided the information. Given the emphasis on transparency in Audit Procedures, this was bizarre.

25 I ask the committee to add to the bill provisions ensuring transparency, such that the Secretary of State shall publish the full rationale for each exercise, showing how the data and metadata he demands derives from the relevant legal frameworks, and the full logic underpinning the requirement for individuals to be subjected to an investigation.

26 Such provisions would give people - and participants - a reasonable expectation of exercising the apparent right provided for in the present code of taking steps if they feel that uses and interpretations of data within the NFI are inappropriate.

27 The ICO or some new body needs powers to investigate and comment on data mining exercises.

28 We need a clear avenue for appeal and complaint.

Why do the same NFI exercises throw up huge numbers of ‘hits’ time after time?

29 The ICO submission comments on this fact. However, as a creature of statute he states that he cannot on any individual matching exercise. Even if a data mining exercise were predicated upon nonsensical assumptions, he could not act. This, I think, prevents him realising why some exercises routinely provide so many hits. I have one answer. In respect of some NFI exercises, there is no actual inconsistency. If you compile hit lists based on membership of a group perceived as high risk, you inevitably produce large numbers of hits every time and these high numbers will not reflect actual loss of revenue through error and/or fraud.

NFI Data matching in respect of a ‘sole occupant’ discount which does not exist in law.

30 As there is literally no such thing as a sole occupant discount, NFI figures for amounts of this discount allegedly found to be incorrectly claimed or awarded have to be taken with a pinch of salt.

31 The following account of CT law is consistent with a legal briefing produced for the NFI team by Leah Griffiths of the Audit Commission Legal Department in February 2009 [7] . This was in response to complaints to the Commission. My information on electoral registration comes from an Electoral Commission handbook.

32 Referred to in the law as ‘the appropriate amount’, and set at 25%, this discount applies on any day when, put briefly, only one countable adult has his or her ‘sole or main residence’ in the dwelling.

33 The law specifies who is disregarded, including apprentices and students, i.e. a very large proportion of newly adult voters and young people on the electoral register as rising 18s. These make up a large proportion of abortive investigations.

34 One understands why the nickname ‘single person’ discount arose. However, in recent years the term has give rise to a wave of misunderstanding and misinformation. I repeat, therefore, that the law contains no such thing as a ‘single person’ or ‘sole occupant’ discount as these terms are commonly misunderstood.

35 CT bills are issued at the start of the tax year. The law, reasonably, provides that these are estimates and sets out assumptions that the council must make a) before making any calculations and b) on issuing the bill. These include two assumptions about continuing entitlement to a particular amount of discount. Where a council has reason to believe that a 25% discount applies, it must issue the bill on the assumption that the same rate and amount will apply on every day of the coming year.

36 Any council issuing the bill on the assumption that, say, one adult and no disregarded adults are resident would be breaching of the regulations. The NFI encourages such maladministration via its guidance.

37 There is no obligation under CT law to inform the CT department about changes in circumstances per se, e.g. any new resident. The obligation of the discount recipient is to correct the assumption that the same rate would apply throughout the year. Merely telling the council that you are no longer entitled to a 25% discount discharges the obligation. I mention this because dishonestly failing to provide information required by law is fraud, so clarity is important. There no obligation to tell the CT department about changes that ‘might’ affect entitlement. Many councils say there is. Claim forms, press releases, and web sites and are common sources of such misinformation.

38 It is improper to assert that mid way through the tax year, a recipient of this discount is making any CT ‘claim’ about who lives in their house. Such assertions are made, to the prejudice of entitled taxpayers on NFI hit lists.

39 CT computer systems are set up so that discounted accounts are given at the outset a code such as ‘single’ or ‘disregard’. These codes may fairly be said to represent information the taxpayer stated to be correct on a certain date, presumably the date of application. But they must not be the basis on which the bill was calculated, or tax was demanded. Regulations 15 and 20 make this improper. There is no obligation on the council or the taxpayer to keep these codes ‘up to date’.

40 The CT definition ‘sole or main residence’ comes from the Act. Case law clarifies that in deciding sole or main residence no single criterion shall be used. This rules out using the electoral register to decide sole or main residence.

41 Some adults, including most students, may lawfully register to vote at more than one address. This makes it doubly illegitimate for a council to use its electoral register to decide sole or main residence.

42 Obviously, the electoral register cannot determine whether anybody is disregarded.

43 In short, one cannot tell by matching against the electoral register whether or not there appears to be entitlement to a 25% council tax discount or that people are claiming a discount to which they appear not to be entitled.

44 The AC appears to have sought legal advice on the use of the electoral register on the basis of this fallacy. A range of documentary evidence supports this analysis. The AC appears never to have sought counsel’s opinion on what it is actually doing. Had it taken the simple step of reading a few valuation tribunal appeals, or looking up the case law it would never have made this mistake in the first place.

What happens in the NFI so-called SPD reports?

45 The NFI insists that councils attach the inappropriate code ‘single’ to certain CT accounts and upload these codes for data mining. Where more than one adult is on the electoral register the family is supposed to be subjected to an investigation - despite there being no evidence of irregularity. It provides distressing model letters. Some councils have refused for various reasons, including financial ones, to investigate. Given that the output provides no evidence of maladministration, some have argued that the exercise is an unjustified attempt to interfere in how they work. When the code is changed from ‘single’ to ‘disregard’ the NFI claims an ‘incorrectly awarded or claimed discount’, neatly maligning both councils and taxpayers, multiplies the number by 3 and produces inflated figures for money saved for the public purse such as those in the explanatory notes to the draft bill. The majority of these people were entitled to their discount all the time, and no additional income whatsoever arose from the alteration of a code whose relationship to the law is misunderstood. The statistics are unprofessional and highly misleading. They are based on the ‘cancellation’ of a discount which does not exist and its replacement by another which does not exist. These are both the same discount, one of ‘the appropriate percentage’ as per Section 11(1) of the Act.

46 The results are misunderstanding, large numbers of abortive investigations, maladministration, successful tribunal appeals, loss of trust, and much needless distress.

November 2013