The principal reason for our meeting is to take evidence on the “Administration of Scottish income tax 2021/22” report, which was produced on 12 January this year by the Auditor General for Scotland and is, in turn, a commentary on a report that was prepared by the National Audit Office.
I am pleased to welcome our four witnesses, who are here in person this year. Stephen Boyle, the Auditor General for Scotland, is accompanied by Mark Taylor, audit director at Audit Scotland. I am particularly pleased to welcome Gareth Davies, Comptroller and Auditor General at the National Audit Office, who is appearing in person before the committee for, I think, the first time. He is joined by Darren Stewart, audit director at the NAO.
We have a series of questions to put to the witnesses, but I begin by inviting the Auditor General for Scotland to make a short opening statement.
Many thanks, convener. Scottish income tax remains a key part of the package of new financial powers that was implemented as a result of the Scotland Act 2012 and the Scotland Act 2016. The purpose of today’s session is to look at the administration of Scottish income tax.
The reports before the committee relate to 2021-22, which was the fifth year in which the full amount of non-savings, non-dividend income tax that was collected by HM Revenue and Customs was payable to the Scottish Government. It is also the fourth year for which HMRC has published Scottish income tax outturns in its accounts. Those outturn figures relate to 2020-21. The difference between actual United Kingdom and Scottish tax outturns and the amounts that were forecast at the time is adjusted through the 2023-24 budget. That is known as the budget reconciliation process. The reconciliation for the 2020-21 accounts results in an increase of £50 million to the 2023-24 budget. That is noteworthy, as it is the first time that there has been a positive reconciliation in respect of the Scottish budget.
HMRC’s annual accounts also include an estimate of Scottish income tax for 2021-22, but that is for reporting purposes and does not affect the Scottish budget.
HMRC collects and administers Scottish income tax as part of the UK’s overall income tax system. The National Audit Office audits HMRC’s accounts, and the Comptroller and Auditor General is responsible for reporting to the Scottish Parliament on HMRC’s administration of Scottish income tax. I then report to this committee to provide additional assurance on the NAO’s audit work, in line with a recommendation from your predecessor committee in 2014. I also explain what the findings mean for the Scottish budget.
In summary, my report says that I am satisfied that the NAO’s audit approach was reasonable and covered the key audit risks. I am also satisfied that the findings and conclusions in the C and AG’s report are reasonably based. The C and AG has concluded that the outturn on Scottish income tax was fairly stated. That, therefore, provides the Scottish Parliament with valuable assurance over that aspect of the Scottish budget.
As ever, my colleagues and I look forward to answering the committee’s questions.
Thank you very much. I invite Gareth Davies, the Comptroller and Auditor General from the NAO, to make a short opening statement, after which we will move to questions.
Thank you for the invitation to give evidence today. I am delighted to be here in person.
As the Auditor General has explained, the content of my report follows the requirement, as set out in legislation, that I report to you on, in this case, the outturn for 2020-21 and HMRC’s estimate of revenue for 2021-22. We look at the rules and procedures that are in place to administer the system, and we test how those are operating through our audit work. We also cover the costs that are recharged by HMRC to the Scottish Government for that specific work. Our report sets out our conclusions in those three areas. As the Auditor General said, I have concluded that both the outturn for 2020-21 and the estimate for 2021-22 are reasonable.
It is worth pointing out the context for the work and the years that the report covers, because 2020-21 was the first full year of the pandemic. There was a small increase in the income tax revenue that was collected for that year, which might be a surprise to everybody who, at the outset of the pandemic, expected a significant decline in tax revenues. However, we have seen that pattern across the UK, and it is largely attributable to the various income support schemes that were put in place to ensure that people were still in employment through that very difficult year.
The committee will have noticed that the estimate for 2021-22 shows a sharp increase from the previous year. That reflects the reopening of the economy and the extra activity that everyone was able to see during that year, but, for this time next year, it also provides a very good test of the estimation methodology that is used by HMRC. Clearly, when it is estimated that there will be a significant increase of more than 11 per cent in one year, it will be very interesting to see how the outturn compares with the estimate when we have next year’s report.
I look forward to answering the committee’s questions on the report.
Thank you very much. I will kick off. You make the point that HMRC estimates that Scottish income tax revenue in 2021-22 will be £13.2 billion, which would be an increase of about 11.3 per cent on the previous year—a year when the economy was in lockdown—but it is also estimated that the increase in the UK will be 13.2 per cent, which is significantly more than in Scotland. Why is there an expected difference between UK performance and Scottish performance?
Stephen Boyle might wish to comment on that but, first of all, I point out that, as set out in the report, the Scottish outturn figure for the previous year—2020-21—also slightly lagged behind the UK-wide figure. For that year, there was a 1 per cent increase in Scotland and a 2 per cent increase in the UK as a whole. In a way, it is not a surprise to see that pattern replicated in the estimates for the following year, because the same methodology is used to roll forward.
I do not have any more detailed information in order to elaborate on that, but Stephen Boyle might want to comment.
We know that there is variation, increasing divergence and volatility in relation to the factors that are set out in the NAO’s report, with various parts of the UK recovering from the pandemic at different rates. We should factor in other external events, such as the war in Ukraine and the residual impact of the UK leaving the European Union, and there has been volatility more recently as a result of the cost of living crisis and the pandemic. Those issues, coupled with early signs of variation and divergence relating to Scottish income tax, are all factors.
As Gareth Davies said, it will perhaps be the most significant and interesting outturn when we eventually see just how reliable those estimates are. We are aware of those early signs and that combination of factors. That probably pushes us to the boundaries of what we are able to give precise detail on. HMRC, as the owner of the methodology, will be able to set that out in more detail, should the committee wish.
I draw the committee’s attention to the fact that there are a number of estimates for what the tax position is likely to be during the subsequent year, and the Scottish Fiscal Commission, whose estimates are used for the Scottish budget, provides commentary relating to its forecasts. It also provides very helpful and useful commentary on some of the drivers that result in the differences between the UK and the Scottish positions. If it would be helpful to the committee, we could point you in the direction of some of that detail.
I think that that would be helpful. One of the factors that we considered last year but which has possibly been brought more to the fore in this financial year is the rate of inflation. Inflation has led to higher wage settlements and, therefore, people’s incomes are rising. I presume that that means that the income tax take is going up. Is that a fair assessment of what is going on?
It is hard to disagree with that, convener. People’s incomes are rising faster than would have been anticipated. We are seeing that through some of the public sector pay conversations. The assumed rates have been higher in reality, and that will eventually feed through to the income tax that HMRC collects.
They might be higher in reality, but I am not sure that they are higher in real terms. My point is that people’s incomes going up—albeit that they lag behind the cost of living and the consumer prices index—presumably leads to an expansion of the revenue that comes in to HMRC. In turn, that will lead to an increase in the resources that are available to the Scottish Government.
That is a UK-wide effect, which we are seeing in the rest of our work. Obviously, one of the big effects of that is that it pulls people into higher tax brackets more quickly than might have been expected; that is called fiscal drag. That presents a challenge in the accuracy of estimation to HMRC, because it is trying to predict a faster moving target. In relation to the difference between the estimate and the outturn, it will be very interesting to see whether HMRC will be able to get as close as it did in 2020-21.
On the convener’s comment about the impact on the Scottish budget, in overall terms, at a UK level, raising more taxes means increasing budgets, but what matters at the Scottish level is the interplay between the increase in Scottish taxes per head and increases in taxes in the rest of the UK—or, to be strict, in England and Northern Ireland. It is the interplay of those two factors that drives the Scottish budget. If taxes per head increase at a faster rate in Scotland than they do at a UK level, there will be more budget. However, even though taxes might be increasing, if that rate is slower than it is at the UK level, the budget will be squeezed.
Yes. We will come to discuss that in more detail.
Gareth Davies, you mentioned methodology a couple of times. Again, that is something that exercises the committee’s interest, and other committee members will have more detailed questions in that area. Scottish income tax is now well established, but we have heard some concerns that the methodology does not give us the level of data that we think is necessary to make informed policy decisions.
I will cite an example. In the report, you identify some of the methodological flaws, if you like, or recognised areas in which there is not the level of data that we might like. The second point about the methodology in your report states that
“The methodology combines the calculation of PAYE and Self Assessment liabilities for the UK such that the amount apportioned to Scotland does not reflect the differing proportions of each type of taxpayer between Scotland and the rest of the UK.”
Do we know, for example, what the balance is, in England and Northern Ireland versus Scotland, between people who are on the pay-as-you-earn system and those who are on the self-assessment system? Are there more people on PAYE in Scotland than there are in England and Northern Ireland, or vice versa?
I do not have the data with me on that proportional split, but it is possible to derive that data. HMRC calculates the estimate by interrogating and taking extracts from its self-assessment and pay-as-you-earn systems, driven by a Scottish taxpayer indicator. It should absolutely be possible to derive the proportional split that you talk about.
Therefore, that could be built into the data that is available, but currently it is not. Even on some kind of sampling basis, it would presumably be possible to gauge the relative balance between PAYE and self-assessment liabilities in both jurisdictions.
09:15Another point strikes us in relation to the methodology, which the Auditor General for Scotland reflected on in last year’s report. Paragraph 16 of this year’s report, which notes the NAO’s report, states that the
“methodology has remained broadly the same since the prior year.”
This time last year, we were having conversations with Scottish Government officials, as well as with HMRC, about the methodology, the service level agreement and the quality of data that was being procured, because there is obviously a financial cost to it.
Last year, Gareth Davies commented:
“Clearly, if you had a more highly specified agreement that got into some of the areas that we have been talking about today”—
why we cannot get more qualitative data and fewer estimates—
“it is likely that HMRC would require more costs to be covered. However, on the basis of the agreement that is currently set out, we think that the cost is fairly stated.”—[Official Report, Public Audit Committee, 3 February 2022; c 25.]
In other words, we were getting into the terrain of suggesting that, if the Scottish Government was prepared to pay a little bit more money to HMRC, it could perhaps get some better-quality data.
In her evidence last year, Alyson Stafford from the Scottish Government said that
“the committee’s consideration is extremely timely for us to take your observations and comments into account”
and that the Government was
“looking at data breadth and depth.”—[Official Report, Public Audit Committee, 12 May 2022; c 41.]
However, nothing appears to have changed. Is that a fair estimate of what is going on here? Nothing.
I do not think that anything in the system has changed fundamentally since last year’s report.
May I return to your previous question?
Sure.
It might be helpful if you could point to the reference in our report that you mentioned in relation to PAYE and self-assessment taxpayers. The outturn figure for 2020-21—in other words, the actual figure being reported—is based on Scotland’s PAYE and self-assessment taxpayers. I just want to check the reference that you quoted to make sure that I am not missing a chance to help the committee.
Sorry—I can look it up in your report. I think that there were four areas in which reference was made to methodology and you looked at what could be reasonably assessed. The report states that
“the use of sample data introduces sampling uncertainty”,
so the figures are based on samples. Paragraph 1.22 in your report has four bullet points. The second point is:
“the methodology combines the calculation of PAYE and Self Assessment liabilities.”
That is the estimate for 2021-22.
Yes.
The outturn figure—which is, if you like, the real figure that relates to tax actually remitted to Scotland—is always based on actual taxpayers in Scotland. The 2020-21 outturn figure that we report on is therefore not an apportionment between PAYE and self-assessment. It is not based on an aggregate of the UK; it specifically relates to the Scotland figures. That is an important point. You can therefore rely on the accuracy of that number for Scotland.
Paragraph 1.22 deals with the estimate for the year. By definition, the estimate is, of course, made before people have completed their tax returns and before HMRC knows exactly what is due from each taxpayer. To get that estimate, it uses the apportion method that we describe in that paragraph.
I hope that that is a helpful distinction. You can rely on the outturn, but the estimate is inherently more uncertain.
That clarification is helpful—thank you for that.
Obviously, the estimates are in the domain of the decision-making process, so they are quite important to us. That is why we have taken the view that having better data would give us a clearer sense of where policy should go and what will have the most impact in relation to raising revenue or redistributing the burden of taxation.
Gareth Davies’s distinction between the outturn and the estimate is important. The estimate is used to support decision makers—policy makers in the Parliament and the Government—in making future spending commitments.
By drawing on the NAO’s conclusions, we can note the volatility that you have described and the increasing divergence between tax rates in Scotland and those in other parts of the UK. In our view, it is for the Scottish Government to be satisfied with the methodology and associated data that it receives to inform its budget-setting process. Broadly speaking, we have reached the point at which, given the complexity and the divergence, it is for the Government to reaffirm its satisfaction with the current arrangements and to test whether the quality of the data is still fit for purpose in supporting its forecasting requirements.
The report mentions the omicron variant of Covid and the war in Ukraine, and it is suggested that those would have an impact on taxpayers’ ability to meet their tax liabilities. Will you elaborate on that, please?
Two points apply to HMRC’s work across the whole of the UK, including Scotland.
The pandemic impacts included the omicron wave, which, you will remember was the last very big disruption to life in the UK. Two big factors affected HMRC. First, it reduced its compliance activity during the pandemic. As you may remember, HMRC was tasked with standing up the big programmes for paying furlough and self-employment income support. Literally thousands of HMRC staff were therefore redeployed on to those tasks—quite a lot of whom came from the compliance teams. The level of compliance work was therefore lower. We have reported separately on that for the UK as a whole. I would be happy to share a copy of that report with the committee, if you would find it helpful. Not surprisingly, that resulted in a reduction in the level of compliance casework and a reduction in compliance recovery. In the estimate, Scotland is bearing a share of that reduction in compliance recovery.
Secondly, for the same sorts of reasons, tax debt in the UK has gone up sharply. Taxpayers owe more to HMRC because of their inability to pay on time. Some of that was caused by HMRC’s deferment of payment dates during the pandemic, but not every taxpayer has been able to catch up on their payments to HMRC since the pandemic phases came to an end. Now, therefore, there is a fairly significant challenge for HMRC in collecting the outstanding tax that is due. Again, that has had an impact on the estimate for next year, in the way that we set out in the report. Those are two important UK-wide factors to be aware of.
Thank you. Later this morning, we will go on to talking a little more about debt collection. For now, the deputy convener, Sharon Dowey, has some questions.
Good morning. HMRC uses the same systems to administer income tax whether a taxpayer is from Scotland or from elsewhere in the UK. However, it operates additional rules and procedures for Scotland, such as residency checks and the application of “S” tax codes if individuals are identified as Scottish taxpayers.
The NAO report states that, as has been the case in previous years, one of the main administrative challenges that is faced by HMRC is the maintenance of an accurate and complete record of Scottish taxpayer residency addresses. That is because HMRC relies on taxpayers to notify it of any change of address. What can be done to improve on that? Would a legal requirement to notify HMRC of address changes assist with that administrative challenge and with overall compliance?
Making sure that the tax is allocated to the correct country in the system is a key area of control. Intuitively, the more divergence there is, the greater the pressure on that system of compliance and the more incentives there are for people to manipulate their address, for example, as a way of avoiding what would otherwise be a higher tax payment.
Obviously, HMRC is fully aware of that risk. We know that it discusses that topic with the Scottish Government during their regular conversations.
The question of what would represent useful increases in the level of control over those risks is really one for those at HMRC, rather than us, as they are the operational experts, and they will give you a much more informed answer than we can, as auditors. Having legal requirements to notify HMRC sounds plausible, but the experts would need to advise you on whether such requirements would have the desired effect or whether they would create any kind of perverse, unintended effects.
The most recent scan of taxpayer records to
“identify missing or invalid postcodes which would result in incorrect residency status being applied”
was in June 2021. It identified 25,488 missing or invalid postcodes, of which 3,031 were updated.
Reported postcodes, whether missing or invalid, represent 1 per cent of the 2.5 million total income tax population in 2020-21. A small number of missing taxpayers can potentially equate to a large amount of revenue, especially if that group contains high-net-worth individuals.
Has the NAO received an update from HMRC around the monitoring of that situation, and specifically on the causes of missing or invalid postcodes? What might the impact be on revenues, and what, if anything, is being done to address the matter?
I ask Darren Stewart to pick up on that question, please.
Starting with the point about the causes of missing or invalid postcodes, that is clearly a key area of focus that we return to every year, as part of our work and in reporting to you. An invalid postcode can be missed because of things as simple as keying errors in tax returns and things of that nature. The scale of missing and invalid postcodes has fluctuated over time, and it is a key feature of the assurance work that HMRC does. An important part of that work is to interrogate those missing and invalid postcodes to identify where there are residents within those postcode areas who are liable for tax. The analyses that HMRC has done often identify that there are not taxpaying individuals within the households concerned. That could be for various reasons: there could be dormant accounts, or the people might be economically inactive—things of that nature.
You asked about any follow-up that we have had since that exercise in June. We have not had that, but that is something that we will pick up as part of the forthcoming year’s work. However, that remains a key focus of our work—it is really important.
Good morning, Mr Davies and Mr Boyle, and welcome. It is good to see you in the flesh, rather than on Zoom.
I want to delve a bit deeper into the monitoring and evaluation that you can do around the compliance of Scottish income tax payers, which is becoming a live issue, given the increasing divergence between the Scottish tax system and the UK tax system. Indeed, that is referred to in a report by the Institute for Fiscal Studies, which is covered in today’s Daily Telegraph.
I return to the issue that the deputy convener raised regarding postcodes and where people are resident. The NAO report says:
“HMRC has not identified any significant or widespread instances of taxpayers changing their address to obtain a tax advantage.”
However, today’s newspaper report points to David Alexander, the chief executive of Scotland’s largest letting agency DJ Alexander, saying that he is already seeing signs of high earners leaving. He said:
“It’s a natural situation that people think, ‘actually why would I reside here when I can move not that far and pay substantially less tax’.”
Given that HMRC says that it is not seeing any significant or widespread instances of that, can you elaborate on that? What would be significant or widespread, bearing in mind that Scotland has a very small number of upper-rate and high-rate taxpayers?
09:30
You would definitely have to ask HMRC what it meant by those definitions. Those are its terms, and that is not an auditable bit of the process on which we could help shed light. I am sure that HMRC would be able to help with the detail behind those comments.
We know that it uses its database to run the kind of computer checks that you would expect in this situation, so it could run a report to list the taxpayers who have changed their postcode from a Scottish one to an English one in the past 12 months. That is not a dauntingly enormous list, but of course, there would be a lot of people on it in any one year, although I do not know what the exact number is. It is a meaningful audit process to then look for any signs of patterns of behaviour.
I would be very surprised if that is not what the HMRC refers to. I suspect that it uses the exception reports that it runs on a regular basis to assess the volume of cross-border transfers. Of course, there will be some in the other direction for work reasons, personal reasons and all sorts of other reasons. I assume that its comments are based on its analysis of the data, but it would have to give you the detail behind that.
The divergence is around 2.1 per cent in relation to the rest of the UK, which is getting into what I perceive to be a significant sum. From your perspective, would it be reasonable to expect that there is an incentive for individuals faced with the prospect of higher tax to consider legitimate tax avoidance measures, such as converting income to dividends or moving south of the border? Is that a reasonable expectation that HMRC should be aware of?
You are inviting me to get into policy areas. I can comment on the audit of the numbers, but those are judgments for elected politicians rather than for auditors.
The Auditor General wants to come in on that point.
Mr Hoy, you have picked up one of the central themes of our report. Some of the HMRC’s assessment and the recent commentary from the Scottish Government, such as the Deputy First Minister’s evidence to the Finance and Public Administration Committee, explored some of the emerging risks of changing taxpayer behaviour as a consequence of the divergence in tax regimes between Scotland and other parts of the UK. It is probably early days, but that is the picture that we are in. HMRC says that there is limited evidence on the potential for behavioural change.
You mentioned high-net-worth individuals. In relation to the analysis in the NAO’s report and the comparisons of tax implications for different income tax earning rates, as people earn more than the top rate and go much higher beyond the £150,000 threshold, the differential is less. The more significant impact is on who we might call middle earners; for whom the differential is £50,000.
The recommendation and judgment that we make in our report is for the Scottish Government to take a view on what kind of additional compliance activity it wishes to ask of HMRC, based on HMRC’s position that there is not enough evidence generally to suggest that that sort of behavioural change has an impact. Whether that is a sustainable position is something that the HMRC and the Government need to keep under close review and discussion.
There is a cost benefit analysis to all that. Additional compliance activity for Scottish income tax purposes will come at a cost, and that will need to be weighed up against the benefit that it will produce in further tax take.
Is it difficult to do that modelling in a predictive sense? Do you only realise that the flight has happened once you look back over the output for a year?
As Gareth Davies mentioned, it is difficult for us as auditors to take a position on that. The specifics and reliability of the modelling is something for HMRC and its advisers to keep under close review. The Scottish Government will want to be satisfied that the compliance activity that it asks of HMRC is consistent and proportionate for its requirements.
You came forward with various recommendations in your report. Do you think that more third-party data checks and compliance activity should be undertaken in Scotland?
I am sure that Gareth Davies and Darren Stewart will want to say more about third-party data checks. Looking at the results of the third-party data checks, we have reached the position that there is an opportunity for the Scottish Government to request more and perhaps more frequent third-party data checks in order to satisfy itself about the accuracy and consistency of the tax base in Scotland, which leads to the totality of the outturn and the reliability of the estimates. NAO colleagues could probably say more about that.
That is absolutely fair. More than once, we have made recommendations on the frequency of that third-party data exercise. In the hierarchy of HMRC’s work and the assurance that it provides, third-party independent checks are clearly preferable. When last prompted, HMRC and the Scottish Government settled on a two-year frequency, which they were comfortable with because they thought that it struck the right balance between timeliness and cost. We would always recommend that they keep that under review.
Going back to an earlier point, where we attribute to HMRC the view that there have not been instances of significant or widespread manipulation, that refers very much to the manipulation aspect. It is important to make a distinction between the actions of people who to choose to legally move across the border or change their tax arrangements versus those that would clearly be of interest to HMRC because they stray into the areas of manipulation, non-compliance and evasion.
So, you are saying that if people are doing it, perfectly legitimately, over a number of years, HMRC would not necessarily be particularly activated about it.
It should be interested in it; I am not sure that it would have the powers to do anything about it. That is the distinction that I would make.
Thank you. There is a whole other area that we want to explore, which includes the tax gap and other data issues. Colin Beattie will lead on that.
Over a number of years, I have been fairly critical of the quality of information on the Scottish rate of income tax and all the estimates and so on that are around it.
Last year, I drew up a list of 32 areas in which there were estimates—or, more accurately, guesstimates—of what the figures might be. I have not yet had time to do the same with the report that is before us, but I will do so. Looking at it, though, I can see clear evidence that the separate Scottish figures are inadequate to enable an accurate calculation of the tax. At times, there is conflation with UK figures because separate Scottish figures are not available, yet the mix of taxpayers in Scotland is quite different from that in the rest of the UK—particularly when we take London into account, which is hugely distorting. Are the figures that we have on Scottish tax just fantasy?
I return to my earlier answer to the convener. When we talk about the outturn figures through the system—in this case, those for 2020-21—we can see that they are definitely not fantasy; they are auditable. Our opinion on them is based on the fact that PAYE and self-assessment form the vast bulk of the amount, and that 99 per cent of the figure is not an estimate but based on tax that has actually been received. We can therefore be very confident in the accuracy of the outturn figure. That is quite a strong statement for an auditor to make—we do not normally go that far—but in this case there is a strong evidence base behind the figure.
The concerns that you have just described do apply to the estimate, as we discussed earlier. The estimate is not a figure that affects the amount that is collected for Scotland and delivered to the Scottish Government. It is a relevant one in the planning system, though, for the reasons that the Auditor General explained earlier, so it is important to work on its accuracy. However, that is not the same as questioning whether we can trust the amount of income tax that Scotland is receiving. It is the outturn, so in that case we can.
I hope that that was a helpful starting point to the discussion, but perhaps you will want to go into specific areas of the estimate in more detail.
There are one or two areas that I would like to examine more closely. Again, though, there are an awful lot of estimates in here that must distort the final figure.
Paragraph 15 of the NAO report states:
“There are no risks identified in the Scottish SPR which are specific to Scotland as HMRC assesses that compliance risk in Scotland is consistent with the rest of the UK.”
Given the different mix of taxpayers and so on across the rest of the UK, is that statement accurate?
It is an accurate statement of the view of HMRC, and our job here is to give you a clear picture of its view. You would have to ask HMRC for—
However, as the auditor, you would have a view on that.
In the same paragraph, we point out that, if further divergence occurs between the Scottish and UK tax rates, the risk assessment will merit a closer look. It would be surprising if that did not start to feature in a more prominent way in the strategic picture of risk, as it is called. That is why we make the point, as we have just discussed, of returning to the question whether the risk is sufficiently well understood, given the changing policy picture. When the risk is understood, is the level of investment in the controls for the mitigation of those risks proportionate to the level of risk that is being run?
That is a constantly moving picture, so I do not think that the committee should ever be satisfied with that; you should continue to probe into whether the Scottish Government and HMRC have arrived at the right balance, in the way that the Auditor General described earlier, between cost and rigour of control. However, at the moment, that is an accurate picture of how HMRC is running its system of risk assessment, and we are pointing out that the risks are changing, so that will need to be kept under review.
In connection with the compliance yield of £280 million, paragraph 16 states that
“HMRC calculates these figures as a proportion of the equivalent UK figure, rather than ... Scotland-specific data to quantify the risks.”
That does not seem as though it could be accurate.
The figure of £280 million is, in essence, the return on compliance activity, which is otherwise known as chasing up suspect cases of, for example, underpayment.
As we said in the report, HMRC uses the same process in Scotland as it does in the rest of the UK. It does not adjust its compliance approach for Scotland, so it thinks that it is a fair estimate to apply the proportion of net losses through the compliance approach—and, in that case, the yield from its compliance work—on that basis.
Again, the committee will have to keep probing because, as circumstances and risks change, that assumption might become unsafe at some point. However, HMRC’s current assessment is that that remains the best approach.
However, paragraph 17 says that
“HMRC ... cannot easily track and monitor compliance activity in Scotland and this affects its ability to collect performance data about the extent of Scottish non-compliance”.
Without data, it is not really possible to project accurately into the future.
Intuitively, without knowing the ins and outs of all the systems in great detail, it does seem surprising that it is impossible to isolate Scotland compliance activity from the UK as a whole. However, the fact is that HMRC’s compliance systems are built on a UK-wide basis and were never designed to be applied differently in different parts of the UK. That might become inappropriate at some point, so it is a really important point to stay on as the facts and risks change. You are right that it is intuitively surprising that HMRC’s compliance systems are not able to identify the country of the taxpayer.
I move to paragraph 1.5 of part 1, which mentions
“The reduction in Scottish income tax outturn arising from these adjustments”,
which are estimates for tax due, uncollected amounts and tax reliefs. It goes on to say:
“In some areas of the calculation, HMRC does not have data available in sufficient detail to identify income tax liabilities, reliefs or other adjustments relating to individual taxpayers.”
Again, HMRC is using UK figures to work out some sort of calculation, but the pattern of Scottish taxpayers—versus those in the rest of the UK—is very different, as far as the information that we have seen here is concerned. How accurate can that estimate be? What kind of distortion comes in there?
The figure that you refer to is the one bit of the outturn that is estimated. As I said, 99 per cent of the outturn is based on actual figures, and that is the remaining 1 per cent that is based on estimates.
Figure 3 in our report breaks down the £119 million adjustment into categories of estimates. That includes estimated self-assessment liabilities and estimated PAYE liabilities, and there is a category of “Estimated further liabilities”. That includes tax relief on pension contributions, gift aid and so on. There are then some deductions from revenue based on pension contributions. It is possible to break down that estimate into its constituent categories.
09:45We point out that some of the estimates are of liabilities that have not yet materialised in their final form. For example, some self-assessment taxpayers have been very late in concluding their tax return, so the figure for that is not known and cannot be accurately entered into the calculation. However, HMRC knows that, UK-wide, there is a level of late filing of tax returns that it can predict reasonably accurately. It cannot separate that assumption between Scotland and the UK; it is applying a UK-wide proportion to the figure to derive the Scotland amount. That is the process that HMRC is in at the moment. The question is whether HMRC should be able to accurately forecast such things as late filing of tax returns, and how much tax will arise from those late returns. You can see how that is not straightforward, as there is no actual number that can be attributed to Scottish taxpayers. However, the whole system needs to be kept under review.
Darren Stewart might wish to add to that.
That is a fair summary. We have spoken about this matter with the committee before, and I agree that it is an area to continue to probe. There is a high level of assurance around the net outturn for those residual areas of estimation and adjustment. The judgment that HMRC and the Scottish Government reach in deciding whether to go further in this concerns the precision of the outturn figure versus timeliness. In theory, a compliance case on a self-assessment tax return could take up to six years to settle, for example. We could wait until the end of that six-year period and have absolutely clarity on the amount that is due, but that would have consequences for budgeting, planning and so on. The judgment that HMRC and the Scottish Government have reached is that there is a balance between timeliness and the precision of the outturn that is produced.
Continuing on deductions from revenue, in paragraph 1.15 of your report, you note that HMRC has used
“historical data for the UK as a whole to determine patterns of uncollected liabilities and then apportioned an amount relating to Scottish taxpayers. HMRC calculated this to be £97 million”.
In paragraph 1.16, the report refers to gift aid—which is not insignificant. It says that HMRC calculated the deductions
“by estimating the Scottish share of ... tax relief claimed across the UK using historical data. HMRC calculated the Gift Aid deduction to be £114 million for 2020-21 and pension contributions to be £155 million.”
However, it is unclear to me what the pattern across the UK is. It is not really the same. It becomes one amorphous figure when we put it all together, but, when we try to separate things out, take one part and compare it with the other, we find that it cannot be accurate. There must be a margin for error here.
That is absolutely true. As with any estimate, one thing that you can be pretty sure of is that it will not be absolutely the correct figure, almost by definition.
All those questions come down to a balance of cost and timeliness. How important is the accuracy if it is going to be late or expensive to get the figure? That is what it all comes down to, really. You might get a bit of extra accuracy, but at what price, and with what delay? Clearly, there cannot be a long delay, as you need the figure to complete the Scottish Government accounting process.
Those are completely valid questions about how the balance has been struck in each case. Some of them seem more surprising than others to the layperson. For some of them, it is hard to understand why Scottish data is not available straight off the system. In other cases, it is easier to see why it is difficult to arrive at that.
As the systems improve, as they are doing over time—systems are becoming more sophisticated and the ability to interrogate the data is stronger—it is important to keep challenging the Scottish Government and HMRC on why they have not made a different call on some of the estimate areas. I expect that, over time, some of the figures will become quite cheap to produce and should be routinely fed into the calculation rather than estimated, whereas others might continue to be problematic and expensive. It is a live conversation to maintain with policy makers.
If I go through the report and compare what I regard as anomalies with what I took out last year, will I see any improvement?
Which anomalies were you thinking of?
I said at the beginning that I took out 32 cases in which parts of the revenue and liabilities are estimated. The extent of the issues is such that, if you take one, it is not that big but, if you take 32, you are looking at a bigger distortion. I hope that, over a period, that will gradually reduce with experience and an understanding that there are areas to be sorted.
We have not seen any big changes in the approach in the year that we are reporting compared with the previous year, as we said, but you are right. Over time, you should see a reduction in the length of the list of estimated areas as it becomes more possible to use the system to home in on the actual figure. In some of them, you will still need to include an estimate because you are estimating something that has not happened yet. However, even in those cases, it should be possible to use Scotland-only data rather than an apportionment of the UK figure.
In those two areas, you should expect to see improvements over time as improvements to the system allow.
As the auditor, have you had those conversations with HMRC and the Scottish Government?
We have not had them with the Scottish Government because our remit does not take us there, but, as the auditor of HMRC, that is the stuff of all our conversations with the agency. We audit it for the accuracy of its accounts and exercises of the kind that we are discussing here, and we carry out value-for-money audits on HMRC as we do on every other UK Government department. The Public Accounts Committee in Westminster regularly holds hearings based on those reports.
We challenge HMRC all the time on whether it could be doing more to ensure that the revenue that is due is collected, to minimise the compliance losses that we have been discussing and to reduce the tax debt figure that has grown significantly since the pandemic. At the heart of all that, of course, is the system of information technology and data collection. One of the biggest IT operations in the country is at HMRC. It has hugely sophisticated systems nowadays, and there is a programme to continue to improve those. We are close to the planned improvements in those systems and challenge HMRC regularly as its auditor on what it is doing to address the risks that arise using the better data that it has available.
Darren, you are HMRC’s auditor, so do you have anything to add to that?
I return to the point about evidence that suggests that the quality of the estimates and data is improving over time. It is valid to continue to reflect on the residual areas of estimation in the outturn.
I draw the committee’s attention to the overall variance in HMRC’s estimate of what the Scottish income tax outturn would be for 2020-21 and what the outturn was. In absolute terms, there was a difference of £87 million. In pounds and pence, that is a significant sum. However, in percentage terms, the variance is 0.7 per cent, which is the lowest that it has ever been in my involvement in this work. It has shown steady improvement over time.
Although the methodology has not changed substantially, there is some evidence that the quality of the overall estimate is improving over time. It is important to reflect that.
Perhaps the Auditor General for Scotland could comment on engaging with the Scottish Government on that.
Ultimately, looking at the reliability of the estimate informs the Government in relation to the budgeted proposals for its public spending. As we have seen over a number of years, it has resulted in fairly large negative budget adjustments and reconciliations once they are tallied up with the subsequent outturn.
In our report, we note a positive budget adjustment of £50 million, but that will not necessarily be a reliable indicator of future positive reconciliations, given the volatility that we have covered already.
The primary vehicle for conversations with the Government is our additional assurance report that the committee has before it. In that, we set out our view that the Scottish Government has to be satisfied with the robustness of the methodology, the associated compliance activities and the cost benefit arrangements—we have already spoken this morning about what the Government is asking of HMRC—and that the data is robust enough to support overall budgeting for the Scottish budget. It is something of a moving feast. As we look at the totality of today’s reports, we see more volatility, more divergence and change, and therefore the Scottish Government must take a view on whether the existing arrangements are sufficient for its purpose.
In the reports from the National Audit Office and us, you have probably got the sense that the increasing complexity and volatility might lead to the Scottish Government asking HMRC for more compliance and more requirements.
I will stop there, because a lot of my questions are for HMRC to respond to. I will simply say that there seem to be huge inaccuracies in the calculations for the Scottish rate of income tax and a great deal of work needs to be done to get an accurate Scottish figure.
I have something to ask before we leave this theme altogether. Paragraph 2.32 of the NAO report says:
“The tax gap is the difference between the amount of tax that should be paid and what is actually paid.”
The report goes on to say that HMRC
“does not currently produce a Scotland-specific tax gap”
and
“has limited information on total compliance activity undertaken in Scotland.”
Do the Auditor General for Scotland and the Comptroller and Auditor General at the NAO have a view on whether it would be useful to have specific Scottish tax gap data?
I will say something about the UK approach to calculating the tax gap. By tax gap, HMRC means what would be collected if everybody paid the tax that was intended by the Government or Governments across the UK compared with what is collected. It is measured in the tens of billions. Typically, it is £30-something billion each year. That is how we measure the UK-wide tax gap and we close it by improving compliance with tax rules, minimising avoidance and so on.
The way that is calculated is essentially through a statistical exercise that is carried out in the UK. It is quite a controversial methodology. There is a lot of academic challenge to it every time that it is used, and plenty of people have ideas about how it could be improved. There is a live debate about it between HMRC and the community of people who are interested in the accuracy of that number.
HMRC’s view is that it is hard enough to come up with an acceptable figure for the UK. It struggles to see how it can come up with a Scottish-only version of that statistical estimate, because so many of the figures used are UK-wide assumptions. Of course, in theory, there is nothing to prevent somebody from trying to do that, but HMRC has said that, because of the cost and effort involved in doing it, and the likely methodological challenges that it would face, there is no good case for doing it at the moment. Like all of these matters, it needs a discussion between HMRC and the Scottish Government.
I do not have a terrific amount to add to that. Gareth has set out the complexity of the issue and the reasons why that has not been done. The Scottish Government would have to be clear that the benefits would outweigh the costs in assessing whether the exercise would derive further benefit in establish a more robust estimate for its purposes, and whether it would be satisfied that the extent of tax available to Scotland is, in theoretical terms, clearly stated.
I am sitting on the fence about it somewhat, convener. It is for the Government to weigh up the pros and cons as to whether that would support its wider understanding of the methodology and feed through to its estimate of Scottish tax.
10:00
It goes back to our earlier conversation about the service level agreement and whether, if additional payment was made to HMRC, that could elicit, notwithstanding some of the methodological challenges, the kind of data that, it seems to us, would be fundamentally useful to have.
Before I bring in Willie Coffey, I raise the issue of debt collection, which, so far, has been mentioned only in the passing. Somebody mentioned the regular reporting to the Public Accounts Committee from HMRC, the NAO—of course—and other Government bodies. The UK Government’s response on the 48th report of session 2021-22 was a reflection on where things had got to with tax collection. It highlights:
“In addition, from September 2022, there will be a new contract through which HMRC places debt with private debt collection agencies (DCAs). This will allow HMRC to increase placements with DCAs by around £1 billion a year without increasing the cost to the Exchequer.”
Are those debt collection agencies doing that for free—given that it is said that there is no cost to the exchequer—or are they taking a percentage of everything that is collected, which is normally what happens in such situations? Are those agencies operational in Scotland? Are you aware of whether the Scottish Government has a view about the deployment of private debt collection agencies, and who are they, principally?
I do not have those details to hand, but Darren Stewart might be able to help.
We are looking at those arrangements as part of the current—2022-23—audit round. I can say with relative certainty that they will not be free, but I am not party to the specifics of the commercial arrangements. Your question on the distribution of that activity between Scotland and the rest of the UK is probably more for HMRC, but we will certainly be interested in that as part of our audit.
Thank you. Obviously, the committee will consider what its next steps are on that, but there would be interest in understanding whether private debt collection agencies are now being deployed.
Convener, you asked whether the Scottish Government had a view on the possible operation of private debt collection agencies in Scotland. I do not think that we know the answer to that. However, it is perhaps also worth checking with the Scottish Government whether it is able to have a view on that, given that it might just be a case of HMRC collecting income tax through its own arrangements across the UK. That is probably for the Scottish Government to confirm.
Yes, it poses some wider questions—including, again, some policy questions.
Good morning to the panel. I want to drag you back to your opening remarks and the two big figures that are in the very first pages of the report: HMRC’s final outturn calculation at £11.9 billion, and the estimate at £13.2 billion. The difference is substantial—it is £1.3 billion. On behalf of the man and woman in the street, my simple question to you is: ultimately, is that real money, which the Scottish Government will see further along the line, or is it not?
It will be, if HMRC’s estimation accuracy is maintained at the level of the past couple of years. As we have discussed, the estimates have been quite accurate in recent years—increasingly so. The difference between the outturn and the estimate was only 0.7 per cent in the previous year.
This is a more challenging test of that estimation accuracy, for the reasons that I set out at the start. Clearly, a big swing of 11.3 per cent in one year is substantial. There are a lot of moving parts in there, and the risk of the estimate not being as close to the outturn must be higher. However, we do not have any information to point in either direction, so I cannot give the committee a steer on whether it is optimistic or pessimistic.
We know that it is a sterner test of the estimation methodology than was the case in the previous year. Clearly, there is a lot more riding on this than simply an academic interest in an estimate, but it will be interesting to see, and a much better test of the methodology that is used, given the scale of the increase in one year.
It is a substantial difference.
The reason that we have not flagged it up as incredible—as not worth believing—is because there is a similar picture in the UK as a whole. As we have pointed out, there is, in fact, an even larger overall increase in the UK as a whole. That represents the shift in economic activity between 2021 and 2022 coming out of the harshest of the lockdowns.
Interestingly, on the inflation discussion, it was post lockdown but pre heavy inflation for most of 2021-22, which is why we think that it will be quite a hard one to call. Inflation was just starting to take off towards the end of that financial year; it is only in 2022-23 that we have seen the highest levels. Given how fast things were moving at the end of that year, we think that it is a very challenging estimate to get right.
It sounds as though there could be a potential windfall further down the line. When can we expect to see the real figure emerge?
This time next year. Next year’s report will include the outturn for 2021-22, which we will be able to compare with the estimate of £13.3 billion that we show in this report.
Clearly, HMRC will get indications of that much sooner than this time next year, because we have passed the deadline date for self-assessment returns to be submitted to HMRC, which was the end of last month. HMRC is therefore finalising the tax liabilities for all those taxpayers, so we will have an emerging indication as we go through this year.
Is the £50 million figure—the actual positive differential that the Auditor General mentioned—an example of that?
That relates to the £11.95 billion relative to what had been assumed. As we discussed, it is the first time that there has been a positive reconciliation to the Scottish budget, which is significant in itself.
I will return to the estimate point. Mark Taylor may also want to say a bit more about this in a moment. It is hard for anybody to be categoric that the estimate that is set out here will be what is actually received once HMRC collects all the relevant tax. Nonetheless, if we are looking for evidence and corroborating positions, the Scottish Fiscal Commission’s estimate is very close to the estimate that is set out by HMRC. I think that that provides more confidence to the Scottish Government, when it is setting its budget, that it will not result in hundreds of millions of pounds of reconciliations either way. However, there is always a degree of uncertainty with any estimate, and especially this year, given all the external variables that we have discussed.
To develop that point, paragraph 23 of the Auditor General’s report quotes the most recent figure from the Scottish Fiscal Commission. I will make two points about that.
First, the Scottish Fiscal Commission regularly updates its forecasts from the initial one that it prepares, which hits the Scottish budget, as to how that is playing out in its assessment. Second, its approach is very different from that of HMRC. Its approach is founded in the administrative data that it receives on an on-going basis, and it is much more granular in its overall assessment, as underpinned by its overall modelling. The fact that those two numbers are very close should give the committee a degree of assurance. Whether that continues to be the case is the big question that sits alongside that.
The other point to pull out is that both the Scottish and the UK tax take matter; it is about the interplay between those two. Again, the Scottish Fiscal Commission will look at the initial assessment of that difference and what it means for the Scottish budget. That is what plays through individual Scottish budgets. In the final reconciliation, it will also continue to forecast how that is likely to land.
We have seen the first positive reconciliation, and a big negative reconciliation is currently forecast for the subsequent year. After that, you will see a degree of volatility in that. The system is designed in such a way that it will sometimes be up and sometimes be down, because it is based on the interplay of two estimates. However, from a financial planning perspective, the Scottish Government needs to be very aware of that big negative reconciliation that is forecast for the next budget year, which is 2024-25, and to plan for that. It is forecast to be in the order of £750 million. Part of the reason for that is that Scottish ministers had discretion, as a result of the timing difference of the UK budget, to take an earlier or later figure. It took an earlier figure, which gave more money up front but, as the reconciliation plays through, that is likely to add a significant challenge to next year’s budget-setting process.
Thank you for helping to clarify that.
On the S codes and their application by employers in Scotland, the number of employers that do not apply the S tax code to their employees appears to be going up. The paper in front of us says that there were about 39,000 cases of that, and that it is now up to 41,000. Do you have a view on why that continues to be the case? Is it the same sectors that are repeatedly not applying the code, and do you, or does anyone else, have any information that could help the committee to understand what is going on and what work is being done to fix that problem?
That position has improved over time as employers have become more familiar with how the system works and the requirements on them to apply the S code. That number is still high, and it needs to be continually borne down on—that is absolutely right.
My understanding from the work that we have done is that HMRC will target routine offenders. If employers are consistently getting it wrong, it will reach out to them and intervene directly. However, my understanding is that the instances of that have reduced over time. HMRC also issues things such as employer newsletters, which are called nudge campaigns, to remind employers of their responsibilities in running their payrolls.
The overarching thing that I would say about that is that there is a high level of assurance around which taxpayers should have the S code applied and therefore, if employers are not getting it right, when HMRC comes to reconcile its records with what employers have applied, there is a high level of assurance on the accuracy in that regard and how that feeds into the Scottish income tax outturn. It would be really unfortunate for the individual employees who have not had the correct code applied, because that could mean that, for a period of time throughout the tax year, they have been either underpaying or overpaying their tax, which could clearly be a surprise, and an unwelcome one at that.
The number is going up, not down. There were 39,000 cases of employers in Scotland not applying S codes, and there are now 41,000. Are you able to dig any deeper into who these groups of people are? Is it the same employers year on year? Why is the number not significantly coming down by now?
We look at that as part of our work, but I go back to your point that it has increased year on year. That is true, but the trend has decreased over a period of time—that is what I was pointing to.
Is there any further information that the committee could get, either from you or HMRC, about the categories and groups of employees—
HMRC could certainly elaborate on what is driving that.
Convener, I think that we asked HMRC for that last year, but I do not recall whether we got a response. Clearly, the committee is interested in why and whether employers are habitually not applying the S code to their employees’ tax returns. Can you add anything on that, Auditor General?
That is primarily a question for the National Audit Office and HMRC, Mr Coffey. If you make further inquiries, it will be interesting to see whether HMRC’s position has changed. As the NAO set out in its report, HMRC’s position is that the number of employers that routinely use an incorrect tax code is very low, and it is not routinely tracking that information. Perhaps that takes us back to the cost benefit point—whether, given its compliance activity and, indeed, the compliance activity that the Scottish Government asks for, HMRC would look for more information on that area.
None of that detracts from the overall point that you make, which is that the reliability and robustness of the Scottish tax base are fundamental to the outturn and the reliability of the estimate. That package of measures allows the Scottish Government to take a view on whether it is satisfied with the overall reliability of the estimate that produces the numbers that support the Scottish budget.
10:15
I would certainly be interested—I am sure that colleagues would be, too—in whether it is a few employers, with high numbers of employees, that are habitually not applying the code. I am sure that it would be of interest to the committee to help resolve that. We do not know, because we do not see that level of detail. Perhaps we could follow up on that if we get the opportunity.
My final question is about parliamentarians and the correct application of the S tax code. You will probably recall that, in the early days of the Parliament, 45 of our dearly beloved colleagues were not regarded as Scottish taxpayers. Is that problem completely resolved now for Scottish members of the Scottish Parliament and Scottish members of Parliament who serve at Westminster? Do we know who they all are, and are we applying an S tax code to them all?
You will not be surprised to hear that, since that happened a number of years ago, what used to be a manual process is now automated and has more scrutiny from HMRC. We certainly look at that on an annual basis, and we are keenly interested in it. Since that time, I have not identified or been made aware of any issues around MSPs’ coding.
Auditor General, I want to touch on something in your report. In paragraph 42, which is in the section on taxpayer behaviour, you say:
“In my view, the publication of the income tax behavioural analysis and the development of a dataset to track taxpayer responses to income tax changes over time is a positive development.”
Can I check with you where that is? I have not seen anything published yet, but is there an expectation that something will be published later this year? Has work been commissioned to get that data set and put it into the public domain?
I will touch on a couple of things. In the preceding paragraphs, we refer to an analysis of Scottish taxpayer liabilities and behaviour over time, which was published by HMRC in December 2021. That explores the behavioural change and incentives that we have covered in part this morning. Some of that analysis is about the addresses that people use for tax purposes. Other parts of the analysis that have not featured prominently today include incentives to behavioural change, which encompasses the extent of hours worked in Scotland relative to other parts of the UK to mitigate tax paid relative to income.
That section of our report also mentions HMRC’s work with devolved Governments in the UK. Scotland is further ahead than Wales is on applying its divergence from the rest of the UK in income tax rates. HMRC is working with devolved Governments as part of a working group to establish the reliability of data sets and anticipated outturn statistics.
All that points to an on-going need for the Scottish Government to satisfy itself—fundamentally, as a customer in this exercise—that it is getting reliable estimates from HMRC, that the outturn is consistent with the methodology and that it can produce a Scottish budget that is as close as possible to what it will ultimately receive in income tax returns.
That work is happening. What we want to see next is that the Government is satisfied that what has been produced, with more to come this summer, is consistent with its expectations in this changing environment.
On that note of great clarity, I will draw this morning’s evidence session to a close. I thank our witnesses. The evidence session has been very useful, and it provides us with a platform on which we will build.
I move the meeting into private session.
10:20 Meeting continued in private until 10:49.