Agenda item 2 is consideration of the Auditor General for Scotland’s report, “Administration of Scottish income tax 2023/24”. I am very pleased to welcome our witnesses this morning. We are joined by the Auditor General, Stephen Boyle. Alongside the Auditor General are Carole Grant, an audit director, and Richard Robinson, a senior manager, both from Audit Scotland. I am also very pleased to welcome once again to the Public Audit Committee Gareth Davies, who is the Comptroller and Auditor General at the National Audit Office. Alongside Mr Davies is Darren Stewart, who is His Majesty’s Revenue and Customs financial audit director at the National Audit Office.
I think that you are aware that we are quite tight for time this morning. We have some important areas of ground that we want to cover with you, so if you are able to make your answers concise, that would be helpful—and I have leant on the committee to encourage concise questions. Before we get to those questions, I invite both Stephen Boyle and Gareth Davies to make opening statements. I will begin with you, Auditor General.
Many thanks indeed, convener, and good morning, committee.
As you mentioned, convener, the purpose of today’s session is to look at the administration of Scottish income tax for 2023-24. HMRC collects and administers Scottish income tax as part of the United Kingdom’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 report to the Public Audit Committee to provide additional assurance on the NAO’s audit work, in line with the recommendation from the previous Public Audit 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, and that the findings and conclusions in the C and AG’s report are reasonably based.
The C and AG has concluded that the outturn for Scottish income tax was fairly stated, which provides the Scottish Parliament with valuable assurance over this aspect of the Scottish budget.
This is the sixth year that HMRC has published Scottish income tax outturns in its accounts. Those outturn figures relate to 2022-23, and the difference between actual UK and Scottish tax returns and the amounts forecast at the time is then adjusted for the 2025-26 budget—this is known as the budget reconciliation. The reconciliation for 2022-23 outturns results in an increase in the 2025-26 budget of £449 million. That is the largest adjustment in the budget to date resulting from a budget reconciliation.
HMRC’s annual accounts also include an estimate of Scottish income tax for 2023-24, but that does not affect the Scottish budget.
My report highlights that while income tax revenues are continuing to rise, the impact on the Scottish budget is reduced by relatively slower economic growth in Scotland compared with that in the rest of the UK. In the period between 2017-18 and 2022-23, Scottish taxpayers paid an additional £3.4 billion due to tax policy differences, but relatively slower economic growth in Scotland meant that the net benefit to the budget over the same period was £629 million.
Taxes and the economy are important pillars of the Scottish Government’s approach to fiscal sustainability. The current economic performance gap underlines the importance of relative economic growth to Scotland’s public finances, which needs to be a key focus of the Scottish Government in the coming years.
I will hand over to Gareth Davies, but of course, Carole Grant, Richard Robinson and I look forward to answering the committee’s questions.
Good morning, committee, and thank you for inviting me and my colleague to give evidence on our work on HMRC’s collection of Scottish income tax.
As in previous years, the content of our report follows the requirements of the legislation and covers the outturn for 2022-23 and, to a lesser extent, the estimate for 2023-24. We have looked at the rules and procedures that are in place to administer the system at HMRC and the costs recharged by HMRC to the Scottish Government under the service level agreement between the two parties. The approach taken by HMRC in calculating the outturn and estimate has remained broadly the same as that taken last year, and I have concluded again that both are reasonable. Both the outturn and the estimate demonstrate growth in the Scottish income tax take, as the Auditor General has already said, reflecting—among other things—the impact of inflation, but also the fact that more taxpayers are falling within the income tax system because of slowly increasing thresholds.
HMRC has corrected some discrepancies that it found in how it calculated the outturn for Scottish income tax in previous years. We cover that in our report in a bit of detail. Based on our work, we are satisfied that the discrepancies are not material to the context of our audit in any of those years, but I am not satisfied that an error of that kind was present in the system, even if it did not result in material errors in the calculations. We would be very happy to explain the work that we have done to test the adjustments that have been made, but also to ensure that similar errors do not occur in the future.
Although the divergence in income tax policy between Scotland and the rest of the UK widened in 2023-24, and will widen further in 2024-25, HMRC continues to assess the risk of non-compliance resulting from that divergence as low. In response to recommendations made previously by us and by the committee, I am pleased to see that HMRC and the Scottish Government are exploring what further compliance activity can or should be carried out to test that assumption that the non-compliance risk remains low. That work is on-going and we will be tracking it closely through our work. I am sure that you will want to explore that in this morning’s session.
My team and I again worked closely with the team from Audit Scotland all the way through our audit, and I am grateful for its co-operation in completing the work.
I am very happy to answer questions. Thank you.
Thank you very much indeed. I will turn to a couple of issues that both of you raised in your opening statements. In particular, Auditor General, you referred to something that is in your report. Exhibit 3 shows that £3.367 billion of additional income tax revenue has been raised from the Scottish population, but that that converts to additional budget spend of only £629 million. Presumably, that is a product of the fiscal framework. Do you have any observations about how the fiscal framework is working? What are the implications of that? As I calculate it, for every £5 raised, only £1 is available for the budget.
You are right. Exhibit 3 looks to set out the impact of relative tax policy and economic performance of the Scottish budget as it relates to Scottish income tax. The exhibit tracks that from 2017-18 through to the most recent available data, which is from 2022-23. Fundamentally, we drew on the work of the Scottish Fiscal Commission and its published analysis when compiling the table. I will hand over to Richard Robinson in a moment to set out some of the mechanics of how this works, if that would be helpful for the committee, but it illustrates perhaps one of the core points that we make in the audit insights section the report. I touched on this in my opening remarks: there is a complexity to the fiscal and economic levers that are available to the Scottish Government and the Scottish Parliament.
One of the key tenets of the fiscal framework is about relative economic performance. Perhaps I can set out for the committee some of the observations that the SFC made in its fiscal update from August 2024, where it explains the differentials in a bit more detail than we have done in our report.
The SFC notes:
“Scottish average earnings have grown 3.1 per cent lower than in the rest of the UK between 2016-17 and 2022-23 while employment has grown 3.2 per cent lower.”
Of course, income tax is based on the number of people in employment and what those people are earning.
There is some very useful analysis in the SFC’s update, but our purpose in referring to that in our report is the continual necessity not for me to make what would be a policy judgment on the appropriateness of the content of the fiscal framework, but to ensure that the complexity is understood, that the trends of data and differentials are also well observed and noted, and that the Scottish Government, especially, is content and satisfied that it is tracking and managing the risks related to fiscal and taxation policy.
I will maybe bring in Richard Robinson, who can give a couple of points of detail that might be helpful to support the committee’s scrutiny.
At a high level, the fiscal framework does two things. The first is that it allows devolved decisions to be made on the appropriate tax rates and tax balance—the tax policy side of things. Also built into the fiscal framework is incentivising economic growth in Scotland, and, therefore, the rewards and the risks that come with that fall through into the figures that will affect the budget.
How does that translate? We can see that the tax policy is bringing additional funds to the budget. For a number of years, the SFC has been highlighting what it calls the “economic performance gap”. What that means is that, because the tax is devolved, the UK Government will make an adjustment to the block grant—as we show in exhibit 1 of our report—to represent what it estimates it has forgone in taxes by devolving the tax to Scotland. That is based on UK rates and UK rates of revenue growth on an indexed per capita basis. In a situation where the growth in revenues per capita and earnings is greater in the rest of the UK than it is in Scotland, you will see that economic performance gap coming through. As the SFC said, that is acting as a drag on the amount that is hitting through into the budget on the bottom line each year. In the report, we say that that is one of the areas that the Scottish Government should consider.
The other point to raise is that it does not necessarily have to be this way. The SFC had put into its December 2023 forecast its expectation of faster growth in Scotland than in the rest of the UK. We can see in exhibit 3 a reduction in the 2023 economic performance gap to £624 million from £663 million the year before. However, in its more recent forecasts, the SFC has suggested that perhaps that economic growth—or relative economic growth—will end in 2024-25. That means that the Scottish Government has to actively consider its economic situation and policy, along with the work that it is doing on migration—we refer to that in the report—and how things will translate into what hits the bottom line for the budget.
I said at the start we are quite pressed for time, but can I ask you to confirm my calculation? The figure in exhibit 3 is a cumulative figure—I draw attention to it because it is one of your key messages. For every £5 that has been raised through additional income tax in Scotland, only £1 of that—or less than £1 of that; it is about 18 per cent—finds its way into the Scottish budget.
Yes, that is correct, convener.
Okay. Let me move on to another area that your report draws attention to and which you mentioned in your opening statement. The reconciliation figure for 2025-26 is £449 million—nearly half a billion pounds—which is the largest reconciliation to date. Is there a likelihood that that level of reconciliation will continue in the future? How does that affect the Scottish Government’s budget setting?
You are right again. I will hand over to Richard Robinson, rather than both of us talking about some of the detail of that. For the committee’s information, we set out at exhibit 2 some of the detail behind the reconciliation—effectively, the difference between the final outturn and the forecast. As I mentioned in my opening remarks, there is also the catch-up nature of budget reconciliations from previous years and the time lag. I will bring in Richard Robinson to set it out in a bit more detail.
I will be brief. We see two things happening in exhibit 2. One is the correction for a forecast error. When the SFC and the Office for Budget Responsibility gave the forecast for the budget, they did so during quite a volatile period. They predicted that it would reduce by £190 million—that needs reversing out. The forecast did slightly better on outturns—revenue was greater than the block grant adjustment, adding another £259 million. That takes us to the £449 million that is shown in exhibit 2.
09:45I do not have to hand the details of exactly what the SFC is saying for future years, but it is expected that the differential tax policy and the fiscal drag from the differential structure of bands will lead to an increased amount coming through into the Scottish budget. What we set out in exhibit 4 is that it is slightly less than was originally expected as a result of revisions in the forecasts that have happened between December 2023 and December 2024.
You mentioned volatility. One of the things that happened in the medium-term retrospective look that we are taking is, of course, the pandemic, which was a severe shock to the economy. How does that affect some of the estimates and assessments?
Briefly, convener, the SFC is very clear in its reporting on the significant variability from both the downturn in economic performance caused by the pandemic and then the sharp recovery. Of course, that relates to a point that I suspect we will come back to a number of times today about the relative economic performance of the Scottish economy compared to that of the economy in the rest of the United Kingdom and how that impacts on the outturn of tax revenues. Undoubtedly, there is a lag effect. You will see that in a number of spaces, both in our own analysis and very certainly in the NAO’s commentary. When there is a forecast and then tax is collected, and there are associated estimates, there is a degree of complexity. As we will perhaps touch on during our discussion, there is uncertainty and volatility throughout much of the estimates and forecasting.
Mr Davies, you might want to respond to that question, but I want to ask specifically about the point that you made in your opening statement about the error, which you have taken a fairly dim view of, I think. Can you talk us through that?
On the broader point, the degree of change in the tax estimates is a UK-wide issue. It is certainly not just a Scotland issue. The very large growth in income tax year on year coming out of the pandemic has been a very big feature of the tax system. We did a report on HMRC’s overall cost of implementing UK taxes a couple of weeks ago. That report showed how much more complex it has become to administer the UK tax system as a whole, even leaving aside divergence in Scottish income tax rates, because far more people are now paying higher rates of tax across the UK and far more people altogether are in the tax system because of either no or slow increases in the threshold for paying tax. Volatility is a feature of the system UK-wide.
Your second question was about the error that I mentioned in my opening comments. It would be helpful for Darren Stewart to explain a bit more of the detail but, broadly, when HMRC compiles the estimate of tax due for Scotland, it uses the two big databases that it has, which are the pay-as-you-earn system—people in employment paying taxes—and the self-assessment forms that are typically filled in by the self-employed. The point is that it is not as simple as that. Some people who fill in self-assessment forms also have employment income and it is the accuracy of the process for eliminating any double-counting between the two systems that is the issue. HMRC discovered that it had not been eliminating all the double-counting. Perhaps Darren could explain in a bit more detail what has happened and what we have done about it.
Absolutely. I will start by reiterating what Gareth said earlier, which is that while the errors are not material, we take them seriously. We have done considerable work with HMRC to get to the bottom of how the error occurred—Gareth has already captured how that happened—and then think about how we can avoid it being repeated. As Gareth said, in producing the outturn calculation, HMRC brings together information from the PAYE system and the self-assessment system. The self-assessment system necessarily includes employment income. That is just the nature of submitting a self-assessment return. HMRC has always had an adjustment to remove duplicates where it is expecting a self-assessment return from somebody who has employment income. The error arose where people submitted self-assessment returns unsolicited, so HMRC was not expecting a return and had not adjusted for that.
With HMRC, we have looked at the adjustment process and at how to avoid the error happening again. Rather than focusing on the adjustment that HMRC used to make, the process now is that we bring together the two sets of data from the PAYE system and the self-assessment system. Through a computer coding approach, using our digital expertise, we basically compare the datasets line by line, taxpayer by taxpayer, to ensure that all the duplicate entries are completely and accurately removed. I hope that that gives you a good level of assurance that this will not happen again.
That might come back to haunt you, Mr Stewart, but we will see. I hope that you are right
My final question is on the overestimates. Your report speaks about calibration adjustments and a calibration adjustment was required to be made in 2021-22. Do you expect HMRC to have to make a calibration adjustment for the tax year 2023-24?
That is a necessary and important feature of financial modelling. A calibration adjustment is about looking at recent experience and applying that to the current forecast to make sure that any under or overestimates in previous years are adequately captured and factored into the current estimate. I would expect HMRC to make calibration adjustments every year. I guess that the specificity and the accuracy of those is the key question, but it should make them every year.
Over a number of years now, I have repeatedly brought up my concerns about the figures that are behind the tax collecting. Gareth Davies mentioned working on some aspects of this, but consistently, right across the board, there seem to be an awful lot of assumptions and estimates and so on made and it is quite difficult to see how they would result in an accurate figure. Just to remind you, the service level agreement states:
“HMRC will identify the Scottish taxpayer population and collect from it the correct rates of SIT to ensure the Scottish Government receives the correct amount of income tax revenue each year.”
There are lots of other clauses on the same thing, but to me it all rotates around the key word, “correct”. I will take a high-level look at the issue and go through a few of the points of concern that I have picked out. I would be interested in your comments. It is not that making an estimate or an assumption is wrong; it is the sheer number of them. On occasion, you make an assumption based on an estimate, which to me must multiply the divergence and the inaccuracy of that figure. For example, in paragraph 12 of your report you say that HMRC successfully matched 72.9 per cent of address records. What is the equivalent figure south of the border?
That is a Scotland-only exercise, partly to calculate the Scottish share of income tax. It is not done south of the border in the same way.
You do not do it?
No.
So you have no idea?
No. Obviously, there is a lot of compliance work that goes on south of the border, but that exercise is specifically to ensure that people with an S code, which means that they are liable for Scottish income tax, are matched to a Scottish address. That is a Scotland-only exercise that HMRC carries out.
In paragraph 14, you point out that tax policy in Scotland has increasingly diverged from that in the rest of the UK—that has been so for a number of years now—but HMRC has not yet assessed the impact of that or what it would cost to do so. Why?
Obviously, it is for HMRC and the Scottish Government to answer the why question. I think that you are absolutely right: the greater the divergence, the stronger the case for close evaluation of the impact of that, as well as very close attention to the accuracy of the application of those differential tax rates to the taxpayer base. Since I was here this time last year, I have noted that HMRC has for the first time published some of its early research into any evidence for the impact of differential tax rates on taxpayer behaviour. That is a welcome development. It has taken a long time to get to that point and you could say that that work needs to accelerate but, clearly, all that boils down to a question of resource prioritisation, both for HMRC and the Scottish Government. I agree that it would be desirable to have better information on some of the areas of estimated figures so that the calculation can be as accurate as possible.
That is important, not just in retrospect, but it is important to have the estimates as accurate as possible for the future because, as we have heard, it has an impact on Scottish budget setting. For all those reasons, I think that accuracy is very desirable here. The question is how much resource both HMRC and the Scottish Government agree is necessary to put into that level of research to improve the methodology and to understand better what is happening to taxpayer behaviour in the light of such changes so that the information that you are all working on to set the Scottish budget is as accurate as possible. As the auditor, obviously I cannot answer that question about the right level of priority and resource, but I think it is very important that it is addressed.
You mentioned one aspect there, which is the behavioural changes, but there must be a process already in place across the UK to assess the behavioural change that any taxation increase or decrease might have.
Not as much as we at the NAO would like. One of our big critiques of Government generally, including HMRC, is the surprisingly low level of investment in evaluation of the impact of different policies. I think that it is a general problem in Government and HMRC is no exception. In a lot of NAO reports, one of the main recommendations is, “This needs to be better understood if you are going to make better decisions about the allocation of resources.” That applies in this area too.
I am mentioning all these issues because, in the aggregate, they are a concern. For the past two years, HMRC has been having difficulties in importing land and property transaction data. Two years?
That is right. We have reported on that consistently. My understanding of the nature of the issue is that Scotland has a data set on land registry information that would be useful to HMRC, but it is collected for a different purpose and stored in a system that is not compatible with HMRC’s systems. It is inherently difficult for HMRC to bring its data together with the land registry data and use the information in the way that it intends. That would be the preferred option, but HMRC has not been able to pursue it at this time.
As I say, I am just highlighting examples. In part one, paragraph 1.5, the report says that
“HMRC does not have data in sufficient detail to identify income tax liabilities, reliefs or other adjustments relating to individual taxpayers.”
Is that not a bit of a concern? The report goes on:
“The gross total of all the estimates and adjustments made by HMRC totalled £1,077 million in 2022-23”,
so it will be more now. That is not a small sum of money. Is the chance of error and incorrect projections based on that lack of data not a concern?
Yes. At the heart of that issue and the previous one is the quality of the systems that HMRC is using. Although there has been quite a big investment in upgrading things like the real-time income system, which has been a very important step forward for lots of reasons, HMRC is still using incompatible databases for different elements of its work. The cost of bringing all those systems up to a modern standard is very large. That is essentially the main constraint on its ability to do that, but it will clearly be necessary.
10:00If we want an efficient tax system, capable of answering the kind of basic questions that you are raising, we absolutely need a system where data-matching is much easier, where close analysis of the data elements that you mentioned is possible and the system is fit for purpose. Behind all the issues for the calculation of Scottish income tax is the fact that the systems were nearly all designed before Scotland had the ability to vary its rates of income tax, so a system not designed to deal with that is having to cope as best it can, and it lacks the design features to make that efficient and straightforward. Behind all this is the need for long-term investment in modern systems that can deliver the job that is now being expected of this information.
Again, in self-assessment liabilities, it would appear—and I am interpreting this, so correct me—that the deduction of
“£57 million to estimate Scotland’s share of other relevant Self Assessment balances where specific data are not available”,
is presumably based on a UK-wide average.
Is this in figure—
It is in paragraph 1.9. There are a number of other cases—I will not bore you by going into them one by one—where UK-wide data is being used. That has to be incredibly skewed because the south-east of England skews everything.
I think that the general point is that where HMRC needs to make estimates, it is generally due to two issues. The first is the timeliness of the data becoming available. An example of that would be where HMRC has to estimate how much Scotland is entitled to through the outturn for taxpayers who have not yet submitted a tax return where one was due, so it needs to make an estimate of that. The other, as you quite rightly say, is where Scotland-specific data does not exist. HMRC need to attribute an element to the Scottish outturn and it does that in a number of different ways. Sometimes it is based on income, sometimes it is based on headcount. You are quite right to point out that where that data does not exist at the Scotland level, an estimate has to be made. That is where HMRC does that.
That estimate must get distorted by the sheer wealth of the south-east of England. That would also be so for any area of England, but when you are comparing Scotland, the divergence is huge. Again, in paragraph 1.13 you are talking about basing calculations on UK averages, which, as I have said already, is a bit of a concern.
Excuse me; I am flicking through the report because I have highlighted all the bits that I am concerned about.
Here is one: the Scottish taxpayer population. To me, it looks as though there has been a fall in lower-rate taxpayers, but there does not seem to be any assessment of why that is or how that is working. You have fewer people paying less tax within the lower tax bands, based on previous years. Does that mean that they have moved into a higher tax band and are paying more tax now?
Could you say what paragraph you are asking about?
Sorry, it is 2.16.
I think that that is factual. It is basically the income tax distribution, based on income band.
There is no real analysis of why. Those are significant movements within the broad bands. You would hope that HMRC would be providing some data as to how that has arisen, what the consequences are and where we are going with this. There is nothing behind it.
I think that those are questions for HMRC and the Scottish Government. Clearly, the scope of our audit is not to get into what is changing the number of taxpayers at different levels, so you will not find that kind of analysis in our report on this exercise, but clearly they are very relevant questions for HMRC.
In relation to the population, the C and AG is right that that is a core part of the future analysis and consideration that the Scottish Government, working with HMRC, will need to undertake to understand the risks from Scotland’s tax approach increasingly diverging from that in the rest of the UK. The C and AG’s report quite reasonably observes that there has been an increase in compliance activity and that more consideration has been given to some of the risks, and we certainly make that observation.
We recognise that there will be a time lag. It might be 18 months to two years before the Scottish Government is clear about some of the implications of the divergent approach that it has taken, but there are firm steps that the Scottish Government, working with HMRC, now needs to take.
I will not go through the other issues step by step, because my colleagues will want to drill down into some of those areas, but, given the overall number of assessments, estimates and assumptions that are made, we cannot have an accurate tax figure—we just do not know.
Clearly, the ideal position would be to have full information that allowed us to identify every taxable individual in the country and apply the correct rates across the UK, with no estimates. However, getting to that position from where we currently are with our systems would involve a long journey, so it comes down to priorities and investment plans to create a completely different database from the one that we have now.
At the end of the day, will there not always be a major discrepancy between projected tax revenues and what we actually receive?
When it comes to the outturn calculation, our exercise shows that we can take material assurance on the accuracy of the figure, most of which is based on actual records of Scottish taxpayers. However, while we have our current systems, an element of it will always be an estimate. Reducing the uncertainty even further will require significant investment in more accurate Scottish data.
I keep coming back to the point that using any assumptions, estimates or projections about the UK as a whole will lead to a big discrepancy, because there is such discrepancy between income levels in Scotland and those in the south-east of England, for example. That will distort every tax figure across the country.
That does not affect the basic facts about income tax, because the information is taken from PAYE and self-assessment forms, which we know are accurate for taxpayers in Scotland. Most of the estimates that affect the figure do not involve a UK-wide application of a percentage across the board.
The estimates that remain to be addressed include those on the level of compliance returns for Scotland-based income tax compliance work by HMRC. In other words, when HMRC pursues taxpayers who have not paid the right amount of tax, that is called compliance work, but HMRC’s systems do not currently allow identification of Scottish taxpayers who are subject to compliance processes separately from the rest. That is when HMRC applies estimates about the productivity of its compliance work. In that case, Scotland might benefit from UK-wide application, because HMRC’s compliance work might generate bigger returns in the south-east than in Scotland.
That illustrates that it is difficult to draw an overall conclusion about the effects of some of the estimates. However, there is a strong argument for better data so that estimates do not have to be made in the first place, which is, I think, your point. The constraint relates not just to the use of existing data but to implementing a new system that provides the data in the first place.
To be clear, I am not suggesting that we would get more money out of the tax system if we had accurate figures, but, if we are to base our budgets on anything, the figures must be reasonably accurate in order to give us certainty for the future. My concern is that we will get lumpy adjustments every few years to take into account changes in all the estimates and assumptions, which is not desirable.
On that note, I will move things along. I invite Graham Simpson to put some questions to you.
I will follow up on some of the issues that Mr Beattie asked about. In your report, Mr Davies, you say that HMRC
“has not tested its assumption that non-compliance in Scotland is the same as in the rest of the UK.”
I guess that you would stick to that. You have said that there is an issue with systems—I guess that you are talking about computer systems—and that they need to be upgraded. Is that basically the position?
Yes. HMRC’s compliance teams use a system that does not have the separately identified data records that are in the main PAYE and self-assessment system. Darren Stewart can explain the position in a bit more detail.
As you correctly stated, our general view is that HMRC has not tested the assumption that the compliance risk in Scotland is the same as it is in the rest of the UK.
In our report, we say that the Scottish income tax board was due to meet in January to start looking at what it could do to move things on. Part of that conversation is about how it can update its compliance system to allow Scottish compliance cases to be flagged and better analysed. Another consideration is whether there is a cost-benefit argument for producing a Scottish tax gap, as is produced for the rest of the UK. Both those steps would be important in having a better view of compliance risk in Scotland.
Last week, we met Scottish Government and HMRC officials to get an update on that work. We understand that a decision was not made at the board meeting in January and that the issue will be brought back to the board meeting in April, which is, I believe, around the time in the cycle that the committee normally speaks to HMRC, so you might want to ask about further developments at that point.
Sorry—a decision on what?
A decision on what additional compliance activity the board might wish to do.
In paragraph 14 in your report, you say that the compliance working group
“expects to report to the Scottish Income Tax Board on the first phase of this evaluation in January 2025”—
which you mentioned—
“including the likely costs of additional compliance work, after which the Scottish Government must decide on the merits of funding any additional activity.”
There is therefore a role for the Scottish Government.
Absolutely.
Is the Scottish Government waiting for the Scottish income tax board to report to it?
The Scottish income tax board includes representatives from the Scottish Government and HMRC, and my understanding, based on our recent conversations, is that it will look at the issue again in April.
Okay. Auditor General, in your report, you say:
“The completion of the additional third-party data exercise in 2024, one year earlier than originally planned, is a positive development, however the results of this exercise have not yet been published”.
Have they been published now?
My NAO colleagues might be able to share the most up-to-date position on that.
On the wider point, as the C and AG referenced in his earlier remarks, some additional compliance activity has been undertaken since the committee previously considered the reports. That is welcome, and it speaks to the wider point about having as complete an understanding as possible to inform the Scottish Government’s understanding of the position on tax revenues and the economy.
Timeliness matters. The committee will not be surprised to hear me make the point that we do not yet have an up-to-date medium-term financial strategy from the Scottish Government. I understand that the Scottish Government’s target is that that will be published in the spring. Getting all these dates aligned matters in order to support understanding and parliamentary scrutiny of not just one year’s budget but the medium-term financial picture.
I will pass to anyone who wants to say more about timeliness in that regard.
10:15
The results of the address assurance exercise were not available when we published our report in January, but they have subsequently been made available to us, and we will report on the exercise in detail in our equivalent report this time next year.
The analysis that we have seen to date shows broad consistency in performance. The key metric, which you will see in figure 8 of our report, is the number of records that are unable to be matched through that exercise. The report shows that that number was 8,540, or 0.2 per cent of the total records that were identified, and that it has reduced slightly to 7,017, or 0.1 per cent of the total records.
There are two key points. First, the fact that the exercise is being carried out annually is a welcome development, and the committee has been asking for that for some time. Secondly, the relative stability or marginal improvement in the results is positive, but that position needs to be maintained.
I am looking at figure 8—the writing is very small—which says:
“In January 2024 HMRC identified 45,809 cases where ‘S’ prefixes were not correctly applied to tax codes.”
The S prefix issue is a different issue. The exercise that I have just described in relation to figure 8 relates to the third-party address issue. The S prefix issue relates to the incorrect application of tax codes by employers.
It is still an important issue.
Absolutely.
HMRC wrote to employers to ask them to update their records. Do you know what the result of that was?
HMRC does not gather data at the employer level. The committee has asked about that previously, and I understand that HMRC has committed to giving you an update on it.
We know that, over the course of this year, HMRC has been exploring how it can nudge employers and employees to ensure that tax codes are correct. That can be done through personal tax accounts, with that being administered through the app online, and through targeted communications to employers when repeat offenders are identified.
In the past, the committee has been particularly interested in the broader point about data collection to find regular offenders. As far as I understand, that work is still in development.
It should not still be in development, should it?
I tend to agree with you, but that is more a question for HMRC.
Yes—it sounds as though there are a few questions for HMRC.
I want to ask about one other issue. Mr Davies, in relation to tax relief on pension contributions, paragraph 2.13 in your report says:
“HMRC must identify Scottish taxpayers so that tax relief is correctly allocated. Pension administrators claim tax relief at source on behalf of their members and add this to their members’ contributions. HMRC applies tax relief on pension contributions at the basic rate of 20% for all taxpayers. Scottish taxpayers paying a tax rate above 20% can claim the remaining tax relief through a Self Assessment return or by contacting HMRC.”
I have asked about that issue previously. It strikes me that a lot of people will have absolutely no idea that they can do that and will not know how to go about it, so they could be missing out.
First of all, do you agree with my assessment? If you do, do you have any idea how many people are affected?
That is a UK-wide issue—that feature of the tax system affecting pension contributions does not affect only Scottish pensioners—but the additional complexity is that Scotland has different tax rates at the higher level. The Scottish issue is just a variation of a UK-wide issue.
I do not think that we have any observations on whether that is the right system for ensuring the accuracy of tax relief for pension contributions, unless Darren Stewart has anything more to say.
I do not. I agree with the broader point that people understanding their entitlements to reliefs and applying for them is not a Scotland-specific issue but a feature of the tax system, which is inherently complex.
Yes, I absolutely accept that. I think that most laymen and laywomen have no idea that they can do that and are probably not made aware of it. However, there is a Scottish element, because Scotland has a different tax rate—that is where we come in.
I have no further questions.
Thank you very much. I will now pass over to the deputy convener, Jamie Greene, who I know has some questions to put.
I do, convener.
Good morning, gentlemen and lady. I want to cover a few areas. First, I draw your attention to figure 7 on page 21 of the HMRC report, which is probably one of the more helpful tables in what is quite a lengthy and detailed report. It is certainly more visual, from my point of view; I was wanting to get my head around the analysis that has been done on the Scottish tax base in general, and this table paints a picture that we can look at.
Just to make sure I am being accurate, can you tell me whether the table is saying that higher and top-rate taxpayers in Scotland accounted for 13.3 per cent of taxpayers in 2018-19, and that in 2022-23, the latest year we have available, that figure had risen to 18.1 per cent of the tax base? That is the percentage of the tax base, but alongside that—and what is more important—is the percentage of tax that they are paying, which for the same group has risen from 58.6 per cent to 64.2 per cent. Is such a shift normal? If the percentage of taxpayers in any tax band increases, does the amount of tax that they pay also increase proportionately and at a similar or the same rate?
It is certainly a pattern that you would observe UK-wide, as well as in Scotland. The fact that a relatively small proportion of taxpayers—that is, the ones paying at the higher rates—pay a large proportion of the total tax is a UK-wide phenomenon. In that sense, then, the answer to your question is yes, that is what you would expect.
Does that present any risk at all? For example, when we look at the top rate, we see that less than 1 per cent—0.8 per cent—of Scottish taxpayers paid 17.8 per cent of all tax. In other words, less than 1 per cent of taxpayers account for nearly a fifth of the Scottish tax base. Again, the relevance of this questioning will become clear in my later questions on tax behaviour. You have said that it is a UK-wide phenomenon, but irrespective of that, does that present any risk to any Government when it comes to trying to forecast how much revenue it will get in any given year from its tax income as opposed to from the block grant?
I can see that you are heading into policy territory that is probably not for me, as the auditor, to talk about. Our issue here is accuracy of the data rather than the policy drivers behind it. Clearly, though, understanding the behavioural impacts of changes in the higher rates of tax is, as we have said, a crucial factor in understanding the future finances for the budget. However, it is probably not really for me to say any more than that.
I agree with the C and AG. Part of the reason for the change that you are seeing in the percentage of people in either the top rate or higher rate comes down to the pretty well-established phenomenon of fiscal drag. As the bands stay the same and people’s earnings increase, more people emerge into these new groupings.
Some of the consequences of that have been the change in ratios and some of the skewing effect. As the NAO has quite starkly set out, 18 per cent of people pay 64 per cent of tax. I come back to that last point that the C and AG quite rightly made; when we think about the increasing divergence of tax rates, we should bear in mind that those who pay the top rate of tax, in particular, are likely to be the most mobile of taxpayers. Therefore, there needs to be a clear understanding of the costs and benefits of the different rates, and they need to be as reasonably and accurately reflected as possible in future projections. That allows me to labour the point that the medium-term financial strategy is ever more important, given the variable context.
I would just note, without straying into policy areas, that the numbers themselves show that a relatively small group of people on that top rate are responsible for funding a huge chunk of Government revenue.
When you say that these taxpayers are more mobile, are you talking about their ability to move their residency to another part of the UK, or even outside the UK, or about their ability to shift money around in different ways that results in, say, their paying less tax? What do you mean by “mobile”?
The NAO might want to express a view on this, should it wish, and it is probably conjecture on my part, but people with more disposable income perhaps have more choices when it comes to where they reside, their tax status and so forth—the factors that you have alluded to.
While I have you on the line, Auditor General, I want to ask about paragraph 56 of your own report. I have read it about 100 times and still cannot get my head around it, so perhaps someone in your team can talk me through it.
In that paragraph, which relates to the budget year 2022-23, you state that
“The forecasts originally used ... reduced the budget by £190 million, the net difference between forecast tax foregone by HM Treasury and forecast Scottish Income tax receipts.”
On the next page, the report says:
“Outturn data shows that there was an increase of £259 million, a positive difference of £449 million from the forecast reduction.”
I have no idea what any of that means, so please talk me through it. It seems quite stark, whatever it is.
I will again pass over to Richard Robinson, but first of all, please forgive us, deputy convener. In some of the report’s terminology, we naturally mirror the language used by HMRC, the Fiscal Commission and so forth with regard to forecasting. Richard Robinson can set this out for you, but this is about the difference between the forecast using the best available information at the time and then what actually happened, as shown in the outturn data.
The convener referred to this earlier when he mentioned the context behind the increased variability in some of these results. Our coming out of the pandemic, inflationary changes and the war in Ukraine were all part of the context for the assumptions that were made at one point; however, although they were made with the best available information, they did not quite materialise in the results. Hence, we have a larger budget reconciliation—£449 million—than had been assumed when the estimates were first made.
In paragraph 56, we effectively try to explain exhibit 2 in words. When budgets are set, neither the SFC nor the OBR know exactly what the income tax receipts will be for the coming year, so they both make forecasts. According to the forecasts that were made, and which are set out in the first line of exhibit 2, the revenue that the SFC expected to be generated from 2022-23 was less than the amount of BGA that the OBR forecast would be taken from the block grant. The adjustment that was made for that in the 2023 budget effectively reduced the 2022-23 budget by £190 million. In other words, spending power was reduced by £190 million.
Going forward in time to when we knew the outturn, we were able to say, “Actually, that is not exactly what happened”, for some of the reasons that the Auditor General has mentioned. For example, higher inflation led to different wage bills, fiscal drags and all those types of things. When people looked at the actual figures, they found that, instead of the budget being reduced by £190 million, the revenues came out higher than the block grant adjustment; as a result, the £190 million needed to be given back, and then the difference between the outturn figures, which was £259 million, needed to be added to that.
With the reconciliation of the 2024-25 budget, we get both those things. We get the return of the £190 million that we did not have to spend in 2022-23, plus the difference between the final outturn and the final BGA, which we now know, because we have the outturn figures. Those two figures make up the £449 million.
10:30
Can you talk me through exhibit 1, on the block grant adjustment, just so that we can get our head around this? A Barnett-determined block grant is allocated to the Scottish Government, but that is not what we actually get, due to adjustments based on devolved taxation. Can you, in very simplistic terms, talk me through how we get from the block grant allocation, through a net adjustment up or down, to what the Scottish Government actually gets?
There is more detail on this in our previous “Operation of the Fiscal Framework” publications, which obviously contain more detail than I will go into right now. However, in general terms, exhibit 1 shows that, before anything was devolved, we had just the block grant. Now that non-savings, non-dividend income tax is devolved, the money that we are talking about and which has been audited for these reports will be added to the block grant adjustment. That is the revenue from NSND Scottish income tax.
To reflect the fact that the UK Government no longer collects that tax, and no longer has it for its own budget, an adjustment is made to the block grant. Those are the two things that are happening: the tax forgone is being removed from the block grant and the actual tax is being added on. What hits the budget is the difference between those two amounts, which, as you can see, is set out in exhibit 3 of the report.
Moving on to exhibit 3—this will all make sense in a moment—can you tell us about the relationship between any increase in taxation that is received through Scotland’s tax policy differences and what is described as the “net position”? Let me take the year 2021-22 as an example, as, according to the table, it was perhaps the starkest with regard to tax divergence. In that year, £749 million extra in taxation was raised, but the net position—however you describe it—was only £85 million, or 11 per cent of the tax raised, which is a tiny amount. Indeed, it is even starker than the figure of a fifth that the convener referred to. Again, can you talk me through the effect of that on the original Barnett-derived block grant versus what the Scottish Government gets? Is the Government getting the tax receipts, or is it getting this 11 per cent net figure?
What the SFC has set out in these charts and tables, which come from its fiscal updates, is the net effect of the two boxes that are shown in exhibit 1. Instead of showing the full amount of the revenues coming in and the full amount of the block grant adjustments being taken out, it sets out the net amount and then the amount that one would expect to collect through tax policy differences if economic performance were exactly the same. It then calculates the economic performance gap between the net figure coming through that is being added to or taken from the block grant adjustment, based on outturns, and the expectations based on tax policy differences between Scotland and the rest of the UK.
I picked that year, because in an exchange that I had about taxation with Alyson Stafford from the Scottish Government on 25 April last year—we were talking about inward migration to Scotland increasing tax revenues—she made the following statement. This is why it is important, I think, that we understand the numbers, because she said:
“In 2021-22, taxable income in Scotland increased by £200 million as a result of the positive inward migration of taxpayers.”—[Official Report, Public Audit Committee, 25 April 2024; c 7.]
Presumably, that £200 million is part of your £749 million in exhibit 3. If we use the 11 per cent calculation that I highlighted earlier, it would mean that that £200 million would result in only £20 million of net benefit to the Scottish budget. Is that assumption correct? I did not have this data to challenge what was being said at the time, but, nearly a year on, we now have the ability to look at these numbers and see what they mean in real terms to the Scottish Government’s budget.
Perhaps that is a question for the Auditor General.
I am not sure how definitive I can be on this, deputy convener. I would quite like to track the various figures that are being quoted and, just for completeness, refer back to Ms Stafford’s evidence, so that we understand what was said. As ever, it might be another line of inquiry that the committee might wish to pursue separately with the Scottish Government, with the benefit of hindsight and some of the numbers that we have.
In the National Audit Office report, there is more detail on some of the changes with regard to taxpayers within Scotland as well as some analysis from HMRC on the issue that you highlighted of inward migration to Scotland. What matters is not just the numbers of individuals concerned but the tax band that they are in, which brings us back to some of our earlier conclusions in today’s report with regard to ensuring that we have an understanding—a trend analysis—that arises from not just one year in isolation. Given some of the complexity involved, this needs to be known, understood and monitored over a far longer period of time.
One of my concerns is that, although taxation has been devolved since 2017-18, there still seems to be no long-term analysis of tax behaviour. In fact, I think that one of the reports—or perhaps both of them—refer to 2027 as the year when we might have a better understanding. That is a decade on from the taxes being devolved, and it is not helping Governments either today or next year make decisions about tax policy.
I am not asking you to comment on the policy itself, but surely the source of this data—that is, HMRC—and the people who analyse it, including the Scottish Fiscal Commission, would be in a better position to inform Government policy if we had that sort of analysis. How can we not know this? You have talked about net inward migration figures being on the increase, for example, but knowing that is irrelevant if we do not know how much extra revenue is being brought in. The number of people is not important; what are important are the amount of money being paid and the net benefit. Is there more that we should be doing in that respect?
The C and AG might want to comment on this, too, but what we have set out in our report brings us back to the divergence point and the increasing complexity and changes that we are seeing in the Scottish income tax system compared with other parts of the UK. The impact of that matters, and it needs to be known and understood, especially in the context of the Scottish budget.
However, both reports show that there has been a response to some of the committee’s previous concerns, with increased analysis and understanding of compliance activity. We do not have all the results of that yet; decisions on it have not yet been taken; and there will be a further time lag before all of this is known. Ultimately, it will be up to the Scottish Government to weigh up how much additional activity it wants to ask HMRC to undertake and the cost benefit of doing that extra work. I have been listening carefully to the C and AG’s own analysis and observation of HMRC’s systems and ability to do something that informs policy but which represents good value for the Scottish Government.
Clearly the committee is one of the main interested users of this data. After all, your job is to hold the Scottish Government to account for how it is using public resources and how it is using that information to make policy decisions. Articulating your demand for more information on all the areas that we have discussed this morning will be extremely helpful. Your sister committee in the Westminster Parliament is equally interested in evaluation information on the impact of all sorts of tax policies, not just income tax rates, but tax reliefs of various kinds and so on. Articulating that demand for better information to inform your scrutiny as well as the Government’s decision making will be a very important part of the process.
I have one final question. Auditor General, at the start of the session, you talked a lot about Scotland’s economic performance relative to the rest of the UK. I want to delve slightly deeper into that. When you talk about slower economic growth, what metrics are you specifically talking about? Has there been any analysis of the effect of that on the Scottish economy?
Exhibit 3 in the report specifically references as the source of that information the work of the Scottish Fiscal Commission, the Scottish Government’s own forecasting body. As I have mentioned, we have referred to paragraph 4.18 of the Scottish Fiscal Commission’s fiscal update of last August, which cites various factors. The reasons for the economic performance gap being what it is are complex, but the commission notes that, between 2016 and 2023, Scottish average earnings were 3.1 per cent lower than those in the rest of the UK, and employment rates in Scotland grew by 3.2 per cent less than in other parts of the UK.
Of course, there will be other factors. Some of the analysis of gross domestic product growth in recent times highlights examples of the Scottish economy performing better in more recent years than other parts of the UK. However, those were the specific sources that we drew on, particularly as they related to income tax and the performance gap. Certainly, if it would be helpful, we can share some of the detail of our sources with the committee.
I was just challenging you on this, because paragraph 60 of your own report makes quite a profound statement. In that paragraph, you talk about the net benefit of the devolution of tax being significantly less than the taxes that are paid, but then you state:
“This is due to the relatively slower economic growth in Scotland.”
It is important that you substantiate that claim, because, as you have said, we will be challenging ministers on that comment.
Thank you very much. I have no further questions.
I think that that concludes our session. I was particularly struck by Gareth Davies’s point about the importance of evaluation, especially when it comes to the impact of taxation, both in advance of those taxes being introduced and after their introduction. I think that we would all benefit from that.
You are right to highlight the fact that we, as the Public Audit Committee of the Scottish Parliament, have a particular interest in how public money is spent. However, we are also MSPs who will vote on budget decisions. As a result, this is not just retrospective consideration of the way in which things have gone; it is also about the contemporaneous decisions that are being made on tax policy.
I thank Richard Robinson; Carole Grant; the Auditor General, Stephen Boyle; the Comptroller and Auditor General, Gareth Davies; and Darren Stewart of the NAO very much for their time and their evidence this morning. That evidence is very valuable to us.
I suspend the meeting.
10:43 Meeting suspended.