Finance and Constitution Committee

Meeting date: Wednesday, February 7, 2018

Official Report 551KB pdf

---

Scottish Fiscal Commission (Revised Forecast)

Good morning and welcome to the fifth meeting in 2018 of the Finance and Constitution Committee. I remind everybody present to switch off their mobile phones, or at least to put them into a mode that will not interfere with our proceedings.

Agenda item 1 is an evidence session on the Scottish Fiscal Commission’s revised forecasts, which were produced following policy changes that were announced last week at stage 1 of the Budget (Scotland) (No 2) Bill. We are joined for this session by Professor Alasdair Smith, who is one of the commissioners; David Wilson, who is also a commissioner; John Ireland, the chief executive; and David Stone, the head of economic and income tax forecasting. I warmly welcome our witnesses to the meeting. I think that Professor Smith wants to make a short opening statement.

Yes, thank you. Good morning. We are grateful to have this opportunity to discuss our updated income tax forecasts following last week’s stage 1 debate. As you said, during that debate the Cabinet Secretary for Finance and the Constitution announced a number of changes to the budget bill that had been based on the draft budget published in December. Those changes included a reduction to the higher-rate threshold for income tax, an extension of the Government’s public sector pay policy and a number of additional expenditure commitments totalling £137.8 million in 2018-19.

Last Wednesday, the Scottish Government published its provisional estimate of the additional revenue of around £220 million in 2018-19 from the income tax policy. That is the income tax policy in its entirety. It includes an additional £55 million, which is associated with the reduction in the higher-rate threshold.

The Government noted that the Scottish Fiscal Commission would provide the official revenue estimate, and that is what we did yesterday afternoon when we published a supplement to our December forecast. That timing was agreed with the Parliament and the Government before Christmas. The written agreement was published on our website.

The Government’s final income tax policy is forecast by the commission to raise £219 million in 2018-19, which is £55 million more than the policy that was announced in December. In addition, the extension of the Government’s public sector pay policy is forecast to raise a further £7 million in income tax revenues.

We have been told that the cabinet secretary intends to put it on the record at stage 2 that the additional spending will be funded from tax revenues plus a combination of Scotland reserve drawdown and underspends. The Government does not intend that the additional expenditure that was announced by the cabinet secretary will require additional borrowing or impact on the position of the non-domestic rates rating account, which is also known as the NDR pool.

In light of the announcement, we see no need to change our December assessment of the reasonableness of the Government’s borrowing plans. Our next forecast will be published in the spring to accompany the proposed medium-term financial strategy focusing on the longer-term sustainability of Scotland’s public finances. In May, we will review both our assessment of the reasonableness of the Government’s borrowing projections and its scope.

That is all I want to say by way of introduction. We look forward to your questions.

Thank you. That was a very helpful introduction. We are also very grateful for the Fiscal Commission’s ability to turn round its forecasting so quickly, given that stage 1 of the bill was completed only last week.

Your report suggests that the extension of the 3 per cent public sector pay award from those earning more than £30,000 per annum to those earning up to £36,500 is expected to generate an additional £7 million of tax revenues in 2018-19, rising to £8 million per annum by 2022-23. How did you arrive at that figure of £7 million?

May I ask David Stone to talk you through the detailed modelling on that?

Yes.

Our starting point for those forecasts is the survey of personal incomes, which provides detailed income tax payer data sets. That allows us to identify individuals who we think are working in the public sector. We take the detail of the Government’s extension to the public sector pay policy and, using information provided to us by the Government, work out the number of people who are in the scope of that 3 per cent pay award and the number of people who are in the scope of the 2 per cent pay award.

We then have a specific adjustment for public sector pay in our income tax model. Using the information that we were given, we saw that in 2018-19, before the extension, public sector pay would have increased by an average of 3.2 per cent; with the extension, that average increase is now 3.3 per cent. Running those two numbers through the model and finding the difference gives us the £7 million figure. We increased the total amount of income tax liabilities in 2018-19 by the £7 million difference that was introduced by 3.3 per cent growth versus 3.2 per cent growth.

Can you unwrap what you mean by “public sector”? Does that include people in local government or those parts of the public sector that are not under the direct control of Scottish ministers, or do you identify the public sector as a whole?

We have to use the data that we have, which is the survey of personal incomes. There is a broad sectoral classification for each individual in that database, which can be things like healthcare, education or finance. We look at that list of sectors and work out for each of them whether the workers will be primarily public sector or primarily private sector. That gives us our breakdown in our data set between public sector and private sector.

For those sectors in the model that we identify as having primarily public sector workers, we apply the public sector wage growth rates; we work those out in a separate exercise, based on information that the Government gives us.

Can I delve a wee bit deeper into that? If we are then looking at the public sector, the issue of the 3 per cent public sector pay award being extended to those earning up to £36,500 can only really affect the core bits of the Scottish Government that it is responsible for.

Yes.

How do you account for the difference between that core group and others in the public sector? How do you deal with it methodologically? That is what I am really asking.

In a sense, the Government is very clear about who its pay policy covers. You asked about local government, and the policy does not cover local government apart from teachers. We use the information that the Government gives us on the scope and we can, therefore, make an appropriate estimate of the size of the coverage of the pay policy.

Patrick, do you want to follow up on that?

Yes. I am not sure that I quite followed the answer to the convener’s supplementary question. Are you saying that you have made a judgment on who is directly covered in the Scottish Government’s pay policy and that the £7 million projection of additional income tax revenues is about those people who are directly covered by that pay policy?

Yes, that is correct.

If local government achieved a similar pay policy, or perhaps even something better, would that generate additional income tax revenues on top of that £7 million?

Yes. We did not include local government in the scope of this.

That is helpful. I want to ask about the process and protocol that has gone on over the past few days.

Let us cover the behaviour stuff before we move on to the process. We will try to get through all the issues about the numbers first.

I have a quick question on the behavioural response. Were you surprised that the cost of it is almost 20 per cent of the additional amount raised? Was there an anomaly between the decreases from behavioural responses in the top two rates and the number of taxpayers in those rates, which you say will increase over the next five years? Is there an anomaly, in that you have revenue on that reducing but numbers of people increasing? Can the behavioural response be broken down into people who will manage their affairs differently but remain in Scotland and people who will actually leave because of that?

The overall approach that we have taken to costing the new approach that the Government announced at stage 1 is exactly the same as the approach that we took back in December. That is the key thing to emphasise. As you put it, it may look as though the behavioural impact of the change in policy is much less than the behavioural impact of the initial policy that was announced back in December; that reflects the particular nature of the announcement.

Our overall approach recognises that when there is a change—for example, to the additional rate for very high tax payers—that is likely to be an incentive to adjust tax affairs for many people who might have the ability to do that. That is reflected in our modelling approach, which leads to the reduction between the static impact, as we call it, and taking into account a behavioural impact.

The new proposal and the adjustment in the personal allowance will affect a large number of people, but it will do so in a relatively limited way. The Government’s numbers are that it will, in effect, increase the tax payment by around £170. We do not expect that to have a very significant behavioural impact compared with the initial proposal. The incremental behavioural impact of the new proposal over what was announced in December we would expect to be more modest, based on the assessment that we have made.

10:15  

I would like to touch very briefly on the behavioural stuff. I may also follow up on the issue about local government in order to get clarity on that and the wider public pay impact. I thank the witnesses very much for coming along and for the excellent work that they have done on this.

I am looking at table 10 in your updated income tax forecasts. If I have understood it correctly, you are saying that you would have expected—on the static costs and the Government’s recent changes to the threshold—to have raised £61 million, but that you are losing £5 million of that because of behavioural impact. It is something less than 10 per cent. Is that correct?

I would like to clarify a point that was made earlier. The reason for that figure being lower than your previous forecast is that the change hits a number of people who are in the range of the £43,000 or £44,000 threshold and who are the most likely not to be able to change their affairs dramatically. It hits a lot of people, but it hits them to such a small extent that you do not expect it to drive any change in their behaviour. Is that assessment correct?

It hits them by a fixed amount. If someone who is earning over £150,000, for example, reduced their time worked or slightly adjusted their affairs, the maximum amount that they could save would be limited. It is the difference between an average tax and a marginal tax. We would not expect the behavioural impact to be anything like as great as, for example, the impact of a change to the additional rate of tax.

Have you done that using the numbers that you gave us before for the tax income elasticities, which were 0.35 up to 0.75 per cent?

It is the same approach—the same model.

Thank you for clarifying that.

The convener and Patrick Harvie asked whether the assumption of £7 million in tax revenues applied only to Scottish Government employees or to the broader public sector. I think you said that teachers were included in that, but not other workers. If the increase was applied across the piece, and if local government and others awarded those increases as well, how much extra would that raise over and above the £7 million? Can you quantify that?

To clarify, we got advice from the Government on who would be included in the scope of the policy; it also told us who it believed would be aligned with this policy. The Government said that that would include around 260,000 public sector workers, and that would include the national health service, the police, the fire service and teachers. It is up to the Government to clarify exactly who the pay award will cover, but that is the advice that we received. It covers around 260,000 public sector workers in Scotland out of a total of—I cannot remember the figure—400,000 or 500,000-odd. It applies to a fair chunk of all individuals in the public sector, and then we break down the increases in pay for that group.

Based on what you have just said, that sounds as though it applies to roughly half.

Yes, but it does not apply to local government. Local government workers are excluded from that 260,000 figure.

Correct, so the £7 million will become a number that may be double that. I am trying to see whether you can quantify that number for us in any way, shape or form.

We have not done that quantification because it is not stated Government policy that the increase will apply to local government.

Okay. That is clear.

If all local government employees were to receive that level of pay award, those would be the order of magnitude numbers. However, we have not done that quantification.

That is clear. It could be double, give or take. Thank you very much.

That whole process becomes even more difficult to make a judgment on because of the number of United Kingdom civil servants who are employed in Scotland. We do not know yet where they will be in terms of the final pay position. UK civil servants might not achieve the same levels as individuals who are directly affected by the Scottish Government. The picture is pretty complicated at this stage.

Yes, it is a complicated picture. Estimates could be made but, to be clear, that is not an estimate that we have made or which would be within our broad remit to make.

That is fair enough.

On the behavioural effects, you have stated:

“The greatest behavioural response comes from those in the higher and top rate bands.”

In discussion with Ivan McKee, I think that you agreed that there is likely to be a larger effect at the highest end of the income scale, but you are still saying that the higher-rate band is where the greatest effect will be. What is the evidence base for the extent of the effect of behavioural changes around tax avoidance at the higher-rate band? In particular, I am thinking about the bulk of higher-rate taxpayers who may be at the bottom end of that range. On what evidence have you based your assumptions?

We set out in our papers that were published in December—we also published an earlier paper—our overall approach to forecasting income tax. There is academic literature that sets out the experience of countries and areas that have—as far as such comparison is possible—comparable variations in income tax, but that literature is far from comprehensive or definitive in terms of the conclusions that can be drawn from it.

The principal source of analysis—or the best example that we draw on—is the work that was done by Her Majesty’s Revenue and Customs going through the experience of the change to the higher rate of tax in 2010 to 2012 in the UK, when there was an increase to the 50p rate of tax that was subsequently removed. That experience has enabled people to assess the likely behavioural impact. HMRC published in March 2012 an assessment that I think has been quite influential on how people think about behavioural impacts. The Institute for Fiscal Studies has also subsequently produced assessments.

I will quote from the HMRC study, which I think is quite informative. One of the key things that it says in its conclusion is that

“Behavioural responses to tax changes are often large and highly uncertain.”

That is our starting point: we should expect, especially for higher-rate taxpayers, a large behavioural effect, but the precise level and impact will, inevitably, be uncertain. A conclusion of the assessment of that experience was that there was

“considerable behavioural response and a substantial amount of forestalling”.

We have attempted to take advantage of work that has already been done. We have developed an overall approach, which we have published. We have stated what our approach is to behavioural response through tax income elasticities, and we have stated our assessment of how we think forestalling would work, depending on policy options. We have set out our overall approach and have evaluated the stated Government policy against that background.

Finally, we plan to publish a paper on 7 March setting out an even more detailed approach to modelling income tax, including behavioural responses.

That will be very helpful; I look forward to it. It will be a couple of years before we can reconcile your projections with reality. Will we know earlier than that what the reality is? Will you be in a position to refine your projections for next year based on information that you will gather during the coming financial year about the amount of tax revenue that has been generated and the amount of tax avoidance behaviour that has taken place?

There are two things to say. We will undertake a forecast evaluation: we will assess annually how well we have done and what we have missed, and we will publish a report setting that out. If that evaluation suggests that we should adjust our overall approach or our elasticities assessment, we will of course do that, and we will set out the reasons why.

One element to be cautious about in looking at the data and the outturns as and when we get them is that there will inevitably be a delay. It might not be in the next year but in a subsequent year that we get the detailed information. Disentangling the likely behavioural effect from the change in tax from all the other factors—economic change or wider changes in Government policy—will also be a challenging exercise, but we will do it as best we can. We will set out all the reasons and the assessments that we make.

I was going to ask a couple of things beyond behavioural effects.

Does anyone else want to ask about behavioural stuff? I know that James Kelly is interested in some of that. Is your question more about the process?

Yes—my question is more on the process.

We might as well bash on. What you have been saying has reminded me of what Robert Chote said to the committee about forecasting being like a spot-the-ball competition but somebody is always moving the ball. No matter how much we want to make this an exact science, we cannot and will never get it exactly right. The reality is that it will never be bang on.

I am grateful that you are saying that, as well as us.

It is the reality. We all have to remind ourselves of it, being so close to the process. You are making the best judgments that you can make: your assumptions are based on the information that comes in front of you.

It is very reassuring to know that we are all getting it wrong one way or the other, at the end of the day.

You have said that there is an assumption that the higher-rate threshold will increase by inflation in the absence of any other policy change. Increasing the higher-rate threshold by inflation is itself a policy decision. Am I understanding the commentary in your update report? Is that a policy decision that the Scottish Government has informed you of, and will you work on the basis of that decision?

John Ireland will give you a more specific answer on this, but there is a distinction. There is official stated policy for this year, but we have to make assumptions about what might happen in the future to enable us to undertake five-year forecasts. We look to the Government to give us an indication of what it thinks its future approach will be simply to enable us to do the assessment. In terms of status, the official stated policy for this year is of a different order to the best assumption about what policy might be, going forward.

We need to make assumptions about what thresholds will be in future years, and we try to make them as neutral as possible so that we are not introducing policy changes. It is not realistic to leave assumptions fixed exactly as they are. We do not have such a tax history for Scotland, but in the UK the higher-rate threshold has, over the long term, increased in line with inflation. Even if there had not been policy decisions, there is statutory indexation in the UK that ensures that that happens. In discussion with the Government over time, we have established that the neutral starting point for us in assessing where thresholds will go over the next few years is to increase them in line with inflation. We would adjust for policy around the thresholds on top of that.

I will add to that, to be crystal clear. Paragraph 11 of our report says:

“Therefore, while not a policy, the Scottish Government suggested a set of assumptions for further years”.

The Government has made it clear that it is comfortable with our assumption, but it is not a Government policy.

I presume that it would be helpful for you if there was a policy intention. For example, if it became Government policy that tax policy should reduce economic inequality year on year, rather than that it should continue to give stability, would it be helpful for you if that was a stated policy that you could base your longer-term assumptions on?

It is helpful for our forecasting that the Government be as clear as possible about its future policy, whatever that policy is.

Okay.

I was also going to ask about the process and the protocol.

You have had a good crack at it, so I think I will let James Kelly come in on the process stuff first, then we will come back to you.

On the process, you published a report after the publication of the draft budget, and then the Cabinet Secretary for Finance and the Constitution announced changes last week. You have had, I guess, interaction in order to be able to produce your report. Can you describe the process between the announcement in the chamber last week and the subsequent publication of your report?

Yes. After the announcement last week, we were sent, by the Government, formal notification of the policy, including a description of the policy. That notification also contained some background information about other aspects of what had happened, including the expenditure announcement, and it said a little bit about how the Government intends to find the money to pay for the additional expenditure.

10:30  

After we had received that, we turned around our costing and sent the Government the preliminary results of that on Friday. On Monday, we sent the Government a near final draft of our report for fact checking. We offered the Government a challenge meeting so that it could discuss our costing with the commissioners. The Government thought that it had had enough conversation with us, so it did not take us up on that opportunity. Yesterday, we sent another version of the report for fact checking, and by about 2 o’clock we had a small number of comments back from the Government. We published at 3 o’clock.

That is helpful. You mentioned that a challenge meeting was offered, which the Government did not take up. Did such a meeting take place before publication of the first report?

There was a series of challenge meetings before publication of the first report. We published on the commission’s website a protocol describing the interaction with the Government. The challenge meetings looked at individual taxes and the economic forecasts. At various points we provided preliminary forecasts to the Government. It looked at the assumptions and came back with methodological points about how we had done our forecasting. That process is very clearly explained in the protocol.

In table 9 you outline a position where the end behavioural impact is minus £56 million. If the Government disagrees with that after all the discussions and challenge meetings, is it able to use its own figure or is it legally bound to use the figure that you calculate?

The Scottish Fiscal Commission Act 2016 requires that if the Government wishes not to use our figures, it must write to, I am pretty sure, Mr Crawford to explain why. The Government is able not to use our forecast figures, but if it goes down that path it has to explain why it has made that decision.

This will be my final question. If the cabinet secretary were to decide again to change tax policy, say the weekend before the rate resolution debate on 20 February, how would you be able to deal with that in terms of going through a process such as you have just described happened in the past week?

We have had preliminary discussions earlier in the process with the Government about eventualities like that. We have agreed a rough timetable with it for any further changes. The next obvious point at which the Government could change policy is at the rate resolution. If the Government intended to make a change at that point it would need to tell us in advance, and we would go through a similar process.

Patrick Harvie will follow up on the process stuff.

I was preparing for—believe it or not—a relatively consensual debate yesterday afternoon about some aspects of the budget process, and was looking at how it used to be done, when there was a six-month budget scrutiny process for what were much simpler budgets. I am thinking about my reaction when I read the protocol: a stage 1 vote on Wednesday; notification to the SFC of policy changes by Thursday; an amendment deadline of Friday; then, I presume, you work over the weekend on your analysis, which comes to the Government for a potential challenge meeting on Monday morning, followed by deadlines at 1 o’clock and 2 o’clock; and publication at 3 o’clock. It is a bit of a breakneck process, is it not? What can be done to improve that and ensure calmer reflection on what are not trivial questions about Scotland’s finances?

Through the process that John Ireland described we had an extended period of engagement with the Government, in accordance with the protocol, before the December announcement. The details of our income tax model are well understood by the Government and are clearly set out by the commission. A change in the Government’s policy that was announced to us last week could be processed quite quickly through a model that we and the Government understand. My colleagues will say more. The work is done speedily and efficiently. It is not fair to say that it is “a breakneck process” or that people are working under stress. There would be application of an established model in respect of a change to previously announced policy.

I will add to what Alasdair Smith said. What is really important—as was the case with the report—is that we have sufficient time to do quality assurance. I am very grateful to say that we did not work over the weekend, and we made sure that we had sufficient time to do quality assurance. During the process, David Stone and his colleagues spent an awful lot of time checking the calculations. We checked with Government analysts, because they were using a similar model and we could compare notes on how we and they had done the work. We also have an internal checking process within the commission, and people who were not involved in the work checked our numbers as well. We are pretty confident that the quality assurance process that we have in place and the time that we have for it are sufficient.

On the wider point about the budget process, as you know there is a budget process review group that worked during the course of last year and produced a report that made a number of recommendations. Those include recommendations on the work that we are doing and about moving towards the SFC making a second forecast in late spring, in May. Our working through this, and recommendations being accepted or evolving, will change the whole nature of the process—it will elongate it.

That is helpful.

Politicians will still get in your road with last-minute decisions and make it difficult for you. The question for us should be to ask whether we can make decisions come quicker. That would probably be the right way to think about this in the round. I am not asking you to comment on that.

In your paper at paragraph 18, you have a calculation of the impact on income tax liabilities. You come to the conclusion that taxpayers with gross incomes below £26,000 will have a reduction in their tax liabilities of up to £20 a year, and taxpayers with gross incomes above that threshold will have higher tax liabilities. The cut-off point that the Scottish Government has cited is £33,000. It would be helpful if you could explain why you arrived at a figure of £26,000.

When considering the costing, we compare what taxpayers would have paid in 2018-19, in our baseline assumption. I go back to our earlier discussion about how we grow the higher-rate threshold without a policy change, compared with what would happen with a policy change. Tables 5 and 6 above paragraph 18 show what the parameters would previously have been in 2018-19 and what they will be with the policy change. In that comparison, the cut-off point is £26,000. The £33,000 figure I believe is based on comparing what taxpayers paid in 2017-18 with 2018-19. That is where the difference comes from. I think that the Scottish Government’s published income tax fact sheet makes that distinction, as well: it has both sets of figures.

Willie—do you still have questions on forecasting or have they been exhausted?

No, I have a number of questions, if that is okay.

I will ask more about your five-year forecast, which you have illustrated in table 12 in your paper. We see that your estimate for tax take for the year ahead is £12 billion, and for year 5 it is about £14 billion. Can you explain why there is such a significant increase? That is a significant jump in tax take in the five-year period.

There are a number of reasons for that. One is that inflation’s impact will be an increase in the tax take. It will be a particular factor in the current circumstances. Another factor will be the increasing tax take through public sector pay increases. Thirdly, an increasing amount of tax will be received by virtue of the fact that, as thresholds change, there will be an increase in tax take built into the tax system. Through those factors and others, we expect the overall tax take to increase by the degree that we suggest. So, inflation is a big factor and the cumulative impact of decisions by the Scottish Government this year and last year also contributes. John Ireland and David Stone might want to add to that.

We are forecasting economic growth at around 1 per cent a year, which will also add to income tax revenues. As people’s incomes increase, income tax revenues will increase. The numbers are in nominal terms and include the impact of inflation, which also adds to people’s incomes. I think that those are the main driving factors. Population growth also contributes through growth in the number of people in employment. Many factors will lead to that increase over time.

I will ask about margins of error in forecasting. I know that it is difficult territory to get into. Are you more on the cautious side or on the optimistic side? As we go through the five years for which you have forecast, are we likely to see a divergence between your estimates and what the Government of the day might produce for itself?

As a general point I say that our central forecasts are our best estimate of what we think the numbers are going to be. One is always careful with the assumptions one makes, but the assumptions are not pushed downwards in order to be deliberately cautious.

On forecasting error, I think that people are quite used to seeing fan diagrams that show confidence in forecasts over time. Because the Fiscal Commission has not previously produced forecasts—this is our first set of forecasts—we do not have historical forecasting errors so we cannot produce fan diagrams. When we have enough information about forecast outturn, we will produce such diagrams so that the committee will have a much clearer sense of the confidence intervals.

We have—and we did this in the December report—produced a number of sensitivity analyses. One can look back to that December report and get a sense of the impact on the income tax forecasts of moving some assumptions.

No one else has said that they want to ask a question. I thank the witnesses very much for coming this morning to give evidence. We are very grateful for it: it has been helpful and has provided some clarity.

10:42 Meeting suspended.  

10:45 On resuming—