Public Audit and Post-legislative Scrutiny Committee

Meeting date: Thursday, June 15, 2017

Official Report 544KB pdf

---

“The Administration of the Scottish Rate of Income Tax 2015-16”

Agenda item 2 is further consideration of the report by the National Audit Office entitled “The Administration of the Scottish Rate of Income Tax 2015-16”. In March, we heard from the National Audit Office and the Auditor General for Scotland on the report and agreed to take evidence from Her Majesty’s Revenue and Customs. I welcome Jim Harra and Sarah Walker, both of whom are from HMRC. As there is no opening statement, we will move straight to questions.

I have pretty much the same question that I have asked in previous evidence sessions on the issue. It is about the identification of Scottish taxpayers. I believe that there is no definitive record or data set of Scottish residents against which to test your success in identifying Scottish taxpayers. In previous evidence sessions, we have tried to pin down the error margin and find out whether it is 5 or 10 per cent, but nobody has been able to tell us that. What reassurance can you give us that you have identified all the taxpayers?

We have three tasks. First, we have to ensure that everyone in our database who has a Scottish address is properly flagged as a Scottish taxpayer. We then have to check the accuracy of our database and make it as accurate and up to date as possible. Then, we have an on-going job of maintaining that and continuously improving on it.

We have taken a number of steps to check our database against third-party databases. Where there is a difference between those, we have looked to see whether we can learn anything about our database management. For Scottish addresses on our database, we have run corroboration exercises with a number of third-party databases, such as the electoral roll and the databases of credit reference agencies and major retailers, and we have been able to corroborate between 98 and 99 per cent of the addresses that we hold. That does not mean that the remaining 1 to 2 per cent are wrong; it just means that they are not corroborated by anyone else.

We have also gained some reassurance that our processes for keeping our database up to date mean that it stands up to scrutiny compared to those other databases. In fact, where we found discrepancies between our database and the electoral roll, in the majority of cases, ours was the more up-to-date and accurate record. It is a question of continuously improving, but you are right that there is no definitive benchmark against which we can judge whether our database is at 100 per cent or 99 per cent completeness.

It seems that, mostly, you rely on postcodes. How accurate is that approach?

We have a set of parameters that identify Scottish addresses in our database. Initially, we relied too much on postcodes and what we call the PAF—the postcode authentication field. We have since extended that scan, and we pick up addresses in lots of different formats. We have also carried out an exercise to identify addresses on our database that are incomplete—for example, those that do not contain a full postcode. We have done a cleansing exercise to bring that up to date.

A small number of postcodes straddle the English-Scottish border, so we cannot rely on postcode alone to determine who is a Scottish taxpayer. In those cases, we have taken further steps to identify which addresses in the postcode area need to be flagged as Scottish addresses.

You have mentioned checking against third-party databases—the electoral register, for one. Could you remind me what other databases you check against?

Sarah Walker will give you some details about that.

We used a commercial firm that specialises in that sort of thing. It has a number of address lists—commercial credit agencies and places where people are registered to take deliveries from mail order companies. There are mobile phone and other commercial providers to which people have to provide an address, and those providers make their address lists available to commercial companies, so they are then available to compare with our lists. If you can identify somebody by name and address, you can see whether the same name and address appear on other lists.

Every time we order something by mail order, you guys are checking on that.

That is not what I am saying. I am talking about what is made available to commercial companies. We do not get all that information ourselves.

You purchase that information.

You are right, in that your footprint with retailers, mobile phone companies and utilities is building a record in databases, which those commercial organisations all share, and we use that record to verify and validate our database. We do not just take addresses; when our record does not corroborate theirs, it does not mean that theirs is more accurate than ours, but it is something for us to investigate.

That is obviously an on-going process. How often do you go through it? Is it one big exercise every year, or is it on-going?

We are talking to the Scottish Government now about exactly how regularly we should do that comparison. We expect that we will do it every year. It is quite a major exercise. It costs money and it takes time to do, so I do not think that we would do it more often than that.

I was just about to come to the question of cost. Is it included in what you already charge the Scottish Government?

Yes.

It is not an additional cost.

No, it is not an additional cost.

I understand that 80,000 employers had problems with tax codes, but that the issue is now fixed. How was it fixed, and what was the real problem?

As well as identifying people as Scottish taxpayers, we require employers to deduct the correct amount of tax from them. This year, 2017-18, the amount of tax is different from that in the rest of the United Kingdom for the first time. We send a coding notice to both the employer and the employee telling them what code to apply, to ensure that the correct amount of tax is deducted. We then have a programme of employer compliance checks to ensure that employers comply with their pay-as-you-earn obligations, apply the codes correctly in their payrolls and deduct the correct amount of tax.

When we go out on employer compliance visits, we find a very high level of compliance, with the lowest tax gap out of all the taxes that we administer. There are problems, however. Employers sometimes make errors. In the case of the Scottish code, we have found that, in some instances, the software programmes that employers use have not been updated to take account of what we call the S codes—the Scottish taxpayer codes. We have taken action both with employers, where we have found that they were making errors, and with the software providers, to ensure that they have the correct specification and update their products for their customers.

How widespread is that error?

09:00  

It has been very small in percentage terms, but it is there. I will explain what its impact would be.

We run a reconciliation exercise at the end of the year for each employee. At that point, we would pick up whether there had been an underdeduction of tax and correct it. There would be no impact on the Scottish Government’s finances, but it would be undesirable from the taxpayer’s point of view because, at the end of the tax year, they would find that they have an underpayment when, ideally, that would have been taken out of their salary over the year.

Despite the fact that we have a fail-safe reconciliation at the end of the year, the ideal is that the system operates correctly over the year. We have very high compliance rates and we are continuously improving on those. The key is to make sure that software providers keep their products up to date and that employers use the latest versions of those software products.

I am interested in the situation whereby 420,000 taxpayers did not receive the information that the other 2.45 million taxpayers received. The Comptroller and Auditor General said in his report that that

“may have created a less informed group of taxpayers”.

Do you recognise that concern, and is it well founded?

It is a possible concern. The 420,000 Scottish taxpayers who were not picked up in the initial scan received a written communication, but it was a different communication from the communication that was received by the other 2.4 million people. We sent letters out to 2.4 million people in December 2015, and the other 420,000 people received a pay-as-you-earn coding notice in early 2016-17, which specified that they had been identified as a Scottish taxpayer. That is our standard means of communicating with people on pay as you earn about the code that we apply to them. Everyone received a communication, but they were different communications for the two population sets. We have not seen any evidence that that has had a practical impact on the taxpayers.

Communication is clearly important so I am interested to hear about your wider communication strategy and its cost.

Good communications are key to maintaining and improving the accuracy of our address database, so we continuously remind taxpayers, employers and tax agents of the need to notify us of changes in addresses. That is the key compliance activity that makes a difference.

There was an initial communication exercise that involved unpaid and paid communications, and that communication continues. For example, we use our regular contact with taxpayers and employers to remind them of the need to update their addresses. We used Twitter just two days ago to tell all our followers of the need to keep their addresses up to date. Sometimes that is specifically about Scottish addresses, but sometimes it is a general message to all taxpayers, which the administration of Scottish income tax benefits from, as does the administration of tax in the rest of the UK.

We can do a lot of effective communication that does not involve additional cost, and we can give advice on the additional benefit from paid communications, so the question is how much money the Scottish Government wishes to spend on paid communications. Most of the paid communication that we did in the early stages was targeted at specific groups of Scottish taxpayers whom we thought we needed to reach to make sure that they heard the message about keeping addresses up to date—they included people who live in areas of the Borders. It might not be good value for money to target groups that are more widespread than that.

You mentioned Twitter, which can be a very cost-effective way of getting a message out. How many followers does HMRC have on Twitter? Do you use Facebook and other social media channels?

I think that we have 28,000 followers on Twitter, many of whom are professional associations and tax agents who spread the message more widely, and we also use other social media.

What assessment has been made of the communication strategy and the cost incurred to ensure that it provides value for money?

We did an evaluation of the publicity that we conducted at the time of the launch of the Scottish rate of income tax, which was when we sent out the letters. The evaluation suggested that a significant proportion—around 30 to 40 per cent—of the target population, who were people who had recently moved, people who live in the Borders and students, were aware of the campaign that we had conducted.

There was an increase in the number of people updating their address details on our website. We think that the campaign was effective at the time. We continue to use cost-free methods such as Twitter and our contacts with employers and agents, and we are talking with the Scottish Government about whether we should be doing more targeted communications to pick up people who move house.

If we are fairly confident that we have a good picture of what the static population was when we started, we need to make sure that we capture the changes—the people who move into and out of Scotland. We are thinking of using the data that we get from stamp duty, HM Land Registry—where people register the purchase of property—estate agents and other places to prompt people and remind them that if they have just moved into or out of Scotland, they have to tell HMRC. We are looking at what is the most cost-effective way of picking that up.

We also use our own digital services. We are conducting publicity to encourage people to sign up for a personal tax account online, through which they can manage all their tax affairs. That includes making sure that they are aware of the address that we hold for them and making it easy for them to update that information.

That is helpful, thank you. I am sure that our witnesses already follow the committee on Twitter, and I am sure that my colleagues will endeavour to follow you back by the end of the meeting.

I do not think that I have ever knowingly wanted to follow HMRC on Twitter—forgive me for that.

Good morning. Is it not a bit disconcerting for the public to hear that, to a degree, HMRC relies on mail order companies and mobile phone companies to compile a definitive list of taxpayers?

That is not what we do. I am glad that I have an opportunity to clarify the point. We regard ourselves as holding the database of who pays tax, but it is important that we check its accuracy against all other available sources, maintain its quality and learn and take action if any corroborative evidence tells us that we have a problem. Those things are a source of information that helps us to check what we have, but we are not reliant on them to maintain the database.

In no case have we replaced an address that we hold with an address that we have taken from that sort of database. The addresses that we hold, which we get primarily from the taxpayer and sometimes from employers, are always our primary source. The comparison with the commercial databases would lead us simply to raise a query, if you like. If they have a different address for someone, we assess whether we need to query that and make more inquiries, but we would not take the address in the commercial database in preference to the address that we hold.

An outcome of the exercise that we have done is that we have not found any database that we would wish to use to overwrite ours. In the majority of cases, our database proves to be more accurate and up to date than the databases with which we compare it.

There are no cases where data that you have got from mail order companies or mobile phone companies has given you additional information to corroborate the information that you have on a Scottish taxpayer—you had the information anyway.

We looked at whether the evidence from other databases tended to corroborate the address that we had, whether it caused us to raise a query about the address, or whether it did not tell us anything at all, which meant that the address that we had was uncorroborated and we had no evidence either way.

Where the other databases tend to corroborate what we have, we regard that as evidence that our databases are of good quality. Where they say something different from what our database says, that raises a query on our part, as Sarah Walker said, to go back to the taxpayer and check exactly where they live. Where the databases do not corroborate either way—a significant proportion provide no corroboration—we look at other factors to assess our data. For example, if we have an address for someone for whom the other databases do not show anything at all, but we know that they are employed by a chip shop in Aberdeen, the chances are that they are a Scottish taxpayer and our Scottish address for them is correct.

We learn from such comparisons—that is why we spend the money on the exercise—but they have tended to corroborate our Scottish addresses. As we said, between 98 and 99 per cent of them have been corroborated by that exercise and we have yet to identify a database that we think is a much better database than ours.

You mentioned addresses with partial postcodes in some of the data. How did you resolve that issue?

We work with Royal Mail in such cases and we have teams of people who go through individual addresses. If we have a road and a house number, we will work with Royal Mail to make sure that we have the right postcode on the record.

Why would that not have been done as a matter of course, to make sure that the data that you have is always correct? It only seems to have arisen because of the requirement to identify separate Scottish taxpayers.

It is a continuous process. We do that work year by year to make sure that we are keeping our addresses accurate but we have millions and millions of addresses; it is a constant process. This is the first time that someone’s address has had an effect on their tax, which makes it even more important that those addresses are up to date, so we have put extra effort into making sure that the Scottish addresses are corrected.

What is to stop the problem continuing if people move around, as they generally do—particularly in relation to cross-border addresses? What is to stop those anomalies continuing to be part of the system?

It is a never-ending task to maintain and improve the quality. It is not a static, one-off exercise. We have assessed that there are about 80,000 moves a year across the England-Scotland border—roughly half in each direction. If we did nothing and did not update our databases, they would degrade over time, so it is a continuous exercise.

We get a continuous flow of data—for example, taxpayers contact us all the time to tell us that they have moved house; employers contact us all the time to say that they have taken on a new employee and to give the employee’s address; and we periodically run those corroboration exercises against the commercially available databases, which throw up cases for us to query. It is a never-ending process.

Lastly, in the dataset that you have, is the solution not as simple as allowing people themselves to identify the country that they live in?

That is exactly what we want people to do.

Are you doing that now as a matter of course?

Yes, we rely on people to give us their address and that is how—

Yes, I know that, but do you ask people to give you information about which country they live in—Scotland, England, Wales or Northern Ireland?

For self-assessment, we will. Where people are filling in a tax return, there will be a specific question asking, “Are you a Scottish taxpayer?” We need to have people’s addresses for all sorts of purposes and that is our main source of information for this. If they are prepared to tell us that they are a Scottish taxpayer, they ought to be prepared to tell us their address because we need it anyway.

Yes, I know that—I understand what you are asking there but that is not what I am asking. Are you asking people to give the country that they live in? They may be a Scottish taxpayer or they may think that they are not a Scottish taxpayer, but do you ask them whether they live in Scotland, England or elsewhere? Do you ask them that information so that you can record it? You were talking about S codes—

It is important to understand that there are only about 9 million taxpayers in the whole of the UK who have to complete a self-assessment return and therefore are under an obligation to tell us that information. They will be asked for that information on their self-assessment return.

For the vast majority of taxpayers, their affairs are administered between HMRC and their employer and there is no obligation for them to get in touch with us. Therefore we have to look for data that tells us where they live. The address is the key thing. We have not gone down a route of asking all taxpayers to declare the country in which they live for tax purposes; we have asked them to keep us up to date on the address where they live and then we apply that information to determining their taxpayer status.

If, for any reason, people feel that the process comes out with the wrong answer for them, they have the right either to update the data if they think it is wrong or to challenge the decision if they think the data is right but they do not think they pass the test.

Would it not be easier just to ask them the country they live in and gather that data?

There is a test for determining whether someone is a Scottish taxpayer and it applies over the whole of the tax year, so it is a slightly complex thing to ask some people, particularly if they are mobile. We feel that, if we know the address where you live, that will enable us to determine it accurately.

09:15  

The test that you have applied so far missed 420,000 of them.

Yes, there was an error in our original scan. Although we had a Scottish address on our database for those people, the technical parameters of the computer scan did not pick those addresses up, and that was an unfortunate error, for which I am very sorry. I am glad to say that it was identified and fixed very quickly, but it does not go to the accuracy of our data about where people live. It went to the accuracy of that one-off job of running the scan against our system.

If you do not mind me pressing you on that, what does it mean when you say “running the scan” and that the scan did not pick up the fact that they were Scottish? Do you mean that it missed a field where there should have been a postcode, or it was in the wrong field, or there was only a partial postcode, or there was no country name? How did it miss it?

We had all that address data on our systems, but we had not in the past flagged people as Scottish taxpayers, because there had been nothing in the tax system that required us to do that. When the Scottish rate of income tax came in, we had to identify all of those addresses that we held on our database and flag that record as a Scottish taxpayer record. There was a technical specification of the scan that ran across our whole address database to pick that up, and some of the parameters of that scan did not pick up some of the input methods that had been used for our addresses. In particular, it overrelied on postcodes and the way in which the addresses had been input from postcodes. In fact, there are other means by which addresses can be input, which that initial scan did not pick up.

Two things should have happened. First of all, the people who designed the scan should have had a deeper understanding of how addresses are input into our system and should have got the scan right. Secondly, there should have been more testing of the results of the scan to identify the error, and unfortunately that mistake got through. When it was picked up we fixed it, but if I was doing that exercise again—which we will be doing shortly for Wales—I would be looking at the lessons to be learned to improve the quality and testing.

Is not the solution simply to have the word “Scotland” or “Wales” or “England” in a column in the data attached to every address? That way you will not miss any—if the software works, that is.

Until now, because the country in the UK in which you live has not been relevant to your tax affairs, that has not been a field that we have required people to capture. With the exception of a small number of postcodes that straddle the border—about six, I think—the postcode tells you which country the address is in, and the postcode is what we have always had on our address list because that is what is relevant for mailing. We feel that that is the best way, from the data we have, of identifying Scottish taxpayers. With hindsight, if we had known when we were starting to collect addresses from people 30, 40 or 50 years ago that there might have been a differential, we might have asked people for the country, but that is not where we are and I do not think it is necessary to do it in order to get at the identification we need, because the address data and the postcode data pretty much do it.

So there will be no errors next time round.

I hope that, after the equivalent scan in Wales, I will be able to say that there were not. We have certainly learned lessons from that exercise, which Scotland will not gain from, because it was a one-off exercise, but Wales will. We have also learned lessons about the on-going need for assurance, which have strengthened our assurance activity, and you will benefit from that.

The estimated costs of implementing the Scottish income tax powers keep reducing, which on one analysis sounds quite good, but how can a £5 million saving—I think that that is the latest figure—be made?

The story is that the initial estimate, which I think was £40 million, was given back in 2010. That was a long way ahead of implementation. We were doing a lot of new things that we had never done before, and I think that we were cautious in the estimates that we gave.

Over time a number of things have happened. We have been able to get a better idea of how to do the information technology changes, and the costs of those changes are coming down. However, the big reason that the estimates have come down is that we thought that there would be a lot more costs for contact with the public—the mail-shot that we did and staff and operational costs in our business from dealing with changes and correspondence.

As it happens, because we had good time to plan for it, we managed our communications. We did work at database level to check our addresses. We did not need to make as much contact with customers as we expected and we had a lot fewer contacts and queries from them than we thought we would have to deal with. A lot of the cost reduction is because of less contact with the public and less administrative work than we thought that we would have to do.

The recent reduction is, again, largely because there is less administrative and operational work. We also have a clearer idea of the IT requirements for working with pensions relief at source; we have had to put in a separate system to make sure that pension schemes that are claiming relief at basic rate from HMRC are able to identify Scottish taxpayers, so that if the basic rate is different, they will claim a different amount of relief. That work is on-going. It is the last big chunk of work that we have to do.

This committee often sees estimates going the other way. By one analysis, the reduction is all quite encouraging, but there is a concern in my mind that significant overbudgeting has consequences as well for how one makes provision for it. Are you aware of how the Scottish Government made provision for the £40 million in terms of its own funding, and whether in effect it said, “We haven’t got £40 million to play with, so let’s put it over here”? If so, what impact has that had on day-to-day running?

I cannot comment on how the Scottish Government manages its budget. We have been keeping in touch with the Scottish Government on our reassessment of the costs, but we have also been working closely with it on the short-term forecasting. Each year we have given it an estimate, which is the best estimate that we can give. It has been involved in our programme governance, so it has the same access to cost forecasts that we do and it sees all our programme and project papers. It has as good an understanding as we have.

I agree that the estimate of £40 million in 2010 was pretty speculative. I think that it was done well before we had done a lot of work on it—we were asked to give an estimate and we did. We erred on the side of caution because we thought that this was a big, novel project and the cost was difficult to predict at the time.

We have looked at an error on the number of Scottish taxpayers. There is an impact on the tax taken and there is a rectification cost.

I understand that this time around HMRC has borne that cost. If there were to be a further error, there would be a rectification cost, but presumably there would also be potential for rectification of the opportunity cost of not having taken all the tax that should have been taken. Is it reasonable to assume that HMRC would pick up that cost again, if there were a further error?

That is hypothetical.

We have in our memorandum of understanding with the Scottish Government a presumption that it will bear all our implementation costs, which follows Treasury rules for public expenditure. In the case of the error that led to 420,000 people being left off the database, there was no effect on tax take, partly because the error was rectified quickly and partly because there was no difference between the Scottish and UK tax rates at the time. For that particular issue, we agreed that we would bear the costs because it was clearly our error, but we would want to look at future instances according to the circumstances at the time.

To be clear, if HMRC was to make an error in the future for which there was a cost, one could not automatically assume that HMRC would pick up the cost of rectification. Is that fair to say?

It is fair to say that you cannot automatically assume that. I would expect us to follow the approach that we followed in the instance of the 420,000 taxpayers—we would approach it in a reasonable manner. In that instance it was clearly our mistake and we took it on the chin financially. However, as Sarah Walker said, we would have to look at each instance. We hope that there will not be any, but if there are we will have to look at them individually and assess what we want to do, and we will have to enter into a dialogue with the Scottish Government—I can pretty much understand where it will come from in that debate.

I would like to explore a final issue. The Scottish rate resolution needs to be agreed by the end of March. This committee’s predecessor committee heard that if the Scottish rate resolution is agreed right at the end of that period it could give rise to extra costs, because you guys have to change people’s tax codes at the start of the financial year. Do you have any idea what those extra costs will be and whose costs they are?

The costs are borne by the Scottish Government. The extra costs will depend on the nature of the change. The change this year to the higher rate threshold was relatively small; it affected only a minority of tax codes. A change to, say, the basic rate later on in the process would affect a large number of tax codes, so there would be a much greater cost.

Are you able to give me a ballpark figure?

I cannot off the top of my head, but we may be able to write and let you know.

That might be useful. Thinking aloud, I presume that the Scottish Government has to factor that into its consideration: if we change the tax rate, there will be a cost.

It is a very small proportion compared with the yield from the tax change, obviously.

The analysis is predicated on the Scottish rate resolution not being agreed until right before the end of March. That could incur additional cost. How far before the end of March would the rate resolution need to be agreed to avoid incurring the additional cost?

We do an annual tax code update for the whole of the UK, which runs around the end of the year—December and January. The agreement that we have in our memorandum of understanding with the Scottish Government is that by the end of November we need an assumption, if you like, to use in the codes that we issue in the main coding exercise. We will have to review that, because potentially the UK budget and the Scottish budget will be announced later in the year, so getting an assumption by the end of November may not be realistic.

We have been able to accommodate the Scottish budget announcement in December and reflect that in our main coding exercise. However, if a change comes after January, as it did this year, that means that we have to do an additional exercise to redo some of those tax codes, and that is where the extra cost arises.

Even if it happened right at the end of March, we would still do an additional tax-coding exercise. However, that would probably happen after the start of the new tax year, and it would have a bit of knock-on effect on employers, who would have to update codes at that time. However, it is possible to do, and there are precedents for our changing tax codes after the start of the tax year. A whole range of different things can happen, and it is always possible for us to accommodate late decisions, but a range of costs might be incurred.

09:30  

I would not mind having more clarity on those costs. At some point, the Parliament will debate the rate resolution, and it might want to know that there could be a cost implication of not getting something agreed by a certain time.

We can certainly go away and look at what we can come back to you with. As Sarah Walker has explained, there is a whole range of different scenarios, each of which would have to be looked at individually. In any case, we would expect to liaise closely with the Scottish Government on any particular instance or issue. We will see what in general we can give the committee, but there is obviously a wide range of scenarios, each with its own separate impact.

Thank you.

I want to pursue three supplementary questions on who is and who is not liable to pay the Scottish rate of income tax. First, what if I had a house in Scotland and a house in England and I told you that my main house was in England but I had a house in Scotland? Would I pay the Scottish rate of income tax?

It would depend on which address was your main home, and we have a set of criteria that we apply to determine that. For example, we have identified just over 2,000 cases of the nature that you have just described, in which people with an address in Scotland and an address elsewhere have told us that the address elsewhere is their main address, despite the fact that the Scottish address is a correspondence address for us. We have done some compliance work to understand the accuracy of what we are being told.

The wealthier you are, the more likely you are to have two homes and, indeed, to have scrutiny by us. Affluent and high-net-worth people are basically man-marked by a relationship manager in the United Kingdom who tracks their residence status, but Sarah Walker can probably give you more details of the test that is applied.

We are looking at, say, where your children go to school, where your family is, where you are registered with a general practitioner and that kind of thing. We would expect people to have a main home with those characteristics.

If Scotland were to introduce a 50p rate of tax, and I were to say to you, “My main home’s no longer in Scotland—it’s south of the border,” would you automatically accept that or would you apply these tests and say, “No, we don’t accept that”?

Although this is a new scenario for us in the UK context, we are very used to it in the international context. We have lots of taxpayers who tell us that, for tax purposes, they are resident in France or wherever while appearing to have a presence in the UK, and we are very used to policing that sort of thing.

We take a risk-based approach to compliance. In other words, we assess the risk of someone misleading us, and the level of intervention that we take is based on that risk. Relevant factors include the tax differential between Scotland and the rest of the UK; in the current tax year, that differential will be a maximum of about £300, and our assessment, therefore, is that the risk of people deliberately not complying will be pretty low. Obviously, though, if that differential grew in future, we would have to reassess that risk, and we intend to do that exercise annually and share the results with the Scottish Government.

In the circumstances that you have described, we would probably have assessed the risk to have grown significantly and therefore it would be something that we would have to actively police. That is exactly the type of case that we might well decide to investigate.

It depends on the taxpayer and their level of income, and what we know about their tax behaviours generally. For example, we will keep a close eye on a high-net-worth individual who we know has actively engaged in avoidance schemes in the past. On the other hand, if we see evidence that someone appears to have moved house, we may decide that, in that particular case, an intervention is not required. We would certainly be constantly assessing risk.

So you will monitor it.

The level of compliance intervention that we would expect to take would, subject to discussion with the Scottish Government, be relative to the risk assessment.

As part of the compliance assessment, do you look at people who are getting free tuition because they are allegedly staying in Scotland?

Factors like that would all pour into the risk that someone would actively mislead us. We would look at all the factors that might cause someone to mislead us about their address and which would have a tax consequence. Clearly, a range of factors can drive people’s behaviour—it is not just about the tax benefit, but that may be part of it.

Let us say that I have a son who is 19, and I am living in England and a taxpayer there, but he has told the authorities up here that he is resident in Scotland, and he is getting free tuition and not having to pay fees. Would you pick that up?

Obviously we are not directly accountable for administrating the risk of fraudulent access to tuition fees—

Who is?

I assume that it will be someone in the Scottish Government, in the department for education or whatever. We have ways to share data with other agencies that are investigating potential fraud. For example, if the police in Scotland are investigating fraud, there are gateways through which they can share HMRC information. We collaborate closely with other agencies on tackling fraud, because if someone commits fraud in one area, they are quite likely to do so in another area, including ours, so it is in our interests to do that.

And you would share the necessary information with Scottish Government enforcement or compliance officials.

Provided that there is a legal gateway for us to do so. We have to comply with data protection rules and the rules that say that we have to keep taxpayer information confidential, but there is an extensive range of gateways that allow us to collaborate with other law enforcement agencies. For example, in Scotland, we have HMRC staff embedded in Gartcosh who are both working on joint operations and arranging to share information. We always check that we have the right gateway.

If there is a risk in an area where there is no gateway to enable us to share information, we and our administrators have a very good record of acting quickly to create the necessary gateway.

I want to ask about another scenario, which is quite common, not necessarily just among high-net-worth individuals. This is not about someone who is trying to defraud the system; I am asking about how we define a Scottish taxpayer for the purposes of income tax. Let us say that I live in Scotland, but I work in London. I go down on a Sunday evening or Monday morning, I work in the City and I stay in London five days a week, and I come back up to Scotland on a Friday night. For all intents and purposes, I live in Scotland. Does my employer in London deduct the Scottish rate of income tax?

I believe that those people are known as WILLIEs—

WILLIEs?

WILLIEs?

Someone who works in London, lives in Edinburgh—is that right?

I never knew that.

I am glad that I got an opportunity to explain what I was saying there. [Laughter.]

The answer to your question is yes. We would have that person flagged on our system as a Scottish taxpayer, and the fact that they were working for an employer in London would not alter that fact. Their employer would be given an S code to operate.

That clarifies that matter—thank you.

I will go on to an area that has nothing to do with fraud or people trying to avoid or evade tax. As you probably know, there has been a big debate in Scotland about whether we should increase the 45p rate to 50p, and what the behavioural impact of doing so would be. I know that there are many other scenarios that you could paint not just for the top rate of income tax but for the standard and low rates. However, for the purposes of this discussion, can you tell us how much work HMRC has done on the behavioural impact of tax changes such as increasing the 45p rate to 50p in Scotland?

There was a 50p additional rate in the UK in 2010-11, but it was in place for only a year. In March 2012, the department published an analysis of the behavioural impact of the measure, and you can find that online or, if necessary, I can send you the link to it.

Obviously it was quite a difficult analytical exercise to carry out, because the rate was in place for only one year, and having a series of data might give you a better understanding of the on-going behavioural impact. However, the broad outcome was that there was a very significant behavioural impact on the tax take in 2010-11 that we had underestimated in our forecast of the yield from the 50p rate. In particular, we saw significant forestalling, with people who had the option to do so bringing income into the previous tax year when the rate was lower. That had a consequential impact on the yield in the following tax year.

The analysis drew what conclusions we could from having that rate in place for a year but, nevertheless, it indicated a significant behavioural impact, and you would have to look at whether and how you could constrain that impact if you decided to introduce higher rates in future.

Based on your analysis, then, is it fair to say that a 50p rate might result in our receiving less revenue because of likely behavioural impacts?

I am not sure that it is safe to draw conclusions for what would happen in Scotland; after all, this was a UK-wide rate and, as I have said, we had only very limited data, because the rate was in place for only a year. However, in that instance, we saw a large behavioural impact that we had underestimated at the outset and which meant that in the year in which the rate was in place less revenue was raised than had been forecast. The analysis says very little about what the on-going effect would have been, because obviously we were not able to test that.

It was also not a permanent rise. Had it not been announced that it was going to be abolished at some point?

No. From recollection, it was introduced as a final act of the Labour Government before the coalition Government came in in 2010 and repealed it. It was not announced as a one-year measure, but in practice it was in place for only a year.

But had it been place for longer, would the behavioural impacts have been greater, smaller or neutralised?

We do not have the data to test that but, to the extent that people took income early or delayed taking it, there might have been a different impact over time. People might be able to do that kind of thing for a period of time, but they might have less of an ability to do so after a while. The analysis does not get into that issue because, as I have said, there was data for only one year.

Have you done any more general work on behavioural impacts?

We do it all the time with regard to policy costings for our ministers, but I cannot share with the committee what policies they ask us to cost.

Would you do that kind of exercise for the Scottish Government, if asked?

It would be for the Scottish Government and the Scottish Fiscal Commission to carry that out.

But you would co-operate with them.

We would certainly provide them with the data sets to enable them to do policy costings and forecasts. In the UK, the Office for Budget Responsibility uses the survey of personal incomes as a key data set, and we have essentially given the Scottish subset of that data to Scotland so that the Scottish Government and the Scottish Fiscal Commission can make their own forecasts. We do not have an agreement to provide an analytical service, but we certainly provide data.

Thank you.

I want to pursue this a little bit further, because you are absolutely right to point out that it will be the Scottish Fiscal Commission that will provide us with some of the modelling of behavioural impacts for our budget discussions. We saw forestalling when the land and buildings transaction tax was introduced, as you might have noted, but I wonder just how much of this work you can do in advance when planning a tax system. If the Scottish Government indicated that it was considering a 50p top rate of tax and it or the Scottish Fiscal Commission asked you to do some modelling in advance, is it not conceivable that some anti-avoidance measures could be put in place before that measure was brought in?

09:45  

If, as a policy maker, you knew that a policy measure under consideration was likely to have a big behavioural impact, you would look at whether other measures could be put in place to manage people’s ability to make such an impact. You might be able to close off certain avenues, but there might be other avenues where you felt the options for closing off were more limited. That is certainly the kind of thing that you could look at.

So in effect it would be possible to minimise the bulk of the impact in advance.

Undoubtedly there might be some behavioural choices available to taxpayers that you would have options to close off, but there might be others that you would have fewer options to do anything about. People with wealth and high incomes can plan their affairs, including whether and when to take income or whether and when to sell an asset and realise a gain. You might be able to legislate to constrain some of those behaviours, but there might be others that you decide you cannot.

That is helpful. I recall as a member of the Finance Committee when we were discussing this very subject one of your senior colleagues telling us that anti-avoidance measures had improved over the piece and that he did not necessarily foresee any problems in that respect. Is that your feel for what is going on on the ground?

The large raft of anti-avoidance measures that has been put in place in recent years has significantly stamped out tax avoidance as an activity. It has not been stamped out completely—and probably never will be—but we have seen a very significant reduction in the use of tax avoidance schemes.

Moreover—and I should point out that this is not tax avoidance—the ability for people to put their money into something that is subject to relief has in some instances been capped in recent years. As I have said, a lot of work has been done to close off some avenues that we would regard as mischiefs, because they are about tax avoidance.

As far as tax planning is concerned, a key pressure on the tax system at the moment is what we call tax-motivated incorporation, where there is differential tax treatment of employed and self-employed people or companies. We are certainly seeing a behavioural impact where people incorporate, because one of the reasons for their incorporating is to take advantage of that differential. Whether you call that tax avoidance or taking sensible and legitimate steps, we are nevertheless seeing that kind of behaviour, and if those differentials were to increase, people would have greater incentive to do that kind of thing. There are still areas of tax planning where certain opportunities are available, and not all of them have been closed off by any means.

Finally, I am aware that you have had a series of mergers or consolidations—call them what you will—that have affected tax offices in Glasgow and Edinburgh. Has that had any impact on your work on the SRIT?

No. We have a programme of reducing the number of small offices that we have over the coming years and concentrating our resources into 13 regional centres plus, I think, six specialist offices and a head office. We think that that will enable us to deliver a better service more efficiently. Not all of that has been implemented—it is in train over the next few years—but there has been no impact on our implementation of the Scottish rate of income tax.

As there are no more questions from colleagues, I thank our witnesses very much for appearing before the committee. I suspend briefly to allow a changeover of witnesses.

09:49 Meeting suspended.  

09:52 On resuming—