National Assembly for Wales

I'm going to turn to our witness, and welcome Robert Chote. Croeso mawr. A very warm welcome to the committee.

Thank you.

We're very grateful that you're here with us. It's an exciting time for tax devolution Wales when we're looking at our first tax rates. We've just been looking at that in detail and, of course, devolution of income tax in a year's time as well. So, we are grateful to you for helping us explore some of these issues and some of the work that you've been doing on the devolution of taxes and forecasting as well. If you're happy for us to start just with some questions and so forth—

Yes, fire away. My pleasure. 

Excellent. Can I just start, then, by obviously highlighting that we've just had the budget? Your office produces a substantial part of the papers underpinning the assumptions in that budget, looking at forecasting, and an element of devolved forecasting as well. I suppose the big picture at the moment, just before we get into some of the detail—the big picture stuff was the attention you brought on productivity forecasts and also the impact of Brexit. So, I just wondered if you could sketch for us your office's assumptions and forecasting in that area and perhaps some of the implications that we might need to consider in the Welsh context as well.

I'd be very happy to do that, and thank you very much indeed for the invitation. I think you're right, certainly in the sense that, if you look at the way in which the forecast for the devolved taxes has moved since March, essentially speaking, it's the changes in the UK forecast that are dominating this rather than changes in the specifics of the Welsh shares. Broadly speaking, on the overall forecast, what has changed since the March budget is that the economy has been growing a little more slowly than we had anticipated then. There was the expectation that things were going to slow down into this year, partly because of the rise in inflation putting a bit of a squeeze on consumers. That has come slightly earlier than we anticipated in March and slightly later than we anticipated in the previous November, but the broad story is there.

What's been more significant in the forecast is the judgments we've made about, as you say, the medium-term outlook for the potential productivity performance of the economy and, therefore, for potential GDP growth, the potential size of the economy more broadly. That matters because the smaller the potential size of the economy, the smaller the pool of spending and income that we tax and, therefore, that implies a weaker outlook for receipts. The reason for the judgment on productivity—. I should say at the outset that it's nothing to do with a specific re-evaluation of the likely impact of Brexit. It's much more a question of looking back at the continuing puzzle of why the reality of productivity growth has been so much weaker over the period since the financial crisis and, arguably, since a little bit before the financial crisis, relative to the 30 or 40 years that preceded that. So, we've seen productivity—which is, in the way that we look at it, the output per hour that workers work—rising by 0.2 per cent, 0.3 per cent a year over the last decade compared to more than 2 per cent a year on average over the preceding four decades. 

We have always assumed, and this was common to most other forecasts and to the Treasury's view when the financial crisis hit, that you would never make up the lost ground and have a period of very rapid economic growth that got you right back to the path you would have anticipated if the financial crisis had never happened. But we had always assumed that the rate of growth in productivity at least would get back to historically more average rates—not immediately, but after two or three years. That was a judgment based on the fact that most of the explanations for weak productivity growth that were put forward initially were related to the financial crisis and therefore were inherently the sort of things that you might expect to have a temporary effect. But, as time has passed and as the financial crisis has receded further into the past, those explanations inevitably look less powerful for why productivity growth remains weak. It may have been right at the time, but it's hard now to be saying, 'Well, there are particular problems in the financial sector, firms are clinging on to workers because they think it's a short-term downturn.' That might have worked as an explanation a while back, but it doesn't now.

So, the view we've taken is essentially to assume that productivity growth will pick up from its current rates but not to anything like the pre-crisis historic average by the end of our forecasts—not to rule out that it could do that beyond that horizon, but essentially, what we've done is ended up with a judgment that you get productivity growth improving to somewhere that's about halfway between the weak performance of the last decade and the stronger performance of the previous three to four decades. There is not, as with most people doing these forecasts, a great deal of science to that, because it's very hard to explain why this productivity puzzle has persisted as long as it has.

One striking thing is it's not unique to the UK. If you look at the changes that the Congressional Budget Office has made to its forecasts for potential productivity growth and potential output growth for the United States, the overall picture is stronger, but the surprises relative to expectations have been very similar. So, that's the judgment that we've taken on that, and that has essentially dominated the fiscal forecast, implying, despite a somewhat stronger than anticipated starting point, weaker growth in most tax receipts over the course of the five-year period, with consequent implications for the devolved elements, because you're dealing, again, with labour income for income tax and activity in the housing market for land transaction tax, obviously. So, there are common features to that, but that's basically the big driver of what has changed in our forecasts in November.

Okay, thank you. I think there are a few questions coming in on this. Mike Hedges first and then Neil Hamilton.

Just two quick points. Don't you think part of the problem is the relative cheapness of labour? And the other one is that we've got a lot of movement into sectors where it's very difficult to increase productivity. In the care sector, for example, increased productivity is not necessarily a good thing. If you've got people visiting and having 10-minute visits instead of half-hour visits, productivity has trebled, but the quality has disappeared. Do you see that as part of the problem? In making motor cars, for example, which we used to do a lot of in the past, increasing productivity was easier to do—bringing in automation and reducing the amount of labour needed to do these things. But do you see that part of it is the movement into sectors where it is difficult to increase productivity?

I think you're absolutely right. Things like relatively labour-intensive sectors—social care would be a good example of that—are likely to have, as you say, less potential for productivity growth. The sort of documentary stories about being able to introduce robots into this area—we're some way away from that. I think it would be hard to explain much of what has been a really decisive change in the historical pattern, from the pre-crisis to the post-crisis period, as being the consequence of a really major shift in economic activity in those sorts of activities, but it is certainly a reminder that there are parts of the economy where it is inherently more difficult to achieve productivity increases and those may be growing in importance. But that's not going to explain very much of the huge shortfall relative to where we would've been on the pre-crisis trajectory.

The idea that, as you say, wage growth has been weak and therefore labour has been relatively cheap—that can clearly be part of the story. Of course, there's a question here that the causality of that moves in both directions, because one of the arguments will be that the reason labour is as weak as it is is because productivity has been weak, and fundamentally, the working assumption is that you would expect to see real wage growth broadly moving in line with productivity growth over the long term. But of course, the causality can work in the other direction as well. A particular reason is that, at the moment, in addition to the relative prices of labour and capital, in situations where people are uncertain about the outlook—so, for example, those firms that would be affected in different ways by different potential outcomes of the Brexit negotiations, or they may be uncertain about things for other reasons—then there can be a temptation then, if you are going to expand output, to do it by hiring more people, rather than buying or building more machinery, because if you've made a mistake, it's less costly to reverse it if you've taken people on than if you've made some large irreversible capital investment. So, that story could be there as well. So, both points are well taken.

Neil Hamilton.

Isn't there an inherent conceptual problem in relation to productivity? Just to extend the argument that Mike has started. We've moved away from manufacturing widgets into a predominantly financial service dominated economy, and whilst you can measure profitability, measuring productivity in these areas in statistical terms is actually impossible. You're pursuing a mirage. At the moment, for example, bitcoin miners must be vastly more productive this year than they were last year, but does that mean anything? So, is there any point in calculating or attempting to calculate productivity figures at all, because they can't really tell you very much about the potential profitability, necessarily, of the businesses in which people are employed, and, therefore, you can't extrapolate from that the potential revenue that might be raised from the taxes that the businesses that employ them will be paying?

Well, you're certainly right that there are sectors in which just simply getting to the point of measuring the output, which is obviously half of the equation in terms of measuring that—. Indeed, one explanation for the relatively weak performance of recent years may be that the measurement techniques used to measure value added and output in financial services may have been overstating the contribution prior to the financial crisis, and, therefore, there's more of a normalisation rather than a decline from that. The measurement of value added in financial services is, typically, a relatively rough and ready judgment between the rates at which people are borrowing and lending. So, that's a good point.

That said, I think the fact that, again, you're trying to explain this relatively dramatic change from what had been a fairly consistent pattern over many decades to what has followed subsequently—. Pinning much of that to the fact that the measurement problem has got substantively more difficult since then—it's harder to nail it down to that. I think the Bank of England has thrown a lot of person power at this sort of analysis, and I think their estimates are that measurement difficulties can maybe explain about a tenth of the gap that you're ending up trying to explain.

You also point to a broader issue of the difficulty of identifying output growth, and are we missing something in terms of whether the economy is just generating more growth and we're not measuring it. That matters, obviously, quite a lot if you're thinking about the living standards and the goods and services people are consuming. From the perspective of being a fiscal forecaster, it's less helpful, because what we're not seeing is tax revenues coming in far higher than anticipated, implying that there is some taxable activity out there that we are not capturing. So, it may very well be, for the reasons you describe, that there's even greater difficulty now in taking the total amount of cash activity in the economy and dividing it between how much is a genuine change in the volume of goods and services versus the prices for those things, but what you're not seeing is, as it were—we are puzzled by the fact that tax revenue is ploughing in and yet we can't measure something that is generating it. So, even if that is part of the story, and the statisticians come back in 10 or 20 years' time and decide that the split between real GDP growth and whole-economy inflation looks very different, they're not going to come back and say, 'Actually, there was a whole lot of cash under the carpet that you didn't spot at the time or Her Majesty's Revenue and Customs didn't collect.' So, that problem it doesn't resolve to the same degree.

I see.

Okay? Thank you. We'll move on with Nick Ramsay, then, thanks.

Thanks, Chair. You've touched on this already: how have you improved methodologies for forecasting Welsh devolved taxes since the March forecasts?

In depends, as, obviously, the issues arise with different taxes. What you will tend to see, obviously, is, within the income tax side, you're moving to new bases of the survey of personal incomes, which is basically the information set you have available to work out what the shares are for Wales versus the rest of the UK, and to draw conclusions about that, going forward. Those are updated year by year, which, in one sense, is an improvement, because you're dealing with more up-to-date information. However, one should caveat that: the most recent SPI on which data was based—the 2014-15 one—is one where the difference between that and the actual outturn in terms of the revenue that you've brought in has been rather larger than it was in the previous year. So, I would live in hope that moving to 2015-16 will actually be more of an improvement than the move to 2014-15. Hopefully, that will be a bit different.

What we don't know, of course, is the specific Welsh difference between what that survey shows and the rest. We know that, in UK terms, it has been somewhat less of a good guide in that year. Whether that's a pattern across the UK is not entirely clear.

So, that's something we'll only get with time, is it?

Well, that changes from year to year. In some years you get one that is a good reflection, and in some years you get one that is a less good reflection.

I think, on the income tax side, looking forward as well as what we've seen recently, one set of improvements you will see is that we are at the moment relying heavily on this survey, but we will move forward to a period at which Welsh taxpayers are flagged explicitly, where they didn't have to be before. Therefore, you will get a much more robust estimate of the Welsh share, which would be an improvement on that side. More generally on income tax, which is something that I think holds out the promise of improvements UK wide as well as with Wales, there is the continued use of what's called real-time information, which basically means that employers have to interact with HMRC on a much more real-time basis pretty much every time they're paying people, as distinct from monthly or annually. We're already starting to see some useful information on richer and more up-to-date figures on differences in wage growth in different parts of the income distribution, and that may be possible geographically as well, but we're still at early stages there. HMRC is reluctant at this stage to be publishing a great deal more of this information until they've had a chance to quality check it. But that's quite useful and potentially important.

I think another area where we might look for improvements going forward would be on tax-motivated incorporations. One of the puzzles of recent years has been the fact that—and this is a UK-wide story—you have seen more people becoming, as it were, single directors of their own companies and choosing, essentially, to be paid dividends in that form rather than being employees or being self-employed. Now, that may be partly because, in the modern world of work, more people find it convenient to organise their affairs in that way, but there are also tax advantages to doing so. I think it will be useful—there is work under way with some Welsh-specific modelling to see whether the importance of that differs for Wales from the rest of the country. That work has already been done for Scotland, and it didn't actually show a terribly different pattern from the rest of the UK, but I don't think you should rule out the possibility that the story for Wales could be different. It's worth doing that work and I think next year we should hopefully see the fruits of that.

But they've obviously got the Fiscal Commission in Scotland. We don't have that here. In terms of the Government's methodology and the role of Bangor University in evaluating that, are you supporting Bangor in that process or are they doing that independently?

They're doing that independently. They've been taken on to do that job, and I've looked at the output that they've done. I think that's a very sensible arrangement, to have some sort of external oversight there. Obviously, there are issues for the Welsh Government about how they want to organise that over the longer term, but I think that's sort of additional scrutiny. Clearly, the Scottish operation is a much more labour- and resource-intensive exercise, and they've chosen to go down that route. They're doing quite a lot of forecasting in-house that we would instead deploy the tools of HMRC, for example, to do. At the end of the day it's still our forecast, and they have to implement the changes that we want to implement. In the Scottish context, more of it is happening in-house rather than them relying on a long-established Scottish tax authority, because there wasn't a long-established Scottish tax authority. 

A good reason not to. Finally from me, what work is being done with Welsh Government to prepare for independent forecasting of devolved taxes for Wales, and what are the timescales for such arrangements? 

Well, we're talking to them on an ongoing basis. Clearly, the issue about what role they would like us to play and what role we would be able to play in the steady state once the devolution has continued further, and how that fits in with the budget timetable, is an important issue. The finance Minister has said that he's, I think, keen on having or exploring whether the OBR should play that role rather than a new independent body, and within our—

Would you be happy to?

Within our resource and time limits, we would be happy to play that role. There are particular challenges for us around the coincidence of the timing of your draft budget timetable plus a whole set of other documents we have to produce in the UK, because it coincides with the full set of national accounts, the first set of decent UK-wide public expenditure numbers we get in. So, I think the chances of us being in a position, even with more resource, to be able to present you with a fully bells-and-whistles publication ahead of a draft budget timetable is quite difficult, but I think what would be possible would be to provide scrutiny support, and then obviously to return to that more comprehensively when we do our November forecast and you get the equivalent of the devolved taxes publication, which could probably be expanded somewhat.  

But basically, capacity issues aside, if they could be satisfied, then you'd have no problem with that.

In principle, we're happy to play that role. As I said, there is an issue I think for the Government and for you as an Assembly as to what degree of scrutiny and information you need when, and, as I say, what I wouldn't want to do is to over-promise our ability to deliver, just partly because of this coincidence of timetabling. You know, early October is grim for us as it is in terms of other stuff that we can only do at that point, and just providing some extra resource only goes so far because you basically need the same sets of experts over the same sets of people, so having a dedicated person or two people to do that is not the right solution to the right problem.   

Can I just ask about that? Setting aside for the moment whether the Welsh Government decides to develop something independent, as the Scottish model, or use your office in a different way, you remain a UK body so you remain, I would assume—to be fair, you've already told us—interested in how tax devolution works throughout the UK, and particularly, of course, with income tax coming into force in Wales. You've already talked about how HMRC will be flagging up Welsh taxpayers much more accurately than the estimates that we have now, which are more of a survey kind of estimates. Do you expect to have more real information over the next year from HMRC from this process, to the extent that you are producing forecasts for Wales and that will allow you to  produce hopefully more accurate forecasts in that regard?  

I think one has to be realistic about the amount of time you need for the system to bed in and, of course, one important issue in terms of how difficult the forecasting task is is to what extent—this is true for devolution wherever it happens—or the degree to which the policies are differentiated, because, clearly, the more they are differentiated, the greater the challenge it is to identify what changes in behaviour, which can go in all sorts of directions, may result from that. You can see that already to a degree with the way in which LTT versus LBTT in Scotland versus SDLT in the UK. You're seeing the systems moving apart somewhat, and the greater those sorts of differences, the greater the challenges get. Clearly, there will be some areas where, if rates and rules and allowances move apart, it might not have a great deal of information, but if you're doing it in a way that affects relatively mobile, relatively well-off people who are well advised and flexible, then that will inject a greater degree of uncertainty there. 

As I say, over time, we have reasonably high hopes for real-time information on the income tax side as something that would be useful, but, at this stage, you don't want to place too much weight on the signals that you're getting out of that which might, for example, be showing different Scottish/Welsh shares from the mainstream data, but I wouldn't want to tread too much on the ice for fear of falling through the lake prematurely. 

It might be a good time of year to have that analogy. 

Chair, can I just—

Sorry, yes. I'll bring in Jane Hutt. 

In a sense, this is an important point where you can perhaps make the case for wanting to be able to engage in this way because, obviously, the Scottish Government decided not to set up a resource-intensive Scottish Fiscal Commission, but it would just be helpful to know: do you feel that you could make a case for being able to—? I know there's a resource capacity for you, but we want to know what's best for Wales, and whether the OBR is the right vehicle and outfit to do it.

In the end, that's obviously your choice to make. Certainly, many of the methodological and data issues that you're dealing with are common ones, and we have an expertise that we can deploy in that area. The way in which we have been approaching the task of the devolved taxes so far is obviously to try to—and, again, this applies to Scotland as well—have the discussions of the forecasts while they are being put together done in such a way that we can have, and we do have, a good interaction with the Welsh Government, with the Scottish Government and the Scottish Fiscal Commission. Because clearly, one of the advantages from our—. We will continue to have to forecast the whole UK, which includes all of those elements, but the devolved administrations, governments, whatever their institutional format, are clearly going to be able to devote more time and interest to providing additional information that we can use and deploy. And, in some cases, obviously, it's very specific. Take landfill or investments in lumpy infrastructure; if you suddenly have a new plant or something like this, this can make quite a difference. And that's the sort of thing where local information is indispensable in doing that, and obviously, if we do end up doing this task, you would obviously want to organise yourself in such a way that you continue to exploit that information as best you can. 

Obviously, as well, it depends on what the Welsh Revenue Authority—what sort of capabilities and degree of information and additional data they will be able to bring in as well. But that would affect both. If you had an independent institution and us doing it, if you had a revenue authority that is generating more data, then that can affect the methodologies you use. So, in the case of LTT and LBTT, the choice of methodology—the way we do the taxes in Scotland—has been based on the lesser availability of very detailed microdata. But, as time goes by, and the revenue authority there develops its work, you end up with more of that and maybe you want to move your methodology back to something that can exploit that. And I suspect the same issues would arise with Wales as well. So, as I say, if we're asked to do this, and within those constraints, we will do the best we can, and we will exploit information on the ground as best we can. But, at the end of the day, it's for you to judge whether that's preferable to—. As I say, what's in Scotland has ended up being a—I don't want to say over-resourced, but a necessarily well-resourced operation. 

'Resource-intensive' you said. Thank you. 

Neil Hamilton.

When Mark Drakeford, the Cabinet Secretary for Finance, gave evidence to us on the draft budget a few weeks ago, he said that the OBR was looking to improve the Welsh-specific data available for the new purposes that we need it for in the era of devolved taxes. So, can you perhaps tell us a bit more about how your Welsh-specific data has become more available, how you're using it, and how accurate it is?

Well, as I say, I think, obviously, the story varies from tax to tax. On the income tax side, the big changes will be flagging and the real-time information—real-time information being a UK improvement with spillover benefits—and on the other side, more Wales-specific. On the tax-motivated incorporation, again there's an opportunity to drill down more into the specifics of what information we have on how that operates there. Now, this is an important one, because if there is a growth in tax-motivated incorporation, you're essentially reducing income tax receipts, but increasing corporation tax receipts and dividend tax. And it's the income tax, of course, that is the Welsh receipts, but the other is reserved to the UK. So, it matters in that context, which is why, I think, it's certainly worth us and HMRC investing the effort in at least exploring whether there appears to be a different pattern. It may turn out, as with Scotland, that there isn't, with their score on that. On the other ones, I think, with the availability of explicit additional information—. So, on landfill tax, we don't have particularly good administrative data. The reporting of this is often at an England-and-Wales level, so the degree of specificity there is not all—

It's quite a small revenue raiser as well.

It's quite a small revenue raiser, which again comes back to the results issues. There are cost-benefit analyses to be done about how much is it worth the time and effort. Is it worth getting more information, if you're not going to make an enormously important adjustment at the end of the day? The irony, in one sense, is that, for the land transaction tax, you actually have reasonably good administrative data, relative to the others, but it's probably the one where the dangers of large forecasting errors are amongst the greatest, simply because of the nature of the sort of tax it is. With the land transaction tax, people transact properties relatively infrequently. It's not a constant stream of income. You are affected both by your ability to predict prices and transactions. Often, one would find that if you have differences of view over forecasts, it's often down to the assumptions about transactions rather than the more high-profile ones about price, and transactions are often forecast on the basis of: you have some assumption you make about, in the long term, how often a house will turn over, and you're a way away from that, so how quickly do you go back to it? So, there is scope for improvement, but it's not necessarily, ironically, the case that those taxes for which you have the best locally specific, administrative data are the ones where the forecasting task is easiest. 

To come back to the cost-effectiveness point referred to in relation to the landfill tax, do you have a view on what further data it would be worth having and would be cost-effective to obtain in relation to Wales to support your modelling and forecasting of devolved taxes in the future?

Clearly, in situations where the devolution has been in place and you have had a run of years and you are able to draw more conclusions from outturn data as distinct from relying on what you get from the survey, that will come over time. So, it's an improvement, and it will be new data, but it's not something new you're asking for, in that sense. We can certainly come back to you if there are any specific asks. I don't think we have anything where we've said, 'You've really got to come up with this'. But if we have any particular asks, I'll certainly get back to you on that point.

Good. Thank you.

Okay? Back to Jane now.

On the question about how you're going on to the devolved taxes now and the changes in your forecasts, I think it would be interesting if you talked us through that in terms of the changes made to the November forecast for LTT for residential and non-residential. You commented on it in your earlier opening remarks, but can you talk us through those impacts?

Absolutely. Because the Welsh Government forecasts were necessarily and sensibly based on the March economic inputs, I don't think—. If you look at the difference between the forecast that the Welsh Government has produced and the forecast that we produced in March, there are differences, but none of them look to me to be hugely glaring or things that you would be surprised by. The patterns are broadly similar. There will obviously be some differences in the precise way in which particular statistics are treated or the information that can be drawn upon. But more broadly, I think the difference between our November view and the Welsh Government's March-based view is more down to changes that have taken place to the UK forecast as a whole. The key ones there, I think, are the fact that if you assume weaker potential productivity growth and weaker actual productivity growth, the main consequence of that for the economic forecast—you're obviously bringing down overall GDP growth—is that it shows up in weaker growth in earnings. So, that is a major change and basically means that, if you take the example of income tax, receipts were higher last year than anticipated, so you end up with a boost in the early years, which then erodes and moves negative as weaker earnings growth cumulates over the period.

Another change that is relevant is that we had a new set of population projections from the ONS in October, which is one of the explanations for why our forecasts have moved between March and November. One implication of those is that population growth remains weaker in Wales and Scotland than it does in the rest of the UK, but not to the degree that the previous set of population projections suggested. So, you have some positive gain as a result of the—if I can try and find it—. So, if you compare our March and November forecasts, you're getting a boost of £20 million to £30 million a year on the changes in the population projections. You end up, as I've said—. If you think about the underlying forecast for non-savings, non-dividend income, which is dominated by wages and salaries, which is hit by productivity, it's higher in the near term, but then the gains go and it moves negative in the longer term. Then, there is another set of changes that we've made that affects the Welsh share specifically, and that arises from having re-costed policies that have previously been announced at a UK level.

One consequence of that is that we now estimate that more revenue is going to be raised from measures that take revenue from relatively well-off people—taxation of partnerships, for example. There are a higher proportion of those people in England than there are in Wales, so that has an automatic effect of reducing the share. Similarly, changes to the costing of things like the marriage allowance—while it's a UK-wide policy, if you change the cost of that, it can move the share around. So, if you look at why we've changed the forecast, say, for 2021-22 for income tax, between March and November, overall it's down by £40 million. The population change is actually in Wales's favour. It has pushed it up by about £40 million, and then the underlying changes to the UK forecast and these Welsh-share specific factors have each knocked it down by about £40 million. You have things moving, as ever, in complicated and different directions, but with, at the end of the day, a relatively small overall effect. I think it's fair to say that the measures announced by the UK Government in the most recent budget don't really affect the Welsh picture terribly much, relative to anybody else anywhere else.

Your report states that the impact of those changes to UK stamp duty taxes will reduce the Welsh LTT in 2018-19 by less than £0.5 million. One of the issues that the Cabinet Secretary has raised with us is the impact of forestalling. He's looked at it from two perspectives. Do you think this is significant? Have you taken that into account in your modelling, particularly in terms of putting it off and behaviour changes?

We always take forestalling carefully into account, because it can move quite a lot of money across years quite easily. But I think in this context it doesn't look quantitatively significant. So, you're not talking about a very large change at all. Forestalling is much more—. It can matter when you have significant changes in the structures that are moved from the slab to the slice. It's much more an important issue if you have anything that affects taxation of individuals who can choose when to take their income as dividends on the income tax side, but it's not been enormously important in this context. 

Obviously, one thing that we will have to do is to come back in the March forecast and look at the Welsh Government's announcement on not, as it were, proceeding with the first-time-buyer relief, but instead the alternative change to the tax-free allowance and the rates there. But again, it's not clear. We haven't done a formal costing of the proposal here. We'll do that in March, but it doesn't look as though you would be opting for a dramatically more or less expensive option than if you had proceeded with the UK proposal.  

Well, you saw the announcement on Monday by the Cabinet Secretary.  

Indeed, and the sort of implications you would get from that sort of policy are—you know, it will increase the concentration of receipts you get from the relatively top end, because obviously you're taking people out at the bottom and increasing the marginal rate somewhat. So, that might imply a greater degree of uncertainty and change in the forecast. But, on the other hand, you're avoiding some of the incentives to evade or avoid that are created by having to specify whether you're a first-time buyer or not. So, as ever, swings and roundabouts. 

Can I specifically ask on that, because I think your initial UK report had something specific to say about the impact of a first-time buyers' relief on a UK level, which had some attention at the time, but, of course, the Government in Wales is proposing that we don't go down that route at all but have a more general relief for everyone. Are you able—not today, obviously, because you've only just heard, but, by March, would you be able to actually model that, or is that still too subtle to be able to model even at that stage? 

Well, when we get to the March forecast, we will look at the same sort of—. One of the things you take into account is whether tax changes of this sort are likely to generate price changes that then have consequences for the receipts. 

Which is the case with a—you argued that a first-time buyer relief does have a price implicaton.  

Exactly, and there is a tendency with property transactions prices that when you have changes in the structure, to use the economist jargon, the impact is capitalised upfront in price changes. So, unlike a lot of policy in the property development area, where you don't quite know what impact it's going to have and it can take ages before you know whether the policy is implemented and what effect it's going to have at the end of the day, these are the sorts of changes where, again, the magnitudes may be hard to estimate, but there are good reasons to expect, if there is an effect on price, it will be quite swift. So, in the case, as you will have seen, of the UK policy, we assume that there would be a reasonably substantive price response that, basically, the change in the flow of the tax payments is crystallised 2:1 into the price, because you're not only benefitting the next person who transacts the property, but future first-time buyers who transact the property, and that comes in as well. So, clearly, you don't have the same issue with the changes you described. As I say, what you do have is nudging the system more in a progressive direction, which obviously—you know, it's the decision you want to make about the distributional consequences of this, but it is also the nature of progressive tax changes of that sort that more of your eggs are in the relatively well-off basket, and you're therefore in future more affected, positively or negatively, by developments in the economy or elsewhere that move the top end. 

I think that's a good quote that the Cabinet Secretary would be interested to hear. Just one final point and question from me: are there any significant differences in our modelling arrangements— forecasting modelling—between OBR and Welsh Government?

Certainly, none that I would have any concerns about. I don't think so. Over time, I think there will be an issue, as all the systems settle down, as to how best we do SDLT, LBTT and LTT on that, of to what degree you are able to draw upon basically a very detailed micro distribution of house prices and therefore how you assume that evolves over time, versus a slightly more rough and ready approach that basically draws a shape that you think is consistent with what the underlying pattern would look like, if you knew what it was. In the case of the data availability, we and the Scottish Fiscal Commission use that more sort of rough and ready approach, rather than the micro approach that we currently use for SDLT in England and Wales. As time goes by, it's not clear that we and the Scottish position will still be doing that once there's more micro data available there and whether we go through the same sort of modelling transition here. But in neither case have the numbers turned out to be terribly different. It's as much about how tractable and useable and flexible these things are to do the analysis. But I think if you look at the differences between the November forecast that we've produced and the most recent Welsh Government forecast, they're primarily explained by what's changed in the economic determinants between our March forecast and our November forecast, not that we're taking a dramatically different approach to the Welsh Government. 

Thank you.

Nick Ramsay.

Can I just ask a mischievous question, which I'm sure Robert will appreciate? You're within your rights to say it's policy so it's nothing to do with you. 

It's a mischievous answer—[Laughter.]

You mentioned the Scottish Fiscal Commission a few times and you obviously work with them. The decision's certainly been taken in Wales for the foreseeable future not to have a Welsh fiscal commission. Would you appreciate having a Welsh fiscal commission to sound off or to work with, or do you think it can be done with—? I'm not really asking whether you want one or not, I'm just saying: would you appreciate being able to work with one if there was one in the future?

I think we—. At the end of the day, I think as long as there is a good working relationship and flow of information and trust, that's what matters. 

It doesn't matter what the structure is.

And we have had, in the Scottish context—you know, there's a tale. We are kind of like each other and therefore there's a natural commonality of purpose, et cetera, but we have also had a very good working relationship with the Scottish Government both before and after they were there, as we do with the Welsh Government. Clearly, whatever the institutional arrangements are, what you don't want to do is to get into a world where people don't want to share information or a degree of trust. Obviously, at one level, you have to respect some confidentiality. We cannot tell the Scottish Fiscal Commission, the Scottish Government or the Welsh Government what we know about what Philip Hammond is going to stick in his budget when we're talking about the substance of the forecast beforehand. That's inevitable and there's no reason why—I can quite understand in the context if a devolved administration didn't want to tell us something that they were thinking of doing in policy development. But once you get to the point at which we all have to understand what the hell's going on, the more good working relationship or the exchange of information there is the better, and so far we haven't had any problems on that in any part of the UK. So, that's, I think, fundamentally more important than who precisely—you know, what's on the nameplate when you're dealing with them.

Mike Hedges.

Two quick questions—on land transaction tax, of course. When the economy goes into its next recession, all your predictions—Welsh Government's predictions—are going to be going that way and the thing is going to go that way, isn't it? So, I'm a great believer in looking at predictions but treating them with a level of scepticism, because we will have a recession at some time. The other point I would make is—

Can I just there pick up the first one you've just said? You're absolutely right, history suggests that there is an evens chance of a recession in any five-year period looking forward. It is the nature of forecasts that they are predicated on the assumption in the UK context that the Bank of England sets monetary policy in order to keep demand and supply and the economy roughly in line. So, you don't tend to predict recessions until the roller coaster is starting to come over the peak—that's inevitable. But what it does do is that it means that it's very important from a policy-planning perspective to think about the risks and not just the central forecast. So, we produced a horrifyingly large document on fiscal risks back in July that obviously makes this point and points out the issues that the economic cycle can have. The other thing that we did in the back of the main report that we did at the budget time is, given the importance of this judgment that we've taken on productivity growth, to say, 'Well, what difference would it make if the last 10 years is actually the new normal and we never get back to anything like historically decent productivity growth, or at least not over the next five years, and what happens if we snap back to something that's more in line with the historical averages?' So, from a policy perspective, no-one should ever bet the farm on any particular economic forecast turning out to be correct. You should be thinking about what could go wrong. There are some things that could go wrong with the forecast that don't matter very much in pounds and pence, in terms of revenues; there are other things that matter a deal more. It's not just, also, the ups and downs of the economy, it's the composition of economic activity as well. Different sorts of income and spending are taxed at different rates, so, if the mix between wages and profits is different, if the mixture between consumer spending and business investment is different, that can matter. So, it's not just how big the economy is at any given moment that's a puzzle, it's lots of other elements as well.

Thank you. That leads me nicely into my last question. You talked about risks around the central tax forecast for land disposals tax. Even if you were 10 per cent out, it would only be just over £2 million in 2021-22; if it was 20 per cent out, it would be just over £4 million in that time. Are those numbers so small that they are really de minimis?

You're better placed to say what is material as far as the Welsh Government's planning and the Assembly's planning. What you do have to bear in mind is that property transaction taxes are inherently volatile and uncertain because of both the prices and the transactions elements.

No, land disposals tax I was talking about.

Oh, sorry. So, on landfill—

Yes, landfill tax.

Yes, well, there's a different set of issues there. There are uncertainties around the quality of the data that you'd base the central forecast on, and, as I say, we don't have great administrative data on that score. You can think of some risks arising, for example, from—. The Welsh Government has included, as we haven't, an assumption about the potential taxation of illegal activity. That has been taken on board in the UK-wide context, but we've given it what is called a very high attrition rate—you assume that you might get some revenue from that to start with, but most of it has disappeared after a relatively short period of time. So, you're taking what you might think is a sort of prudent approach to not banking all of that and spending it straight away.

As I say, the other issue with this—with landfill—is, as we were discussing earlier, if you have the introduction of new infrastructure, that can make quite big differences over time. So, if a particular plant or incineration activity comes in earlier or later than expected, or not quite as efficiently as expected, all of that can move things around as well, so the same principles apply.

The point I was making is that, if you're 20 per cent out on that, you'll be talking about £4 million out of a Welsh Government budget that's somewhere over £15 billion.

Indeed, and what you're dealing with, with the devolved taxes, is a relatively small number with a widely different—. Clearly, income tax is much the most dominant share. The transactions tax is by no means insignificant, but also probably the most volatile and uncertain of them. Then the other two, while being more certain, are considerably smaller—or the other one, and then the aggregates levy would fall into a similar category if it gets over the line.

It's even smaller, isn't it, the aggregates levy?

Exactly. So, yes.

Okay, I think on that we will conclude. We look forward to hearing more from you in future around income tax, I think is the conclusion from that. Thank you very much for coming to the committee this morning.

Well, thank you very much for the invitation—it's a pleasure to come, as always.

We're very grateful to see you in person as well, and wish you a very happy Christmas—Nadolig llawen, as we say.

Thank you very much indeed, and to you too.

Nadolig llawen to everyone else on the committee as well. We shall see you in this committee in the new year. Okay. Diolch.