Public Accounts Committee
Oral evidence: Fraud and error stocktake: progress review, HC 490
Wednesday 20 July 2016
Ordered by the House of Commons to be published on 20 July 2016.
Watch the meeting http://www.parliamentlive.tv/Event/Index/70766935-20fe-4aa0-a0e1-0c900e100b02
Members present: Meg Hillier (Chair); Mr Richard Bacon; Deidre Brock; Caroline Flint; Mr Stewart Jackson; Nigel Mills; John Pugh; Karin Smyth; Mrs Anne-Marie Trevelyan.
Sir Amyas Morse, Comptroller and Auditor General, Adrian Jenner, Director of Parliamentary Relations, Max Tse, Director, Claire Rollo, Director, John Thorpe, Director, National Audit Office, and Marius Gallaher, Alternate Treasury Officer of Accounts, HM Treasury, were in attendance.
Questions 1-74
Witnesses
I: Sir Robert Devereux, Permanent Secretary, Department for Work and Pensions, Jon Thompson, Permanent Secretary, HM Revenue & Customs, and Nick Lodge, Director General Benefits and Credits, HMRC.
Witnesses: Sir Robert Devereux, Jon Thompson and Nick Lodge.
Q1 Chair: I welcome our new witnesses, Nick Lodge, director general of Benefits and Credits at HMRC, and Jon Thompson, the permanent secretary at HMRC. It is perhaps worth acknowledging that there is a new Minister for HMRC. We have already had a chat, Mr Thompson, you will be pleased to know.
I will kick off then. I wondered, Mr Thompson, what progress you have made in reducing fraud and error in personal tax credits. They have levelled off a bit this year.
Jon Thompson: Tax credit fraud and errors were reduced from 8.9% to 4.8% over the last six years.
Q2 Chair: But in the last year it’s been quite static, hasn’t it?
Jon Thompson: You are right. It is the same year on year between 2013-14 and 2014-15. I don’t think that means that we have been in any way complacent about that. Since March 2015 we have introduced a range of further changes and interventions—I am happy to go through those—that we think should make a further difference in 2015-16 and 2016-17.
Q3 Chair: I will ask you to go through those in a moment, but in broad terms—before we go into the detail—is it just that you have done the relatively easier stuff and now you are getting into more difficult issues? Is that the simple reason why it has levelled off?
Jon Thompson: It could be that. We publish what we think are the six key risks. I think you are familiar with those; they are set out in the National Audit Office standard report for HMRC for this year. We know we need to make progress on all of those, but particularly the No. 1, which is undeclared partners—I guess you will want to come to that in relation to Concentrix at some point. We need to continue to push on all fronts to try to get the level of fraud and error as low as we possibly can.
Q4 Chair: Okay. So how much further can you go? And do you perhaps want to give us some examples of the things that you highlighted and that you will be doing?
Jon Thompson: So, we think there are five areas where we can and have gone further since 2015: first, increasing the number of interventions that we make in relation to claims; secondly, changing and clarifying rules; thirdly, considering who we work with, most notably further working with the Department for Work and Pensions and the Home Office; fourthly, increasing the number of penalties and sanctions in the system; and lastly, we have a series of changes in 2016-17, including introducing new risking rules, bringing in RTI changes in-year and so on. We have a list of those changes that we have made since March 2015, which have produced the reduction to 4.8%.
Q5 Chair: So how much further can you go below that 4.8%?
Jon Thompson: I gather that the Comptroller and Auditor General believes that it is possible to construct a model of what is theoretically the lowest possible level that you could go. I was informed of that last week; I haven’t had the chance to talk to him about whether it is possible to calculate that. I don’t want to be complacent about it—it needs to be lower—but we also need to be frank that it is probably never going to be zero. So I’m dodging around your question—
Chair: You are dodging, Mr Thompson.
Q6 Mr Bacon: It doesn’t have to be zero. It could be immaterial, though, couldn’t it?
Chair: What would be “immaterial”?
Mr Bacon: Nobody’s asking for zero; what one is asking for is that it is not material.
Chair: Where are you aiming, Mr Thompson?
Jon Thompson: I think that we should aim to continue to get it lower—on that, I’m really clear. You know what the overall strategic Government policy is here: to eliminate tax credits eventually and for those benefits to be paid through universal credit. So for the time that we remain responsible for tax credits, we should continue to strive to make the level of fraud and error as low as possible. But I am not going to sit here and say, “I want to get it to 3%”, or 2%, or 1%”, because that would be a fool’s errand, frankly.
To be really clear, in 2014-15 Ministers wanted the level to be no more than 5.5% and we delivered 4.8%. The target was then reduced to 5%, which I think we will meet. And in April of this year the Chancellor agreed 5% for one year for us to do some further work and then re-engage with him on 2017-18.
Q7 Chair: You mentioned in a previous answer this issue about couples living together. The upper tribunal—well, the judiciary there—have made scathing comments, according to evidence we have received, about HMRC’s standards of decision making for the data matching and the very poor quality of submissions to the first-tier tribunal. We know that this is an issue for both DWP and HMRC in trying to assess what cohabitation is in a modern world, but do you want to respond to those criticisms from the upper tribunal judiciary?
Jon Thompson: Well, let me just say one thing, and then perhaps you would like to get Nick to give you a more detailed answer, given that he is the expert on this. The question “What does cohabitation mean?” is quite problematic, because my understanding is that it is defined as being in a loving relationship, and establishing exactly what that means can be problematic. We are exploring how you might exactly test that—
Mr Bacon: You just need a love-ometer, don’t you?
Caroline Flint: Is that your own relationship?
Jon Thompson: My own relationship is fantastically good.
Mr Bacon: You’re referring to your relationship with Norwich City Football Club?
Jon Thompson: Indeed, yes. That is a lifelong love.
That is the problematic definition; what exactly does it mean and how do you translate it? It is perfectly possible for people to be living in the same house but not to be in a loving relationship. People live complicated and interesting lives, do they not? We have to try to determine the truth of what exactly is going on in order to administer this benefit.
Q8 Chair: Perhaps Mr Lodge wants to answer that particular point?
Nick Lodge: As Jon said, the undeclared partner area and whether a claimant should claim as a single claimant or as a couple is very complex. We have done quite a bit of customer research with claimants and it is clear that the current rules are difficult for them to understand in some cases, depending on the complexity of their circumstances. Later this year we are going to trial writing to claimants, not in the sense of opening up an inquiry into their award, but in the sense of writing to single claimants to explain what the rules are, in the hope that they will recognise that they might be in a situation where they should be claiming as a couple. We are going to assess that and see if we should take it further.
In response to the points about tribunal cases, you are absolutely right that we have had some issues with the quality of submissions to tribunals. We are working very hard on two things: first, improving the quality—I myself am not aware of any very recent issues, although it is true that there have been some in the past—and secondly, the speed with which we get those papers to the tribunal, which is improving all the time. We have made some progress but I have no doubt that there is further to go on that.
Chair: It is always a challenging area for Government, operationally or policy-wise, metaphorically going into people’s bedrooms. Caroline Flint?
Caroline Flint: I’m good.
Chair: You’re good? Okay. Richard?
Q9 Mr Bacon: Actually, I do want to come in. I was thinking about something else. I would like to start by talking about what your expectation of the profile of fraud and error is, in the light of the switchover? Over the next five years where would you expect to be?
Jon Thompson: Sorry, can you define “profile”?
Mr Bacon: The proportions. I recognise that you said trying to adhere to a particular target and see it moving for your present need to administer tax credits is probably a fool’s errand, but are you expecting, once it is all done and you have moved completely over to universal credit, that fraud will be higher or lower?
Jon Thompson: I think that question is for you.
Sir Robert Devereux: To be clear, in a world where tax credits are finished, he won’t have a problem; I will. Will the amount of fraud and error in the system be higher or lower consequent upon making this transition? Our view is that it will be lower. In saying that, I need to be clear that we will be applying what I am going to call “benefit rules” to how we do this. We have just had a long conversation about monthly assessment and monthly calculation. The tax credits system down not work on that basis. A whole load of changes that I would class as a mistake are not classed as a mistake by my colleagues. They report a concept called “overpayments”, which is different in their terminology from fraud and error. In the real world, when they start to be my claimants, I am going to have to calculate that.
Our assessment, on a like-for-like basis, is that the design of universal credit will be such that the levels of fraud and error will be lower. That is why, in the business case, we banked a substantial sum for precisely that. A large part of it is to do with using RTI.
Mr Bacon: It is to do with?
Sir Robert Devereux: It is the use of HMRC data to check other people’s earnings and to determine earnings.
Jon Thompson: It is the use of the PAYE system, RTI, which is the live feed of PAYE from employers.
Q10 Mr Bacon: Obviously, we are talking about error rather than fraud. It is often customers simply making mistakes, and you are trying to do things to reduce the mistakes they make. The briefing talks about how just having a prompt to ask people whether they are claiming childcare vouchers, which is something they often forget about, can make a significant difference. What other things are you doing to reduce customers’ propensity to make basic mistakes?
Sir Robert Devereux: I have a couple that I could tell you about. In respect of universal credit itself, one thing that we know is that when certain circumstances occur, some other circumstances often occur, so we now prompt people. If they tell us about one change of circumstance that we think is normally correlated with another one, we say, “Would you also like to check that there is nothing else in this box to tell us?” That is one.
Another is housing costs. It is a complicated area, but we now make sure that we ask people what sort of tenancy they are in, in order to finesse the questions we ask about the nature of their rent. We have found that that makes a huge improvement in the way that people declare.
Thirdly, we have spent a lot of time on the declaration. At the end of all these applications you have got to sign to say that you have understood everything. We found—you can test this online—that people weren’t spending very long on this page, despite the fact that there were many paragraphs of text. We have worked with customers and our own people to reduce the amount of text, and we now find that people spend longer on it, so they have a better comprehension of what they are signing up to, including a responsibility to tell us about any change in circumstance. Those are just three areas where we are consistently trying to say, “What can we do to help you get your declarations right?”
Q11 Mr Bacon: The evidence so far is that it is producing larger underpayments and overpayments, rather than smaller. Admittedly, the sample is relatively small, but that is correct, isn’t it?
Sir Robert Devereux: Yes. So we have—
Mr Bacon: Sorry, was that a yes?
Sir Robert Devereux: I was going to say that we have published our first numbers, and the first level of fraud and error for universal credit is higher than it was for jobseeker’s allowance. I don’t think it is for the reason you adduced about people telling us their circumstances. If you go back and look at the individual breakdown that we published, you will see a couple of things. First of all, you can see very clearly that there is a positive benefit to having universal credit determined by the PAYE feed I get from HMRC. From memory—if I can get the right page—I think fraud and error in universal credit from earning is in the order of 0.6%, and I think it is one and a bit in JSA. Let me find the right page, because I don’t want to misquote.
Q12 Chair: Members of the Committee will have to go and vote in a moment. Can I go back to the issue of undeclared living together? HMRC has managed to reduce overpayments to undeclared partners, but the DWP has not reduced fraud and error related to living arrangements and housing. Why, and are there any lessons you can learn from HMRC?
Sir Robert Devereux: Fair enough. Let’s get the scale right on this story, first of all. We have consistently been reporting about £180 million in overpayments on living together for the last four or five years, and it has been stable.[1] HMRC have indeed made huge improvements—they have halved their level—but their level today is £345 million, so they are currently three times higher than I am because, in my view, they have got an in-work benefit, and in-work benefits have more changes in circumstance.
Q13 Chair: So this will probably tip over to you.
Sir Robert Devereux: It will do, but in terms of whether or not their improvement is somehow or other not reflected in my numbers, I have already got this to a good level. We are obviously doing lots of work on it, but one of the things we are particularly doing is making sure that whatever magic is being done in terms of data matching and all that sort of stuff in HMRC, we are using the same sort of rules.
Q14 Chair: Do you think it is just the wrong definition in the modern world? Would you have any advice to Ministers about setting a policy on this that relates to the practice?
Sir Robert Devereux: We have looked at precisely that, because it is a perfectly reasonable question. Is there a better way of doing it? We spent quite a bit of time on this with colleagues in the Cabinet Office and at HMRC. To be honest, we have not found a better definition. I know that sounds a bit disappointing, but you can construct relationships as economic relationships and end up catching brother and sister, or other sorts of arrangements.
In essence, the thing we are trying to do here is that the Government are determined that the amount of benefit that somebody in a couple needs to get is not twice as much a single person, for perfectly obvious reasons, but if they are not in that sort of relationship, you can’t. We have tried to think of other ways around it, but it has not proved particularly fruitful.
Q15 Chair: So are you saying that what you have reached now is your limit?
Sir Robert Devereux: No. I was answering the question about the definition. I don’t think changing the definition is going to make this any easier.
Q16 Mr Bacon: I have found the figures now: £36 million of payments and 7.3% overpayments for universal credit; and £1.4 billion of payments and 4.8% overpayments for tax credits. It was said of tax credits that it was inherent in the design that there would be overpayments. Is it not also true that it is inherent in the design of UC that there will be overpayments, so that the intrinsic problem is still there?
Sir Robert Devereux: I don’t think so, no. A couple of things are true—
Chair: Sorry, the vote is now at 4.28 pm—I did say it wasn’t an exact science. You have a bit more time, Mr Bacon.
Mr Bacon: You have been not saved by the non-bell. You have plenty of time, Sir Robert.
Sir Robert Devereux: To be honest, I am trying to remember the question.
Mr Bacon: I’ll restate the question. It’s really about the intrinsic—
Sir Robert Devereux: I’ve got it now. I’m going to tell you that I think it is intrinsically better. Why is that? Because even in a world of tax credits, people are still claiming housing benefit. The one intrinsically better arrangement is that I’m going to offer one benefit with one authority managing, rather than several. That is an improvement. Secondly, a system that is based on, “We’ll settle up three or four months after the end of the year and hope that the payments were right,” is not conducive to an attitude of, “Actually, I’m going to tell the authorities what my current circumstances are—Yes, I’ve got a job. Yes, I had a child. Now I’m a partner of this person.” The way the benefit system works is that it is constantly asking people to confirm it. When you go in to see your universal credit payment, it is saying, “Are your circumstances still the same?” month after month, not once in a year. All of those things have good properties, relative to tax credits—
Q17 Mr Bacon: Hang on. On that first point, when you are asked constantly to confirm that your circumstances are the same or whether they have changed, are you asked to iterate the various parts of your circumstances that might or might not have changed? Do you have to respond in the correct way to a number of different questions, or is it a universal “I confirm” click that nothing has changed?
Sir Robert Devereux: I will have to go and check, because I don’t know the answer. I think we play back to people the principal elements of what they have declared to us.
Q18 Mr Bacon: Can you confirm that in writing to us? That is really quite important. If it becomes a very simple, formulaic process in one click, you can imagine how people will just click and that’s that.
Sir Robert Devereux: I understand that. In part because of the nature of the benefit, we have a continuing relationship with people, because I am trying to get them into work and I am trying to get them more earnings. It is not just an income supplement, which is what HMRC are delivering. So you would expect that to be better.
In addition, within the rules we have made a number of changes that ought to make a difference. For example, we are saying to people who are self-employed, “You probably can’t be self-employed unless you’re earning at least X pounds. That can’t be a proper job.” There are a lot of people sitting in the tax credits system with very low levels of income—whether that is a real job, I don’t know, but it gets you out of being on JSA and having to look for something else.
We have made lots of changes. The long and short of it is that in the business case for universal credit—you read out all the benefits—one of the numbers that has not been changing iteration after iteration is the assessed benefit of how these rules drive down fraud and error, which in steady state is in the order of £1 billion a year.
Q19 Mr Bacon: So why are the overpayments higher, rather than lower?
Sir Robert Devereux: The short answer is because two things are going on. One is that you can see in the data we have published that because we are using RTI in the way we have said, the losses because of incorrect earnings have gone down substantially. It is 1.5% in JSA. It is 0.6% in universal credit. Before you ask me why it is not nought, that is because RTI only gives me employed earnings; it doesn’t give me self-employed earnings and it doesn’t give me people who are paid cash in hand.
Q20 Mr Bacon: I wasn’t going to ask you that. I was going to ask you about what the rest is that is not the 0.6%.
Sir Robert Devereux: I thought I would get that bit in first. The most substantial increase is in a box called “lost contact with the claimant”. When you proceed to unpack what is going on here, you find an interesting transitional arrangement to do with culture. In a world of JSA, you only get paid if you turn up for your fortnightly signing. If you don’t turn up, I don’t pay you—very simple. So if you get a job or you decide JSA is no longer for you, the system automatically corrects—“You’re not here. I don’t pay you—fine.” In a world of universal credit, I positively want to encourage people who can go from being out of work to being in work to continue to be supported by universal credit.
What we found by going through this very carefully is that for every 20 people we lose contact with, it turns out, when we chase them up, that around 17 of them have gone into work but they did not tell us. That is a cultural question, so we need to improve declarations, but universal credit has been paid exactly correctly: they were adjusted, we have got their earnings and we paid them properly—
Q21 Mr Bacon: Due to RTI?
Sir Robert Devereux: Due to RTI. Because of the definition.
But within those 20, we have found three people who are not in work. We do not know quite what it is that they are doing at the moment, because this data has been published only relatively recently. We are trying to explore what that is. But it is that piece—the people we have lost contact with—that we need to think about how to manage. That is what is generating the apparently high number in universal credit rates of JSA.
To be clear, it is in our gift simply to stop it by saying, “If you don’t show up, I will just stop your benefit”, but in a world in which 17 people do not need their benefit stopped, that is the wrong answer in the short term.
Q22 Mr Bacon: How do you know that there is not the possibility of large-scale, or even small-scale, criminal attack? I drew this to the attention of HMRC many years ago, when I had a very helpful brown envelope come through the post. It turned out that people were coming to this country, setting up an English or British job, with a British bank account, getting tax credits set up so that the payments were flowing, then returning to their home country elsewhere and withdrawing the money with an ATM card every week. The payments kept flowing, at least for 12 months. How do you know that that is not the reason you cannot make contact with them?
Chair: Sixty per cent.
Mr Bacon: This was a real example—it made page 2 of The Sun, Mr Thompson. I have yet to make page 3.
Sir Robert Devereux: One of the pilots we are currently running with the banks—actually in a pension credit space—is to see if we can detect when people are having a financial transaction in a foreign country, although the rules of the benefit do not allow you to be there.
Chair: Sorry?
Sir Robert Devereux: One of the pilots that we are just setting up now, with the banks, is to try to establish where somebody who is in receipt of a benefit whose presence abroad should be very limited—a week or so—is persistently having a financial set of transactions there. My expectation, if that turns out to be a thing that works, because the banking system can detect this, is that we would then use it in the same way with universal credit.
Q23 Chair: To be absolutely clear—correct me if I am wrong—60%[2] of the 7.3% of universal credit that is overpayment is because of the loss of claimant contact.
Sir Robert Devereux: Exactly.
Q24 Chair: So what is it? Is it three phone calls and one letter over a month—or is it six weeks—before you decide? How do you assess whether they are still alive? In order to check that—perhaps you will just run us through the process. Is it three phone calls and a letter? I cannot remember.
Sir Robert Devereux: The process at the moment—this is the thing that does require some looking at. It is not impossible that the policy needs a bit of thought in this area. The policy at the moment is that if you did not turn up, then on the face of it we may need to sanction you, because you are still out of work and you have not been looking for work. However, it is equally possible—given 20 years of JSA experience—that you have not got the message that not turning up and going into work requires that you tell us something. In my example I have 20 cases, but 17 are in work.
So the process that we are running at the moment is to be slightly careful not to throw the book at people immediately and cancel their payments—to go back to the same sort of character in Cornwall—but the work that we are trying to do at the moment is to say, “Hang on. Given that I want to get people into work, and I don’t want to cut off their benefit if they have done exactly the right thing, what is the best way to run a policy and process to get to a sweet spot here?” I do not want 60%[3] losing contact and sitting on benefit, but on the other hand I positively want all these people to go into work. This is a piece of the “test and learn” world. This has told us something. We are on it now, and I expect to be able to fix it. [Interruption].
Chair: We will now have to break because of the vote. We will not start again before 16.38, so you have 10 minutes if you need to lose any facilities. We had better go. Apologies.
Sitting suspended for a Division in the House.
On resuming—
Q25 Chair: Welcome back to the Public Accounts Committee. Sorry, that vote took a bit longer than we thought; we got the back of the queue there. We were just discussing the issue of the 60%[4] loss of claimant contact in the 7.3% overpayments of universal credit. I want to be clear that when you have gone through the process of three telephone calls and a letter to check if someone is real, you are counting that as fraud and error.
Sir Robert Devereux: I misunderstood you. I thought you were talking about the benefit process, but you are talking about the measurement process.
Chair: Yes, I was not clear. I think I was a bit anxious about the vote coming.
Sir Robert Devereux: You are right; that is the process. We try to get hold of somebody and then we write to them and, if nothing else happens, we stop the benefit.
To expand a bit more on what I was saying previously, I told you that for every 20 we found 17 in work. In absolute numbers, we had not got contact with 200 people, and 30 of them were the ones who have turned up in the figures. We ran through the process you described, and by the end of it we had made no contact with them, so we scored all the money as a loss and we also scored it all as fraud.
However, since then, because we are interested in this, we have found the 30 people. It turns out that we think that, at best, three are actually fraudulent, and most of the other 27 are self-declaring that they don’t want to be claiming benefit any longer—they are at home with their parents and that is their story. Our fraud people have looked at them and decided that they are not trying to spin us a yarn. That’s quite an interesting fact. Again, in JSA, if you decide you don’t like it and go and live with your parents, that’s fine—we stop your payments. In universal credit, we don’t.
Q26 Chair: What lessons have you learned from doing that work?
Sir Robert Devereux: There is clearly something about how quickly we can get to people to tell us, in real time, what they are doing. There is also the question of what processes we want to put in, because we are not chasing them the day after they don’t turn up.
The second point is more philosophical. Because of how universal credit is designed, by law, if you do not turn up on the second week of a four-week period, you are still entitled to the payment at the end. In the way the statistics have been done, all of that has been declared null and void, which is arguably not what the policy is. We have been working with Ministers on whether the policy is right and whether there is a way of making sure that people are at least telling us what they are doing. We then need to peel away the people who are genuinely making a mistake that isn’t actually putting any taxpayers’ money at risk.
Q27 Chair: What do you predict the impact of what you have discovered will be on your figures?
Sir Robert Devereux: Although the figures are higher than JSA and therefore not what I wanted to start with, they have not in any way shaken my confidence that I can get this to the number I thought it would be, in due course.
Q28 Chair: So in future you will not count them as fraud and error in such large numbers.
Sir Robert Devereux: To be fair, we are going to have to agree whatever we do with the National Audit Office, but I am telling you in almost real time what this analysis is. We have published the figures based on the three phone calls where we couldn’t get hold of them. We have done the subsequent work. We have found that there is this large population of people who have just chosen not to be there. We now need to decide the best way to classify that. We also need to decide, if it is true that you are legally entitled to this payment whether or not you have taken the last two weeks back, how that works. That is a bit of work to do, but my expectation is that whatever noise this little piece of the story has generated, we will be able to get that out of the system before we get to the roll-out.
Q29 Chair: I want to bring in Sir Amyas Morse.
Sir Amyas Morse: That all makes good sense—I did wonder whether there were some distortive factors, so that is interesting. I guess that some larger sampling and so on will be included in your forward action, just to understand what the error rate settles down to. You may well conclude that that 60%[5] is wrong and come out with something significantly lower. If that quite small—as you said—sample turns out to be indicative, it will be at a much lower rate.
Sir Robert Devereux: Yes. Most of what is being reported that is adverse, as opposed to the beneficial stuff that I have already told you is in the figures—better on earnings—is to do with loss of contact. If this loss of contact is not actually a problem, as I have just described, the numbers will go down.
Sir Amyas Morse: Sorry—I understand that and do not have any problem with any of that. It all makes perfectly good sense. I was saying that if the 200 is quite a small sample, you are not really satisfied with that as representing the population, so you will want to establish a more stable model.
Sir Robert Devereux: Okay, I get it. We have a sampling methodology whereby as more people are on universal credit, more people will be picked up in the sample, so it will be slightly bigger. In respect of whether I sampled enough people do get a decent estimate for universal credit, I think I did last time around, but I will have more people next time because there will be more people on universal credit.
Sir Amyas Morse: Good last-minute news to hear.
Sir Robert Devereux: Well, I told you we would test and learn, and that is what we are doing.
Chair: You don’t often get plaudits like that from the Comptroller and Auditor General in public, so I hope you are revelling in it, Sir Robert.
Sir Robert Devereux: I will write it down.
Chair: I’m sure it is on tape, which we can provide.
Sir Robert Devereux: I’ll play it back for my parents.
Sir Amyas Morse: He knows what I mean by that.
Q30 Chair: I’m sure you’re not resting on your laurels, Sir Robert.
Before I go to Nigel Mills, who is going to ask about pensions, I wanted to ask about pension credit, where fraud and error has increased from 4.6% to 5.6% this year. Can you explain that? How is your new strategy on pension credits going to start, hopefully, to drive that down?
Sir Robert Devereux: Okay. I will try to make this as simple as possible. It is a slightly complicated story, but you will follow it.
The big new thing that has gone on since we last met is the consistent use of RTI data for cleansing. We have 1 million referrals based on RTI data to go and check. In the world of pension credit that amounts to 86,000, so 86,000 cases have been checked.
When we first started doing this, the operational colleagues took those referrals and positively chose the highest apparently valued ones to chase after. They were finding on average a £35 to £40 adjustment to make, because it was a good target. That showed up in the 2014-15 statistics. If you remember the way our statistics work, we provide provisional ones based on sampling of the first half of the financial year, when we didn’t have any RTI, and then we do final ones in the second half, when we did have RTI. You will see, if you look at the figures, that pension credit fraud and error went from 5.2%, with no RTI, down to 4.6%. It is very clear that RTI is making a difference.
Then what has become apparent—we have done further work, as I have just described—is that at the start of 2015-16 our colleagues started to prioritise those referrals in a different way. There is another part of the policy system that we are trying to do, which is to bring to an end a thing called assessed income periods, where you are basically allowed to be on pension credit and not declare a change to your occupational pension for five years. They started to prioritise those instead of just the biggest value. The consequence of that is that the average correctness dropped from £35 or £40 to around £10. That meant that we were getting less efficacious RTI hits, not because RTI isn’t good, but because of the way we were choosing them. That then did show up. Part of the reason why you are ending up with a higher number is that we have not yet optimised the RTI piece there.
That isn’t the only story in pension credit. The other things that went up overall were around loss of contact and stuff to do with people being abroad. In pension credit, you are not supposed to be abroad. As I said to you earlier, one of the things we are trying to do to tackle that is the work we have been doing on data trials. We have done several of them, but in one of them the banks are trying to detect people we know have got a UK account with benefit going into it, but none the less the banks know—
Q31 Chair: When you say fraud for being abroad, how long do you have to live overseas before it is fraud?
Sir Robert Devereux: I am not going to be able to remember. I think, even on benefits, that we don’t restrict it to the English channel. You can be abroad for a bit, but it is for a limited period. I am guessing it is a week or two at most.
Q32 Chair: Are pensioners on pension credit going to be hammered because they’ve gone on a long holiday? I just want to be absolutely clear what you count as fraud.
Sir Robert Devereux: Let me write to you and tell you the rule. I would be surprised if it is very generous, because this is as means-tested benefit. If you have got enough money to have a long period abroad, then potentially we may think—
Q33 Chair: Well, except that in constituencies like mine, people have family abroad. They weren’t born in the UK, but they are British citizens.
Sir Robert Devereux: Let me just take you back to that. Whatever the law is, I will write to you and tell you it. There is a limit on how long you can be abroad in receipt of a means-tested benefit.
Q34 Chair: I have very poor constituents who spend chunks of time abroad. They plan to do so every three or four years when they can afford it.
Sir Robert Devereux: There may be a policy reason for thinking about the rule, but the rule is the rule and at the moment we are finding that too many people are being—
Q35 Chair: I remember that there was a change to the cap a few years ago. Perhaps you can write to me. I may want to pursue it as a constituency interest.
Sir Robert Devereux: I will certainly do so.
Q36 Nigel Mills: While we are on pensions, when we last looked at this we had an exchange about what would happen to the level of error and fraud in the state pension, given the change that happened three and a half months ago. As I recall, you were quite confident that that new pension would not adversely affect the level of error. Is it still your belief that there is no particular fraud and error risk to the new state pension?
Sir Robert Devereux: It is still my belief, but we are way too early to have any data. We only started on 6 April. At least in principle, the old state pension still had two parts to it—the basic state pension and the state second pension. The state second pension is an immensely complex thing. Even on just simple error, we would have had errors in that. It no longer exists, so by definition, so long as I know you have 30 years of national insurance contributions or whatever it is—ping—you get the answer. It is a perfectly well founded belief to imagine that the incredibly low levels of fraud and error on the state pension will be even lower in the future.
Nigel Mills: That was one ping, wasn’t it? There was quite a bit of media coverage of another ping: if you had been opted out for some or all of those 35 or more years, you would get a reduced pension, and there was a question about discrepancies between data at DWP on the length of the opted-out period and what employers had as opted-out data, which meant there was a question about whether the calculation of how much single-tier pension a lot of people were going to get was going to be done right. Is that something that you have looked at?
Sir Robert Devereux: I have not looked at any changes since 6 April because we haven’t got enough data to look at the changes. The problem you describe is a problem that existed in the old world and exists in the new world. Fundamentally, we have a view when we make our payments of how many years you have been contracted out based largely on my colleagues having the right records. It is very often the case that occupational schemes’ data is behind the event. One of the things that has gone on as part of the state pension programme is to give a short period—I think it is measured in two, two and a half years—in which all schemes need to get up to date as of 1 April in order that this problem, which has been a problem for a long time, goes away. We just need to know, based on your earnings and position up to 6 April, where you were, and that is work that is in train within HMRC at the moment.
Q37 Nigel Mills: I suppose what we were exploring last time was how, when you have a major change—you have a plan and a set of targets and you think, “This is new to the people receiving it and to us processing it”—some of these calculations are really quite complicated to get right. Is there any risk that the formulas have been misapplied or we have the wrong data? You might envisage quite a high error rate at the start, which you would review quickly and check that what was going wrong had settled down. I am not quite getting the feeling that we are in that strategy.
Sir Robert Devereux: That is partly because there is no new complexity in this. The problem you are talking about is a problem that existed in the old world; in the new world I need to know what your historic entitlement to the old state pension was up to 6 April. But with every passing week and month, that is now quite a long way in the past, and I will eventually clean it up with the occupational pension.
There is no new problem being created, because in the current year and the year after I do not need to know whether you opted in or not, because you can’t be. There is no new class of problem that has been created by abolishing the old regime. The pre-existing problem is there until such time as everybody’s records for the period up to 6 April 2016 have been properly brought to book. Nine times out of 10, by the way, it turns out to be that HMRC got the right data straight away and it is the pensions industry that was behind.
Jon Thompson: It is, of course, perfectly possible for you to log on to your personal tax account and check your national insurance history. All UK citizens can do that right now. It will also give you some estimate about when you are eligible for the state pension and how much.
Q38 Nigel Mills: I take it the hit rate on that is quite low.
Jon Thompson: We do get calls from people saying, “I’ve looked through my national insurance record and it appears that 17 years ago I have a gap. Can you give me some information?” We do get those curious people who want to put it right. It is good that we are transparent about that data. You can log on, see it and raise a query.
Q39 Nigel Mills: As of this stage, three months in, you are not thinking, “Oh God, there’s a problem with this new pension.”
Sir Robert Devereux: No, I’m not.
Q40 Nigel Mills: You are not struggling; you are quite relaxed.
We had a little exchange at the start of the last session about setting individual targets for fraud and error for individual benefits. I thought we would have some agreement that you were quite tempted by this and it might make the whole thing more meaningful from your end, especially around having perhaps separate targets for pensioner benefits, state pension, and the universal credit as we move on. The Treasury minutes did not suggest you were as keen on that as we thought.
Sir Robert Devereux: I wrote to the Chair on 8 July and in that letter I presented to you forecasts by four benefit groups. I don’t know if you’ve got that letter.
Q41 Nigel Mills: Is that a forecast or is that targets?
Sir Robert Devereux: What we have done here—just in case we get on to a conversation about net versus gross—I have shown you gross numbers and net numbers for universal credit as one class, all the benefits that universal credit replaces as a second, pensioner benefits, and then other benefits, which are primarily disability and carer benefits. I have shown you the actual recorded numbers in 2013-14, which is the base for the current period of the target, and I have shown you our current projections—I am using my words carefully here—about where I think each of those benefits will get to. I have not set them as individual targets, for a rather simple reason, which is that projecting fraud and error is actually rather more difficult than you may imagine. I am relatively comfortable that at the aggregate level I can make a sensible suggestion and set a target. I am being transparent about what my workings show, but just to be clear, the further down this barrel you get, the less likely it is that the precision surrounding each number is the same as the precision for the higher order number.
Q42 Nigel Mills: The reason why we were quite keen on those separated targets was that I think there was a bit of a feeling that you had hit the previous target because there had been an increase in the level of the state pension as a proportion of the total of the benefits that you pay, and that skewed the overall fraud and error rating. That was perhaps a little misleading, so we wanted to be able to measure on individual benefits, rather than—
Sir Robert Devereux: That is what I have given you.
Q43 Nigel Mills: But you are still saying that you do not want to set individual targets.
Sir Robert Devereux: We are arguing a little bit about semantics. I have given you 12 numbers that represent my best assessment at the minute of where each of the four classes of benefits will be. You can call it a target; I will call it a projection. The fact is you are going to hold me to account for it. I get that. I just want to be clear that you can ask ever more precise questions ever further out into the distance, and the chances of being accurate are getting quite low because this is not a precise science.
Q44 Nigel Mills: It is not unreasonable, is it, for us to be thinking that if you are projecting a universal credit net loss of 3.7%, you will presumably be doing things in the course of the next two years to try to make that less than you are now projecting? Presumably you would quite like that to be lower, and therefore perhaps you will be setting yourself a target to bring that in lower than you are now projecting it. That is, I guess, what we are after.
Sir Robert Devereux: Okay, so I have tried to be transparent here about what I think the combination of all the things I am trying to do over the period until I get to the target year will have achieved. It is not an opening gambit because I know I can beat it; effectively, the construct of the total number is based on the sum of those four sets of numbers. I am telling you that this is what I think all the work yet to be done is going to have to do to achieve this. It is not a sort of, “Oh, well let’s just pluck a number out of the air.”
Q45 Nigel Mills: Okay, so these are projections and if you end up bringing it in a lot higher than that, we can obviously hold you to account for that.
Sir Robert Devereux: I am sure you will.
Q46 Nigel Mills: That is the reason I was being a little unduly cynical, clearly, Mr Thompson. It just appeared that you had set a target for error and fraud in tax credits slightly higher than your current run rate, which made us think that you perhaps were not being overly stretched. I think the explanation was that you thought you would lose all your easy cases to universal credit and so you would end up with the harder ones where there was more fraud and error. Have I remembered that correctly or is the hot day addling my mind?
Jon Thompson: Let’s just be really clear: the target is for it to be no more than 5%, so it is perfectly right that we should strive to get the number as low as possible. We could argue that it should be set at 4.5% or 4%. Ministers decided and agreed on 5% for this year. We need to re-engage with Ministers on future years. That is where we are.
Q47 Nigel Mills: What do you think that should be, then? It seems a bit incongruous to set a target that is worse than you are currently achieving. That is not normally how you use targets.
Jon Thompson: I think you are misunderstanding. The target is not for it to be worse than it currently is; the target is for it to be no more than 5%. I have given you my commitment that we are striving to get as low as we can.
Q48 Nigel Mills: When you are hitting 4.5% or 4.6% or something, do you not think you would want to set 4% or 3.5% or 3% or something as a target?
Jon Thompson: Those options were available to Ministers, and they decided on no more than 5%.
Q49 Nigel Mills: When you engage with them for the next year, will you be saying, “Look, we’ve done 4.5%. Actually, we think 4% is realistic,” or, “We think 3.5% is realistic,” or will you be saying, “Actually, we quite like 5%, because life’s easy”?
Jon Thompson: I am perfectly content to engage with Ministers on future targets that are more demanding than where we currently are, but this target was set for 2016-17—for one year—at no more than 5%.
Q50 Nigel Mills: I suppose I have always found that what gets measured gets done, so if you set a target that is not very demanding or stretchy, it might not get done. You would think that these things would gradually reduce, wouldn’t you?
Jon Thompson: You seem to be implying that we might be complacently trying to hit 5%, which I have to say I refute entirely. I have been to Liverpool. I have spent all day with the team. I have followed claims from opening the brown envelope through to prosecution. I do not find an organisation that is in any way complacent about where we are with error and fraud.
I have set out a whole series of areas where we have changed since 2014-15. I continue to find a team that is very motivated to get this as low as possible. I was there sitting with the caseworker who discovered an £11,000 fraud through a data-matching exercise. They have on the wall of the Bootle office a target for the year of how much they need to save, which is £88 million. On the day I was there, it notched up to £64 million. This is a team that is incredibly motivated to try to do as much as possible, but I am at the minute with a target that says no more than 5%. As I have said to you, I am perfectly content to engage with Ministers on setting a lower target and I have every confidence we will beat no more than 5%. I can see where you are coming from.
Nick Lodge: I wonder whether I might just add something. Our track record does bear out that we will try to bear down on it further. Our target for 2014-15 was 5.5% and we achieved 4.8%, so there is definitely no resting on our laurels.
Nigel Mills: I guess we will look forward to seeing the fruits of your engaging with Ministers on what that new target should be.
Q51 Mr Bacon: Mr Lodge, while you are on, I had a case involving child payments to a single parent who allegedly had children in school. The parent without care came to see me to explain that he knew for certain that his children were not in full-time education because he knew where one of them was working and the other one had left school at a quite young age, yet because there was an assessment for child maintenance, HMRC was in a position where it couldn’t do anything other than reply, “We have to make this payment.” It took several years and, in fact, my raising the case with you across this Committee floor some years ago, at which point I am delighted to say the case was solved extremely quickly and my constituent got a significant sum of money back that he should not have been required to pay. It worked out very well. What interests me is, systemically, what would prevent that happening again now? HMRC was just unable to interact with the child maintenance people in a way that worked at the time.
Nick Lodge: I remember the case well, actually. I believe you wrote to the NAO and asked them to have a look at the processes that applied to such cases and we did do some work to make sure that we changed our processes so that we would liaise much more closely now with other agencies—the child maintenance agency. There are more robust processes in place. We have also changed the rules around full-time non-advanced education, which helps in general. It doesn’t bear exactly on the issue that you have touched on, but we now do check every year and ask people to reconfirm that their child is indeed in qualifying education and is remaining in qualifying education.
Q52 Mr Bacon: I think in that case the parent with care had been confirming wrongly that the children were in education, but then nothing else in the system worked, including my constituent’s pointing out that there were what the French Foreign Minister would call “lies” going on. That was the problem. The weight of the other evidence just was not put in the balance. Are you confident that now couldn’t happen in the same way over such a long period of time?
Nick Lodge: I would not say it could never happen, but what I could say is that we have definitely strengthened our processes and we have speeded up those processes to take those kind of assertions that are made to us from time to time forward quickly and to act on them. They are not always correct, of course. Where they are, I would hope that we are much better at taking the required action.
Q53 Mr Bacon: Okay, thank you. Sir Robert, can I just return to this business of claimants you have sort of lost? I think you said that once you have done the relevant number of phone calls and letters, you then classified it as fraud. Is that correct?
Sir Robert Devereux: Yes.
Q54 Mr Bacon: Just for the avoidance of doubt, do you also at that moment stop payment?
Sir Robert Devereux: Yes.
Q55 Mr Bacon: You do.
Sir Robert Devereux: I found a bit more information when you were voting, so let me tell you the story. I told you a story of 20 people who suddenly had a job. Let’s do it in whole numbers, because it makes more sense.
There were 200 people we lost contact with, and 30 of them were in this category. We went through the process that your colleague described, which includes stopping the benefit—the benefit was stopped. All that had happened, and that was the knowledge we had at the point at which the statistics were published.
Subsequently—after they were published—we went after the 30 people and there were only three of them who we had any fraud suspicions about. In future, we need to find a way to make sure that, if that is likely to be true in the real world, the statistics reflect it. This is not a fraud problem with 30 individuals; it is a potential problem with three individuals—an order of magnitude smaller—and another problem of people who have not told us. We need to think, “If you are going to go back and live with your parents, is that actually a mistake under the law?” We need to check that.
Q56 Chair: Sir Amyas wishes to come in here.
Sir Amyas Morse: I am just going to go back to HMRC. As we said in our standard Report, over the last four years the reduction in annual payment in regard to children has been very creditable: it has gone down by £520 million over four years. We recognise that that is quite an achievement. There have been other improvements, but that is probably the standout. We give credit for that in our Report, so it is worth—
Chair: I see incredulous looks. You have given two compliments on the record today.
Sir Amyas Morse: I am just giving balanced reporting.
Chair: Credit where credit’s due.
Mr Bacon: Let’s see if we can end on a bad joke!
Sir Amyas Morse: I am simply quoting from our existing published Report, actually.
Sir Robert Devereux: It is not new, then.
Chair: We will just pause for a moment before I bring Richard Bacon in.
Q57 Mr Bacon: I only have a couple more. This is to both Mr Thompson and Sir Robert. Can you give some indication of what you think is a realistic level of fraud and error by the end of this Parliament?
Sir Robert Devereux: My colleague has had a go at this, so let me have a go, because I have given it quite a bit of thought. In the annual report, as you know, we have diligently played back that in response to the NAO’s request we have been trying to establish the concept of the lowest feasible level of error. We all agree it cannot be nought, so it is perfectly reasonable to say “Let’s have the lowest feasible.” We have presented to the NAO four possible ways of getting to the lowest feasible number, and I have got some clever people thinking about it.
In practice, to be honest, we seriously doubt whether that produces a methodology that you, the Committee, or I, the accounting officer, will put any weight behind, because you have to make so many assumptions about what may or may not be the case as you progress towards the lowest feasible level that I do not think it stands up as a methodology. However, we have agreed with the NAO to have another go with it in September and if by some magic, despite my profound doubts, this assumption-based system is worth what it is written on, I will endeavour to write to you. But I am keen to make a decision by the early autumn: either this has some legs, and let’s diligently do it, or if it is in fool’s gold territory, let’s not spend any more time trying to do it. Personally, I fear it is in the fool’s gold territory.
Chair: Sir Robert Devereux, also known as Harry Potter.
Q58 Mr Bacon: There is a growing management theory that says you should scrap targets, or at any rate not worry about them, and instead spend more time concentrating on your understanding of how the work works, so that you get greater fluidity and economy of flow. By concentrating on that permanently and continuously, you get bigger improvements in performance than with any target one might have dared to set. Do you recognise that characterisation? Does that characterise your approach?
Sir Robert Devereux: It does characterise my approach in the real world. Just like Jon, I have very diligent people doing all this stuff. We have run through some of the new things we have been doing since we last met. We are on the case, trying to drive this down. The thing I am putting in a box called “I am just not so sure about this” is trying to work out some academic, theoretical possible future number. I am not too sure what the benefit of it is. It certainly consumes resources—hours of resources—and I don’t know that those aren’t better spent actually doing proper analysis about chasing something today, which I think would probably be a better use of my analysts’ time.
Q59 Mr Bacon: Or spending those same resources analysing bumps in the road or flaws, such as the one I was discussing one with Mr Lodge earlier, that have manifested and ironing them out.
Sir Robert Devereux: I have measured universal credit. I found a fact out, as I have described today. That required analytical efforts to get in behind all that. I would much rather my analysts did that than go after some spurious number, because I do not believe it will be found.
Q60 Chair: Of the initiatives to deal with the bumps in the road you have talked about and said that you are developing, how do you determine which ones you should focus on? What is your analysis?
Sir Robert Devereux: We set out in page 58 of the annual report, which I copied to you in my letter, the fact that we know clearly what our main problems are. They are to do with earnings, household formation and capital. I set out in three bullets underneath what we are doing about all those.
Q61 Chair: How do you determine which ones to pursue?
Sir Robert Devereux: I start with what is losing most money.
Q62 Chair: Okay. That is clear.
Sir Robert Devereux: The thing that has been losing most money is earnings. I have a profoundly effective regime through RTI. I have done that in two ways. One is in respect of changing the definitions. Universal credit is better for lots of reasons because it is based on that. I am not asking people to declare it; it just is the answer. I am using RTI to wash through the old data, such as pension credit and housing benefit. I have got that one straight.
On capital, which is the second one in this list, the best I can say is that we have finally made a breakthrough with the banking system, and we now have a partner who is trying to work with us to find out, “If I tell you which bank accounts benefits are paid into, are you capable of telling us what is suspicious about it in terms of lots of interest flows going in which are commensurate with capital they should not have?” That trial is under way and could make a profound difference on capital.
That leaves us with living together—household formation—which is genuinely quite hard to deal with for all the reasons we have been through.
My answer on how I prioritise it is that I look where the losses are and I try to find mechanisms to do it. I think I will have a good story on RTI. I expect the 2015-16 numbers, when they are finalised, to be better than the provisional ones, because I know stuff about how housing benefit is working that we have not had time to talk about today. I expect that the trial will potentially be helpful. I hope it will be helpful in terms of straightforward capital. That is how I do it.
Q63 Chair: I will come back to that in a minute.
Sir Amyas Morse: I am sorry to boringly have to put this on the record again, but I just listened to Sir Robert’s characterisation of our interest in discovering practically what the feasible lowest level of fraud and error can be and saying it is some theoretical model. None of that is true, and he knows that perfectly well. He is just deliberately mischaracterising what I have said. I have explained it to him many times. I have explained it to Lord Freud and lots of other people.
At the end of the day, it is what I require to make my decision—my decision—on what is regular and what is not regular. Don’t make me regret having tried to come up with a more forthcoming approach to this than my predecessors had. That is what I have done—not that I have had much response to it. In fact, your Department has done some good work trying to understand the reasons for fraud and error in particular allowances. We have made progress with that, but constantly we make a bit of progress and then we stop trying to work on it. That is really disappointing.
If you are planning to do a bit more work on it—I doubt it will be something that you will be able to determine in five minutes, to be quite honest—but if it is being approached in good will, I am always ready to have the discussion.
Sir Robert Devereux: The thing that has happened since we last had the conversation is that it has been clear—it is not an unreasonable point of principle for the C&AG to say, “I will qualify your accounts until you get to or within 1% of a reasonable level.” That is an entirely reasonable position and I am not arguing with it.
Since then, we have tried to work out how you would determine it. It sounds good, but how would you do it? I have brought with me the 15 pages that we provided to the NAO that went through four possible ways of doing it—my guys are endlessly innovative. When you come back and say, “What assumption have you got to make to make those go?”, you suddenly find that the lowest reasonable level is a product of your assumptions as much as it is a product of science. I am not persuaded that this is going to turn into something that I can nail. It is not going to me who is being quizzed. You are going to quiz me and say, “Is this a really good number?” You are going to ask me questions.
Sir Amyas Morse: I might be going to do that, but since you have not actually shown me any of this stuff, this is just—
Sir Robert Devereux: I have shown you all of it. Your colleagues have had all this stuff, Amyas.
Sir Amyas Morse: Well, they’re telling me no, so as far as I am concerned, this is just rather a theatrical conversation.
Sir Robert Devereux: I am quoting some facts about—
Sir Amyas Morse: I am actually seriously interested in finding out where we can get to on this.
Sir Robert Devereux: The position I reported to the Committee was that we have done work with your colleagues we have not previously done. I have agreed that we will have one more roll of the dice in September to see whether we can make something go, but I am observing it is taking a lot of time and unless it produces something that the Committee believes—
Chair: This is not the place to have this discussion.
Sir Amyas Morse: No, it’s not up to the Committee. Sorry. You seem to be under a misapprehension; this is not up to the Committee.
Chair: But you brought it up, Sir Robert. I just think we have to be clear. We are here for a recall on the issues. We know you have this long-standing concern about this.
Q64 Mr Bacon: Chair, can Sir Amyas just repeat his last point, because I didn’t hear it, about a misapprehension?
Sir Amyas Morse: Let me repeat it. This is a question of what my opinion on the irregularity of the DWP accounts is. It is not a question for the Committee to decide anything on at all. It is a question for me to decide at my sole discretion.
Sir Robert Devereux: I understand that, and to be clear—
Sir Amyas Morse: Robert knows that. We have had a series of campaigns on this subject. This is just another of them.
Sir Robert Devereux: To be clear, if I do not think we can make this work, I am resigning myself to Amyas consistently qualifying my accounts—
Q65 Mr Bacon: Or you could do this. I have had this discussion with—I can’t remember how many permanent secretaries there have been in the DWP in the last 15 years, but there have been a number. You could turn round to your Ministers and say, “I can’t implement this policy in a way that’s satisfactory, and in order for you to get me to implement the policy, you’re going to have to issue me with a letter of direction.”
I think it was with Sir Richard Mottram, one of your many predecessors, that we had the discussion about housing benefit fraud. It turned out, five years after the National Audit Office had previously looked at housing benefit fraud, that the Department was still unable to say whether housing benefit fraud was going up or down, and this was in an environment where the accounts had been qualified since 1988, I think. I have actually stopped having this conversation with accounting officers because it just gets boring after a while, but this has been going on for a long time. There is an issue around the complexity of our policy making it unimplementable—I think Richard Mottram used a phrase not dissimilar to that in relation to tax credits—at which point your role as an accounting officer ought to require you to do more than you are doing, in terms of asking Ministers to direct you, but I don’t think that has ever happened.
Sir Robert Devereux: It has not happened. It is an interesting thought process, but I am not sure I get to the same conclusion, because the reality is that I regard the system I am trying to run as really quite complicated. Parliament and Ministers have created endless rules. We are trying to do our best on this. We don’t ever get figures about what the banks regard as fraud, or insurance companies, but I can assure you that if they had numbers like ours, they would be tickled pink. But we are in public, so we are constantly on a quest of, “Could it be a bit lower?” Personally, I think that we need to keep pressing down—that is what my colleague said. I am going to keep pressing down on it, but I do not feel myself as if I am somehow wasting taxpayers’ money, because I am doing the right things and in the meantime running a system that is putting money in the hands of 22 million people, which Parliament want me to do. They want that done. They also do not want fraud.
I think we are in a position where, so long as I keep pushing this down, which has been done, that is not an unreasonable position for the accounting officer. That is not an accounting question; that’s for the C&AG, but if four of my predecessors—actually, way more than four—have had their accounts qualified because of the way the accounts are done, I’m not sure that I can be worried about that.
Q66 Mr Bacon: To pick you up on that one point about insurance companies, I am not quite sure that it is the case that the figures are not available. I have had many meetings over the years with Aviva, a very big insurance company and a big employer in Norfolk—they used to be called Norwich Union—about precisely these questions of fraud, and as far as I am aware, their accounts are not qualified. They are obviously very exercised by fraud.
Sir Robert Devereux: They are not qualified, but they do not have the same problems. Amyas is not arguing that my accounts are misstated, as in the numbers. He is saying, perfectly properly, “Parliament didn’t ask you to waste this money on a fraudulent claim. Therefore, since I don’t know how much fraud is reasonable, I have to put it down for regularity.” This is not, “Are the numbers right?” My accounts are as good as Aviva’s accounts in terms of, “Are the numbers right?” But in public life, I have another concept—is it what Parliament intended?
Mr Bacon: That is why you are called an accounting officer, and we are very glad about that.
Q67 Chair: And why the Comptroller and Auditor General has his constitutional role, which obviously we on this Committee completely uphold. We can end this discussion here because we do not want him to be questioned in his legal and proper approach.
Sir Robert Devereux: That is his judgment for the accounts—
Chair: Absolutely. This should not be an issue for disagreement. He is the Comptroller and Auditor General and, in the end, it is his call. Sir Robert, like it or not, you, as accounting officer, as Mr Bacon says, have to live with that.
Q68 Mr Bacon: I have one more question. Who is the senior responsible owner within DWP for fraud and error?
Sir Robert Devereux: Jon Fundrey.
Mr Bacon: Thank you.
Q69 Caroline Flint: I want to ask a question about how you decide what you review, because carers’ allowance has not been reviewed for 20 years. How do you know what is going on there in terms of fraud and error?
Sir Robert Devereux: We are investing quite a lot of money because even now we are having arguments about what precision these estimates come with. We are spending quite a lot of money doing the measurement. We have been doing the ones with most of the problems in because they are means-tested ones, and that is why we are doing them continuously. We have done some, in all honesty, somewhat sporadically. So I cannot tell you the answer to the question “Exactly what is going on with carers’ allowance?” but I have focused my attention on the ones where, just by cause of the definition and the benefit, you can imagine issues to do with income, partnerships etc. are relevant.
Q70 Caroline Flint: Do you have plans to institute a more regular review of carers’ allowance? Because it is going to be a growing area, given the demographics.
Sir Robert Devereux: Yes. I am not currently sitting on plans to do that. I am sitting on a plan, though, to review personal independence payment, which is the successor to disability living allowance. One element of which, just to come back to the design process, a substantial amount of fraud and error in disability living allowance turned out to be people’s conditions changing, so we have changed the definition of personal independence payments and every so often we will review your case. That is a deliberate policy choice to try to address a known weakness. But we have done that to create the benefit and I am now going to measure that in due course rather than go back and remeasure something about a benefit that is not going to have any long-term life.
Q71 Chair: Mr Thompson, when you appeared in front of us before we recommended that you work with the Fraud, Error and Debt Steering Group and you refused in your response to us. You cited the customer survey you do as more useful in determining what to do. Can you tell us what that customer survey told you about tackling some of these issues?
Jon Thompson: I do not think it was actually in the hearing. It was in relation to the Treasury minute from before, which I have got in front of me. I’ll be up front: I have changed my mind about the recommendation. This Treasury minute was put in front of me on day seven, I think, with a note about why we should essentially say no. I have now been there for three and a half months, met the team and spent quite a bit of time with Nick and the team, and we do an awful lot of research in this particular area. I have a list here of the nine areas of research we have done, but I see there being nothing wrong with having some further research in line with the recommendation you made, which I think was 5.1 in the TM.
Chair: Okay, thank you.
Mr Bacon: An object lesson, Sir Robert.
Sir Robert Devereux: I would like to suggest that the letter I sent you also gave you three things you asked for that were not in the Treasury minute, so we are all trying to be nice.
Q72 Chair: We appreciate a spirit of helpfulness.
Sir Robert, underpayments are now at the highest ever recorded rate for your Department. That is a lot of vulnerable people. You have now set this target. Your Treasury minute refused to set targets. We have had this with the written statement on universal credit and we now have it for underpayments. Do you now acknowledge that we were on the right track to suggest that targets be set? Because you have, in two actions, taken on our views, after all.
Sir Robert Devereux: Yes. Rather like my colleague, reflecting on it, I thought, “We’ll do this.”
Q73 Chair: That is good, too. It is nice to know that our views are taken on board. My last couple of comments are around tax credits. Jon Thompson, what progress have you made on that since you were last in front of us, which was just over six months ago?
Jon Thompson: On tax credits in general?
Chair: Yes.
Jon Thompson: I am slightly struggling with what the question is. What progress have we made on tax credits in the last six months?
Chair: Sorry, on the underpayments.
Jon Thompson: Do you mean in relation to the recommendations you made?
Chair: Particularly in relation to the recommendations, but if you have anything more to add, we are happy to hear it as well. It matters a lot, obviously, to people out there if they are not getting their money. It is a big issue for us and our constituents.
Jon Thompson: Our overall goal is to produce the most accurate benefit we can. I have the standard Report in front of me, and I understand that I am coming back to talk about it in due course. On page 62 there is a rather excellent figure—figure 19—produced by the National Audit Office that shows you error and fraud on one side and underpayments on the other. Progressively, over the years, this benefit has become more accurate. There is a discussion to be had, following Mr Mills’ intervention, on whether we should combine this into some sort of target about the accuracy of the benefit or, if not, whether we should have a separate underpayments target.
Q74 Chair: So, basically, you are trying to get it right first time? You are looking at that as a priority?
Jon Thompson: Correct.
Chair: Good. I was hoping you might say that, and I am pleased that you have.
Thank you very much for a rather marathon session. Sorry that we were interrupted by a vote. You understand that, out there, people are receiving these things and our surgeries are full of people with confusing letters—from all of you, at times—that they don’t understand and that have big figures in them. It is a real worry when people get either an overpayment or an underpayment. We are completely behind you on tackling fraud, and we recognise some of the challenges there. We thank you for your time on nearly the last day of term for Parliament. We will have you back again soon, no doubt. Our sister Committees are looking at some of these issues, too.
Thank you very much. The transcript will be up on the website in the next couple of days, and we will probably write to you as a result of this hearing. We may do a short Report, and we will discuss that in a moment.