Oral evidence: HMRC Standard Report and Accounts, HC 1526

Oral evidence, 5 September 2018.

Witnesses

Jon Thompson, Jim Harra and Justin Holliday.

Chair

Welcome to the Public Accounts Committee on Wednesday 5 September 2018. We are here today to do one of our annual highlights, looking at HMRC’s standard report and accounts. You will be familiar with some of the subject areas, Mr Thompson, and we have given you notice of some of the things we want to cover. Just to recap to anyone else who might be interested, HMRC reported £600 billion of tax revenues for 2017-18—the last year for which we have figures—which is £31 billion more than last year, so I guess that is good news. We are also keen to probe your transformation plans—thank you for the letter you wrote to us about the premises, which we will obviously want to talk about as well—and your preparedness for Brexit, and what all of that is doing to your day-to-day business. I am going to kick off by asking Caroline Flint to start.

Caroline Flint

I want to ask some questions about the country-by-country reporting, because it came into effect in 2016. Have you found any disparity between the volume and quality of the country-by-country data that you have received, against what you anticipated receiving, and how it ultimately presented itself?

Jon Thompson

I don’t know the answer to that question, although I do know the answer to what the result was, which you might be interested in. We sent out 1.64 million records and received back 1.4 million records across the 90 jurisdictions that took part in the first wave in September 2017. We have subsequently had 2,300 disclosures, worth an addition tax revenue of £55 million. There was an update from the Chancellor of the Exchequer in the spring statement about the anticipated revenue is this expands over the years. The estimate stands at £384 million over the next four years.

Chair

So over four years?

Jon Thompson

Yes, between ’19 and ’23. Revenue will rise a bit. Jim may want to add to that.

Jim Harra

To clarify, that is the common reporting standard, which started last year. The second year’s reports will be received this month. On country-by-country reporting for multinationals—I am not sure whether that is what you are getting at—we have obviously had the first round of those, and we have been disseminating the ones we have received in the UK and received from overseas. They have supported our risk assessment of large businesses, but we don’t rely on the data in them for assessing; they inform our investigation.

Jon Thompson

Sorry; I misunderstood the question.

Caroline Flint

Has it helped you better understand how companies and corporations are actually allocating their profits? How has that helped you to be more effective in collecting tax?

Jim Harra

Yes, they are. I am someone who was a bit sceptical of their value initially, and I have changed my view. They give us information up front to enable us to risk-assess. In general, we would have been able to obtain it, but to get it we would have had to open an inquiry and then use our information powers or our exchange powers with treaty partners. Now we get it automatically up front, and it really supports our risk assessment. It is beneficial. It makes a marginal difference, I think, but it will help us to target our inquires better and make them faster.

Caroline Flint

In your experience of this—obviously, it is relatively new—have there been particular problems that corporations have raised with you, in terms of having to put this information together? Has there been commonality between the different corporations, in the way they have provided the information, so you are not dealing with too much variance in the way they use their terminology or present their figures?

Jim Harra

On the latter point, yes, there are obviously some definitions that they have to follow, so they come in a reasonably standard format. On the first, yes, we get a fair amount of moaning about the burden of having to produce them. Obviously, most companies will say, “We are compliant. This is not going to change our tax liability. It is a burden on us.” As I say, it helps our risk assessment. It is valuable, and it should help the companies that are compliant, because it will enable us to better target our inquires and therefore not waste cost asking for information that in fact leads to no additional tax.

Caroline Flint

Obviously, one aspect of this is not only that corporations can demonstrate that they are paying their fair share of tax, but the implications for developing countries. Has HMRC shared the country-by-country information gathered, and were there any reasons why HMRC declined to share this information?

Jim Harra

We certainly have been sharing it. I am not aware of any instance where we have not done so. I would need to go back, check that and let you know. We have also been providing peer-to-peer support to developing countries to help them with their risk assessment and transfer pricing to make sure that they have the capability to use the data they are getting.

Caroline Flint

I understand that this is not public country-by-country reporting, but on that point about supporting developing countries, would you be able to put information in the public domain about how that has helped those relationships and how those countries are able to benefit from the shared information they are now getting?

Jim Harra

Yes, I can. We have a partnership with DFID. DFID actually funds us to do capacity-building work. I know the Chair was out in Kenya recently helping them with their capacity building in Parliament, and we support that country. We can report what we do and how we help.

Caroline Flint

Finally, what action has HMRC taken as a result of country-by-country data shared by other EU tax authorities? How many inquiries were opened as a result of that data? Are you able to say what additional tax revenue was raised as a result of those inquiries?

Jim Harra

I don’t have that information. I will check whether we can get it. It is unlikely that we would be able to point to a case where we opened an inquiry as a result of that that we would not otherwise have opened, or collected tax that we would not otherwise have collected. As I say, it is more about assisting us to risk-assess and target our inquiries and to speed them up. But I will check that and get back to you if there is any information.

Chair

If would be very helpful if you could write to us with any of that information. Forgive me; I forgot to introduce you all because you are so familiar to us. For the benefit of anyone who does not know you, including our new member—I welcome Douglas Chapman, who has just joined the Committee—we have Jim Harra, the deputy chief executive and second permanent secretary at Her Majesty’s Revenue and Customs; Jon Thompson, the chief executive and permanent secretary at the Department; and Justin Holliday, the chief finance officer at HMRC. Our hashtag today is #HMRC.

Sir Geoffrey Clifton-Brown

Welcome back, Mr Thompson. I am sure you did not have a restful summer. I am sorry to raise again an issue I raised in July, but unfortunately Monsieur Barnier said on 2 September, “By the way, infringement proceedings against London are ongoing because, according to the Commission, Chinese textile imports have not been properly cleared.” I have two questions in one. Have you any update since my question and your reply at our session in July? If the EU wishes to pursue this, is it likely to be part of the exit negotiations in terms of the amount of money, or is it likely, as an ongoing case, to have to go through the European Court procedure?

Jon Thompson

I don’t think we have moved any further forward. We dispute it in principle and we dispute the amount. We have done that in an appropriately private way, by responding following fairly significant analysis and legal advice. There has been a response from the UK Government back to the European Union about why we dispute it in principle and, in substance, why we think the estimate would be significantly wrong even if we conceded the point in principle. That is exactly where we are.

Sir Geoffrey Clifton-Brown

So in terms of process, if the EU wish to press it, is it likely that it will be part of the negotiations and within the overall figure, or is it likely to have to go through their justice system?

Jon Thompson

You are right to say that it could be either. I do not think a decision has been made. We are following the normal process: they indicated they wished to lodge an infraction and we responded to that in an appropriate way privately. It goes from there.

Chair

Thank you, Sir Geoffrey. Let us go on to the main report. We obviously look at fraud and error a lot, and we know that it increased in 2016-17, which is the latest period for which figures are available. You gave us a prediction at the end of last year, Mr Thompson, that it would be up by 7% to 8%. It went up by less than that. Why was your forecast so out? What made the difference, and what are you learning from that in terms of your forecasting?

Jon Thompson

This is set out in paragraph 3.11 on page 49 of the NAO’s Report. If you recall, there were two reasons why I predicted that error and fraud would rise: one was the commercial and profitable test being applied to self-employed claimants in tax credits; and the other was that, following the termination of the Concentrix contract in November 2016, we asked for those funds to be kept with HMRC for us to recruit staff, but a decision was made to not do that. Our estimate was that those two would individually both add 1%. As it has turned out, the commercial and profitable test has reduced now to 0.2%. Our reflection on that is that the 1% estimate that was made, and that was given to Ministers at the time, was, frankly, not adequate. It was based on too small a sample to predict the 1% rise in error rate. That is definitely a learning point for us. If you will recall, the trade—the political decision—made at the time by the Chancellor of the Exchequer was that the commercial and profitable test would reduce welfare spending by more than £500 million a year, and that an increase in fraud and error was worth the reduction in public expenditure. Nevertheless, our prediction of 1% was not adequate, and we should be transparent about that. In relation to Concentrix, we still believe that will add a further 1%. Our estimate for the next two years is that we will see error and fraud in tax credits rise from 4.8% to 5.5%, and then the following year to 6%. That is our current estimate, and it is a slight update on the forecast in the NAO Report.

Chair

Why is error and fraud in tax credits not reducing?

Jon Thompson

There are three broad reasons. First, strategically, the Government’s plan is to phase tax credits out on the migration towards universal credit. I guess that is something you may want to come back about, because the Secretary of State for Work and Pensions obviously made an announcement in the run-up to the summer about that migration period. Strategically, tax credits are not a permanent feature of the landscape, and therefore the appetite for changing tax credits is very low, because of the strategic decisions on universal credit. Secondly, there is the question of resources. To be up front about it, fewer resources are now being put into this than when it was HMRC and Concentrix. The Committee went through that is some detail. In total, fewer resources are being focused on that, because the Concentrix funding was not continued. Thirdly—this has been my long-held view, and I have been fairly frank with this Committee and several others—the tax credits benefit itself is systemically flawed, in that it is an annual benefit cycle. It really doesn’t suit people whose circumstances change significantly more frequently than that. People move in and out of employment, have more or fewer hours or move in or out of a relationship, which have knock-on effects for childcare and other costs. There are too many dynamics in the benefit itself, and it is on this annual cycle, and that level of complexity is inherently open to quite significant fraud and error. That is why 19% of cases actually have a fraud or error. I think that the last time I was here I said that in the previous year, although the error rate was 4.8%, the number of cases with an error or fraud was 20%. That is only slightly down at 19%.

Chair

You are anticipating my next question. We were puzzling over why, with real-time information, some of the issues around tax credits that you just outlined have not been resolved. Clearly, some of these issues cannot be resolved overnight, such as childcare or relationships. However, for the income element, every employer puts their real-time information in. Why has that not smoothed this over and seen that percentage of error and fraud go down?

Jon Thompson

One issue that we continue to focus on—in relation to closing the tax gap, actually—is the role of intermediaries, and particularly employers and pension fund providers. It is increasingly our view that the standard of administration of pay-as-you-earn is not good enough and is significantly variable.

Chair

You’re talking to MPs here; we can testify to that.

Jon Thompson

For some employers—we have drawn up a list of particular kinds of employers—the nature of employment and the hours are quite fluid. They are then not administering PAYE to the standard that we think they should be. That then flows into RTI, and then into this. That raises some other questions about what that means for universal credit.

Chair

You have thrown your typical bomb into the middle of the room. Let me pick up on that.

Jon Thompson

But I am being up front about it.

Chair

Absolutely. Do you have a sin list of these companies or organisations? Is it payroll bodies or employers who are in the frame for—

Jon Thompson

No, it is employers. There are broadly two main areas where you can close down the tax gap: one is about the role of intermediaries and withholding taxes such as pay-as-you-earn, and the other is about focusing on error. If you are thinking about closing down the tax gap in systemic terms, you have to focus on those two things. There is already a significant feature of the tax system, which is, what is the role of intermediaries? But we are discovering, through a more analytical approach to it, that the standards are very variable. You can draw up a list of where we think, “You are all really good at it, but you others are not,” and therefore we can make some sort of intervention.

Chair

So have you got a sin list then? You have a sin list internally.

Jon Thompson

Yes.

Chair

Will that ever be public? Are you willing to share any of that with us today?

Jon Thompson

I doubt it would be public. I can tell you that we have gone chief executive to chief executive with a couple of organisations, and said, “Look, you have this responsibility to your own employees.”

Chair

What size are the companies? Presumably, you are not talking chief executive to chief executive at some small enterprise in Hackney. You are very welcome, if you want to—I will happily host you—but presumably it is the big companies that you are talking about. What level of employees are we talking about? What sort of size?

Jon Thompson

It is going to be larger organisations and they are going to be in areas such as retail and food.

Chair

Any in the public sector?

Jon Thompson

Yes.

Chair

Government Departments?

Jon Thompson

No.

Chair

Government agencies?

Jon Thompson

We have narrowed this down. You are trying to pick away at what my list is. Yes, we had some concerns about some public sector employers and we would treat them exactly the same as private sector employers. It is large workforces where there is movement in and out of that workforce, which leads to some poorer administration of PAYE than for a more stable workforce.

Chair

The largest employer in the UK is, I think, the NHS. I think it is bigger than the Indian railways and the Red Army—or at least, the biggest after those two. Is the NHS in the frame? You are dealing with lots of different bodies.

Jon Thompson

There are some trusts that we do not think are good enough at PAYE.

Chair

So when you have these cosy chats, chief exec to chief exec—perhaps not so cosy, if you are doing your job toughly—what happens if they do not improve? What are your sanctions, or is it just gentle persuasion at this point?

Jon Thompson

I do not think we have any sanctions. Again, let me start with the tax gap. If you think about the difference—the tax gap—between the employed and the self-employed, they are significantly different, so there is some significant advantage to us, in administrating tax, in having intermediaries and others. Therefore, at the minute, there are not any sanctions, and neither is there any sort of merit award or gold standard or whatever. We have been mulling over what we can do to both encourage and discourage employers and pension fund providers. I think I also ought to split that, for the purposes of transparency. We are somewhat more concerned about pension providers—those who are paying out pensions. We have some concerns about that group too. If you want to try to reduce the tax gap systemically, that is an area where we think there are some further margins.

Chair

Can you just explain, in simple terms, what the pension providers are doing that is hitting the tax gap? They are paying out too much, are they?

Jon Thompson

Or not enough. The systemic point that I am trying to land is that PAYE administrators—organisations, either employers or pension fund providers—have a very variable standard, and we think we need to do something about that. It is also one of the five main causes of customer contact, actually—the employer gets it wrong, and it is perfectly natural, if you get your payslip and you think, “Hang on a minute, my tax has changed and it doesn’t look right,” that you might ring us or your employer. What we are identifying is that there is a good deal of failure demand, as we are calling it, in this that is driving customers to contact us. So we think it would be better all round for taxpayers if we tried to do something about intermediaries. We are working up some proposals that we want to put before Ministers as part of the spending review.

Chair

So it is quite a big task. You have all these large organisations that you are trying to improve, but you have no actual power to make them do a better job.

Jon Thompson

Not at the minute, unless someone is going to correct me. When you get on to fraud, one of our emerging risks is this payroll bureau question, where we see some cases of payroll bureaux being set up saying that they want to administer payroll for small enterprises but the payroll bureau itself is a fraudulent enterprise and never pays any of the tax of individuals. That is a growing risk, so there are other advantages for us in being more assertive—

Chair

So there is no kite mark or anything at the moment, no basic standard—

Jon Thompson

No.

Chair

So are you saying that you would welcome some sort of sets of standards for these organisations? More regulation—interesting.

Jon Thompson

Look, another point about a tax gap might be that—if you want to reduce the gap significantly—you have to put more burdens on other parties to collect the tax for you, and there would be a cost to that. That needs to be a consideration. We don’t think that there are any significantly free shots anymore.

Sir Geoffrey Clifton-Brown

Referring to the Chair’s questions on your powers, is one of the areas that is causing you concern the business of whether it is an agent or a principal? You have taken this to court five times and lost, so is the concern about who is responsible for the tax here, and is this going to be a continuous source of leakage for you?

Jon Thompson

Yes. I think we have discussed this before. It is of some concern to us that some intermediaries look to the consumer as if they are the principal, but they are not; they are actually acting as the agent. One could trot out who that would be but most of us have a smartphone and we look at the apps on our phone and we assume that is who we are dealing with, and we are not. The step that the Government have rightly made is to force intermediaries to give us information. That has actually been a good step forward for us. We have had the first two—I don’t think I can name them, but they would both be familiar in terms of apps on your smartphone—and they now have to give us information about the flow-through of transactions. I could give you an example if you want, to bring it to life.

Chair

Please do.

Jon Thompson

I was in Leeds the other day, working with a team who are working on exactly that point. Essentially, we had a Chinese restaurant declaring two sources of income: cash, and debit and credit cards. We can check the debit and credit cards because Government have already given us that information—we are able to access that. What the restaurant was not declaring was any further income through intermediaries. When the intermediary—which I am not naming—gave us the information, we suddenly discovered that the turnover was five times that being declared, but the money was flowing into other bank accounts that were not associated with the business. When you see three sources of income—the intermediary income, the cash, and credit and debit cards—you see a fundamentally different business from the one that you thought you were being presented with. Access to that information is a big step forward for us, and we are growing the number of intermediaries that we are getting information from. That is a way you can do it. The Government are also consulting on whether to introduce split payment—which I think is of some interest to you in relation to online VAT in particular—so that the money never flows down the chain and the tax is taken out by the intermediary.

Chair

We will come back to split payments relating to online VAT. I want to go into the issue of income-related error and fraud—from what you said, some of that may touch on this as well. What are the main drivers of undeclared income relating to employment, social security benefits, dividends and rents, which are things that we would think are automatically put into the system under the new digital approach?

Jon Thompson

You would think that, but the—I don’t quite know how to put this. The essential problem is that the claimant has to estimate the income for the 12 months to follow, then the onus is on the customer to tell us about any changes. Therefore, if you change your job or get more or fewer hours of work, you might think that that would flow through from another party to us. We have described some of the difficulty of that, but the other issue is about timing—so a customer will tell us, but they take some time to tell us, and what has happened in the middle is that a debt or otherwise is built up in between.

Chair

What about the relationship between you and the DWP? Theoretically, you would think that benefit income information from the DWP would be in real time. Is that actually happening?

Jon Thompson

Our relationship with the DWP is rather good, but the Report here and our own analysis reflects that it needs to be improved in terms of moving to real time. There is then also the question of what happens—

Chair

What is the lag on it at the moment? Is there is a typical lag?

Jon Thompson

I could not honestly answer that question. Sorry.

Chair

Is it partly to do with the national DWP systems, or does it depend on how—

Jon Thompson

I am sorry. I cannot honestly answer that question. My understanding is that it is more to do with being in employment, moving out of employment, varying your hours, being on a zero-hours contract, moving into a bit of self-employment.

Chair

So it is more to do with the individual’s work pattern than the DWP’s payment systems.

Jon Thompson

Correct. Nevertheless, we ought to have another look at our relationship with the DWP and that flow of benefit information.

Chair

On fraud and error generally, in the two years prior to the ones we were just discussing, fraud and error went down after the final reconciliation—when you have finally done the numbers. So your forecasts have been out quite a bit over time. What are you learning from that, and what are you doing to improve your forecasting? You are never going to get it 100% right, but what are you doing to ensure it is more accurate?

Jon Thompson

The primary reason why it gets amended with the benefit of further information is that you are on an annual cycle and you have to complete that cycle and make another claim. At that point, a sample of cases—I think 5,000—is gone through in some significant detail. Some of those people will be in other processes, appeals and so on, and you have to let those play out. Then those cases flow back in. Actually, what that is doing is reducing—

Chair

We know what it is doing. But is there any way you can capture that certain percentage of people going through appeal or whatever and work out if there is a trend over time that you can factor in? There will never be an exact science, but are you doing that?

Jon Thompson

We could do that. I think you have seen us take some further steps on child benefit in terms of, “You can see a pattern, so can you build that pattern into your estimate in the first place rather than retrospectively amend it?” Now you are getting me into, “How do National Statistics standards handle that?” I would have to check whether National Statistics would allow you to put that estimate in. In responding to your question, it is possible, though, that we could be proactive and produce two estimates: one based on the same and one that takes into account these other, less quantifiable but predicted factors, as we have done on child benefit. So yes, I am happy to take that on board.

Chair

Okay, that is very helpful. We have obviously looked at the Concentrix contract failure. I will not go into the details, but why have you not measured the impact of that on fraud and error?

Jon Thompson

Because the contract finished in November 2016, so it will not flow through until you get to the fraud and error estimate for—

Chair

But you will be including it. And separating it out as well?

Jon Thompson

Yes. We can separate it out. There is a forecast given in, I think, figure 24 of the NAO’s Report that puts it in the 1%. That is the remaining primary driver of why we think it will rise from 4.8% through 5.5% and ultimately to 6%.

Chair

It is important for all sorts of reasons that we see that separated out and individuals can see it.

Jon Thompson

I am happy that we should do it.

Anne Marie Morris

Mr Thompson, we conflate the fraud and error, but clearly the motivation for fraud and error are completely different—one is an offence and the other is a mistake. Why do we conflate them? Is there any way of separating them?

Jon Thompson

We could separate them, and in big handfuls it is 80:20 error to fraud.

Anne Marie Morris

Okay. Given that, would there be any benefit to the tax take from specifically separating them? Clearly how HMRC try to improve customer behaviour, give training and provide support for where it is error needs a completely different approach from fraud.

Jon Thompson

We do do that privately. If you are asking if we could give that sort of information publicly, yes, I suspect we could in the narrative that goes with the statistics. We could do that, yes.

Anne Marie Morris

Okay. That would be very helpful. Specifically in terms of trying to help the individual taxpayer on error, rather than fraud, what steps are you taking to enable that individual not to make mistakes? I think you had a plan about ensuring that this was an easier process, but I am a bit worried that it might have been overtaken by the delays imposed on the transformation programme.

Jon Thompson

We would have proceeded with some projects in transformation that would have assisted with claiming tax credits and made it somewhat easier to notify us of change of circumstances, and we will not be able to do that now because of transformation, but let me just step back. In the light of Concentrix and the various parliamentary inquiries on Concentrix, we stood back from that, invited in a series of customer and third sector groups and said, “Our approach to fraud and error is obviously, in the light of Concentrix, not working. How can we do more to promote customers telling us in a more timely manner about changing job, changing hours, moving in or out of a relationship, childcare changes or whatever, all of which make a difference to the benefit?” Therefore, there has been more work on promoting something called “Tell us”—“Tell us about the fact that you’ve moved or you’ve moved in or out of a relationship. Try to do that as much as possible.” We did that by fairly extensive work, in the light of Concentrix, with claimants’ groups and with voluntary sector organisations, saying, “Could we do more about that? What is our approach? What do we want you to do in responding? How can you be more on the front foot?” We also then changed our approach to what we call “Respond”, which is where we select a number of cases that we review for fraud and error—how we go about doing that, how we communicate, how we pick the cases and all that. There are a huge number of lessons to learn from Concentrix and they fall into those two groups: promoting more actively telling us about changes, so there is less error, and how we respond to that. Is it making any significant difference to the headline percentages? At this point it is not.

Anne Marie Morris

Okay. So as and when the transformation systems do come online, what impact will that have in financial terms on tax take?

Jon Thompson

To be transparent about it, they will not come online, because new claims for tax credits will stop and therefore we have to make a strategic decision—“Am I going to invest in this, given the other choices and challenges I face?”—and the decision was made to prioritise other things and not this.

Anne Marie Morris

So this may fall off the project altogether?

Jon Thompson

The project has stopped.

Sir Geoffrey Clifton-Brown

In view of what you have said about dropping this particular aspect from the transformation programme, are you accepting that there is going to be an element of fraud and error from the benefits system going forward? Do you just accept that there is a level that you are not going to be able to control?

Jon Thompson

The frank answer to your question is yes, but one of the interesting debates that Sir Amyas and I have had is that, if you create a welfare system, the complexity of any welfare system—unless it is really simple—will have some inherent fraud and error built into it. Some people are bound to get it wrong or they get it in the wrong time. The question is what that is, versus where you are. Could you estimate that what you will always inherently get is—let’s make it up—2.5%, and therefore take some action that would reduce it from where you are to 2.5%? That is the kind of conversation we have been having. I am not complacent about it in the slightest, but in the end it comes down to: “Strategically, where is this? Is it a priority or not, and are we going to put any resources into it?” In the light of everything else HMRC is doing, those decisions have been made by Ministers and we are where we are. We can continue to learn, but at best it is going to be an incremental change.

Chair

So in short, tax credits is an orphan policy that will not get attention?

Jon Thompson

I am not going that far. I am saying it is not as much of a priority as other things. We will maintain it as best we can under the current plans until 2023.

Chair

It’s limping to its finish line, perhaps; on the opposite of limping, let’s bring in Caroline Flint. That is a very bad segue.

Caroline Flint

Thank you very much. On customer service, I have to say, Mr Thompson, that I was a little bit disappointed, because the last time we talked about this, one of the things I raised was about how HMRC could be more positive and thankful to taxpayers who pay their taxes on time, and whether or not something could be reflected in the letters that go out to customers and taxpayers when they do their tax return online and meet all the demands. Have you thought any more about incorporating that in your communications? I read through your foreword to the annual report, which was great and you thanked lots of people, but you did not thank the British public for being probably one of the most responsible nations in the world in paying their taxes. Are we going to see any change on that front?

Chair

Mr Thompson, you can rectify that omission now.

Jon Thompson

I am very grateful for people paying their taxes, and the more they do it, the better it is.

Caroline Flint

Was that so hard to say?

Chair

Sucking lemons would be easier.

Jon Thompson

We have a system that is based on consent and a good deal of trust, and so on and so forth. If we did not have that system, and we reflect on some other international variants of that, there would be a significantly different pattern to this conversation. Let me pick a country at random. If you pick the United States, everyone does self-assessment. It would be reasonable to speculate that if everyone did self-assessment, the level of fraud and error in that would be significantly higher than where we are, because of the role of intermediaries and employers and so on. Now, let’s come back to your question. If we are not doing it in letters, we do send out 5 million annual tax summaries to citizens to say, “Thank you for paying your tax, and this is what the Government used it for.” That has been debated quite hotly with Ministers—I think that is a fair way of putting it—and has a mixed reception. I think that is the best way of putting it. Some people like the fact that 5 million citizens get a letter saying, “Thank you very much for paying your tax, and this is how it is spent.” Some people like that; some people think it is a complete waste of time and money. Opinion varies significantly. It is not quite right that we do not do it; 5 million people do get that. There is a slight risk that one of you may have received one of those.

Caroline Flint

Perhaps HMRC could ask that question through social media or something, which would not necessarily cost anybody anything. Just do something on Twitter or through social media asking people what they think about being thanked for paying their taxes on time.

Jon Thompson

We could do that. To try to embrace your point, I am happy to see what we can do about that. I have to say that I am not a great fan of social media.

Caroline Flint

All right. Well, you could email.

Chair

Wise to be cautious, Mr Thompson.

Jon Thompson

I have had two death threats through social media, seriously, for doing this job.

Chair

You are in a room with people who have also received those.

Caroline Flint

Just going on to some more specifics, last time we had quite a discussion about people ringing up and trying to find out about services, and the amount of time they wait. You made the case for why you would not include the amount of time it takes for people to go through the automated response, and that is fair enough, but in the interests of transparency, would it not be helpful to actually explain and put in some information about just how long it takes for people to go through the automated process before they actually get to speak to someone, so we have a more transparent sense of how long people are spending on the phone?

Jon Thompson

Sure.

Caroline Flint

Could you tell us today how long the automated questions and responses take for people, on average?

Jon Thompson

It depends on how many of the options you follow and whether or not you are registered for voice biometrics. If you registered for voice biometrics, that saves 25 seconds. If you are not, you have to go through further questions. It also depends on what you are looking for and therefore how far you have to go through the IVR.

Caroline Flint

That’s a common-sense answer. If you are open to including in the next annual report some transparency information about the waiting times on that automated response service—

Jon Thompson

Sure.

Caroline Flint

Before we get to that point could you write to us about how you might think about presenting that question in the next annual report, so we can understand that? I get your point that if you want to go to one place you are going to say one thing, or ask one question. That might take more or less time depending on what the issue is. It would be useful before we get to next year’s annual report to have an indication of how you are going to measure this and present it.

Jon Thompson

I am happy to go and do that. In the spirit of transparency, our average is going to be two minutes to get through the IVR—maybe a little more than that. So you have some sense of whether this is a 10-second thing or a five-minute before you get to the queue.

Caroline Flint

That’s quite interesting, because the final paragraph on page 32 of the annual report and accounts says that “for customers who need to speak to an adviser, our average speed in answering calls was 4 minutes and 28 seconds… This is within the five minute target…but was slower than the 3 minutes and 54 seconds we achieved” the year before. So if you add in those couple of minutes, it is all adding up to a considerable amount of time, isn’t it?

Jon Thompson

It is. I think Mr Rowley asked me some of these questions before, and if you will indulge me, I would like to apologise to you, because the last time that we had a series of exchanges, you asked me a question and I fundamentally misheard what you asked me. I think I responded fairly emotively to a question about prioritisation where I heard “expensive” but you had said “extensive”, so I am sorry about the way I reacted to that. There are five broad dynamics in play in terms of giving you the overall picture. It is a very simple measure: what is the average speed of answer? And we understand what is in that measure and what is not—not in the IVR, but in the queue. There are five dynamics. One is that demand from customers varies significantly by time of day, day of the week and week of the month, and there are two annual peaks—the July tax credits peak and the January self-assessment peak. That is our dynamic. It also depends on when you ring, and obviously, we can plan for some of that. Let’s say you ring between 8.30 and 10 o’clock on a Monday morning, when there is significantly higher demand then than there is at other times of the week. We can plan for some of demand to some degree, but for some of it we cannot plan. The second dynamic is what we have termed failure demand. Let’s say we do not serve you, the customer, well in the first place. I write you a letter. That letter is not very clear, or you think, “Hang on a minute. I really don’t agree with HMRC.” It is no great surprise that you ring up. We have to be better at serving customers in the first place, so that it is clear what the situation is and then they won’t ring up. We have to take that on board, and we have done some fairly analytical work about what is demand driven by customers and what is demand driven by what we have called failure demand.

Caroline Flint

How are you thinking about doing that? In the changes in terms of reprioritisation, infrastructure projects and customer services are now being prioritised last, so where does that fit in with what you have just said?

Jon Thompson

Changing standard letters so that, for example, they are in plain English, as opposed to using technical language.

Chair

I’ve just written down “plain English”.

Jon Thompson

Thank you. Technical gobbledegook is a matter of some frustration for the three of us, but rewriting a standard letter is not the same as implementing a significant IT system. We think there is a whole series of quite straightforward things, which are normal continuous improvement, that could make a big difference to the communications between us and customers and also in the basic administration of tax, so that we make fewer mistakes. The biggest cause of complaint ultimately, which is set out in the annual report, is a mistake that we make. I am being transparent with you: I have demand from customers; I have failure demand with regard to HMRC; and thirdly, you have the dynamic about employers. Let’s say your employer gives you a new taxable benefit. They don’t administer your PAYE; and ultimately, six months later, they put it through and you get a big tax bill and may ring us up. That is quite often something about the relationship between you and the employer. Therefore that, again, is supporting the drive to how we improve things on the employer side. Fourthly, there is this dynamic. Okay, if I understand demand better, what about the supply of colleagues to respond to that demand? What is our employment model? Is it flexible enough? Does it work well? Are people trained well enough? All of that can also make a difference to the time. Lastly, there is the enormous growth in working with us digitally. We now have a reasonable handle on what would otherwise have been phone calls. How much of that would otherwise have been a phone call? We have a reasonable handle on that, but we also are seeing the so-called worried well, who because they now have much more transparency about their tax and pension affairs through the personal tax account, are ringing us up to query that. They would have done that anyway, but probably not until they retired. To give an example, on your personal tax account you can see your national insurance history, and you can get when your state pension kicks in and how much it is going to be—an estimate of that. Because you have your full national insurance history on your personal tax account, quite often people ring up and say, “Hang on a minute. What happened in 1987? I seem to have a gap.” That is totally right, but we are stimulating that because we are making it more transparent and that is good, but increasingly the way forward on digital, which gives more people more transparency to their total relationship with us and possibly with the DWP, is a fifth factor in this. With those five dynamics, we have to do better at understanding what is happening in demand and what we are stimulating, and the supply of our people and how that works.

Caroline Flint

I will just come back to a couple of examples. Either a letter is sent out or there is a script that staff use in order for them to go through a menu of options with people.

Jon Thompson

Yes.

Caroline Flint

If it emerges that, either in a letter format or in a script, the clarity isn't there and it isn't plain enough English, how long does it take between identifying that as a problem and changing it? How long does that take to work its way into the system?

Jon Thompson

There was a theology, if you like, in HMRC, that it would take a really, really long time to change standard letters because the standard letters are built into an IT system and therefore if you want to change it, you have to change the IT system and so on. To use possibly inappropriate language, it was sort of hard coded in the system that this is what the standard letter looks like and therefore if we want to change it—this week, next week or whenever—it is going to take some time to put that through in change control terms. The advance of technology now is such that there are tools out there that you can lay on top, which extract the necessary data from the IT system and can produce you a letter that is much more flexible and you can change it on a much more regular basis. So we are exploring that in some depth. It is worth saying that the new director general of customer services, Angela MacDonald, has a background in operational excellence. She and the team are now very much more analytical about what exactly is happening, how much of that we are driving and what is really going on, and that is then leading to a whole series of tactical changes—relatively small changes that can make a big difference to the customer, as opposed to a big strategic programme where you change everything and it takes a long time and you spend hundreds of millions of pounds of public money and it doesn’t really make any significant difference because the whole world has moved on.

Caroline Flint

It would be quite interesting, in terms of looking backwards and forwards, to get some information from you about where this problem has emerged and how quickly the turnaround to change it was. Perhaps we could have, between now and next year, some examples of where you have responded to a problem. It could be not just individual taxpayers, but the business community as well, and be examples of where you have thought, “We need to improve the way we communicate and the language we use,” whether that is through a script-based telephone conversation or a letter. Would it be possible to give us a sense of how you have engaged with that?

Jon Thompson

Sure. We can totally do that. We have an operational excellence directorate with 400 people in it that is constantly trying to see how you improve all of these systems. We need to get more out of that function and it needs to make a difference to the customer.

Caroline Flint

But it is also about capturing information. Sometimes on the same issue, the same problem recurs, such as whoever is at the end of it—an individual or a business—understanding what they are meant to provide, causing these problems with the information. Can I ask about the digital customer services targets? They fell this year. Why that is the case?

Jon Thompson

The target is 80%. To be clear that measures somebody saying they are very satisfied or satisfied with the service and excludes the neutrals and the dissatisfied or the very dissatisfied. That is essentially what it measures. What we are fairly quickly discovering from talking to users is that they get into their personal tax account, they do something and then they say, “Well I wish I could do more. It is great, but I really wanted to do more.” They are now comparing our digital approach with what they are using every day on their smartphone or whatever, and we are falling a bit short in that regard. We have gone out and looked—

Caroline Flint

Is that because some of your customers are very tech-savvy and you are not keeping up with their expectations about how you should respond to them digitally?

Jon Thompson

I am not sure it’s about tech-savviness. Some 16 million people have opened a personal tax account now—we have just passed 16 million—and 3.6 million businesses out of a population of only 5.4 million, so well over 50% of businesses have opened one. It has a range of things that you can do, which is good, but then you want to do more. That is basically the big pattern. We have therefore continued in the transformation programme to invest in the personal tax account, if it means that you are less likely to ring us up.

Caroline Flint

I know that other colleagues have questions on customer service, but I want to ask about your interpretation. Looking at the satisfaction ratings you have given in the annual report, among small businesses it was 74%, among large businesses it was 82% and among individuals it was 69%, but for the mid-sized businesses it was only 55%. Could you enlighten us as to why that is the case?

Jon Thompson

To be completely up front with you, not off the top of my head. Perhaps Jim can.

Jim Harra

It is that feedback that has led us to change our mid-sized businesses service offering in the last 12 months. Obviously, they sit between small businesses, which we generally leave to get on with things, and large businesses, which we manage very intensively. We find that mid-sized businesses hit points in their business lifecycle or events where tax becomes quite complicated for them and they cannot be left on their own but they need attention from HMRC. So we have introduced a new mid-sized businesses service, where we temporarily appoint a customer compliance manager to work with them closely if they have hit one of those events—it may be that they are growing, that they are merging, or that they are distressed in some way—to get them over that and then back into the swim. Whether that results in the increase we want to see, we will have to wait and see. But it is a response to that feedback.

Chair

Do you have a target? We are talking about page 42 on the annual report and accounts. The tax gap estimate for mid-sized businesses is £3.9 billion. Those are the 2016-17 figures. Given what you have just said, Mr Harra, if that works, do you have a target for how much that tax gap would go down?

Jim Harra

We do not target the tax gap. We would expect, as a result of that change, their satisfaction levels to go up because they feel that they are getting the service that they need when they need it. We expect that it will help us in managing their compliance, because when they hit one of these points and their satisfaction goes down, it is likely also that they are facing compliance challenges.

Caroline Flint

Actually, the small businesses have the £13.7 billion tax gap, but they have a 74% satisfaction rate. I just think it is quite interesting, because it is often said that in our country we have a lot of small businesses, but often, rather than scale up, they sell out to a much bigger concern or just keep themselves small, presumably because they do not want to take on all the extra responsibilities. So that mid-sized business area seems to be really important to encouraging smaller companies to develop and not be frightened of doing so for whatever reason.

Jim Harra

Yes, and I think our research, when we looked into what was driving the dissatisfaction, found quite a lot of variety in this sector. Some of them are very mature and their tax affairs are smoothly moving along. But there are others that will grow or will face a merger or a takeover and then they will face some tax issues that they have not had to face before, and they find that difficult.

Caroline Flint

I suppose what I was trying to make clear is that if the satisfaction levels were higher because they were supported through a growing process, it should be good for that sector of our business community.

Jim Harra

That would be our general aim. We do also have a new growth service that we have introduced, where we try to target companies that the data tells us are growing rapidly.

Caroline Flint

It will be interesting to see how that develops over the next 12 months.

Anne Marie Morris

Mr Thompson, we have talked a little bit about the quantitative measures, as in the speed of response to phone calls, but not really about qualitative measures. I appreciate that you have your satisfaction surveys and clearly you are well aware that there are quality issues, but have you looked at how you can produce some sort of sensible qualitative measure, which we can compare year on year and which takes on board all of those things you have talked about, to give you a better handle on your improvement and what you need to do to get it even better?

Jon Thompson

We have, and we have committed to produce our balanced scorecard for customer services by the end of the current year. At the minute, there are eight public targets for customer services, but we think they are measuring quite a narrow set of things quantitatively, and they do not really measure quality at all. Actually, the customer survey number is on digital services, not on other services. This year, we are likely to see more transactions in digital interactions than by traditional means—on the phone and by post. We think we will have the crossover point at some point this year. None of that is being made transparent to readers of the annual accounts—you or anyone else. We are not publishing any of that information. We think that we need more measures on transactions in the personal tax account and the business tax account, and there are some other digital services that continue to grow. We need a better basket of what customers actually say. We have committed to Ministers to develop a balanced scorecard for their consideration by end of the current financial year, with a view to setting something for the future financial year, which will be made more transparent. It is worth remembering that we publish this data every month and we would want more information to be made more transparent. So yes, but that is quite a long answer.

Anne Marie Morris

That is very helpful. I have heard a lot about what you are doing for the corporate client and you mentioned the personal client, but as an MP I hear a lot of frustration: people ring up and not only is it hard to get through, but they get a different person each time, understandably. It is hard then to access the file and they do not really understand the position, because it is not up to date. Sometimes, they need someone who specialises in VAT or whatever it is, and the person on the end of the line does not have a clue about VAT. How can you make that a more helpful and useful service in quality terms for the individual as opposed to the corporate, which is often who it is?

Jon Thompson

That is the kind of very good feedback that we need to continue to get about whether we can join it up, particularly for business customers. One of the aspirations for us is to develop a way in which our customer services advisers can actually see the totality of the relationship between a business and HMRC. At the minute, that is really rather difficult, because over the decades, HMRC has built up IT systems by tax product, not by customer. We have to be able to join it together. At the minute, you have may have multiple relationships with us; sometimes you might owe us money and sometimes we may owe you. Therefore, we need to move to a situation where over a number of years we are transparent about it, we can see the totality of your relationship with us and we have all the necessary IT systems and the training to do it. On a tactical basis, that feedback is very good and we can act on that basis. We need customers to tell us that. We also need to reflect on what the complaints system is telling us in terms of processes, mistakes and delays.

Sir Geoffrey Clifton-Brown

Is there an issue here of taxpayer privacy? You say to my colleague that you want to hear more personal information. In Russia, every banking credit and debit transaction is available in real time. How far down that road do you really want to go with personal information from the taxpayer?

Jon Thompson

The more we have a conversation with taxpayers, the richer the evidence base is for what we need to do to improve. I think that was the thrust of the questions as opposed to whether I need to collect real-time VAT transactions, which I think is the Russian system, which they very proudly show off. They are quite different questions.

Lee Rowley

I am happy to hear about some of your proposals for a greater understanding of your end-to-end and automated customer experiences. When do you expect the basket of measures or balanced scorecard to be available publicly for this Committee to review with data in it?

Jon Thompson

The plan is that we will agree it with Ministers by the end of the current financial year, so targets would be set and we could be transparent next financial year.

Lee Rowley

That would be in your annual report at the end of the 2019-20 financial year?

Jon Thompson

2019-20, but the previous Financial Secretary was very keen that we produce that data transparently every month. We now publish it every month, so you could anticipate that you might get something as early as spring 2019.

Lee Rowley

Are you committing, as per your answers to Ms Flint, to have the basket—the scorecard—include elements of the automated customer experience and the overall cross-channel, end-to-end experience for customers?

Jon Thompson

Definitely the first. There are real challenges with the second. We do not have the technology that says, in easy terms, “Mr Rowley rang yesterday, but he went on to his personal tax account today, and tomorrow he fills in an NI form.” We cannot, at the minute, join that together so you have an omnichannel—I think that is the terminology—approach to the customer. We can certainly do the first set that you asked about, but we do not have the technology that joins those things together. One of the customer service advisers’ issues is that you may have tried on your personal tax account to undertake a transaction. That didn’t work, and then when you rang up, the customer service adviser has not necessarily know that you tried to do that yesterday, or that your letter arrived the day before. They can to some degree, but not anywhere near enough to be able to do what I think you are asking about.

Lee Rowley

You can do things through sampling or separate telephony to assess the origin.

Jon Thompson

Yes, and we are looking at what kind of technology could give you that view. Forget the route by which you are coming in. “What do I know about this customer, Mr Rowley? It’s Mr Rowley on the phone. Yesterday, he submitted that form, and last week he rang up about that. Maybe he is ringing up about that again.” We are some steps forward there.

Lee Rowley

Good. I look forward to seeing that information, hopefully as soon as possible. The main thrust of my questions is on something that you referenced at the beginning of your comments to Ms Flint, which was about your IVR and the voice biometrics within that. Is HMRC acting lawfully in the collection of voice recordings within its IVR system?

Jon Thompson

We believe we are.

Lee Rowley

Have you always been acting lawfully throughout the period in which you have been collecting voice recordings—since January 2017?

Jon Thompson

We believe we have. In the spirit of your question, clearly a complaint has been lodged with the Information Commissioner’s Office. We are working with the Information Commissioners’ Office on how this process works, whether there was explicit consent and so on and so forth, so the ICO can make a judgment about that complaint.

Lee Rowley

In that spirit, to understand it in a bit more detail, can you explain how you think the collection of those recordings within your IVR was lawful and what process you went through to validate the lawfulness of it?

Jon Thompson

We believe that the messages about this were clear enough that it was implicit that you were giving permission for us to collect that data. The aim of that is to improve the level of security so you can access the services. There has been a formal evaluation of what difference voice biometrics makes over other means of verifying your identifying your identity, like a password or your mother’s maiden name. That evaluation is there. We believe it was implicit and therefore lawful, but we have, in the light of the complaint, changed the message so it is now extremely clear and explicit that you are giving your permission for us to collect this data. There is definitely a learning point for us in that.

Lee Rowley

Does the new message make it clear that you can say no, if you do not wish to give permission?

Jon Thompson

It does, yes.

Lee Rowley

So where do I derive the implicit consent from the following statement? “For some calls, we are introducing a quicker and more secure way for you to identify yourself. You can skip these questions in future by using your voice to confirm who you are. Voice ID uses the sound and rhythm of your voice to identify a numerical pattern. This pattern is unique to you in the same way your fingerprint is. It works if you have an accent or you are unwell.” I will skip a bit. “It is the fastest and most secure way for us to know it is really you we are talking to. It’s easy to setup. You’ll be asked to repeat the sentence ‘My voice is my password’ multiple times. I’ll need you to say exactly those words. It takes about a minute. So please say ‘My voice is my password’.” Where is there an opportunity in that statement for the individual to know how not to do that, or not to do that?

Jon Thompson

It is not explicit that if you didn’t say that, you would have moved on. We have now made it explicit that, at this point, you can either repeat the word—I’ve got the script here somewhere, but I can’t quite lay my hands on it—or say, “No,” in which case you move through to the rest of the IVR.

Lee Rowley

But under GDPR there is no explicit consent in that statement, and under DPA there is no ability to opt out, so it breaks both GDPR and DPA.

Jon Thompson

Well, you are judging it on the basis—and this is one of the conversations that we are having with the ICO: we believe that it was implicit and we have changed it to make it much more explicit.

Lee Rowley

But where was the opt-out? Do we agree that there was no opt-out in that statement?

Jon Thompson

No, we don’t agree.

Lee Rowley

Where was the opt-out, then? Where was the statement that said you could opt out?

Jon Thompson

If you didn’t say the phrase—I am sorry, I am not going it—you would not have entered into the process.

Lee Rowley

But there was no statement that said you did not have to say it.

Jon Thompson

There was not at that point, no.

Lee Rowley

Okay, so that is about as implicit as you can get—i.e. not being present.

Jon Thompson

Well, we think it was implicit. This is a discussion we are having with the ICO. In the end the ICO will judge whether it was or it was not.

Chair

What is the timetable for the ICO doing it?

Jon Thompson

I am sorry; I am not responsible for the ICO.

Chair

No, but I just wondered whether you have got any idea. I mean, are you going to be waiting weeks, months, or a year?

Jon Thompson

I am sorry, but I think the complaint was lodged in June—it might have been May—and we have been working with the ICO, but I am also completely recognising Mr Rowley’s point, which is that it needed to be more explicit, and it now is more explicit. I do have what is now the update to what you read out, but I cannot lay my hands on it.

Chair

Okay. Mr Rowley, your last quick point on this.

Lee Rowley

Your privacy assessment, which was released initially, was dated March 2017. I understand that you started collecting data in January 2017. Does that mean you went live without full compliance approval?

Jon Thompson

I could not honestly answer that.

Lee Rowley

Could you write to us?

Jon Thompson

I could.

Lee Rowley

The privacy statement that is currently on gov.uk is dated within the PDF July 2017, the exact same date it was loaded up on gov.uk a year later. I find that highly coincidental. Either that date at the bottom of it is potentially incorrect, and the privacy notice was not completed until July 2018, or it seems a huge coincidence that it was completed in July 2017 despite a different form of that privacy notice—an incomplete one—being provided to other organisations in the interim. I would be grateful if you could write on that.

Jon Thompson

Sure.

Lee Rowley

And if it is determined that your implicitness was not sufficient for the ICO, will you delete all of the voice recordings that you have recorded since this process began?

Jon Thompson

We would wait on the ICO judgment and then make some decisions about what to do. To give you some sense of the scale, 6 million people had registered up to the point that we changed it, and a further million have registered since the ICO complaint was lodged, and some publicity was drawn about that. We have had 11 other complaints, so there are 12 altogether, but I think the sensible and logical thing to do is to wait on what the ICO concludes and then decide what we need to do. But at this point we don’t propose to delete it; we will wait for the judgment of the ICO.

Lee Rowley

Last question, or last request: could you write to us with an analysis based on the data you collect of the proportion of people who went through this IVR prior to you changing the recording and did as the IVR suggested, and the proportion that did not, and then the proportion that did and did not after the recording was changed?

Jon Thompson

We can. I have seen that information and I will happily write it down for you.

Lee Rowley

I look forward to it. Thank you.

Chair

I want to touch on the issue about the estate. Thank you very much for your update letter of 31 August, which the Committee has received. Some of it is not surprising, given some of the things that we have discussed before, so we will not go over old ground. I want first to pick up on the second paragraph, where you talk about the assumption that the estate remodelling was based on. That was on 48,000 full-time equivalent posts by the end of 2021. You are now forecasting to have around 58,400 full-time equivalent posts. That is 10,400 more. I wonder whether you could break that down. We have already, in previous sessions, talked about other staff you will be recruiting. Is this additional to numbers you have reported to us before, or is this adding up to all the Brexit jobs that you have mentioned previously, because every hearing we have a different set of numbers, so I just want to bottom that out. Who are these people going to be?

Jon Thompson

Certainly I can break down the 10,000. There are three reasons. One is the 5,000—it peaks out at 5,000 for additional staff for Brexit.

Chair

Is that across all areas of the business?

Jon Thompson

Yes, although predominantly they are in customer services, because we need to expand the national customs hub, or they are in customer compliance, because it presents a different set of risks in the tax system. Since the 2015spending review, there have been a series of fiscal events in which both Chancellors of the Exchequer have given us 2,000 additional staff, almost entirely for customer compliance. The other difference here is about those who are working on tax credits now and who would otherwise have gone to the DWP, but now will not be doing so.

Chair

But will stay with you.

Jon Thompson

Yes.

Chair

Doing what with you?

Jon Thompson

We can give them some certainty until the end of tax credits, but after that it is not clear. To be up front with you about it, I went to Preston, which is the main centre where tax credits happen. We have more than 2,000 colleagues in Preston. I took the temperature by talking to them about how they feel about that. They wanted to know what the future was in the light of the announcement by the Secretary of State for Work and Pensions. We said, “Let’s be transparent about that. Actually, going to the DWP is not going to result in a job. Stay with us. We can give you some certainty until 2025 in some cases or 2023 in others. At this point, is that sufficient?” People welcomed the certainty about at least that period of time, even though that presents them with a different uncertainty.

Chair

Although that is a long way ahead—well, not if you’re likely to lose your job in seven years’ time, but it is seven years to 2025. Is that too far out for you now to have a prediction about what you might be able to redeploy those people into? We obviously don’t know where we are at with Brexit. Does a lot depend on Brexit and what happens there for those people to perhaps be redeployed into other areas of the business?

Jon Thompson

I can find productive work for a whole range of people to do endlessly. We could find people jobs in customer services. You could reduce the average speed of answer. There would be some learning and development and some training, and you could do that. We ought to be up front that natural turnover is between 8% and 10% in any given year. Between now and seven years’ time, a significant number of people will naturally leave the organisation, so there may be other opportunities for us to deploy those people into.

Chair

If we just take the raw numbers, excluding turnover—even if it is not the same people, the number of people could stay the same—they are currently funded because you have got a reason for them, because they are doing tax credits. Would you be able to fund those people in other jobs? Do you have to make the argument to the Treasury that they are going to bring in a certain amount of tax? It is early days, but have you thought through your strategy about discussions? You said that you could deploy them, and I am sure you could; you are very good value for each member of staff you have because of what they bring back to the Exchequer. Are you having that discussion with the Treasury now, or is that some way down the line? I can just see a future Chancellor saying, “There is a bit of fat there. Can we take it out?” It is a terrible way of talking about staff, but I think we both know what we mean.

Jon Thompson

I don’t know how many questions were in that, but let me try and answer—

Chair

I am trying to work out whether you have got a strategy. Seven years seems like a long time. Have you got a strategy now?

Jon Thompson

Under the 2015 spending review, the cost of administering tax credits was transferred from HMRC to the DWP. On the basis of various assumptions about the roll-out of universal credit, and therefore the rundown of tax credits, the DWP pays us back money for tax credits. That is a matter of some dispute—that is probably the best way of putting it—into which there are ongoing conversations between us, the DWP and the Treasury. If we are being transparent with you, at the minute those staff are being paid for by the DWP under the various agreements that were struck in spending review 2015 about the roll-out of universal credit.

Chair

So you have very little control.

Jon Thompson

It is not in our core budget, no.

Chair

That is very helpful.

Gareth Snell

Briefly, there used to be an HMRC centre in the middle of my constituency that has been closed. There were 215 workers there. They were offered the opportunity to move to one of the new regional hubs, for which HMRC would underwrite their travel costs for a three to five-year period. Those who could not move were basically told that that was the end of their time with HMRC because of the nature of it. When you have done your estate planning, how much modelling have you done on estimated costs of redundancies and building in the costs of transport arrangements for new staff? There was nowhere else for these staff to go. They can travel to Birmingham if they want to, but most of them don’t want to, so the cost to HMRC in the short term will be slightly more than it would have been had they stayed there. Where you have done that modelling across the country, how close does it come out in terms of the modelling versus the actual output?

Jon Thompson

We did all that modelling on the basis of the postcode of the individuals working for HMRC, as opposed to where they worked, because obviously some people already travel reasonable distances. I was in Sheffield the other day, and there are some colleagues in Sheffield for whom the nearest regional office is Manchester, as opposed to Sheffield, for example, or Leeds or Nottingham. We modelled it on the basis of the residential postcode of the individual. The broad assumption made in the business case was that 10% of colleagues would not be able to make the journey to a regional centre, and the cost of that was built into the business case for the programme called “Building Our Future Locations”. It is part of the investment funding in that and ultimately saves £75 million a year on the cost of running the estate. The cost avoids £90 million on the PFI with Mapeley—sorry, I’ve got those numbers the wrong way round. It saves £90 million on the baseline and £75 million a year on Mapeley. There is a business case that backs that up. The business case is amended at least every six months so that it is transparent about what we assume. It has been transparent to NAO colleagues and there has been a dedicated hearing—straightforwardly, yes.

Gareth Snell

So where you have done that modelling on the 10% of colleagues who could not travel, where you have started the process of decommissioning sites and moving work to regional centres, how close to that 10% have you been? I know that in Stoke there have been much more than 10%.

Jon Thompson

The 10% was made on a national basis. The 5,000 people who work in Newcastle now will not be going anywhere, so that will be zero. We started with offices that were the furthest away and therefore the percentage is more than 10%. We still think 10% is a reasonable estimate.

Gareth Snell

Just to push you, where you have started this decommissioning, what percentage of the existing workforce have not taken up the opportunity to travel to a new regional centre?

Jon Thompson

So far?

Gareth Snell

Yes.

Jon Thompson

I couldn’t give that off the top of my head, but we still think the overall business case of 10% is sound because the 13 regional centre locations are where the majority of people are now. If it would help you for us to give you an update on the business case, I am sure Justin could fairly easily write to either the Committee or to you with those assumptions.

Chair

It would certainly be helpful if you could give us a list of the new locations, particularly for additional temporary accommodation in Birmingham, Bristol and Leeds, which is imminent, from what you have said in your letter, and due to be delivered in the autumn of 2018.

Jon Thompson

Where exactly they are?

Chair

Yes, what the actual addresses are. We know that colleagues of all parties around the House have raised real concerns about this programme, as you know, so we won’t repeat it. Another question is about the distance from the regional hubs to certain centres. Have you done a full and proper analysis of the likelihood that certain areas will be visited for compliance and profiling of risk? If you take an area that is furthest away from a hub, there will be less chance of a member of staff visiting unless they are going to stay overnight. That is one of the concerns we have always raised about this hub approach.

Jon Thompson

Whatever marginal costs there are for that are built into the business case, but that presumes that our compliance approach requires us actually to visit them.

Chair

I don’t want to go back over that. We know that a lot is coming down through algorithms and other ways, but there is still something about knowing the patch. Have you done any further analysis, now that you are going to go through an interim period, at least with more officers, not just the hubs, about whether that has had an impact on the profile of an area?

Jon Thompson

I don’t think we are going through an interim period. If one might be up front about it, I think there is a relatively active debate about whether you need to know your patch in the same way as an old district inspector used to know Taunton versus Bridgewater, which had different pros and cons.

Chair

I would love to go down this route again, but in the time we have got, we will not be able to. Mr Chapman, do you want to come in on this issue about the offices or are you going to hold fire?

Douglas Chapman

I would prefer to move on.

Chair

We will move on in a moment. If we can get the list of offices, that will be helpful. The Members for Dundee and Preston and their constituents will be happy about the changes there. I will turn now to Mr Douglas Chapman.

Douglas Chapman

A couple of big figures jumped out from the Report. The first was the level of tax due that might not be collected, which was somewhere just shy of £7 billion. For me, however, the eye-watering figure was the question of reliefs. We have seen an increase in the amount there, up 3.2% on last year. There are 105 reliefs that you can cost and that you know how much they cost Revenue and Customs. A further 80 that are nil or negligible, but 239 reliefs are offered for which you have no idea what they cost. Of the 239, how do we know that they give value for money if you have no idea how much they cost?

Jon Thompson

Do you want me to answer the question about the £6.9 billion that we estimate is uncollectable as well?

Douglas Chapman

Focus on the reliefs, if you can, given that time is short.

Jon Thompson

The straightforward answer is that it is about personal and corporate insolvency. We provide for that—it is the driver of the £6.9 billion. On reliefs, you are right: we published a list of the 239 tax reliefs for which we do not collect the data—I have it in front of me. When publishing that list, we also publish the reasons why we do not estimate the return. The primary reason—for 179 of those 239—is because we believe that the cost of collecting it is disproportionate, and it is not required for a tax return. To bring that to life, the potential capital gain on selling your private residence is not subject to any tax. If you buy a house and sell it years later, there is a gain on that, and there is no requirement to give us that data; we do not collect it. It is not relevant to anyone’s tax. That is the primary driver. Similarly, if a relative dies and you receive a life insurance pay-out, that is not subject to tax. We don’t collect that information; it is not required for tax returns. The primary driver of that 179 is that we believe that the cost of collecting it is disproportionate and it is not required for a tax return.

Douglas Chapman

On the 3.2% increase, how do you know what is driving that figure upwards if you are unaware of the value of these other reliefs?

Jon Thompson

Jim, do you want to answer the question on the 3.2%? I think that is very different from the ones that cannot be quantified. You are asking two different questions.

Jim Harra

There are two different questions on how much a relief costs where it is supposed to achieve a policy objective: is its cost value for money, by reference to the policy objective it achieves; and are we sure that only people who are entitled to a relief are getting it, and where we see some change in the tax receipts, whether that is evidence of non-compliance? We distinguish between those two things. Our primary job obviously, as a tax administration, is to ensure that everyone who is entitled to a relief gets it, and that nobody who is not entitled to it gets it. We have our general compliance approach to ensuring that we do that. We also have guidance in place for all our product owners—our specialists responsible for different parts of the tax code—to ensure that they understand the compliance risks in relation to their relief, and that proper arrangements are in place in the Department to manage those reliefs. When it comes to evaluating whether the cost of a relief is good value for money, we do that based on, first, the nature of the relief. Some are structural and are really there to help define the tax base. They are not really intended to achieve some policy outcome, and the question of whether they are value for money does not necessarily arise in the normal way.

Chair

Can you give us an example?

Jim Harra

An example would be—although we obviously measure the cost of this—the personal allowance. That is a policy decision and is not intended to achieve a specific outcome for which you can measure whether the cost is relevant to the outcome. It is basically a political decision. However, there are some that we define as tax expenditures that are intended to achieve a particular outcome. For those it is much more relevant to measure whether they are achieving an outcome and whether their cost is commensurate to the gain that they give. We identify which reliefs we are going to evaluate based on the nature of the relief, how much we think it is costing, what the kinds of risks and issues are about it, and what the political priority is. For example, Ministers will decide which areas of the tax code they do and do not want to look at. There is a programme based on those kinds of criteria. Where we do not have cost data and we decide that a relief needs to be looked at in depth, we can obviously make some estimates that would not be robust enough to feature as statistics but might be robust enough to inform policy thinking, or we might do a one-off exercise to collect some data, say, on a sample basis. The volume of reliefs we have here means that we cannot do that every year for every relief.

Douglas Chapman

Would your thinking be robust enough to consider the possible increase year on year—say over the next two or three years? How do you see that trend going? It was 3.2% for the last financial year. Is there a straight-line graph to say that you are expecting that to increase at the same level? If you are, we are talking about maybe £40-odd million over the next two or three years, where you would be expecting reliefs to rise by that amount when every public service is under huge pressure.

Jim Harra

Particularly for some of the principal reliefs there are forecasts for how much they are going to cost us. The guidance to the product owners is that they must monitor variances against that forecast and they must understand what the reasons for that are. Where the policy is changed and that is going to result in more or less relief being given there is a specific forecast of what the effect of the policy is expected to be. It is part of our guidance to evaluate whether that is the case, so variances against forecasts are looked at both from the point of view of whether that is an indicator that something is going on from a compliance point of view that we need to worry about, and as an indicator of, even if the relief is being correctly claimed, whether that changes the balance of the judgment about value for money.

Douglas Chapman

I am not sure whether it is in your remit, but in terms of the number of reliefs that were in the estimates, I am wondering whether you are proactively looking at that to make recommendations about how effective these reliefs are, and whether the Treasury would think about simplifying some of these or joining them together to make it much simpler for an actual person who is applying and making their tax return at the end of the year.

Jim Harra

As a tax administrator, I am all in favour of simplifying it. The fact that there is an Office of Tax Simplification is a recognition by the Government that the tax system needs to be simplified. However, the reality is that the list of reliefs grows faster than it reduces, so Governments and Parliament add more reliefs than they take away.

Chair

Pesky politicians. Maybe it would be easier without us.

Jim Harra

As ever.

Sir Geoffrey Clifton-Brown

Mr Harra, on this very interesting subject, when policy changes are being contemplated—either change of tax rates or a new tax—presumably you do a forecast for what a tax exemption is going to cost. Does that forecast include behavioural changes?

Jim Harra

Yes. Ultimately it is the independent Office for Budget Responsibility that certifies forecasts for policy changes. Those are itemised in the Budget and there is what we call a TIIN—a tax information and impact note—that explains how that has been arrived at. They do take account of behavioural change as well as the sort of direct effect.

Chair

Can you tell us when you last got rid of a tax relief, or when the Office of Tax Simplification did? Can you think of one that you have recently got rid of? We went through the list and we were surprised to see how tidy it was. I wasn’t sure if it was just because it is tidy on your website.

Jim Harra

Sorry—is this about the length of the list of reliefs?

Chair

You have the list of tax reliefs up online, which is a welcome step forward from a few years ago, but it did seem very nicely sanitised because there were some odd ones around a few years ago. Have they all gone now?

Jim Harra

Some were removed. The Office of Tax Simplification did make some recommendations a few years ago of reliefs that should be removed, and quite a significant number were as a result of that. However, the overall size of the list has crept back up again.

Chair

Sometimes I’m sure you love not having a Minister running your Department, because it must make it easier—except when we get involved in the Budget, and so on.

Caroline Flint

The UK is going to introduce measures to increase transparency and accountability of property ownership by introducing a register of beneficial ownership for properties owned by foreign companies. That is due to come into force by 2021. I understand that will be overseen by BEIS through registration of overseas entities. How is HMRC working with BEIS to understand the likely impact of these new rules?

Jim Harra

We are working closely with them. Obviously, transparency of beneficial ownership to tax authorities is very important in terms of managing avoidance and evasion. There is no doubt that hiding behind foreign entities is one of the techniques that tax evaders, for example, use, but the problem is much wider than tax. Across Government, law enforcement agencies, including HMRC, are playing into that to make sure that we can use the data. It is part of our offshore evasion strategy that if there are UK taxpayers hiding behind a foreign entity, there needs to be no hiding place for them.

Caroline Flint

When we were in Washington earlier this year—we highlighted this in our Report about the visit—they explained to us how they were using geographic targeting orders, which seem to be along the same lines of what we are attempting to do. Another part of it, of course, is cash transactions in property because of money laundering. Does HMRC expect to undertake enforcement activity in this area?

Jim Harra

I am not aware that we will be directly responsible for enforcing compliance with the register. We obviously have an interest in using the data that is on that register for our risk assessment and our casework. I expect there will be some benefits to us in terms of the compliance yield we get because we have better data.

Caroline Flint

Who do you think will follow up on enforcement, then?

Jim Harra

BEIS is leading the project. It is not an HMRC project.

Chair

Sorry—just BEIS generally? There isn’t a specialist team dealing with this?

Jim Harra

I don’t have information about that. I know that HMRC has an interest in the project—

Caroline Flint

Is there a cross-departmental working group on this?

Jim Harra

Yes, and our fraud investigation service is on that.

Caroline Flint

Just going back to the register, will HMRC have a role in validating the information on the register?

Jim Harra

I’m not sure. Not that I am aware. If a foreign entity had to declare its beneficial ownership, that would probably be helpful new data to HMRC rather than data that we could validate.

Caroline Flint

I suppose it is also about just trying to get a handle—not exclusively in London, but particularly in London—on which properties are foreign owned, and on the nature of that. They are left empty and what have you, and you have buying and selling on, which may have a tax implication.

Jim Harra

There are two tax implications for us. One is that there might be a UK taxpayer hiding behind that entity and proceeds of something that should have been tax might have been used indirectly to buy the property. We would be interested in understanding that. The other is that capital gains tax has increasingly been expanded to non-resident property owners, and we want to make sure that neither stamp duty land tax nor capital gains tax is being avoided or evaded through people hiding behind entities. Those would be the two key areas of interest for HMRC.

Caroline Flint

Have you made any assessment of the expected increase in tax revenue as a result of these changes?

Jim Harra

I’m not aware. I can check and let you know.

Caroline Flint

I think it would be quite helpful to know. As with everything, just going back to the tax relief situation, it is all very well having all these things, but something must come out of them that deals with fraudulent or even criminal activity and improves our tax take. You don’t want these measures to fall into disrepute in one way or another—wouldn’t you agree?

Jim Harra

But obviously the benefits of this measure are intended to go beyond tax compliance.

Caroline Flint

I accept that, but there should also potentially be, in dealing with tax avoidance, some sort of sense of what impact that would have on you.

Chair

Relatively, it may be a small amount of the total tax take, but it is the symbolic nature as well.

Jim Harra

I will check whether we have—I know that we are definitely keen to have the data and to use it, but I will check whether we have made any estimate or forecast of what it is we think we will get from it.

Caroline Flint

Thank you.

Chair

This is something we might want to take offline and pursue in other places. I want to go back to the issue of collecting tax, and particularly VAT. We have obviously had a lot of dialogue about online VAT, and in the month that we heard that Amazon’s tax bill had gone down, despite its UK profits rising, we wanted to pursue where you had got to with the online VAT fraud issue. Mr Harra, first of all, can you tell us how much extra VAT you have collected as a result of using the new compliance measures to tackle online VAT fraud?

Jim Harra

You will recall that at Budget 2016, the Government announced new measures to make it easier for us to enforce the—

Chair

How much have you collected as a result of that?

Jim Harra

Up to 30 June 2018, we had received £150 million of additional VAT unprompted from sellers who have now registered for VAT. Also up to that date, we had issued joint and several liability notices that have resulted in assessments of £160 million.

Chair

One hundred and sixty?

Jim Harra

Yes. If you compare that with the forecast for what those measures were expected to achieve, which was £35 million in ’17-’18 and £75 million in ’18-’19, we believe that they are well on their way to achieving or exceeding the forecast revenues that we expected to get from them. Of course, those measures have since been added to in subsequent fiscal events that have not yet come into effect—

Chair

We understand that about 40,000 sellers or businesses have registered.

Jim Harra

Up until the end of June, we had about 43,500 new applications to register from non-EU-based online businesses, and we think the vast majority of those will have been prompted by the Budget ’16 measures and by the risk that they are going to be excluded from the marketplaces if they do not register.

Chair

Given those numbers, are you able to vet those registrations? We had quite a lot of dialogue and quite an interesting exchange with marketplaces about what HMRC has to do to verify those. Do you have the people power to vet those new registrations?

Jim Harra

Yes. As part of the Budget ’16 measures we were given additional resources to police it.

Chair

And that is enough for the volume you are getting through.

Jim Harra

Yes.

Chair

That’s good. So the unpaid VAT that you have so far collected is £150 million.

Jim Harra

There is £150 million unprompted that has been declared by those 43,500 businesses, and then there is £160 million that we have assessed on joint and several liability notices.

Chair

Do you have more unpaid VAT to come? Do you have an estimate of what you might not have got yet? Are you still pursuing people?

Jim Harra

Our last estimate for ’16-’17 was that the gap in this area, from both overseas sellers and UK sellers, is between £1 billion and £1.5 billion a year. We believe that about 60% of that comes from overseas businesses and about 40% from UK businesses, so there is clearly more to do. We obviously have to do more with those measures, but also with the additional measures that have been brought in. The fulfilment house scheme just started in April this year, and we have also extended the scope of the joint and several liability notices recently, which is still to have an effect.

Chair

We had a lot of discussion about whether you force the organisations—the marketplaces—to comply, and how they would work that. How many online marketplaces have signed the memorandum of understanding so far?

Jim Harra

Seven, including the large ones that you would expect to be on that list.

Chair

Seven out of how many? How many would you have hoped might have signed up to that?

Jim Harra

I do not have a figure. Frankly—

Chair

There are a lot of little online marketplaces.

Jim Harra

Yes. The vast majority of the sales go through three platforms, and then there are some small ones. We do not have complete knowledge of who those very small ones might be, but we believe we have already captured the big players, and we will continue working on that.

Chair

You think it is working in practice.

Jim Harra

They are voluntary agreements. They do two key things for us. One is that the marketplaces promise voluntarily to give us data, so that we do not have to use our information powers to extract it from them, which is obviously beneficial to us, and also they promise to collaborate quickly with our inquiries.

Chair

They have promised that, but are you assuring yourself that they are doing what they said they would?

Jim Harra

Yes. Certainly, on what we have seen so far, both in terms of them producing the data that we ask for on an ongoing basis and also acting promptly on joint and several liability notices, we are satisfied. There is a new obligation placed on them in relation to overseas businesses, where they now have the obligation to police it themselves rather than wait for a joint and several liability notice from HMRC. We will be monitoring that they are applying good practice and due diligence in doing that.

Chair

Given that they are providing it before you have to issue notices, are you saving money on the enforcement element?

Jim Harra

We did face the prospect that, if any of them resisted giving us data, we were going to have to go through tribunal and court to get it, or go to our overseas tax authorities and ask them to get it for us. That would have been not so much costly as very time-consuming. It would have held us back in terms of being effective in tackling the underlying problems. So the fact that we are getting it voluntarily, which some of them were already doing before they signed the MOUs but are now committed to doing, is helpful.