Public Accounts Committee

Oral evidence: Fraud and error in Housing Benefit, HC 706

Wednesday, 29 October 2014

Ordered by the House of Commons to be published on 29 October 2014

Watch the meeting: http://www.parliamentlive.tv/Main/Player.aspx?meetingId=16278

Members present: Margaret Hodge (Chair), Mr Richard Bacon, Guto Bebb, Mr David Burrowes, Jackie Doyle-Price, Chris Heaton-Harris, Meg Hillier, Mrs Anne McGuire, Stephen Phillips, John Pugh, Nick Smith

Amyas Morse, Comptroller and Auditor General, National Audit Office, Gabrielle Cohen, Assistant Auditor General, National Audit Office, Rob Prideaux, Director, National Audit Office, Max Tse, Director, National Audit Office, and Marius Gallaher, Alternate Treasury Officer of Accounts, were in attendance.

Witnesses: Ian Adams, Director of Financial Operations and Customer Services, Islington Council, Andrew Stevens, Assistant Director (Customer Delivery) East Kent Services, Robert Devereux, Permanent Secretary, Department of Work and Pensions, Steve Buckingham, Senior Responsible Officer for Housing Benefit Fraud and Error, Department for Work and Pensions, and Mike Driver, Director General (Finance), Department for Work and Pensions, gave evidence.

              Q1 Chair: Welcome. This is a pre-hearing. We are running late because we had a witness who was unable to be with us on Monday who we had to hear today. Apologies for keeping you waiting. This is an opportunity for you to help us understand the issues that we need to put to the accounting officers. It is really your chance to put on the table the issues that are at the top of your agenda. If I just take the widest figure, it looks like fraud and error in housing benefit is on the rise. You guys at the local authority level are responsible for running the system. What is your explanation? What are your concerns? What are the barriers preventing you from improving the performance in the way that the Government want on fraud and error? Who wants to start? Mr Stevens, do you want to have a go? You are from east Kent; I don’t know where east Kent is. What bits does it cover?

              Andrew Stevens: Canterbury, Thanet and Dover. We have a shared service that incorporates three different councils in east Kent.

              We have discussed many times the reasons why fraud and error seemed to be increasing. We focused increasingly on the claimant error part, which is the part that seems to be increasing the most. We have looked at our caseload in Canterbury and in 2008, one in 10 of our people claiming benefit was of working age and was employed. That is now one in five, which introduces a level of complexity into the system. Some of those people are on zero-hours or low wages that fluctuate and therefore the benefit will need to be reassessed a lot more often than it has been before.

              Some of the initiatives introduced by our colleagues at the DWP have been successful. Because they have been successful, you are now seeing an increase in the reported errors because they are finding the errors that existed before, but were not picked up. In the initial results from the RTI exercise that we are doing with DWP—we have only had one tranche of information in—something like 50% of the matches have resulted in somebody’s benefit being changed. That is the whole point of the scheme.

              There are some other wider issues with that. Our caseload has gone up by something like 20% and funding has gone down by about 30%. We are at the bare bones of being able to process people’s benefits claims quickly and accurately. The other activities and exercises that we did before—such as campaigns to try to pick up people’s changes in circumstance—we have not got the resources to do anymore. We just have not got the funding to do that anymore, so they have gone by the wayside to focus on getting money out to people quickly and accurately.

              Q2 Chair: What about you, Mr Adams?

              Ian Adams: I am from Islington council, so one of the London boroughs.

              Q3 Chair: I should say to the members that these are two authorities that were selected because they were both on the list of poor performance, but have improved.

              Ian Adams: Fraud and error, for the last two or three years in Islington, has remained at about 1.8% of what we pay out in subsidies. We pay about £200 million a year and fraud and error is about £3.5 million, which has been quite static over the past few years. We have not seen an increase in fraud and error over the past three years in Islington, but £3.5 million is still quite a high amount.

              I agree with Andrew. I think a lot of the work that has happened through the DWP sharing data has been really important. The sharing of DWP income-related benefits data automatically has been important and has led to finding issues more quickly. The RTI—that is, real time information—around earnings will be important as well, if it works appropriately. To put that into context for Islington, we had 278 targets for RTI out of a caseload of 30,000 so we are not getting an awful not coming through RTI. Early doors, but it has just started.

              Those initiatives help to find fraud and error. The counter position for us is that housing benefit is different from other income-related benefits because the key issues around housing benefits tend to be occupation and household composition. Those tend to make it different from the other types of benefit. You need a resource to check that. One thing is pretty clear about fraud detection; if you knock on a door, you can sort out the present, but it is quite hard to sort out the past when it is to do with occupancy and household composition. It is very hard to prove the past.

              The diminution of resources, particularly around the admin grant, which for Islington has reduced from £3.7 million in 2010 to its current position of £2.9 million, has meant that we have one less fraud officer than we used to have. There is some impact on fraud and error detection as a result of the reduction in grant, but, again, our numbers have remained pretty static at around £3.5 million for the past three years, in terms of quantity.

              Q4 Chair: What incentives could the Government give to encourage your council to resource you more efficiently or with more people? Your grant has been cut, so you lose some staff and do less detection—that is part of the story. What incentives from the DWP would convince your chief executives and political leadership that they should put in more money?

              Andrew Stevens: From our perspective, any incentive scheme must not be particularly labour-intensive to apply for—some of the funding arrangements from the DWP require a lot of time spent on filling out the paperwork to apply for the money in the first place—and must give sufficient incentive and funding certainty to employ staff to deal with the activity. For example, if the DWP said, “Here’s an amount of money. We want it to be ring-fenced and spent on trying to reduce the level of error in the housing benefit system. If you are willing to sign up to that, you can have the resources to do that”, then we would ensure that whoever we employed for a certain period was doing exactly that. A payment by results scheme does not give you that level of certainty. Given that as a council we currently fund the benefit service by about 50% of its value anyway, that would put another business risk on us to employ staff for whom the funding is just not going to be there.

              Ian Adams: The Report helpfully picks up the different incentive schemes that have happened over the years—I have been around for long enough to have seen a few of them. The one that worked best was called SAFE, which was an incentive payment if a sanction was applied. There was a mix between an incentive per sanction, at a set amount, and a percentage based on the amount of money that you saved. From recollection, that was the one that provided more direct incentive, if the money that came through incentives paid for the fraud team. At the moment, the money that comes does not pay for the fraud team.

              Q5 Mr Bacon: You have all the costs and none of the benefits.

              Ian Adams: Yes—well, you do have some benefits, because clearly there are benefits through rent collection and so forth from limiting fraud, so I would not want to give the impression that it is not something that interests us.

              Q6 Mr Bacon: Right, but there is an asymmetry between the costs that you must incur and the benefits that you are capable of getting.

              Ian Adams: Yes, and my opinion is that that was better up to 2006 under the old incentive scheme. The concern about that is how you get the level right so that it does not cost too much. That was always the concern.

              Q7 Mr Bacon: Given the changes that have happened, would it be possible to resurrect what you are talking about from 2006 in some form, if it is more effective?

              Ian Adams: I think that it would be possible to resurrect it. We are moving towards a central fraud investigation service, so the other problem that local authorities currently have is that we are almost winding towards that. For Islington, that is August next year, so obviously anything that happens is for a very short period of time.

              In terms of whether you can apply an incentive if someone is prosecuted or a sanction is applied that creates an overpayment that can be recovered, that is a good thing, because it is clear that there are currently disincentives. For example, at the moment, if there is a fraud overpayment, the Government will subsidise 40% of that, so you lose 60% of it. If someone commits fraud and immediately goes abroad so you struggle to collect the money—we have recently had a case in Islington in which the person went abroad with £35,000 and we are unable to chase that money. We are immediately 60% down on £35,000. There are inherent disincentives in the current subsidy scheme that mean that, in a sense, you would be better not finding that fraud because you would get a 100% subsidy on the £35,000.

              Q8 Mr Bacon: Can you say that again, loudly, for the record?

              Ian Adams: The truth of it is that if we had not found that £35,000 of fraud, we would not have lost 60% of it.

              Q9 John Pugh: Under the SAFE scheme that you mentioned, were you incentivised to apply sanctions whether or not they were effective? Was there any penalty for sanctioning people and then having the sanctions challenged and appealed? You do not want traffic warden syndrome, where the more you sanction people, the better it is for you.

              Ian Adams: My view is that linking it to a sanction is better than its predecessor. The predecessor was called weekly benefit savings, and you got a sum if you saved benefit from being paid. That created a perverse incentive of trying to search for people to stop paying benefit to. You would multiply it by 52 and get that sum. To me, that was wrong because it created a perverse incentive.

              If you link it to a sanction, at least you are linking it to something tangible. I guess there is a wider question about what impact sanctions have. That is probably a fair question and I am sure there would be some research around that. It seems to me to be a fairer way of incentivising activity if there is a tangible outcome.

              Q10 Chris Heaton-Harris: Thank you for coming. As you can see, we are just kittens here at the Public Accounts Committee. We are very meek and mild and never ask any hard questions, so don’t worry about it. We are just trying to prise information out of you.

              The information I am after is essentially how much fraud and error you find in each of your authorities. Do you have figures for the past couple of years or past year that you can give us?

              Andrew Stevens: Yes, I have for Canterbury. Bear in mind that Canterbury is a medium-sized district council. In 2013-14, we found around £115,000 of housing benefit overpayments and fraud, which we went out and found and issued 25 sanctions. Sanctions consist of cautions, admin penalties and prosecutions. We look at a range of sanctions when we find fraud in the system and we issue cautions. Anecdotally cautions are effective in stopping people from reoffending.

              Unfortunately, when we have conversations with our DWP colleagues they tend to focus more on the admin penalties and the prosecutions as a yardstick of performance, rather than look at the number of cautions. That was £115,000 last year and we are already up to £83,000 this year, so we are finding more this year than last. That is encouraging but it is going to make your next lot of figures look bad as well. 

              Ian Adams: I can probably add case numbers and context to that. For us last year, we had just over 500 fraud referrals, 50% of which came through data sharing. They came through data sharing, known as HBMS. That is in the Report and comes directly from DWP. Of those 36% were reactive referrals from the staff or public. Normally, from our team in housing that deals with rent arrears we get a lot of referrals.

              From that we sanction more because we are a bigger authority. That is around 120 a year, of which 20 to 30 are ordinarily prosecutions. The others will be a mix of cautions and something called admin penalties, where you can increase an overpayment by an admin penalty. Those would be the ballpark numbers that Islington would see.

              Q11 Chris Heaton-Harris: Do you feed those numbers up to DWP? What happens to those statistics?

              Andrew Stevens: In terms of the value of the overpayments and the number of sanctions, they get fed to the DWP by using the SHBE data on a separate return. They sometimes write back to the local authority if they think your fraud activity is lower than it should be and ask what steps you are taking for an explanation of why that is. As I said, they tend to focus on prosecutions and admin penalties rather than cautions. Cautions definitely have a place in the system.

              Ian Adams: With the data-sharing that is provided, which is known as HBMS, we provide case-load information on each of those returns. We will provide to DWP the outcome for each of those cases.

              Q12 Chris Heaton-Harris: In return, does the DWP give you best practice from other local authorities—they might have better or worse practice than you—so you roughly know the ideas out there that could help you improve?

              Andrew Stevens: We tend to do that benchmarking ourselves with local authority colleagues in the immediate area. If your performance is below what they expect they would offer the services of their performance improvement team to help you improve in that way.

              Q13 Chris Heaton-Harris: But there is no obvious spreading of best practice.

              Ian Adams: There has been in the past. I would not say it is a continuous and regular activity. It would be wrong to say that there have never been occasions where some best practice has been shared.

              Q14 Chris Heaton-Harris: Can you remember the last time it happened?

              Andrew Stevens: No, not for a few years.

              Q15 Meg Hillier: I am interested in exploring with you the issue of the change from you looking at fraud to the single fraud investigation service that the DWP is bringing in. Following on from what Chris Heaton-Harris was asking about the sharing of good practice, in summary, what is your view of the change and the difference it will make in uncovering fraud in your respective authorities?

              Andrew Stevens: My fraud officers are due to transfer to the single fraud investigation service from December next year, which is slightly after Ian’s team. In an ideal world, I would have liked that transfer to happen when universal credit was up and running. I think that that was the original intention. The problem we have is that we will be transferring our fraud resource over to the DWP, so the function of investigating fraud and the staff will transfer at a point where universal credit is not rolled out and where we are still responsible for thousands and thousands of housing benefit claims, but with no legal power to investigate or prosecute parts of it. My view is that it should have been rolled out at the same time as universal credit and kept to the same timetable.

              Ian Adams: I agree with that. There is some sense in bringing together shared investigation and shared intelligence and working together on a single case, so I would not want to say that the whole concept is nonsensical. The concern for me is that a lot of fraud work is based on local intelligence and local connections, particularly when it is to do with occupancy and household constituency. Although I have a concern about this to do with universal credit per se, a lot of local authorities provide a lot of support locally. For example, discretionary housing payments are awarded to help those impacted by welfare reform, which always infers a personal contract, an agreement with an individual and some conditions attached.

              From my experience, a lot of contact—contact that has something that leads to some sustainable activity—helps with welfare benefits generally, but also with fraud. Continuous contact, that intelligence and the local connection enable you to deal with issues. If it becomes a central bureaucracy, or more of a central bureaucracy—there are London teams, for example—my big concern is how it then deals with the local connection and the local intelligence that local authorities have. That is connected with other things, such as the blue badge, the single person discount for council tax, financial assessments for social care and all the different types of parking permit. All those things require people to give information about themselves. How is that information shared? How does it become an intelligence-led response that enables fraud to be dealt with and prevented?

              Q16 Meg Hillier: Local knowledge certainly plays a part. One of the things that has been raised with me by someone in your line of work is the issue of contrived tenancies, where people effectively fake a tenancy to cover their mortgage costs. It has been put to me that that is much more easily detected locally than through some central body. Do you have any experience of that in your area? Do you have a view on what will happen under the new system?

              Andrew Stevens: We have looked at the cases that result in a fraud sanction, and it is not the highest type, but it is still there in the case load.

              Q17 Meg Hillier: Have you got an idea of the percentage in the case load?

              Andrew Stevens: No, I have not.

              Q18 Meg Hillier: Can you give us the figure?

              Andrew Stevens: Yes, we can give you that figure. Just to echo Ian’s points, if the fraud investigators were centralised in an area that was not in the immediate vicinity of where they work—the local knowledge is important, because they know the landlords and the tenants, especially in the shared service we run. We find the same landlords and tenants operate in the whole geographic area. To lose that would be bad.

              Q19 Meg Hillier: Do you have a quick comment, Mr Adams?

              Ian Adams: I have a quick comment on private tenancy fraud. The best way to deal with that is to go to the house.

              Q20 Meg Hillier: Can you briefly explain something, because we need to get the mechanics clear? When the new system is set up and your offices are moving over, will they still be based locally, or will they eventually be wherever in the country the central fraud office happens to be? There will be a TUPE rule, so they cannot just be dismissed, but in time will that local connection break down or will they be recruited from the boroughs? Do you know? I do not know myself.

              Andrew Stevens: When they first transfer to DWP we have been told that they will use the most of their local estate. We don’t know exactly where they will be based.

              Q21 Meg Hillier: Oh, I see. So they will move out of your borough.

              Andrew Stevens: But over time if they leave and new people start, there is no guarantee that they will be locally based.

              Q22 Meg Hillier: So they could be in Darlington dealing with Islington and Kent, for instance.

              Ian Adams: I suggest that is unlikely. I think it will be London.

              Chair: Yes, I think they will centralise them in a team somewhere.

              Ian Adams: The only thing I would say which links with your previous question is that our visiting officers—we have three visiting officers in Islington—will not go across to SFIS. It is our acronym for them. So we have people visit private tenancies at the outset. Is there an incentive for us still to do that? I think we would want to do it because we deal with a lot of issues early doors, but will we be funded to do that? Secondly, will there be a disconnect between what they find and the SFIS team? Will it be possible to refer very quickly and to deal with an issue very quickly as now? Those are the two issues that I really mean when I talk about local connectivity. How will those sort of things be resolved?

              Q23 Mr Bacon: You mentioned your visitor teams. My local authority talks about its visitor teams, and it says that the training they have means that when they go inside a dwelling they are trained to spot things that help them to assess what the genuine constituency of the dwelling is. I take it that you do something similar.

              Ian Adams: We do and we have gone a bit wider. We have an issue in Islington about very small dwellings being rented out.

              Q24 Mr Bacon: Houses in multiple occupation?

              Ian Adams: Yes, houses in multiple occupation. Some of them are frighteningly small. So we have linked that up with our environmental regeneration department. If we have a concern we immediately take activity there.

              Q25 Chair: That was your case, wasn’t it? What was it? I can’t remember, but it was tiny.

              Ian Adams: There were 30 very small lets. They were smaller than 3 square metres.  The fact that our people are getting in there and seeing that early, even though they are going with a housing benefit purpose, deals with a wider issue as well. That is another concern about losing some of that local activity. It always deals with more than just the presenting issue.

              Q26 Mr Bacon: So if you were to characterise the value of the visitor teams, how valuable would you say they are?

              Ian Adams: It is hard to quantify—

              Mr Bacon: I don’t mean a number.

              Ian Adams: Personally, I feel they are very valuable in preventing other activity that could go wrong and looking at some of the wider issues. Also with social care and concern: the eyes are on the ground. Sometimes they deal with social care concerns as well. It has a wider value than just housing benefit.

              Chair: I need to draw this to a close soon. So, Stephen, David and then that’s it.

              Q27 Stephen Phillips: I wanted to ask you about the incentives provided through the subsidy system. Where there is fraud and error, if you are given a high return then there is an incentive to let claims through, and if they are detected subsequently you will still get the money. If there is a low return then you will be very hard on claims at the outset because you won’t want them to be detected subsequently. So the subsidy regime points in two different directions and is not really capable of being reconciled. Now you are at the front line. What would you do to the system to make sure that only proper claims get through in the first place and that fraud and error claims are detected as quickly as possible thereafter? How would you alter it?

              Ian Adams: Well, they have tried to deal with that by having a threshold for local authority error. If we have let the claim go through, that is a local authority error. I think it is something like 0.45%. You actually get a 100% subsidy on that and then there is a threshold where you get—we have always got 100%—slightly more subsidy up to a level where you lose your subsidy. I think that is a way in which they have tried to deal with that. That felt to me like quite a progressive act. It is a way of saying, “Don’t allow local authority areas to enter the system.” We were very concerned about going above the threshold. Clearly overpayments happen. It is a fact of life, but what you don’t want to do is to lose all of that subsidy. If you lose all of that subsidy effectively you have paid out as a local authority and there is no return whatever. It is very difficult to collect. So I think, personally—I don’t know what Andrew’s view is—

              Q28 Stephen Phillips: I am going to ask Mr Stevens in a moment. I shall stop you there as we are pressed for time. The NAO found that there weren’t strong incentives and have taken that point into account to detect overpayments after the claim has been awarded. Do you disagree with that opinion?

              Ian Adams: I don’t think I disagree with it, because I think the 40% subsidy is not good enough. The Report says that on average local authorities collect about 40%. You are losing out there.

              Q29 Stephen Phillips: The problem is that the moment that that 40% figure gets pushed up, there will be much less of an incentive to detect subsequent fraud or claimant error.

              Ian Adams: That is a problem.

              Andrew Stevens: That is absolutely right; if you cannot collect 60% of the overpayment, you are losing out for finding it in the first place. A more generous initial subsidy, with the reality of how much councils collect afterwards—

              Q30 Stephen Phillips: I am sure that neither of you have ever closed your eyes to fraud or claimant error deliberately, but how does this disincentive imposed by the 40% figure manifest itself? Do you just put less effort into detecting fraud and error?

              Andrew Stevens: No. From our perspective, we put in as much effort—the guys processing the claims put in as much effort now as they ever used to. The problem with the lack of funding now is that we used to have more visiting officers to go out and check claims, as well as more campaigns to pick up mistakes or errors when they entered the system.

              Q31 Stephen Phillips: Is there no chat around the water cooler of, “Don’t try too hard on this one because we will just end up losing 60% of it when we get our subsidy”?

              Andrew Stevens: Absolutely not.

              Ian Adams: No. I am not sure if this bit is included in the Report, but fraud and error used to be separated. There used to be claimant error and fraud, with 25% subsidy for claimant error and 80% for fraud. In that sense, you could clearly distinguish the activity.

              Q32 Stephen Phillips: It is not easy to distinguish between claimant error and fraud in all cases, is it?

              Ian Adams: It is not; that’s the problem. But if fraud is something that is sanctionable and if they had a sanction regime—

              Stephen Phillips: I could try to spend another hour discussing this with you.

              Q33 Mr Burrowes: I have a general question about the implications of the single fraud service. The experience in Enfield, which I think is similar to your own experience, is that the financial challenges are leading to a strategic direction towards the big cases rather than the numerous small cases. Is that the same with you?

              Ian Adams: Yes.

              Q34 Mr Burrowes: There is a shift—not just from housing benefit—into the wider risks of procurement, grant allocations and vetting; they have to focus on where the big financial risks and taxpayers’ money are, more so than housing benefit.

              Ian Adams: Yes, and the wider impacts of council tax support and how that continues to be investigated and dealt with—particularly as it is now not council tax benefit so is completely separated from DWP benefits. Another big point is the resource left to deal with the wider issues.

              Q35 Mr Burrowes: Is that the same for you?

              Andrew Stevens: Yes. The timing of the introduction of SFIS causes us a problem in that we will still be responsible for paying the benefit. We will still be responsible for investigating council tax support fraud but will lose all our fraud staff and our function of investigating housing benefit fraud. The system will become completely disconnected.

              Q36 Mr Burrowes: With the single fraud service, particularly in London, I understand that FALCON teams have been set up to work closely and in partnership with local authorities and centralised fraud teams. How will that fit in, given that it has only been there for a few months? In February, Enfield is moving to a single fraud service. What will the impact of that be?

              Ian Adams: At the moment, I have not seen an impact. I am not sure of the impact. Enfield, presumably, goes before we do. We do not go until August next year.

              Q37 Mr Burrowes: The Met have set up centralised fraud teams that are working alongside local authorities. I wonder if that will be impacted by the disconnect you say there is with local investigators.

              Ian Adams: I honestly do not know the answer to that.

              Q38 Mr Burrowes: Okay. In terms of the transfer of open cases to the service, have you looked at the roll-out and the percentage of those cases that will go over? Will you be able to track what the central service will pick up? Will some be lost in the process?

              Ian Adams: That is a concern. A lot of cases already running these days are joint cases. Ours are 75% joint cases. How do we close off those cases which are not joint or follow those through if we lose our staff? At Islington, we intend to keep a team of three corporate staff because we want to deal with council tax support fraud; we also want to cover blue badge and other issues of concern to us. We have found some resources to do that so we will be retaining that function.

              Andrew Stevens: We will be retaining a compliance function, looking at many council tax and business rate irregularities. My visiting officers do all of benefits, council tax, and business rates. All the fraud stuff will transfer to SFIS when it comes in.

              Q39 Mr Burrowes: Is there a concern that there may be some cherry-picking on those open cases that are transferred? Will there be some proper follow-through to ensure that they are not picked off?

              Andrew Stevens: Yes. We don’t know enough detail at this point about exactly how that migration will take place, but we will obviously be keeping a very close eye on that.

              Q40 Chair: Is there anything you haven’t had the opportunity to say that you think is relevant?

              Ian Adams: No.

              Andrew Stevens: No.

              Chair: That was really clear and extremely helpful. Thank you very much indeed.

              Mr Bacon: Thank you very much indeed. A lot of civil servants could learn from listening to how you give clear answers.

Examination of Witnesses

Witnesses: Robert Devereux, Permanent Secretary, Department of Work and Pensions, Steve Buckingham, Senior Responsible Owner for Housing Benefit Fraud and Error, Department for Work and Pensions, and Mike Driver, Director General (Finance), Department for Work and Pensions, gave evidence.

              Q41 Chair: I will start with a general question and we will come on to the incentives scheme, which is the key issue that we need to focus on. My general question comes out of the evidence and my own experience in my constituency. There is a growth in the number of people claiming housing benefit for all sorts of reasons, whichever figures you look at, whether they are NHS figures or other figures. Certainly my own figures tell me that. There is going to be less money available for it because of the financial cuts that you have to implement to meet deficit reduction. You have a record where the figures for fraud and error are not moving in the right direction. How are you going to have a system that will meet the Government’s objective of reducing fraud and error? It is already creaking at the edges and it will have increased demand. How are you going to try to square that circle, or is it unsquarable?

              Robert Devereux: I don’t think it is unsquarable, but I think it is difficult. You are right that there is a small growth in the overall housing benefit caseload and there is quite a rapid growth in the caseload of those people in work. It is those people in work who fundamentally have the earnings issue, which we will spend some time on this afternoon. Across the piece for the Department, and as the Report records, we have a growth in fraud and error in housing benefit that is pretty much offset by a reduction in fraud and error for other benefits. So it is not as if the constraints on public spending are causing us not to be able to do it right across the piece.

              What I hope we will talk about this afternoon is those areas that are to do with earnings: if you think about the losses that we are recording which local government does not then go back on and recollect, by far the largest is earnings—they come in two or three big buckets. At whatever point you want to talk about it, we should have a conversation about the step change that is possible in a world in which the Revenue is now collecting, month after month, every single person’s earnings and reporting it to them. The whole world of PAYE-related earnings suddenly becomes a vehicle for Government, in a joined-up fashion, to be able to tackle a problem that was previously impossible to tackle. So this is not related to reductions in total spending and costs of administration; this is a big step change in what is possible.

              Q42 Chair: I understand that. The national federation figures I have show that the number of households in work that are claiming housing benefit—I accept that’s a segment—has doubled from 11% in November 2008 to 22% in August 2014. That probably reflects what you said within the wider context. The system is creaking. Are those figures wrong?

              Robert Devereux: You are quoting percentage make-up of the case load. The actual case load working age standards in 2010 was 1 million and it is now 1.3 million. It is up 20% over three or four years in actual numbers. Some of the other lines are decreasing, for example, passported pension housing benefit is going down, so the percentage that the working age appears to be taking seems to be growing but that is not the absolute case load.

              Q43 Chair: Okay, but let’s stick to that. This is a guess from talking to people in my constituency: when it is claimant error, it is because of the nature of the work that they are doing. More people are on zero-hours contracts so they do not know what they are earning and there is not that certainty of earnings, which could lead to overpayment. That is one thing. We did a Revenue thing earlier this week and we saw with the tax take that there is growth in self-employment but not megabucks in earnings, so people are still entitled to housing benefit. The nature of the people coming through your housing benefit door and being overpaid because their circumstances change so rapidly, are so unpredictable or they are self-employed is changing. Is the system you have in place appropriate in that new context?

              Robert Devereux: We have done some quite deep analytical work about the nature of the overpayments, how long they last and what is causing them, and it is clear that over many years—this is the key thing that the Committee needs to clock—people’s earnings in the real economy move around quite a lot. We have a system of rules and of earnings moving around. The second thing you need to know is that these errors rapidly self-correct in many cases because the claimant calls two or three weeks later and says, “By the way, my earnings have changed.” That is a fairly persistent problem over time; it is not simply the advent of a potential growth in zero-hours contracts that has caused the problem. You might hypothesise that that makes some of it slightly more difficult, but the underlying problem is that people’s wages change frequently and we are running a system in which we are relying on them to tell us they have changed, until such time as you harness the power that we now have through the Revenue to know what their earnings are before they know themselves.

              Q44 Chair: So you are quite confident. People’s earnings change and they are maybe going to change more with a growth in this client group; the people who claim housing benefit will be disproportionately zero-hours-contract sort of people, I would have thought.

              Robert Devereux: To be honest, it is not a question of whether they know ex ante what they are earning. The issue is whether they have reported what they have earned when they have earned something; that then derives their housing benefit. You are correctly hypothesising that some people will be less able to know what it might be in advance, but we do not want to know what it is in advance; we want to know what it was when they earned it. As long as they are on their toes and tell us what they earned, whether it was a regular or a zero-hours job, it is the same problem for the housing benefit system. It does not make it any worse.

              Q45 Chair: That is where I think the system may not be appropriate to the changing nature of the way that people earn their money and therefore report to you. If the way they earn their money is so irregular, having a system that is built on expecting them to tell you just becomes really complicated. It would be a phone call every week to the local office to say, “Last week I did so many hours so I earned that.” Is the system fit for purpose?

              Robert Devereux: We will be back here in a month talking about universal credit. You know that, by design, universal credit payments are not made until we have already banked the earnings data that has come from the Revenue. We are not asking anybody what their earned income is to determine their universal credit payment. We are taking the data and using it to do the computation, so this entire class of problem for all those people in PAYE income goes away by strategic design. When I talk about using the RTI data in a tactical, data-matching fashion, I am doing that after the fact, trying to ensure that we keep it right. But housing benefit is fundamentally a system where I am reliant on people telling me. For universal credit, the whole point of it—the reason why we have gone to such trouble to make this connection with HMRC—is to make sure that we define what you are due, based on data that I already hold. This will transform the system completely.

              Q46 Chris Heaton-Harris: It will transform part of the system completely, because it transforms it for those for whom HMRC has PAYE data and so on coming through on a regular basis. That leaves a group of people out there who will have this continuing issue. They report their changes in income by phone or however at the moment, but if 75% or whatever pay by PAYE, does that not suggest that for the future you can concentrate on the remaining group to ensure that there is less error and fewer problems? How will we do that?

              Robert Devereux: Two things are true. First of all, let us be pleased that we can answer two thirds of the biggest problem. On the estimates of how much is self-employment, I think the Revenue is at something like 15%; we think it is probably slightly more prevalent among benefit recipients, so let’s call it 30%.

              The thing that happens under universal credit that does not happen at the moment is that the interaction that we are having with them is on a monthly basis. So many of the things that we are trying to do include having a continuing relationship—even for those in work—in a way that tax credits don’t do at the moment. The interaction with those in work is to have, at least once a month, a prompt to say, “What are your earnings?” It is not a case of, “If you fancy, tell us; otherwise we will assume that nothing has happened.” We will be positively asking, month after month, “What did you earn?” That will be a new system.

              Q47 Chris Heaton-Harris: Who will be asking that question? I thought that local authorities were responsible for policing the system, so will they be prompting once a month?

              Robert Devereux: I was illustrating what will be true under the world of universal credit, which will be us. So given that I have got, on the face of it, a strategic solution coming along and given that I have got real time information that I can use tactically on the earned information, the question that you should be asking is: can I drag forward that regular contact with those who declare themselves to be self-employed in exactly the same way I intend to do with universal credit, but do it earlier? That seems to be a perfectly good thing for us to be exploring.

              Q48 Chair: Mr Devereux, there is so much around the future world when universal credit will be in place, but can we deal with this issue in the current context? When universal credit comes in it will be a different world. We are already getting RTI in, so it is fair enough to bring that in, but we want this money now. Universal credit, if you are lucky, will be 2017, but if you are being realistic you are probably talking about a long time after that, so let’s deal with it in that climate.

              The Report says that you will not meet your target. Is the system fit for the purpose of meeting the target? The Report says you will not meet the target that, presumably, you agreed with Treasury Ministers when you had your settlement. Why not, and what are you going to do about it?

              Robert Devereux: I am sorry; you are talking about the overall target for fraud and error, so we have gone beyond housing benefit now—is that right?

              Chair: We are here talking about—help me, Max.

              Max Tse: It is the 1.7% target.

              Q49 Chris Heaton-Harris: That is for all benefits. You have set the target of reducing fraud and error from 2% to 1.7%. As housing benefit is an area where there is more error than most other benefits, what is the target for housing benefit?

              Chair: So take those two questions.

              Robert Devereux: Okay. At the risk of bringing one fact to the table first, it is true that gross losses on housing benefit are higher than on other things: 5.8%. As the Report says, so much of this is to do with earnings and so much of this is ultimately declared by claimants that local authorities collect most of it back again. So the net position, in terms of the taxpayer’s overall losses, is that housing benefit is not the most significant area of loss; that is actually pension credit.

              The Report records net losses on housing benefit of 3.5%. The measurement changes that we have agreed with the National Audit Office would make that about 3.1%. Don’t get me wrong: 3.1% is still far too much. But in terms of whether it is the worst thing, it is the worst gross, but the very nature of earnings and the fact that most of it comes back to us means that actually we have got quite a bit of it under some control.

              Q50 Chris Heaton-Harris: That is a completely fair point. You have a target for overall benefits, you have obviously identified how you can measure it and you have agreed on the 3.1%. Where would you like it to be in a year’s time?

              Robert Devereux: I think the answer to that is that we have assessed a number of interventions as being worth a certain amount to get us to 1.7% rather than trying to then compute, item by item, what that does to individual benefits. Maybe Mike can add something.

              Mike Driver: Our ambition is still to achieve the 1.7% target at the end of 2014-15. We have a series of measures in place, including, as Robert has alluded to, significant use of real time information, both in the DWP benefits system and the housing benefits system that is administered for us by local authorities. The NAO and the Comptroller and Auditor General confirmed that in their last set of accounts. He doubts that we will achieve the 1.7%, but we are working towards that and I think we will come close to, or potentially achieve, that target.

              Q51 Stephen Phillips: Gentlemen, I want to talk about the incentive system and the subsidy. Who is responsible for it and can answer questions about it?

              Robert Devereux: I am.

              Q52 Stephen Phillips: Right. How has it been designed? Is it designed by economists in the Department?

              Robert Devereux: Is it designed by economists? It has been around for quite a long time so I am not sure I know exactly where it started from, but it has a basic sense in it, which my colleagues in local authorities were talking about, that is—

              Q53 Stephen Phillips: Let me ask a different question; when was it last looked at to see whether it was working?

              Robert Devereux: I cannot recall when it was last reviewed.

              Q54 Stephen Phillips: Has it been looked at within the past year? The past two years? The past five years? If you can’t say, that is fine—you’ll have to write to us—but what is the general picture?

              Robert Devereux: I don’t think I know of a formal review of it. If I can just go back to what my colleagues were saying earlier, you encouraged them to say that the system was unfit for purpose because it was trying to do two things simultaneously. When pressed about the conversation around the water cooler, where they deliberately—

              Q55 Stephen Phillips: I heard the evidence as well, Mr Devereux. You are here to answer my questions, if you don’t mind. You accept that the incentive system points in two different directions.

              Robert Devereux: Do I accept that? No, I don’t actually. What the incentive system is trying to do is two things, but I am not sure I do buy the idea that that is in two different directions. We have tried to create a system in the incentive system where I pay them less if they get it wrong. Separately from that, I allow them to keep every overpayment they have, so there are two things going on here. When a lot of the Report was written, it was all about the incentive system and not about the recovering of the overpayment.

              Q56 Stephen Phillips: But the difficulty with that, as we see from paragraph 2.12, is that although you allow local authorities to keep 100% of the amount they recover, they only end up getting about two thirds, even including the 40% that the Department pays them. So there is still a disincentive to identify fraud and overpayment. That is right, isn’t it?

              Robert Devereux: No; I think your arithmetic is wrong there.

              Stephen Phillips: Help me out.

              Robert Devereux: They declare an overpayment, let’s imagine it is £100, and as a consequence I am only going to pay them £40 of subsidy, not £100 of subsidy. So they are £60 out of pocket.

              Q57 Stephen Phillips: So they get about two thirds themselves?

              Robert Devereux: The overpayment is £100. They are collecting about two thirds of the overpayment. Two thirds of £100 is actually slightly north of £60, so they are actually net in profit.

              Q58 Chair: If. If.

              Robert Devereux: No. They are collecting 67% on average. So on average, the system puts them in money.

              Q59 Stephen Phillips: Let us go back to paragraph 2.14 of the NAO’s Report. You would say, would you, that you disagree with the fact that the system is currently not working because it points in two different directions?

              Robert Devereux: Well, the evidence would appear to be that they have sufficient incentive to collect back more money than I have taken away from them through the incentive, so it looks to me as if it has worked.

              Q60 Stephen Phillips: My question to you, therefore, is: has the NAO got it completely wrong when it says that the incentive system is ineffective?

              Robert Devereux: It is very easy to write a paragraph that says, “Of course, if it went one way it would mean one thing, if went another way it meant something else.” In practice, is this system delivering people’s attention to paying things properly the first time round? My colleagues seem to suggest it did indeed. Are they able to collect overpayments and are they collecting them? Yes, they are.

              Q61 Meg Hillier: I am sorry, may I just check something? We had an example earlier of someone who had had an overpayment of £35,000 and had then gone abroad, so there was no prospect that the council would set about reclaiming that. Uncovering that and declaring that to you would have meant they would have been very much out of pocket, so there is no incentive for them to declare that.

              Robert Devereux: Correct, so I am then giving you average numbers, which would obviously not include that £35,000, but by straight maths there must be lots of easy ones to collect back. To be really clear, most of the stuff that goes on in this benefit is not about complicated forensic investigation; it is about people who misreport things, late, tell the system, and we collect it back, typically by paying them less, or by collecting it. That is the nature of what is going on. I absolutely accept that some cases are difficult and leave local authorities out of pocket, but this is an instrument for dealing with many hundreds of local authorities.

              Meg Hillier: But they are not out of pocket if they don’t find it.

              Q62 Stephen Phillips: That’s really the next point. That is a very good example of an occasion on which it was not in the interests of Islington council to find that and tell you about it, but they did. What work has the Department done to identify the number of cases where local authorities are not identifying these difficult, irrecoverable amounts of wrongly paid housing benefit?

              Robert Devereux: You will see in paragraph 2.27 that the NAO records us, delphicly, as “exploring a range of initiatives”. One thing on which we have been working with local authorities, and on which I am hoping that Ministers will be able to make an announcement shortly, is a scheme that goes precisely to the point you talk about. Given a baseline of overpayments identified at the moment, what system would have to be in place to secure further overpayments out of local authorities? That is currently being worked on. Unhappily, I am not in the position to go through it in detail, but we are trying to find a mechanism that would give some incentive to collect even more overpayments than they already do. As I said, they are already collecting a lot of overpayments—they are already, essentially, quids in on average—but this would give them an incentive, which would be good for them and good for me as the accounting officer.

              Q63 Stephen Phillips: If I may go back to the summary of the NAO’s Report, paragraph 14 says that the scheme “encourages local authorities to process claims accurately and recover overpayments that are identified.” That seems to coincide with your evidence. However, it goes on: “It is not designed specifically to target fraud and claimant error and does not create strong incentives to detect overpayments after the claim has been awarded, which account for 90% of all Housing Benefit overpayments”—or, to put it another way, more than £1 billion annually. What steps is the Department taking to create the strong incentives to detect overpayments after the claim has been awarded that are currently lacking?

              Robert Devereux: I am sorry, but I do not think that the £1 billion annually relates to the 90%.

              Chair: Don’t worry about the figures; answer the question.

              Stephen Phillips: Ignore the number I gave you.

              Robert Devereux: I am sorry, could you ask the question again?

              Q64 Stephen Phillips: Okay, the Report says that there are no “strong incentives to detect overpayments after the claim has been awarded”.

              Robert Devereux: I have just explained that, as recorded by the NAO, we are in the process of sorting out a mechanism that will provide an even bigger incentive to find overpayments.

              Q65 Stephen Phillips: Right. What are the features of that going to be?

              Robert Devereux: It would be an incentive scheme, but short of Ministers finally agreeing and announcing it, I am afraid that I am not in a position to say right now.

              Q66 Stephen Phillips: Right, so it must follow from that that you accept that there has been a problem in the past that the NAO has identified and that you are now moving to tackle.

              Robert Devereux: Well, we have been running multiple different systems over different years and our conclusion is that this would be a timely time to do it.

              Mike Driver: Just to build on what Robert has said, we do recognise that we need to incentivise local government in order to deal with the large amount of fraud and error in the system. There is also an issue with the subsidy arrangement, and we have committed to work with local government to review the subsidy arrangements—

              Q67 Chair: What does that mean—put them up?

              Mike Driver: No, that does not mean put them up; it means to review the subsidy arrangements to ensure that they work appropriately. One of the local authority colleagues talked about the way in which we have a differentiated approach for dealing with the subsidy in relation to official error, for example. So we will look at that part as well.

              It is worth saying that the subsidy arrangement is not there just to deal with the incentive arrangements for housing benefit. There is also a big financial control issue associated with this. We are paying out £25 billion through the subsidy arrangements to local government in terms of housing benefit. We need to ensure that we can demonstrate to you and Parliament that that is properly spent. Part of the subsidy arrangement is to do that.

              There is a classic three lines of defence model operating here. We as the Department set the legislative framework and the controls associated with that. It is for local government to operate those controls effectively. There is then a responsibility of a section 151 officer within local government to ensure that the system is properly administered. This comes to your question about where the incentive is to find fraud and error once a claim is in train. I would argue strongly that there is a responsibility on local government to ensure—

              Q68 Stephen Phillips: We both know, Mr Driver, that auditors do not always pick things up.

              Mike Driver: I am not talking about auditors. I am talking about the administration itself, which is the first line of defence. The second line of defence is the finance director, normally the section 151 officer. The third line of defence is the external auditors through the certification process. Part of that process is to ensure that at each stage we can ensure that there is proper accountability for public expenditure.

              So, there is that side of things; then there is the incentivisation, and part of what we do is incentivise people to get it right through the application of reduced amounts where there is fraud and error in the system. We recognise that we need to review the subsidy arrangements, and we also need to think about how we can better incentivise local government effectively.

              Q69 Stephen Phillips: Thank you. Mr Devereux, when did you realise that you needed to review the subsidy arrangements because they were not working?

              Robert Devereux: You will see from the time series of data in figure 3 on page 17—the data points for several years are there or thereabouts—that when you get to 2012-13, the data for that arrived in May 2013. That was the first point at which there was clear evidence that there was growth in earnings problems. It was in the December thereafter that the Government made an announcement in the autumn statement that we would be using—

              Q70 Stephen Phillips: I am interested in the output that is going to come out of the Department shortly in relation to the change in the subsidy arrangements that you have been talking about. When did you identify the problem that has led to the output that you cannot tell me about? Was it before or after the NAO identified it?

              Robert Devereux: It was before. I think it says clearly in one of the sentences, which I will not be able to find in a hurry, that the Department was well aware of the nature and scale of fraud and error issues.

              Q71 Stephen Phillips: Moving on to a slightly different topic briefly—certification required by the subsidy regime—in paragraph 2.19, the NAO Report states, “Local authorities are not required to check—or provide evidence to local auditors—that claimants’ circumstances have remained the same.” That is right under the current regime, is it?

              Robert Devereux: Yes. Picking up on what Mike said earlier, given that £23.5 billion passes from my Department to multiple authorities that in turn are paying claimants—

              Q72 Stephen Phillips: Mr Devereux, forgive me, but I asked you a very simple question. Is what the NAO says there right?

              Robert Devereux: Sorry, is what right?

              Q73 Stephen Phillips: Is it right that “Local authorities are not required to check—or provide evidence to local auditors—that claimants’ circumstances have remained the same.”?

              Robert Devereux: They are not.

              Q74 Stephen Phillips: Why is that, given that we know from the introduction to the Report that this is the largest area for fraud and claimant error?

              Robert Devereux: Because the certification process is there to make sure that local authorities are correctly reporting to me what they have actually done by way of making determinations on claimants. We are not asking the auditors at that point to go out and check with all the claimants. I am simply trying to make sure that when individual authorities—

              Q75 Stephen Phillips: No, no, no—forgive me, Mr Devereux, but you are missing the point. That is not what this is about at all. Certification does not require local auditors to check that the circumstances of those who have made claims have remained the same.

              Robert Devereux: I don’t think I am missing the point. The certification process is not the process by which I estimate whether there are degrees of overpayment in the system. I have my own statistical methods for identifying the sums and the analysis in the Report. This is the local government auditors checking that when a section 151 officer sends me a bill and says, “Mr Devereux, please would you send me £1 billion for housing benefits?” that has been properly constructed on the evidence available within the local authority.

              Q76 Stephen Phillips: But it has not necessarily been properly constructed, because what is not being certified is that the claim is not a fraud or an error. If a sample of claims is taken and the auditor or certifying officer is not looking at whether the circumstances have remained the same under the certification regime, he is unable to properly certify to the Department that the local authority is doing what it is supposed to do.

              Robert Devereux: Even if I were able—which, by the way, I am not—to measure fraud and error accurately at the local authority level, would you prefer me to say to each local authority, “The stock of fraud and error in your system is this high and I will take all that money off you under the subsidy arrangement”? I am guessing that that is not what local authorities want.

              Q77 Stephen Phillips: It is not about what I would or would not prefer. It is about you providing answers to my questions so that this Committee can form a view as to whether public money is being properly spent.

              Robert Devereux: Yes, but you are running a line of argument that somehow calls into question whether this is a sensible way of doing certification when the only alternative to this way of doing it would be to visit on local authorities the cost of the stock of fraud and error in the system. I am asserting to you that I think it is highly unlikely that they will want that.

              Q78 Stephen Phillips: What I am asking you is why, as part of the certification process, you do not require the auditors to certify—

              Robert Devereux: We are going round in circles. I am asking them to do it in order to organise the payments to local authorities who have already paid money out to claimants. Those are two different things.

              Q79 Stephen Phillips: Right. It follows from that that local authorities are being certified by auditors as having paid out monies, some of which should not have been paid. Is that correct?

              Chair: Everyone around you is saying yes.

              Stephen Phillips: Mr Driver and Mr Buckingham are both nodding.

              Mr Bacon: Nods do not go on the record.

              Mike Driver: That is the £8 million in paragraph 2.19.

              Robert Devereux: No, it is not that.

              Q80 Chair: It is a very simple question. If this is important, should you not be collecting data on it? You could collect the data in a certification process. If it is important to the Government to cut it—they have set it as a Government priority—it does not follow that you cut the subsidy, but should you not be collecting the data to get a better understanding of the problem and then take appropriate action?

              Robert Devereux: I am trying to check what data we think we are collecting. I am already collecting information—the National Audit Office has regarded it as top of the range and world class—about what is wrong with fraud and error in the system. We do not have a mechanism for reporting that at local authority level. Even if I asked the local auditor to go and do it, they would have to sample so many cases to get a robust number that it would simply not be a practical option.

              Q81 Stephen Phillips: But they do not sample very many cases at all. We know that from the figures given in the NAO Report.

              Robert Devereux: Yes. All I am trying to say back to you, Sir, is that in order to get a robust estimate of the actual amount of change going on in claimants’ lives, on which I could then base a subsidy regime, I would have to sample way more than it is credibly possible to do.

              Q82 Stephen Phillips: So your evidence to the Committee is that there is a distinction, for the purpose of robust certification, between the number of files—examples—that they need to look at for the purposes of what is currently certified and the number they would need to look at if they were also certifying that there had been no alterations to the circumstances of a limited number of claimants within a set.

              Robert Devereux: Absolutely.

              Q83 Stephen Phillips: Why would those two data sets be different?

              Robert Devereux: I am afraid that this is to do with the ability to sample complex populations and get robust answers. There is a lot of variation in our system. At the moment, we are running a system that is warranted by the NAO as being one of the best in the world to measure fraud and error. The amount of effort that that requires when replicated by the geography of every individual authority would simply be wildly disproportionate to the purposes it is serving.

              Q84 Stephen Phillips: Okay. Had anyone thought about this before I raised these questions?

              Robert Devereux: Yes. I used to be the fraud director in 1998 and I went through that then.

              Q85 Stephen Phillips: Has there been some work in the Department to say, “It is not worth certifying that there have been no changes in the circumstances of individual claimants within the sets being looked at by the certifying officer”?

              Robert Devereux: There has been work that has established exactly how much effort would be required to produce local area estimates of fraud and error that were worth the paper they were printed on. The answer is that it is way too much effort. This Committee will not see the value of understanding 300 authorities’ information.

              Q86 Stephen Phillips: We are quite prepared to accept that, but I would like you to write to the Committee to explain the numbers and why that conclusion has been reached.

              Robert Devereux: I will very happily write to you.

              Q87 Chair: We think you are talking about over £1 billion of housing benefit fraud and error. If it costs a few million pounds to collect the data, that is not necessarily a bad investment.

              Robert Devereux: Off the top of my head, I think the current measurement system is costing us low tens of millions to run—£20 million or £30 million—to get an estimate that is accurate across the country as a whole. Broadly, if you were to multiply that by 200 or 300, you would get an estimate that is accurate in Enfield. While I could mathematically do that, that is not the purpose of certification, and I do not know what marginal information gain on which decisions and systems could be operated would derive from that.

              Chair: Max?

              Max Tse: To clarify, that is probably for the whole measurement for all the benefits, rather than just housing benefit specifically.

              Robert Devereux: Yes, it would, but it is the same.

              Max Tse: But housing benefit would be the one that has the most local need for understanding of that issue.

              Robert Devereux: It is the same degree of sampling, whatever goes on. Why don’t I write you a note? I have all this information because I reviewed it very recently.

              Q88 Mr Bacon: You said the conclusion that you came to after this was that it was way too much effort and, by implication from what you just said, way too much cost. What is your number for what the cost would be to do this work?

              Robert Devereux: That is the number that I have back in the office.

              Q89 Mr Bacon: I am quite interested in it. You said low tens of millions; that could be £20 million or £30 million.

              Robert Devereux: I was offering you my memory that the current system, which produces national estimates, costs about £20 million. I think it is about that. I will write to you to give you the right figure, but we have simultaneously tried to estimate what it would cost to do this at much lower levels of geography and we have computed a cost. I cannot remember it.

              Q90 Mr Bacon: In a way that is robust?

              Robert Devereux: In a way that is robust. It is very easy to go and find 10 cases in Enfield and says, “Gosh, look, 30% of them are wrong.”

              Q91 Stephen Phillips: I don’t know why we are picking on Enfield, but let’s carry on. I am sure Mr Burrowes will not mind.

              Robert Devereux: We could try Barking if you like.

              Stephen Phillips: When the independent auditor turns up in Enfield to do the certification process, how long does it take?

              Robert Devereux: I do not know. They do not do that many cases.

              Q92 Stephen Phillips: If you do not know, do not speculate, that is fine. You can tell us that in writing as well.

              Mike Driver: The auditor is not paid by the Department. This is part of the process that local government goes through in completing its accounts.

              Q93 Stephen Phillips: It is paid indirectly, in a way, because you are providing administrative support through the subsidy, aren’t you?

              Mike Driver: I accept that point, but the way in which the auditing process works is that it takes several months from the end of the accounting year to move from the subsidy to the certified approach.

              Q94 Stephen Phillips: We are pressed for time and we are getting away from where I want to get to. You do not know how long it takes. How many files does the independent auditor ask for when he goes to Enfield?

              Steve Buckingham: The certification process starts in the April following the previous financial year. The process is due to finish, signed off by the section 151 officer, by the end of November in the same year. That is the timeline. When the auditor goes into the local authority they have three main groups. They take 20 cases from each of those groups. If in one of those individual groups they find one case that is irregular they do a second set of sampling, which means that they then take 40 cases from that same group. In the very worst-case scenario, you would have three cases of—

              Q95 Stephen Phillips: 120 and another three lots of 120; is that right?

              Steve Buckingham: No. You would have 60 across each set of three—

              Stephen Phillips: Sorry, 60 and then 120.

              Steve Buckingham: And then you would have 40, 40, 40 if all three found one irregular case.

              Q96 Stephen Phillips: Let us just assume the bog-standard 120 when the auditor goes in. I don’t really understand why, when he is looking at those files, he can’t look to see whether their circumstances have changed.

              Robert Devereux: Don’t get me wrong, we could all agree that we are asking him, in addition, to go and ring people up and see whether their circumstances are right. He is probably not trained to do that, by the way, but never mind. When he has done that we will have data on 120 cases in the whole population of that local authority. That will not produce a statistically robust estimate of anything at all. You can do it, but it will not mean anything. The information content will not be worth the money—which, as I have just said to Mr Bacon, I suspect we do not want to spend. Don’t get me wrong, I am not being cash constrained in thinking I do not want to spend this—if it was necessary, that is what we’d spend—but there is only any point in spending money to get information if the information has any merit.

              Q97 Mr Burrowes: In terms of investigations, I understand from your website that the number of full-time equivalent fraud investigators has fallen between 2013 and March 2014 by about 4%. There has been a decrease in the number of cases referred to fraud investigation of about 14% in 2013-14 compared with 2012-13, while, as I understand it, the value of housing benefit over payments has increased by 9% and stands at around £1.38 billion. Those are the last quarter figures. Do you see any correlation of those figures?

              Robert Devereux: Not directly, no. As I explained earlier, the increase in fraud and error is being driven by claimant error largely to do with earnings. This is not actually a matter to do with fraud investigators uncovering it. People are ringing up and telling us that they got it wrong and we correct it centrally. We are, indeed, ending up with fewer people, and fraud investigators have fewer front-line staff and fewer corporate staff. The organisation is currently running 33% real less cash than I had three years ago. So I am making choices. In that, I am trying to drive up the productivity of people. So knowing how many people I’ve got is not quite as interesting as whether they are being productive in the things they are doing. With referrals, for example, what I actually want to do is to have really good quality referrals. A reduction in referrals does not of itself mean that I am not getting really good quality ones; I may just have got rid of the less good ones.

              Q98 Mr Burrowes: So the good quality ones, is that in terms of the value of those referrals?

              Robert Devereux: It is do with whether, when the fraud investigator gets the referral, there is something substantial to go after. At one end of the spectrum of referrals somebody rings up and says, “I think my neighbour is doing something wrong—

              Q99 Mr Burrowes: No, no, no. I am talking about those that lead to a sanction and some kind of disposal. You say “quality”. Is that quality where it is a large amount of money or quality that leads to a disposal of some sort?

              Robert Devereux: Okay, we may be talking at cross purposes. When the fraud teams are trying to work out where best to deploy their effort they are trying to do two things: one, go after things that have large overpayments associated with them; and, secondly, make sure where they possibly can that they are getting significant criminal penalties for it because of the deterrent value, whatever the monetary value. They are making it clear that the Department is prepared to chase you down, find you and if necessary provide a criminal sanction.

              Q100 Mr Burrowes: You may have heard the evidence from one of the local authorities in Kent talking about cautions and administrative penalties. That isn’t a performance measure for the Department. Perhaps it is not surprising that the number of cautions offered and accepted in 2013-14 dropped by around 27% compared with the last year. The administrative penalties offered and accepted have also dropped by around 16%. Would you say that is a deliberate strategic effect of how you said you wanted to go for the quality fruit?

              Robert Devereux: Probably not. I am sorry but I am not familiar with those numbers.

              Q101 Mr Burrowes: So why are they dropping? Why are administrative penalties dropping considerably? Why are cautions dropping considerably? You don’t know the answer to that? This is your press release on housing benefit recoveries and fraud national statistics. You must know about them and the fact they have dropped.

              Chair: Who is the SRO?

              Mr Burrowes: You must know these figures.

              Steve Buckingham: I haven’t got them with me.

              Q102 Mr Burrowes: But do you know about these figures? It doesn’t matter whether you have them with you. I have just prepared for this hearing today. This is your full-time occupation as an SRO to look at these figures. You must know about them.

              Steve Buckingham: Yes.

              Q103 Mr Burrowes: You do know about them?

              Steve Buckingham: Not to the level of detail to answer those detailed questions.

              Q104 Mr Burrowes: This is just a press release. It is not a big, detailed document. It is on your website.

              Robert Devereux: I don’t think he does have the information.

              Chair: What does it say? Tell us.

              Q105 Mr Burrowes: It simply lists the number of cases. The number of cautions offered and accepted in 2013-14 has dropped significantly by around 27% compared with the previous year and the number of cases related to administrative penalties offered and accepted has also dropped by around 16%. It goes on. There has been a decrease of around 14% in the number of cases referred for fraud investigation—

              Robert Devereux: So let us write to you. Because I have focused on the Report and not on that, we have not come prepared. I am sorry.

              Mike Driver: Mr Burrowes, are those fraud figures for housing benefit that you are referring to?

              Mr Burrowes: Yes.  The local authority ones.

              Steve Buckingham: Those data are collected from the local authority, which we then use, via the performance development teams, to discuss with the local authorities where we think they might like to focus their efforts.

              Q106 Mr Burrowes: But it is a question about quality and quality cases. We heard earlier evidence from a local authority that they are doing all they can on administrative penalties but there is no real mark or incentive for them to focus on this area of work.

Robert Devereux: No. This comes slightly back to the potential organisation—we have a second tier of government here. We have given local authorities statutory duties to do some things under the legislation, so we are not working on the basis that everything that has to happen has got to have some financial incentive associated with it. I have already accepted, because I have told you I am going to do something about it, that we are going to put some more money in to help them provide more incentives for turning over other things.

              Q107 Mr Burrowes: But we then have the single fraud service, and that is a direct Government initiative and approach. What do think the impact is going to be on these local authorities, in terms of their local investigation teams? What impact will it have on the statistics I have just given you?

              Mike Driver: If we look at the fraud error and debt business case first of all, for relatively modest investment we expect, through to 2021-22, to collect an additional £559 million as a result of the creation of the single fraud investigation service.

              Q108 Chair: How much out of housing benefit?

              Mike Driver: About half of that will be housing benefit. In terms of the single fraud investigation service—and we have started rolling that out, and we have a timetable to do that—I think under the current arrangements, in the best situation, we have had two individuals, one from the Department and one from local government, potentially working on the same case, maybe slightly different aspects of that case. So I think that aligning the approach that we have, so that we can focus one individual using all the data that we have available to us, in order to create a more efficient fraud service, is useful. I think it is also worth noting that by aligning and centralising activity in a national organisation—because there may be criticism that we are centralising—we allow ourselves to deploy resource into areas where we find higher levels of fraud and error. So you may have picked up in the media we have been focusing on the 50 areas in the country where there are the highest levels of fraud and error, through a media campaign.

              Q109 Mr Burrowes: So in Enfield, where four staff involved in the investigation of housing benefit fraud—and we heard earlier evidence of the value of the local intelligence built up, not least being able to see the local recidivists and other local characteristics of fraud—can be deployed: how will you be able to replicate that capacity of four staff who have gone?

              Mike Driver: Those staff, if they are dealing with housing benefit fraud and error, will be joining the Department.

              Q110 Mr Burrowes: No, they are going. In February it is moving over. They are going.

              Mike Driver: They are not going. They are not disappearing.

              Q111 Mr Burrowes: I have spoken this morning to the head of the counter-fraud service. They are going. They are gone. They will be gone by February.

              Mike Driver: They have gone where?

              Q112 Mr Burrowes: They will no longer be employed—there will no longer be four staff focused on counter-fraud in relation to housing benefit.

              Mike Driver: In Enfield?

              Mr Burrowes: Yes, in Enfield.

              Mike Driver: One would question why we give them housing benefit administration in order to deal with fraud issues.

              Q113 Chair: I have to say, Mr Driver, I have kept my cool on this, but you have cut the money to them; you are using any money you have got to set up your national fraud capability—you say, so that you can focus on areas of greatest need. Of course Enfield, or Islington, or East Kent or wherever it is are going to get rid of their staff. You have cut them; they cannot with the current cuts carry on. Do not live in a theoretical world when you are a pragmatic person—

              Mike Driver: Chair, I am not living in a theoretical world.

              Q114 Mr Burrowes: I can tell you a local authority that has made great strides in this area, as well—

              Robert Devereux: Can I just check? It is obvious that they are no longer going to be working in Enfield. Are you saying that the four individuals are also not going to be working for us? Because I can well imagine that they are not there, because the creation of a single fraud investigation service means they are no longer employed by the local authority. [Interruption.]

              Mike Driver: Thirty-two local authorities have now been centralised into the single fraud investigation service, and those staff have transferred into the Department.

              Robert Devereux: They are in the areas they are in because we recognise I need local experience.

              Meg Hillier: Where are you based? Could you be clear on this?

              Q115 Chair: Where will the Enfield staff go, and where will the Hackney staff go—or the Barking and Dagenham staff?

Robert Devereux: As you are well aware, I have got a large network of retail outlets called jobcentres and quite a lot of fraud investigators of my own currently operate out of those. They are not spending their time behind desks; they are doing excellent work—

              Q116 Chair: So they will be taken out of the town hall and stuck around the corner?

              Mike Driver: If I could give you an example of something I do know, the single fraud investigation service currently covers Hillingdon. The person in Hillingdon who was dealing with fraud transferred from the local authority and is now working in the jobcentre in Hillingdon.

              Q117 Chair: Big deal—what is the point in doing that?

              Mike Driver: I just explained the point.

              Robert Devereux: At the moment there are three classes of people: tax credit investigators; local authority fraud investigators; and DWP investigators. Not all of them have got the same powers or the same access to data and yet, as far as you and I are concerned, those three things are all welfare-related issues where we want to crack down on fraud and error. So the SFIS proposition is one that says, “Wouldn’t it be better if we pooled this and did it well?”

              Q118 Mr Burrowes: The head of counter-fraud in Enfield told me that the local knowledge to deal with not least recidivists and others will be lost in the form of those four members of staff. Are you saying that that is wrong?

              Mike Driver: I would be very happy to follow up with you—

              Q119 Chair: No. Is it wrong or right? It irritates me when you give that answer. You must know that, because you are changing it.

              Robert Devereux: We have not created this system in order to have everybody work out of Leeds. The whole point in creating the system is to have local knowledge about local things. I need to go and see people and knock on doors.

              Q120 Chair: Then why play around with what you’ve got?

              Robert Devereux: Because they have not got the right powers, Chair.

              Mike Driver: If you take Enfield, if we have a problem in Barnet, we could deploy those staff who were in Enfield in Barnet. You cannot do that under the local government arrangements.

              Q121 Mr Burrowes: The reality is that you have a Falcon team, headed up by the Met, that works alongside boroughs to deal with those issues across boroughs.

              Mike Driver: By the Met?

              Q122 Mr Burrowes: The Met have just recently set up a Falcon team to work alongside local authorities in collaboration. They are concerned that that may be supplanted.

              Robert Devereux: We may be going around in circles. The Department has made a decision that having 300 teams of three or four doing local work on housing benefit who are not connected with all the people doing benefits or tax credits is a sub-optimal system for the public as a whole. The proposition is not to—

              Q123 Mr Burrowes: But the reality is that housing benefit fraud has different characteristics across boroughs. You will see different flavours of it in Hackney from those in Enfield or Waltham Forest. The people on the ground who are best able to deal with that are those who know who to visit, with the local connections between those who are simply absent tenants and those who are cash-in-hand people. They have good on the ground expertise.

              Robert Devereux: So do my staff. I do not know what your counter-factual is. If you think that we have sucked all this resource out and I am going to sit them in some pristine building miles away from your people, you are wrong.

              Q124 Mr Burrowes: In terms of wider strategy, as has been said to us, wouldn’t it make more sense for this to be rolled out at the same time as universal credit?

              Mike Driver: May I make a point about Enfield? In Enfield, staff are due to be transferred to the single fraud investigation service in March 2015. If you look at the performance of Enfield in 2013, it achieved 0.87 sanctions per 1,000 cases, of which 0.43 saw successful prosecutions. That placed it in the bottom quartile in terms of performance in dealing with both sanctions and prosecutions.

              Q125 Chair: Can I ask you a simple question? You are taking people out of local government and putting them into these new, wonderful Jobcentre Plus-led teams. Will we have the same number at the end as we had at the beginning?

              Mike Driver: We would hope so. If we take Enfield as an example, where there are four people, those four investigators will be spending some of their time dealing with housing benefit fraud and some on other types of fraud. In an organisation like Enfield, it may be that they supplement the four with five, six and seven in order to deal with other types of fraud, such as tenancy fraud, blue badge fraud and whatever else it might be.

              Some very small local authorities may have only one fraud investigator. What we have agreed in those circumstances is not to denude them of the resource that they have got in terms of the individual, so we may have to take a bit of a hit in terms of staff numbers in order to make sure that the local fraud investigation in those other areas is not diluted too much.

              Chair: Can I ask one other question?

              Q126 Mr Burrowes: We did not get an answer to the wider question about the timing of the roll-out of the single fraud service being at the same time as universal credit going online. Is there logic in that?

              Chair: That is just what I was going to ask.

              Robert Devereux: I am not sure. A large number of things that universal credit will bring are already being rolled in in advance. We have put the claimant commitment all through every jobcentre before universal credit. We have started to do some of the service—

              Q127 Chair: Mr Devereux, do not take us off on a tangent. Local authorities are running housing benefit. It seems silly to take out the people who are on the ground trying to identify fraud and error before you have taken the function away from them. It just seems silly. It is not a theoretical thing. You have got these guys running the system at the moment for the foreseeable future. Why not reorganise your fraud function when you reorganise the system? That is the question, isn’t it?

              Robert Devereux: Because even now, in the pilots that we ran that predated the SFIS, it turned out that this was a better thing to do.

              Chair: I don’t want to talk about universal credit.

              Mike Driver: This is the pilots for the single fraud investigation service. We ran pilots in nine local authority areas. They demonstrated greater value, which was supported because—

              Q128 Chair: Greater value in what? Housing benefit fraud detection?

              Mike Driver: Across the piece. Shall we send you the evidence? It seems to me that we are in a position here where the rationale for doing this, and the fact that it has been tested, is clear but it was not part of the Report. Shall we write you a note that simply says, “This is the evidence on which we have made a strategic choice?”

              Q129 Mr Burrowes: We are hearing about the value of pilots, but we have heard from two heads of counter-fraud who have experience of many years—I have also heard my own head of counter-fraud saying this—that it would have made more sense for the roll-out to happen at the same time as universal credit going online. Do you accept that?

              Robert Devereux: No, I do not. I do not see what the connection is.

              Q130 Meg Hillier: Following up on that, I do understand that with real-time information and links to other benefits, there is a point at which the DWP is well placed to have all the computer systems talking to each other so they can easily see. But while housing benefit is administered locally, you do not have oversight of that in the same way. Surely those officers should be based in the local authority. That is the whole reason why they are there now.

              Robert Devereux: Is it worth repeating again that most of the overpayments that this Report talks about are official error and claimant error? They are not fraud. The principal skill of the fraud investigators is to go after fraud, which is a small part of the story, and it is a really important fact that a lot of people who are defrauding housing benefit are probably defrauding JSA, and they could well be defrauding other things. All we have tried to do—this is what I said we have tested—is to say, “Wouldn’t it be a good idea if, instead of having three lots of people all with one hand behind their back because they cannot actually investigate all the other circumstances and they do not have access to all the data, we put them in a position where they have access to all of them?”

              Q131 Meg Hillier: But there is some collaboration, as I understand it, that goes on between those bodies. On fraud, I take a bit of what you are saying, but the thing about fraud is that it is damaging to confidence in the system and it is illegal. It is quite right that people who wilfully, illegally and criminally defraud the system should be sought. Even if it is, as you say, a smaller amount than on other issues, local authorities are in a better place to do that. With real-time information, a lot of the other issues will, as you have indicated, resolve over time. For that criminal fraud, the local knowledge and the ability to visit and take action by the body that pays the money out is surely, as Mr Burrowes has said, just logical common sense.

              Robert Devereux: My own fraud investigators are local. You cannot be a generic national fraud investigator. You have to have local knowledge to do it at all.

              Meg Hillier: But you do not go into people’s homes?

              Chair: Let’s move on.

              Q132 Stephen Phillips: Before we do, to be fair to Mr Devereux, let me ask this: what were the benefits that were shown by the pilots of rolling this out nationwide long before universal credit?

              Robert Devereux: The benefits—we will drop you line about what is underpinning the benefit case—rest on the fact that by enabling people to have systemic access to common data, and systemic powers that are the same—

              Q133 Stephen Phillips: Is it basically that more fraud is detected?

              Robert Devereux: That is what the business case is, yes.

              Mike Driver: Yes. We are happy to share the business case with you. The business case, which is signed off by the Treasury, shows benefits just on the single fraud investigation service of £559 million.

              Q134 Stephen Phillips: Can I ask a related question? As I understand it, 42% of the fraud and error across the DWP budget is housing benefit. Is that correct?

              Robert Devereux: Gross.

              Stephen Phillips: And yet you only spend 8% of the budget for detecting fraud and error on detecting fraud and error in the housing benefit bill. How do you account for the discrepancy?

              Robert Devereux: Three quarters of all housing benefit case load get housing benefit because they already have a qualifying benefit from my Department. If that underlying qualification is wrong and there is some error in that, it can have a knock-on effect on housing benefit. What you are comparing is the precise amount of money I was spending purely on housing benefit, when housing benefit is sitting on top of a large edifice of entitlement. All the people of working age and out of work or sick, and all the pensioners on pension credit, are just passported on to this benefit. If I do not keep that straight, housing benefit would be riddled with its own errors just from that.

              Q135 Stephen Phillips: Tell me if this is unfair, but in fact you would say that you did not need to spend more than 8% of the budget because it is better directed at the other benefits that passport people into housing benefit in the first place.

              Robert Devereux: Nearly. There is a rather excellent table in the back of the Report that explains the multiple causes of fraud and error. A lot of those living together is a problem for DWP benefits and local authorities. There are some classes of problem that are peculiar to housing benefit, for example, tenancy fraud or rent problems, or the adults in your household—we don’t worry about that if you are in out-of-work benefits. We need to ensure that we are putting the appropriate effort into those. The single largest element of the entire shooting match is earnings, which is a particular housing benefit issue, because most of my benefits are paid to people who do not have work.

              Q136 Stephen Phillips: But as you said, most of that ends up coming back to local authorities.

              Robert Devereux: No. What I said was that a lot of the work that local authorities do is to collect back stuff they have been told exists. They are not going out with magnifying glasses and finding it. It is actually quite hard. How do you find people who have earned it? That is why RTI is potentially a breakthrough in the subject of housing benefit.

              Q137 Chris Heaton-Harris: I have a couple of very general questions. We heard from the previous witnesses that they provide a report of a general number of cases and what have they found, which goes to Mr Buckingham at some point. Do you know which is the best and the worst-performing? I would like to think from Mr Driver’s stats that we got earlier that you do know which is the best performing and which is the worst performing local authority in England on eliminating error.

              Steve Buckingham: No. What we do is focus on the performance measurement that we have at the moment, which is based on speed of processing. We use the performance development teams to work with the best-performing and the lowest-performing local authorities, based on speed of processing. That is speed of processing for new claims and for changes.

              Robert Devereux: I should add, because it is a stats angle, that the reason why speed of processing is important is that in a world in which things are of short duration, by the time a late-processing person does it this problem will have gone away, but with about six or seven weeks of problems.

              Q138 Chris Heaton-Harris: I understand the point about speed of processing, but there are two points to this. I am interested to know how you can help local authorities spread best practice between each other. If you are not measuring that in any way, it must be difficult to do that.

              Steve Buckingham: We use the speed of processing measurement and we do know who is the quickest and who is the slowest. We use that information via the performance development teams to share whatever it is that those local authorities are doing with local authorities that we think are underperforming against certain metrics.

              Q139 Chris Heaton-Harris: So the speed of processing directly links to the number of errors, does it?

              Steve Buckingham: No, not directly, but we use it as a proxy to see who are the better-performing local authorities on that measure.

              Q140 Chris Heaton-Harris: So if someone is a brilliant typist, or you are Islington and you have people entering data quicker than in dear old Enfield, which is obviously a basket case, you are saying that Islington will have fewer errors than Enfield.

              Robert Devereux: The way this works is this: imagine the individual who does not declare some earnings for a while and then eventually loses those earnings. The person who is on the case and trying to be on top of what is happening is able to interact with that person faster than someone who is taking a long time to do it. It is a blunt proxy, and it is the lead measure we have got.

              At the risk of irritating you, as I can’t explain this very well, the thing we are trying to establish is, relative to the amount of overpayments they are collecting, how can I incentivise them to collect more overpayments wherever they are in the distribution? That will get me into a different sense of who is good at collecting overpayments.

              Q141 Chris Heaton-Harris: There are a couple of points in the Report that particularly interested me. A fraud, error and debt taskforce has been set up to give local authorities more support and challenge and better info and to make use of existing and planned data matching. How do you know where to send them? Surely you would want to send them to the local authorities that most need the help. Where does this taskforce go? Where has it gone recently?

              Mike Driver: The fraud, error and debt taskforce is a Cabinet Office group chaired by the Minister for the Cabinet Office, Francis Maude, and Sir Jeremy Heywood. It is a jointly chaired group that holds us to account for whether we can achieve the 1.7% MVFE target.

              Q142 Chris Heaton-Harris: You have sent letters out. The Report says you sent 25 letters out in 2013-14 to local authorities that need a bit more help. How did you identify those?

              Steve Buckingham: It is not the taskforce—we get feeds from local authorities via the single housing benefit extract, which gives us information on certain activities that take place within each local authority. We also have the ATLAS system, which translates daily down to local authorities the number of changes that have been effected on passported benefits, so we use some of the analysis around that to identify which local authorities are actioning those changes of circumstances and which ones aren’t. We will use that, plus other data such as speed of processing, to focus where the performance development teams then go and have conversations with local authorities. It is a combination of data going to local authorities and coming from them.

              Q143 Chair: I have a couple of questions, Richard has one and then I think we might be there. Mr Buckingham, how long have you been in post?

              Steve Buckingham: As the SRO, since September.

              Q144 Chair: How many SROs have we had in charge of housing benefit fraud and error in the last five years?

              Steve Buckingham: I believe I am the first one.

              Robert Devereux: We have been running the programme for fraud and error as a whole, of which—

              Q145 Chair: Housing benefit is a part. Okay. Tell me how many SROs you have had for benefit and fraud in the past five years.

              Mike Driver: In terms of the fraud, error and debt programme overall, there have been three SROs, including the incumbent. The first was John Oliver. I was the second SRO, and the SRO now is the financial controller within the Department, Jon Fundrey. In terms of Steve’s role, he is the SRO looking specifically at housing benefit fraud and error.

              Chair: That is a new job.

              Mike Driver: It is the first time we have put anyone on a dedicated basis just to look at that.

              Q146 Chair: Can I ask another question about the other side of the coin? Fraud and error seem to be coming largely out of claimant error. Therefore, if you look at the other side of the coin, underpayment is probably largely due to claimants not recognising their rights. There are some quite stark figures for people not in the housing benefit system at all who ought to be entitled to housing benefit. The total amount of housing benefit unclaimed—I think this is a figure by Age UK, but I may have that wrong—is between £1.85 billion and £3.1 billion.

              My slight concern—I have said this to the NAO—is that the focus is on this part of the equation and not on the other, which matters for my constituents who are really struggling. Can you talk a little about what you are doing about the unclaimed housing benefit to which people in the system are entitled—and those who are not even in the system because they do not recognise that they are entitled to housing benefit?

              Robert Devereux: On the first point, I am looking to see whether I have brought the underpayment numbers.

              Chair: It is about £400 million.

              Robert Devereux: The extent of underpayment is a much lower number, not least because people have a rather bigger incentive to get in touch with us if their circumstances are the other way round.

              Chair: It is still important to those people.

              Robert Devereux: Yes. In terms of what you can do to drive up take-up, part of the problem is people who don’t know they are entitled to it in the first place.

              Q147 Chair: I am talking about two figures—those who claim but are under-claiming, probably because of a change of circumstances, and those who don’t claim at all. Obviously it is a high range—£1.85 billion to £3.1 billion are the two figures for people who aren’t claiming at all. But there are also people who, because of a change in circumstances, do not up their claim, which is the £400 million.

              Robert Devereux: On the £400 million, those are people in the system who we are assessing as not having disclosed to us that, given their circumstances, they could have more. The only way we can deal with that is to encourage people to keep reporting to local authorities.

              Q148 Chair: And what do you do? What I am really saying is that this afternoon you have gone through a whole series of actions—what you are doing, your strategies and your incentives—to get money in that is owed. I am not 100% sure of what you are doing to help poor people who ought to be entitled to it.

              Robert Devereux: That is a fair question. Ironically, some of the things that we have been doing most recently on the overpayment side—particularly because claiming low is an issue—have been on trying to get people to recognise that it is their responsibility to tell us. The most recent adverts we have been running are about keeping your claim up to date. That cuts both ways—keep your claim up to date to get you in as well as to stop there being overpayments. That is not a particularly strong argument, and we are spending rather more of our time on trying to make sure that the £1.4 billion does not go out through the door. On the structural question—

              Q149 Nick Smith: How successful is that? After your advertising, have more people put in claims that they are rightfully entitled to?

              Robert Devereux: The short answer is that underpayments have been pretty much steady for the past four or five years. There is no material change in the level of underpayment.

              Q150 Nick Smith: So that is not working?

              Robert Devereux: Consistent things that successive Governments have done have not improved on underpayments, no.

              Mike Driver: Underpayments in the system are about 0.9% and have remained at about that level, while these are reported in our accounts at—

              Q151 Chair: The other bigger group, which are the bunch—

              Robert Devereux: The other big group—the people who are potentially entitled but do not get anywhere near us—are a more difficult thing for us to deal with, if I am honest. I know you do not want to talk about it, but one of the consequences of universal credit will be a much stronger take-up for housing benefit because it comes as part of the package. There is a lot of evidence that there are people on DWP out-of-work benefits who do not know that they are potentially entitled to housing benefit.

              Q152 Chair: But until that comes—when ever, if ever?

              Robert Devereux: I realise that it may not be the easiest thing to say to you, but none the less it is a part of the story. On both of those, you have a fair point. This is a persistent problem and it is quite difficult to know how best to address it, because you are trying to identify people with whom, by definition, you have no relationship.

              Q153 Nick Smith: Our takeaway from this discussion is that you are not doing anything on that.

              Robert Devereux: On that I am not doing anything other than advertising the availability of state support to millions of people, many of whom take it up. Trying to identify the class of person who has not done it is a rather difficult thing to do.

              Q154 Mr Bacon: Mr Devereux, £23.9 billion is a lot of money. It is over 3% of all Government expenditure. I am going to park the question of whether the money could be better spent, but this is a subsidy for people to pay their rent because rents are too expensive, and the reason that rents are too expensive or too high is because there is not enough housing. As chair of the all-party group on self-build, custom-build and independent house building, I could segue quite easily on to ways that we could better spend the money, but I will not. [Interruption.] Mr Phillips has just pointed that out.

              I am interested in the intractability of this problem. I have in front of me the following text from one of our reports.

              Robert Devereux: Text?

              Mr Bacon: Text—not as in a mobile phone, but as in words. “It is totally unacceptable that seven years after we last looked at this issue, Housing Benefit fraud should exceed £900 million, and the Department still do not have information to show whether fraud is increasing, or all the information they need on the types of fraud, including landlord fraud, and variations at regional and local level. The absence of reliable information must cast doubt over the decisions the Department have taken to invest in anti-fraud work and over the achievements they have claimed…The Department expect to have better information once the second Benefit Review is completed shortly, and they are also looking at the feasibility of measuring changes in the levels of fraud at local level over time”. Would you like to tell me when you think that was written?

              Robert Devereux: I think about 1990.

              Q155 Mr Bacon: It was 1998. I repeated it to your predecessor’s predecessor, Sir Richard Mottram, in March 2003, when I also asked him, “Is housing benefit fraud going up or down?” He said, “We do not know. We think down, but we do not know.” That was 11 years ago. The quote that I gave you was from 16 years ago. You still did not know whether it was going up or down 11 years ago, and you didn’t know five years prior to that.

              Here we are in 2014. I see that the central case is for 5.8% overpayments, with a range from 4.7% to 7%; somewhere between 4.7% and 7% of these payments should not have been made—they are overpayments for whatever reason, whether claimant error, fraud or official error. I cannot think of many commercial organisations that have overpayments of 5.8% and survive for very long, but of course you are not a commercial organisation.

              Plainly, there is a very intractable problem here, but in answering questions about how you deal with it, you made it quite clear that there was a certain amount of effort that was not worth making because of the cost, time and effort involved in trying to get better at it. I am not saying, however, that you have not been doing things to try to get better—we have spent most of the past two hours discussing precisely that.

              Nevertheless, if you simply add up the overpayments over the past 16 years since 1998, the £900 million for claimant error would give you £14.4 billion. There is £340 million because of fraud—the £900 million figure that I mentioned from 1998 was fraud, but if you did it on the basis of £340 million it would give you another £5.4 billion, and if you did it on the basis of £900 million it would be another £14.4 billion. The official error of £150 million going back over the past 16 years at roughly the same rate would give you another £2.4 billion. In total, since 1998, that is somewhere between £22.2 billion and £31.2 billion.

              Now, I do not think that those numbers are likely to be correct—they are done on the back of an envelope—but they are not an order of magnitude wrong or a decimal point wrong. You have overpaid and not got back many, many billions, some of which could have been deployed to get better information of the kind that you say is too much effort. When you are paying out so much money—3% of all Government expenditure on this one benefit—and 6% of it is being overpaid, the result of which is that over 16 years you have paid out somewhere between £20 billion and £30 billion more than you should have, at current prices, without getting it all back, I don’t understand why it is not worth spending considerably more money on getting better information, including at a local level so that you can make comparisons between different local authorities and finally grip this, because it’s been going on for really rather a long time.

              Robert Devereux: Some of the quotes that you read out are about us not knowing how much fraud and error there is, but you are looking at a document that tells us that we have six or seven years of fraud and error data, so we sort of know how to do that now, don’t we?

              When I said it was not worth spending the money, that was on a peculiarly narrow question to do with certification, right? I have tried to explain this afternoon that the principal area where things are going wrong with this benefit is because in a world of fluctuating earnings that are not regularly reported to me, the required system change is to grab hold of those earnings and first of all match them against what is being paid, or better, to calculate the benefit on the right basis to start with. I don’t need 300 levels of measurement to know that that is the right answer—it is the right answer.

              With the greatest of respect, we have been trying to think of the actual strategic things that have to be done to tackle this benefit, and the single largest is—

              Q156 Mr Bacon: But you were thinking about that years ago. “The Department expects to have better information.”

              Robert Devereux: With respect—

              Q157 Mr Bacon: With the greatest of respect, Mr Devereux—I think that in the theology there is a difference between respect and greatest respect, and if you can play that game, so can I—the fact of the matter is that you expected to have better information 15 years ago. Perhaps the Treasury has something to say. You have been giving many billions of pounds over to the DWP to pay out; has the Treasury been content to do that?

              Marius Gallaher: I don’t think that we are ever content about fraud and error—

              Mr Bacon: One might say that you are never content about anything.

              Marius Gallaher: But we are quite happy that the Department has tackled the problems as well as they could with the resources that they have been given.

              Q158 Mr Bacon: That is my exact point: the resources are huge. There are billions and billions of pounds of resources.

              Robert Devereux: I cannot read your quote about better information, but I am guessing that that is actually to do with estimating. The information that is available to the Government today, consequent upon the profound change that HMRC has made in RTI, is in a completely different ballpark from where it was previously. We could have hypothesised—

              Q159 Mr Bacon: But when you say in a different ballpark, you mean better.

              Robert Devereux: Yes, better.

              Q160 Mr Bacon: But none the less, Mr Devereux, look at the NAO’s conclusions in figure 20 in appendix 3. In paragraph 4 on page 50 it says, “The Department’s measurement does show an increase in the central estimate”, and “There are several reasons why we feel this trend raises concerns and indicates that fraud and error overpayments have increased” and goes on to set out what those are. Moreover, it says that “our simple tests under-report the significance of estimated increases.” So, despite the fact that it is in a different ballpark—better than it was—it is still going in the wrong direction.

              Robert Devereux: I nearly said “with respect” there but I won’t, not that I don’t have any. The argument that I was making is that if you look at the problem as it is currently characterised, a third or more of it, even after debt collection, is to do with earnings. All Governments prior to now have had no method of collecting that because there is no centralised system for collecting earnings. This Government have put in place, in the RTI system, something that in the future—not today—will be capable of tackling that problem. This Government have also put in place, although I recognise the Chair’s view that this is some years away, a systemic change to the benefits system, which takes it away altogether. There are two things that have not been possible for anybody previously to assert or do and that are positive on this particular benefit. It cannot possibly remove the aggregate losses of the past decade; I agree with that.

              Q161 Mr Bacon: I want to finish now because I know that Mr Phillips has one more question. Mr Gallaher talked about resources and I am interested in these overpayments because they seem to be the resources that could have been available to be used. I want to ensure that we get a bit better in the future. There would have been a number in each year since 1998 for official error, for claimant error and for fraud. Can you just write to the Committee with a little list going back to 1998 and put those three numbers in each of those calendar years in a little table—for 16 years there will be 48 numbers with a total at the bottom?

              Robert Devereux: Yes. I would be happy to do that because in 1998 I was the fraud director in the Department, and at that time the next four years had significant reductions in fraud and error to the level that you have now seen in the recent past.

              Q162 Mr Bacon: It all went horribly wrong when you left, Mr Devereux.

              Robert Devereux: It obviously did.

              Chair: Thank you very much indeed.

              Oral evidence: Fraud and Error in Housing Benefit, HC 706                            1