Finance Committee, 16 Jan 2007

Meeting date: Tuesday, January 16, 2007

Official Report 283KB pdf

---

“Government Expenditure and Revenue in Scotland”

I move to the slightly more interesting second item on the agenda, which is evidence on "Government Expenditure and Revenue in Scotland", colloquially known as "GERS".

It would be hard to imagine a more appropriate day than the 300th anniversary of the union for the Scottish Parliament to be debating the essential character of the current financial relationship between Scotland and the rest of the United Kingdom. We will soon take evidence from a range of independent experts in a round-table session, which means that members and witnesses will be interspersed around the table. That is why our seating arrangements are a little bit different today.

We begin by taking evidence from a panel of Scottish Executive officials. I welcome to the committee Dr Andrew Goudie, who is the chief economic adviser and head of the Finance and Central Services Department; Fiona Robertson, who is a senior economic adviser and is head of office for the chief economic adviser; Sandy Stewart, who is head of office in the office of the chief economic adviser; and David Stewart, who is the head of the finance expenditure and policy division.

I invite Dr Goudie to make an opening statement.

Thank you. We welcome the opportunity to meet the committee today and to respond to your request to discuss the methodology that was adopted in compiling GERS. As the committee knows, we have sought for many years to provide a significant amount of detail about the underlying methodology, although I appreciate that we can never cover all issues that might be of interest. That is particularly true in the case of GERS, which has been a complex document to produce. Members will also be aware that we have sought over the years continually to develop and improve the document and the approach. That necessarily implies a degree of change, but we feel nonetheless that the improvement in the data is important.

I have three observations to make by way of introduction. First, ministers have long accepted the need for a systematic understanding of the financial flows of expenditure and revenues in the public sector, which is fundamentally what GERS continues to contribute. Secondly, it has also long been recognised that, for a variety of reasons, it is necessary to use a combination of accounting and administrative data, and estimated data. That will continue to be the case. We will also continue to rely heavily on UK data sources in estimating the expenditure and revenue data. Thirdly, GERS is produced by Government economists and statisticians to the agreed standards for national statistics.

I do not think that I need to add any more by way of introduction. The committee is familiar with my colleagues, who are also happy to answer questions on the more detailed parts of the GERS document.

I will kick off questioning on the committee's behalf. Probably everyone is familiar with the central allegation that has been made against the document, so it is important to ask about it on the record. To the best of your knowledge—as someone who has been intimately involved in the process for much of the period since the first GERS was published in 1992—has any minister at any time been involved with, influenced or tried to influence GERS?

I have been involved since the first GERS of 1992. Apart from a short period in the middle when I was away from Scotland, I have been involved in all the GERS documents, so I am pretty familiar with the process. We need to acknowledge that, as with all such Government statistics and Government economic work, ministers decide whether the work is to be undertaken. We also need to acknowledge that the broad scope of any work that we do is under the direction and approval of ministers: GERS is no exception.

That said, the fundamental methodology that is adopted for GERS—given the resources that we have—and the quality of what we produce have ultimately been my responsibility as chief economic adviser. I give the committee a total reassurance that, throughout that time, after ministers' approval for the work and their direction for the work to be undertaken were obtained, ministers played no part whatever in the preparation of the work. I will use as an example the past year, which is typical of the whole period. We provided ministers with regular progress reports—for example, on where preparations had got to and on when we intended to publish—but ministers played no part in the design of the methodology or in the work that underpins our work. The first they see of the document is always the final version.

As the committee's expert adviser's paper makes clear, the GERS methodology has been changed and improved in the 15 years of publication. Will you expand a little on that process and on whether any more plans exist for improving the quality of the data or the validity of the estimates that are used in the process?

It is true that we have been open to improving the approach throughout the document's life. We have always made a major point of setting out in the document the assumptions and the methodology that underlie the approach, with the explicit intention of encouraging debate about that. Over the years, we have made a series of improvements to the document. As I said in my opening statement, the balance is always between continually changing—and thereby breaking the continuity of the work—and wishing to improve. There are many examples of significant improvements that we have made to the document, to which we may return this morning. The work on income tax provides a good example of where there has been significant improvement since the latter part of the 1990s.

You say in the report that the figures on VAT

"should be treated with caution".

You go on to say that the figures on corporation tax

"should be treated with extra caution"

and you say that the net borrowing figure

"should therefore be used with some caution."

Do you agree that those are three substantial caveats on the fundamental conclusions in GERS?

It is important to see GERS in the wider context of the statistics that are produced in Scotland and in the UK as a whole. It would be wrong to see GERS as being a similar document to the UK budget, which is based fundamentally on accounting data that are often audited and are very precise. As I said, GERS is different. We use some specific accounting data, but we must inevitably also use a considerable amount of estimated data. It is probably more appropriate to see GERS in the same light as statistical work that we or other Organisation for Economic Co-operation and Development countries do on gross domestic product, in which a heavy degree of estimation is necessary, depending on the component that we are talking about. That is an important context for GERS.

A more precise answer to the question is that we flagged up the caveats and used the word "caution" at various points in the document to differentiate between the two types of data. I have always thought it important when we have such a mix—particularly in the light of the UK document—to make it clear to people that some parts of GERS are estimated. People need to be well aware of that. Therefore, as with other such statistical data, we include the usual caveats about sampling variation, if we are using sample surveys, and about whether we are using other apportionment processes, some of which do not necessarily give a precise number that would relate to reality if we had the real numbers.

There are caveats in GERS. What you have said is helpful, because it makes my point that great care must be taken with the figures. One change that has been made in the current edition of GERS is a reduction in the estimate of Scotland's share of corporation tax from 9.5 per cent to 8.1 per cent, which represents a pretty substantial change in methodology. That estimate carries the caveat that it

"should be treated with extra caution."

In the light of the substantial changes to methodology and the warnings that you have given about "extra caution", is the document interpreted with suitable "extra caution" by ministers?

I will take the question about corporation tax first, after which I will talk about the more general point. I will ask Sandy Stewart to say a little about precisely why the corporation tax methodology has changed, because that will help the debate.

I start by reiterating why corporation tax is necessarily difficult for work such as GERS. The natural unit of account for tax purposes is the key to having a precise figure for corporation tax for the UK. It is a reality that companies do not necessarily see any reason to produce accounts—informally or formally—for any part or region of the UK. I appreciate that they have no legal requirement to do so. Some might, in practice, produce accounts for cost centres and facilities that they have in different regions of the UK, but it is generally true that they have no incentive or obligation to produce accounts for one region. That is the fundamental problem for us in the estimation process. We have done much work on approaching Scottish companies to try to understand whether they informally hold data for Scotland for various indicators that we would find useful, but a fair generalisation is that the data from those sources are very poor.

Did you say that the data "are very poor"?

The data for Scotland that come from Scottish companies are very poor because companies do not produce data on that basis and have no reason to do so. As a result, we must use an estimation process that is based on the hard facts that we know about for the UK. Sandy Stewart can say a little more about the estimation process, after which I will return to Mr Swinney's other point.

The change in methodology for corporation tax resulted from changes that were made to the regional accounts, in which the UK splits among the regions the balance in national UK accounts. Two of the factor incomes that are used for that are profits and holding gains. We take the profits and holding gains for Scotland as a proportion of the UK total, excluding the extra-regio area. That becomes the apportionment methodology to separate out corporation tax for Scotland.

The regional accounts were improved for the blue book of 2006 through the inclusion of employment data from the annual survey of hours and earnings, which replaced the new earnings survey. That is a much better source of employment information and it has fed through the whole of the national accounts and the whole of the regional accounts. Its effect was to increase the share of profits in financial intermediation to London and the south-east.

It has also had the effect of slightly reducing profits on the continental shelf. Those two effects combined to reduce the Scottish share of UK revenues, less extra-regio revenue. That was a significant change and data advancement, but it was a one-off event that was based on a new survey that was used here for the first time and will be used until yet another improvement is made. We do not expect to see a change of that size year on year. The figures were backdated for 10 years and fed into the estimates of gross value added for Scotland and into the GERS calculations. I do not think that we will again see changes of that size in the estimates for corporation tax, using the same methodology.

I will address Mr Swinney's final point, which is important for us because it relates to the whole basis on which we do such work for ministers. I see my remit as being to oversee a team that produces the best estimates that we are able to produce with the resources that are available to us. I do not explicitly see it as being my role to instruct ministers on how to use the document—I would not dream of doing that. As members would expect and as in all our other work, we provide confidential advice on interpretation of the document, but it is very much for ministers to determine how they use it. We believe that the estimates that we make, given the resource base that we have for the work, are the best that we can produce. We are sufficiently confident about the methodology and final statistics that we produce.

There is nothing very confidential about advice to the effect that the net borrowing figure in a document that is published by the Scottish Executive, and which is widely circulated to anyone who wants to look at it,

"should therefore be used with some caution."

Having publicly given advice that the net borrowing figure and various other figures in the document should

"be used with some caution",

do you think that ministers have used them with appropriate caution?

If I were advising the UK Chancellor of the Exchequer, I would be happy to say with great certainty that the UK deficit in 2004-05 was £X billion and was 3.8 per cent of GDP. I would be able to do that with great confidence because the figures are based largely on accounting data. It is an important part of my role to alert ministers to the fact that there is bound to be a degree of variation in Scottish figures because of the estimation process that we use. In my view, the fundamental purpose of GERS has always been to provide an order of magnitude to the flows about which we are talking. We have never sought to give—and, to my knowledge, no member of the team has ever given—the impression that the numbers are absolute to the last decimal point. They are intended to give a broad picture of the flows into and from Scotland, and to give an overall impression of the net position.

Would ministers be misrepresenting your conclusions if they were to say that Scotland has a black hole of £11 million?

The findings of the GERS report for 2004-05 are quite clear—we have a net borrowing requirement of £11.2 billion. However, as I said earlier, I am happy to receive comments on how the document could be improved, because I think that we can improve it.

I want to make an important point—it is not semantic—about one of the labels that we use. For many years we have tried to apply the best statistical conventions of the European system of accounts 1995 and of the UK presentation of budgetary accounts. In the process, we have tried to reflect not just the framework but the terminology that is used. We have picked up the phrase "net borrowing", which we use in the report. In one sense, it is an unfortunate phrase because the transfer from the UK that we describe is not a loan of any sort and there is no net borrowing in the sense in which the term is used of the UK; it is much more a net fiscal transfer of some sort. We could have made that clear in the document. In future years we will look closely at getting the phrase right.

I note that economists appear to favour the use of the word "caution". There is a danger that politicians will interpret that as meaning that figures are inaccurate. Instead of using the word "caution", could you give us an idea of the margin of error that would be acceptable to you in relation to these major figures? Is it plus or minus 2 or 3 per cent? Is it more than that for corporation tax or income tax?

It is very difficult for me to provide a margin of error for the aggregate figures. However, in recent years we have made progress on giving a better idea of the accuracy of some of the components of those figures. A good example is the figure for income tax, to which I have already referred. When we prepared the GERS document for 1997-98, there were substantial concerns about the income tax figure for that year. The Scottish share of UK income tax appeared to fall from almost 8 per cent, which it had been for four or five years, to almost 7 per cent. That caused us great concern and stimulated a great deal of debate about the figure's accuracy. As a result, we undertook a great deal of work with the Inland Revenue, which we encouraged and cajoled into redesigning the key survey of personal incomes that underpins the figure. The sample size was increased dramatically, from 150,000 at UK level to 400,000. For Scotland, that amounted to an increase from about 13,000 to 34,000—about a threefold increase over five or six years.

The importance of the work is that now, unlike previously, the Inland Revenue feels able to describe the data with confidence intervals. It has produced 95 per cent confidence intervals for income-tax shares. The confidence interval that it has provided for 2004-05 is between 7.13 per cent and 7.38 per cent of UK income tax. We have adopted the central estimate of 7.25 per cent. It is important that we now know with 95 per cent confidence that Scotland's income-tax share lies within a range that is equivalent to about £307 million. That gives us some idea of the accuracy of the figure for income tax, which is a good example of the degree of precision that we can give to one component of the aggregate figures. I will not pretend that we can do that for all components—we cannot—but we have made progress for some components on getting the margin of error that the member seeks.

I return to the quality of the information that we can put together. In response to a request to the Treasury from Jim and Margaret Cuthbert, that figures similar to those that the public expenditure statistical analysis—PESA—provides for devolved services in Scotland should be provided for expenditure in England, Her Majesty's Treasury said:

"It would be difficult to produce such a table. The extent of devolution varies in the three countries. Data structures have not been designed to allow such comparisons so estimates might be needed, and it might not be easy to get these sufficiently correct for them to be of publishable quality."

If that is the view of Her Majesty's Treasury on devolved services in England, how can we be so confident of the estimates that have been made in connection with GERS?

Are you referring to the expenditure estimates that we take from PESA?

Yes.

As you know, the starting point for the expenditure side of our work is PESA. The important point for us is that PESA produces national statistics. There are a variety of regulations relating to preparation of national statistics and the national statistician is keen to ensure that those standards are upheld. For that reason, we are confident about using the statistics that have come out of PESA for many years. David Stewart may want to say more about the Scottish data, but that is the fundamental point.

The Scottish Executive provides the Treasury with outturn expenditure figures for its combined online information system database. That database is used across Whitehall to construct the material in PESA. The process is on-going. The production process for PESA 2006 was exceptional in that HM Treasury was in the final part of the transitional phase to the new COINS database. The Treasury has a programme of work to improve the quality of that database; we contribute to that work each year with the material that we produce. As Andrew Goudie said, there is an on-going process of trying to improve and clarify the quality of the data.

In 2003, Dr Goudie, you said that, in your professional opinion, GERS told us nothing about independence. Why did you feel the need to say that?

I cannot remember the precise context, but I have a feeling that the comment is from a press conference for the launch of the GERS document that year. While I would not pretend to remember the precise context, I am happy to say why I think I would have made that comment.

We have to remember why we have GERS. The fundamental reason is to try to capture the actual flows of revenues and expenditures that relate to Scotland in a given year. We therefore focus on the existing context. We look at the actual financial flows, consider the existing constitutional framework, adopt the Office for National Statistics conventions on statistical preparation and use ESA 95—the European system of accounts 1995—standards. By implication, we take the UK macroeconomic framework as it was—the outturn, the policy and the systems. In other words, the whole context of the preparation of GERS is the actual environment of the particular year. We then try to define the specific flows. There is no sense in which GERS tries to consider any other context whatsoever.

My reason for making that comment would have been that, in my view, GERS tells us what a new Administration might inherit irrespective of whether constitutional variations were to follow. GERS focuses on existing flows of revenues and expenditures prior to any change. Any future revenue and expenditure flows would necessarily be driven by the particular political, economic and financial framework that would be put in place under any new arrangement, but GERS does not comment on that.

Before I ask the question I intend to ask, I want to ask about accuracy. Will you confirm that the methodological changes in estimating VAT, for example, during the evolution of GERS have been intended to improve the accuracy of the data rather than to make some sort of political point?

I confirm with absolute conviction that all changes that we have ever made to any of our statistical series have always been made with the intention of trying to improve the objectivity, robustness and accuracy of the data.

Thank you for confirming that.

We have talked about uncertainties in estimates. I come from a scientific background, and I prefer the word "uncertainties" because it suggests that each measure or estimate lies within a range. Has any attempt been made to find out how the maximum and minimum limits of the estimates translate to the final calculation of fiscal transfer?

As I said to one of your colleagues earlier, our ideal position would be one in which we could do what you suggest. At the moment, we do not have the sort of confidence intervals that I mentioned for income tax estimates—

So for the final figure you cannot even give an answer within an order of magnitude?

We cannot give an answer with a specific statistical basis. We could make an intuitive guess—I have not done it, but it would be a guess. It is important to be honest about that.

For specific components, we have a statistical basis for talking about confidence intervals and ranges. However, because consideration of the total means drawing together so many components for which we do not have statistical information, we are unable to give an answer with a statistical basis.

Because of the uncertainties in some of the estimates, is it possible that the fiscal transfer could actually be greater than £11 billion?

I go back to my key point: we are focused entirely on trying to capture the best estimates of the flows. For 2004-05, we have particular estimates for total revenues, total expenditures and total net transfer, but there could be variation in either direction. We would not presume otherwise.

I turn now to the helpful letter that you wrote to the clerk on 15 January, regarding points raised with the committee by our budget adviser, Arthur Midwinter. Will you confirm that the £859 million shortfall—the difference between planned expenditure and outturn—is underspend and slippage and not a consequence of any Barnett squeeze?

I will ask David Stewart to comment.

The letter seeks to point out that the Scottish Executive's departmental expenditure limit budget is only part of the figure. Only part of the figure relates to how the DEL budget has been spent in Scotland. Professor Midwinter's supplementary note helpfully adds to the letter by noting that about half of the figure relates to DEL, some relates to AME—annually managed expenditure—and some relates to UK expenditure.

As far as the DEL part is concerned, of the £859 million difference between the two columns, £322 million is explained by an underspend in relation to DEL. The rest relates to other factors, such as agriculture and UK reserved expenditure.

Good morning. GERS is based on the Treasury's public expenditure statistical analyses—PESA. Have you examined the detailed PESA database, that is, the detailed data at sub-programme level?

No, we have not undertaken such an examination.

The Cuthberts have done so, and have uncovered £5.2 billion of English expenditure on prisons, judicial salaries and nature conservation that has been classified as non-identifiable. That has resulted in an erroneous charge against Scotland. What justification can there be for the Treasury not publishing detailed PESA data as a matter of course?

I will ask Fiona Robertson to comment on the detail, but I will comment more generally. Let me say unreservedly that we are very grateful for the work that the Cuthberts have done. It has brought to light an extremely important issue, which I for one welcome, and we are pursuing the issue strongly. Last year, I wrote to the Treasury to support the general thrust of what the Cuthberts have been asking about, and my team here has been in continual contact with the Treasury to seek improvements.

We hope that PESA 2007 data will rectify many of the problems that have been identified. We had a difficult choice in producing GERS 2004-05: we could defer it until improvement work had been done or we could go ahead but alert people to the problems that were being looked into. In the end, I judged that going ahead would be in most people's interests, provided that we made it clear that issues were under review.

The reason we have not gone back to look at the PESA data as the Cuthberts did is that they are published by the Office for National Statistics and we would not typically go back and double-check such UK work. Partly for pragmatic reasons—we have limited resources—we tend to take such publications at face value and assume that they are of the right quality. The fact that problems have been identified is a concern and I have no doubt that the national statistician is equally concerned about what has come to light.

We are intent on rectifying the problem, and the work is in train. As the Cuthberts have indicated, we are unsure of the precise nature of the problem, but the difference appears to be in the region of £440 million. I do not deny that that is an important sum of money but, nevertheless, to return to my earlier comment, we have always sought to flag up the orders of magnitude of those flows. Although there is certainly a concern—I do not pretend otherwise—I do not think that it changes the fundamental robustness of the rest of what we do and the general story that the figures suggest. I ask Fiona Robertson to comment, because she has been involved in much more detailed interactions with the Treasury than I have.

On your question about the publication of the detailed data, my understanding is that the Treasury will make available on request detailed data from the PESA database. Indeed, I understand that the data were made available to the Cuthberts and that that was the basis on which the issue was uncovered.

It was on the basis of a freedom of information request.

Yes. I understand that.

We have been working with the Treasury for a number of months, since the issue was uncovered, with a view to resolving it before the publication of PESA 2007 in May. As Andrew Goudie said, he has written to the director of expenditure on that basis. Recent discussions with the Treasury indicate that it is on course to resolve the issue that the Cuthberts uncovered in advance of publication of PESA 2007.

In the light of that, how confident are you at this stage that no more problems and material anomalies are lurking undetected?

I refer to what Andrew Goudie said, which is that PESA is a Treasury document and is national statistics. It involves the compilation of data on expenditure from all departments throughout the UK. The Treasury is primarily responsible for pulling the document together. Our efforts will continue to be concentrated on ensuring that the Scottish data in the PESA document are correct. Having said that, we have been working with the Treasury to resolve the issue that has been raised and have highlighted it in the 2004-05 edition of GERS.

Are you saying that it is unlikely that there are other material errors?

I have no reason to believe that there are, but I can only reiterate the point I made: PESA is a Treasury document, is national statistics and is compiled under the guidelines for national statistics.

Given that we have had an error of this magnitude, is that consistent with GERS being published under the banner of the national statistics kitemark?

The GERS source of expenditure data is national statistics. The issue that was uncovered primarily related to English spend being categorised as non-identifiable expenditure whereas in Scotland it was categorised as identifiable expenditure.

In carrying out their work, one of the techniques that the Cuthberts used to identify the errors in PESA at sub-programme level was to look at the identified and non-identified classification in PESA and to compare it with the devolved and reserved classification of the same function within the Treasury funding statement. Have you used similar techniques?

We take our starting point from chapter 7 of PESA, which sets out the context in which expenditure is identified. Identifiable expenditure is that which can be recognised as being incurred for the benefit of individuals, enterprises and communities within particular areas. Identifiable expenditure includes Scottish Executive spend and non-identifiable expenditure—which, for example, cannot be identified because it is deemed to be incurred on behalf of the UK as a whole—includes some elements of reserved expenditure, such as defence.

I come back to John Swinney's point about the Treasury's reservations about much of the data. Do you plan to take any steps to ensure a better alignment of classification in PESA and the Treasury funding statement at sub-programme level, to enable a more effective reconciliation going forward?

For the purposes of Scottish Executive expenditure, that is something on which David Stewart might want to comment.

For the purposes of our own expenditure figure, we have always worked to ensure that, as far as possible, the figures on the Executive's database in respect of economic category, resource, capital and so on are as accurate as possible. Prior to the Treasury's COINS project, we had been through our own figures in that regard, in particular in relation to the split between resource and capital, which is a key control measure for the Treasury and for this Parliament in respect of the budget. We will continue to ensure that the data that we put into the Treasury system are as accurate as possible. Once the data are there, it is the Treasury's responsibility to use them in its own process to produce PESA.

Just—

We have had a good round of questions on this. I am anxious to give other members a chance.

I have one final question.

Given that we have identified the £440 million, do you believe that it should be included in future statements regarding any net fiscal transfer, which I think is the term that you prefer to use now?

I hope that by May 2007, when the Treasury produces the next pizza—I mean PESA—the issue will have been resolved and we will have a clearer view about what the figure is.

Your earlier point is an important one. We had a difficult judgment to make this time about whether to proceed. I judged that we should proceed on the basis of the data that we had, but you will see that I flagged up clearly in the preface that there is an on-going issue. We set out the issues in the appendix. It was in order for us to carry on with the approach, provided that we acknowledged that there is an issue and that work is in train to resolve it. I hope that the issue will be resolved by May this year.

Thank you—

Can I ask a brief question on that point?

All right.

Dr Goudie, did you seek the opinion of ministers on whether to publish?

No. I am pretty sure that there was no ministerial contact at all.

I alert members to the fact that we are incredibly tight for time, so in order to get to the round-table discussion I will be robust in relation to supplementary questions.

I will ask about the basis of the allocation of non-identifiable expenditure. You have used GDP rather than a figure that takes into account the population of Scotland. Is there a reason for that?

For the most part, we use population share. In some areas, where GVA might be a more appropriate measure for non-identifiable expenditure, we use it. An example is some expenditure on agriculture, forestry and fishing, where a sectoral basis might be more appropriate. In such circumstances we use GVA, but for the most part we use population share to apportion non-identifiable expenditure.

I hesitate to use the phrase "one of the more controversial aspects of GERS" to describe the matter that I will ask about, because so much of it seems to be controversial.

On North sea oil revenues, which are politically sensitive, two fundamental and fairly uncertain aspects seem to feed into the revenue calculation. One is the oil price, which we know has been volatile over the past six or seven years, and the other is the exchange rate. In the current version of GERS, we are looking back some years at oil revenues. It is helpful that, as far as I am aware, no one disputes the totality of the oil revenues. However, given that in every budget and every pre-budget report the Treasury is able to estimate and re-estimate North sea revenues and come to oil price assumptions, is the reason that we are dealing here with relatively historical oil revenues to do with other things in GERS holding publication back until you have a complete picture rather than a delay in having the oil figures?

Yes. Perhaps Fiona Robertson can give you details of the lags that are involved. Our fundamental goal is to produce GERS as rapidly as possible following the end of the financial year. However, as has been obvious in this conversation, we depend heavily on UK statistics for several parts of the estimation process and the methodology. The gap between the end of the financial year and when we are able to produce GERS, which I think is about 19 or 20 months, is the shortest time period that we are able to achieve, given the lags in producing the UK data upon which we depend. Fiona Robertson might comment on the disaggregated data that we use in that respect.

Andrew Goudie is right. We produced the publication for 2004-05 on the basis that it gives the most complete picture that we have for that year, based on the data that exist. The timing of the publication and the year that is under consideration are less of an issue in the context of expenditure, because PESA comes out in May each year, but they are a greater issue in the context of revenue. Sandy Stewart might go into more detail on that.

We could vary the timing on the revenue side. However, the survey of personal income data for 2004-05 is still not available—it was supposed to be ready in December, which is why we choose December, on the whole. We could do parts of the work now, such as work on GVA, because we have the regional accounts figures, but we do not have the most compete picture until the end of the year, so that seems to be the most sensible time to publish revenue figures. However, we could produce estimates and forecasts before then.

Given the information that is highlighted in the pre-budget and budget documents, it seems to me—as a layman—that figures might be available and readily identifiable, but you are simply not comparing like with like by considering revenues in the same period. I do not want to put words in your mouth, but that is what I take from your comments.

Are you suggesting that revenue and expenditure are not being considered on the same basis? They are—

I am suggesting that North sea revenues probably are considered on the same basis, because they seem to be specifically highlighted. As I understand it, North sea revenues seem to be available rather more quickly than other revenue figures that you produce.

We are interested in the most complete picture for a particular year. For some elements, it is clear that other estimates are available that might apply beyond the given year, but it would not be appropriate to mix figures on that basis. The exercise is about expenditure and revenue in a particular year and uses actual UK or, where possible, Scottish data. On that basis, the most recent report is for 2004-05.

I am looking at the time-series data. Why has Scotland witnessed such a decline in its share of UK taxation? The share has fallen from 8.9 per cent in 1993, which was higher than Scotland's share of the UK population, to 8.1 per cent in the latest figures for 2004-05, which is substantially below Scotland's share of the UK population.

The figure for revenue is currently 8.1 per cent. That is slightly below Scotland's GVA share, which is 8.2 per cent. In comparing figures for 1993 with figures for 2004-05, you must bear in mind the revisions and methodological improvements in data that have been made, some of which we have discussed. We would not normally compare a document published in 1993-94 with one published in 2004-05; we would go back five years. Sandy Stewart might say more about the revenue estimates.

Definitions have changed: the standard industrial classification codes and European system of accounts classifications have changed, which have a bearing on the figures, so it is hard to compare like with like.

I am acutely aware of the time, so I ask people to be brief.

Okay. I will move on. In table B.1, on page 49 of the GERS document, fuel duties revenue is estimated at £1.31 billion, which is only 5.6 per cent of the UK total—the lowest share of any tax receipt as a proportion of the UK total. However, vehicle excise duties are estimated to be 7.3 per cent of the UK total, which implies that vehicles in Scotland use much less fuel than do vehicles in the rest of the UK. Can you explain that apparent discrepancy?

We have fewer Chelsea tractors.

We still have too many.

On fuel duties, we used estimates from the "Digest of United Kingdom Energy Statistics"—DUKES—which is published by the Department of Trade and Industry, to estimate the Scottish share of inland deliveries of fuel types. There are a number of excise duties—

In the interests of time, it would be incredibly helpful if you wrote to us on that point, which raises an interesting policy issue.

We will be happy to look into the matter and write to the committee.

Various committee members have tried to chip away at the authenticity of the GERS figures. However, in your opening statement you said that the fundamentals of the report are correct and that the basis for independence would be a deficit of either £11.2 billion or £6 billion. Do you stick by that remark?

I stick by my comment that, given that GERS's purpose is to capture the actual flows in 2004-05, we believe that the document provides the best estimates that we can make with the resources that are available. I make no comment about any other constitutional arrangement, other than to say that we can regard GERS as describing the flows that would be inherited by any new Administration, whatever the constitutional arrangements at the time.

I will bring in Derek Brownlee, although I think that we have touched on the matter that he will raise, before I bring in Peter Peacock, who has not had an opportunity to contribute.

John Swinney covered corporation tax in significant detail. The issue has received considerable political attention, despite the fact that, according to the GERS figures, it accounts for a relatively small share of Scottish income.

I will not pretend to understand the fine details of how the Executive produces the regional accounts before it works out Scotland's share of the total figure, but I am considering the profits of the Royal Bank of Scotland, which is a high-profile Scottish company, and the estimated £2.5 billion in corporation tax that is paid. As I understand it, no one disputes the figure for the totality of UK corporation tax. Scotland's 8.1 per cent share of the UK total does not seem to be particularly out of line, but is there another way of validating the figures? Given that the majority of corporation tax is paid by a relatively small number of companies, most of which I presume are in the FTSE 100 index, could an exercise be carried out in which a small number of companies were considered, to identify the share of income from various parts of the UK, and thereby validate the figures from another angle? Would such an approach be conceptually valid? Is that approach already taken in relation to the UK regional accounts?

Scrutiny of the published corporate accounts demonstrates how exceptionally difficult it is to ascertain where corporation tax is paid. Most companies state how much tax they pay—I suppose that is a legal obligation—but because of the global nature of many companies it is extremely difficult, indeed almost impossible, to ascertain where that tax is paid. We know that the big companies produce a significant amount overseas, through subsidiaries or in joint ventures, and that, under the extensive double-taxation agreements that exist between the UK and almost every other country in the Organisation for Economic Co-operation and Development, specific rules govern the place in which tax is taken.

We also know that, despite those rules, there is a great deal of conversation between companies and the tax authorities in all those countries about exactly where that tax will be paid. From our point of view, the complexity of international corporate tax law is such that we would not claim to have the expertise to analyse corporate accounts in that way. Fiona Robertson might correct me if I am incorrect, but we do not have access to individual corporate tax information at the UK level, because it is covered by commercial confidentiality. It is a difficult problem, which is why we use the estimation process that we have adopted.

Okay, I am anxious to—

I have a brief point for clarification. The conceptual difficulties with corporate tax in general that you talked about presumably apply equally to North sea corporation tax.

I am sure that that is true.

Thank you, Derek. I come to John Swinney for one final supplementary.

At various places in the GERS document, non-identifiable expenditure is allocated as a proportion of population and as a proportion of GVA. Taking defence expenditure as an example, the figures give an allocation according to population. However, Scotland's allocation of defence expenditure as a proportion of GVA is 3 per cent compared with UK defence expenditure as a proportion of GVA of 2.6 per cent. That is a difference of 15 per cent and an over-allocation when using the GVA methodology compared with the population methodology. How do you select which measure to use? If we take another example in the report, such as accounting adjustments, the share according to Scotland's population is 8.5 per cent and according to GVA is 8.2 per cent, but Scotland's allocation is 12.7 per cent.

I think that you have made your point, John.

I ask Fiona Robertson to come in on the basis for the apportionment proportion that we use.

Page 44 of this year's GERS report highlights the approach that we take. Again, I refer back to—

Page 44?

My apologies. I was looking at 2003-04.

It was a blank page, which I rather thought answered my question.

It is page 47. I have too many versions of GERS in front of me. The basis for the apportionment is highlighted on page 47. I take the question back to the basis on which PESA identifies expenditure, which is the "who benefits" principle. As we have discussed in a number of editions of GERS and in an article in the "Scottish Economic Report" in November 2004, we apportion defence expenditure on the basis that every member of the UK population benefits from it. We apportion some of the other elements on a GVA basis, that is, on the basis of who benefits. For example, we have apportioned science and technology, transport and environmental protection, and agriculture, fisheries and forestry on a GVA basis. All the others we have apportioned on a population basis.

My point is, why? I see the distinction that you make about who benefits and the proportion of GVA as a measure of economic impact, but why is that not the case with defence?

We do not look at it on the basis of the economic impact of spend; we look at it on the basis of who benefits. For example, with science and technology, it is reasonable to expect that the business sector benefits rather than every individual in Scotland. That is the basis on which the distinction is made. The defence point has been raised a number of times. As I mentioned, we have discussed the way in which we apportion defence expenditure. We have always been clear that it is not about where the expenditure is incurred; it is about who benefits, therefore it is based on population share.

Thank you. I intend to wind up the discussion because we are trespassing on the territory for the next session. I thank the witnesses—it was a useful session on an historic day.

Meeting suspended.

On resuming—

Let us begin the second part of our proceedings. I thank everyone who has joined us for the forbearance that they have already shown. The rules on the declaration of interests require that I point out that I am married to Professor Brian Ashcroft, who is a panel member today. It is not the first time we have had a husband-and-wife team before the Finance Committee—it has happened before in various guises.

I welcome all the experts who are joining us. Forgive me if, given the pressure of time, I do not recount your many and varied achievements. We are joined by Professor Hervey Gibson; Professor Anton Muscatelli; Peter Wood; Professor Neil Kay; Professor Brian Ashcroft; Victor Hewitt, from Northern Ireland; Professor Gavin McCrone; and our own budget adviser, Arthur Midwinter, who has written a paper to brief the committee for today's proceedings.

The aim of today's session is one that we have found useful in the past. Instead of having the question-and-answer sessions that we have had previously, we have decided to have a round-table discussion. In the hour and 20 minutes or so available to us—I hope to wind up the session at around 12.30—I shall allow each of the experts to take five minutes to say a few words of introduction on their view of GERS and its utility. Thereafter, we shall have a general discussion.

I want to alert people to the pattern of the morning's proceedings. When we move into the general discussion, we shall try to address three principal themes. The first is to consider the merits and weaknesses of the GERS exercise, how it might be improved, why there is no GERS equivalent for Northern Ireland and Wales and whether other countries undertake similar exercises. Our second theme will be to look at the debate that has taken place over the past decade or more since GERS reports have been published, and to consider whether the GERS exercise is broadly independent, whether people generally have faith in the estimates that it makes and whether there is anything that strikes people as implausible or unrealistic about it. The third theme, and the one on which it might be useful to focus most of our time, is on what GERS tells us, whether it tells us anything about Scotland's future financial choices, whether it is considered a useful guide to Scotland's economic performance, whether it tells us anything about the net balance and about whether Scotland pays its way and whether it would be an appropriate objective for the Scottish Executive to seek to eliminate that net transfer over the coming years. So, by way of guidance for our expert witnesses, the three themes that we hope to address are: the GERS exercise; the debate that is taking place around it; and what the document tells us.

I now invite each of our experts to make a five-minute contribution. Please do not feel obliged to take your full five minutes. We will be eternally grateful if you feel that you can encapsulate your thoughts more briefly, although as we have lots of professors with us today, I can only live in hope. I now turn to Hervey Gibson and invite him to kick off.

I hope that I can take less than five minutes, because I have already put down three pages in writing, which I hope members have had a chance to read. In that written evidence, I deliberately eschewed the nitpicking that most economists want to do when they pick up GERS—I cannot do that in public, so I shall just pick half a dozen nits very quickly.

It is inevitable with estimates such as those in GERS that new sources of information become available. Air passenger duty is a good example of that, and I am pretty sure that we know the origin of almost every air passenger in the UK, as that information is recorded not only in one powerful database but also in lots of other databases. That could be looked at again. The climate change assumption is that the pattern of potentially environmentally damaging industry is the same as the pattern of consumption. That is one of the small assumptions in the revenue side, and it seems to me to be something that we could easily improve on with industrial data.

Although it is less important now than it used to be, there is a slight asymmetry on product and production taxes, because we tend to subsidise producers but tax consumers. Payments of farm subsidies to Scottish farmers are identifiable, so they get charged to us, whereas the revenue from the production of whisky by Scottish whisky producers is credited to the people all over the United Kingdom who pay whisky duty. Between a product that we want to subsidise and a product that we want to tax, we are adopting slightly different principles. We are doing so for quite good, conventional reasons, but that approach introduces a certain equivocal element about the figures.

GVA ratios have been used in several areas, including capital gains tax, stamp duties and dividends. Either GVA has been used or, in one case, the family resources survey has been used to allocate revenue, and I point out that in those areas GERS moves in the direction of looking at assets—at stocks rather than flows—so the assumptions necessarily become a bit vaguer and more general. That is one of the four main points that I picked up. If we are to look at assets, we want to be able to compare those figures at least with the UK asset figures, which are presented in the national accounts, although the link from PESA to the national accounts is not always clear. The simplest thing that could be made clear is the calendar year issue, as other variables could be made to correspond rather better between PESA and the national income blue book.

The other point about assets—the third point in my written evidence—is the issue of debt interest. GERS is operated one year at a time; in particular, the special oil accounts are considered one year at a time. It is, of course, the case that the interest on the national debt, which is about £2 billion in the current set of figures, is based on the national debt as it stands. It is certainly the case that, if we are trying to use GERS to look at scenarios for Scotland, we will want to look at scenarios in which some of the capital values accumulate—or do not accumulate. Those scenarios would look very different if we accumulated a series of deficits, compared with the way they would have looked in the late 1980s, just before the first GERS report was produced, when we would have accumulated a series of net surpluses. Considering interest payments involves examining assets, and that is something that we cannot do very well at the moment.

My final point is that the public sector is not everything. The two graphs in my paper contain rather old data. They show that what has long been thought to be true for the United Kingdom appears also to have been true for Scotland—that the personal sector surplus and the public sector deficit tend to move in opposite directions. The conclusion that I draw from the first graph is that although we talk about a public sector deficit being financed by fiscal transfer, we could also talk about it being financed by saving in Scotland.

The conclusion that I draw from the second of the graphs relates to a point that Andrew Goudie made—that companies are really difficult to look at in the sectoral balance in Scottish economic accounting. For the personal sector, we more or less know what personal incomes are and what personal capital spending is, so we can estimate the financial surpluses and deficits reasonably well for that sector. GERS now does a good job for the Government sector, and we can look at flows for the rest of the world to a significant extent because imports and exports are now summarised in the annual input-output tables, which cover most transactions with the rest of the world. However, the fourth sector in the sector balance is the company sector, which we do not know about, and that is of the essence in the way in which we manage our economy and look to the future.

I will concentrate on two or three points—if we do not do that, we will end up duplicating what everyone else says.

In terms of the quality of the data, a lot of the issues around which uncertainty arises have been highlighted previously. Given the parameters that the exercise has set itself, what we have before us probably represents the best that can be done. There are huge uncertainties, especially around the revenue data, and some issues around expenditure and the PESA calculations have already been highlighted. However, I do not think that there are easy answers to any of those issues.

The more interesting question is the use that can be made of an exercise such as this. The convener asked whether the exercise reflects what is happening elsewhere. I think that it is more ambitious than similar exercises that are undertaken in other countries because it genuinely tries to calculate all aspects of expenditure and income. In doing that, it sets itself a difficult task.

As an economist, rather than focusing on the metaphysical transfer issue, which is the final headline, I find it more interesting to consider some of the breakdowns of and comparisons between UK and Scottish expenditure averages. In a sense, devolved government is all about local choices in terms of the benefits of different expenditure under various headings. Those comparisons are interesting from the point of view of an economist and from the point of view of the general public, because they allow us to see where the money is actually spent.

I agree with Professor Gibson about the problem of focusing too much on the net transfer issue. That is, perhaps, not the interesting thing that comes out of the data. Partly because it is not a borrowing issue, as Dr Goudie pointed out, the transfer issue does not highlight what would happen under a different constitutional arrangement because, from day 1, things would change quite markedly. It does not explain what would happen in terms of the Scottish current account position because it does not tell us what is happening to the personal and corporate sector in relation to financial imbalances. Focusing on the net transfer issue reduces the value of what is an interesting document in many other respects.

If the exercise is to be improved in order to reduce some of the uncertainty that we have talked about, most of the work will have to be done on the revenue side. Issues such as corporation tax are difficult to resolve, given the way in which data are collected at the moment. There is going to be a huge amount of uncertainty around that area.

I agree with practically everything in Professor Midwinter's paper, which is a masterly summary of the issues.

I also agree with everything that has been said about the quality of GERS. The effort that goes into creating it is considerable and there is a great deal of attention to accuracy. However, we should not be surprised by what we find in GERS. It passes what I would call the reality-check test. We can see that Scotland receives, in the identifiable spending category, about 10 per cent of UK spend—I do not think that anyone really contests that. It is reasonable that non-identifiable spend benefits should be allocated pro rata. Therefore, with the Scottish economy being little more than 8 per cent of the UK economy and with the taxable base of the economy essentially being determined by its size, it is not surprising that expenditure in Scotland exceeds revenue. We would find the same in many other regions across the world that are less prosperous than the national average. One of the benefits of Professor Midwinter's paper is that it shows the long history of people considering these issues.

Given all of that, I will ask a question that is, perhaps, mildly controversial. Has the GERS exercise outlived its usefulness? It is going to tell us the same story year after year, so what is the point of having it? That is not to say that the data that are collected for GERS and the tables that are presented should not be collected and presented; indeed, they should be. To that extent, I agree with Professor Muscatelli and would say that the way in which GERS is presented and the focus on what journalists luridly describe as the "black hole" direct attention away from the more valuable material in the document.

What does GERS actually tell us? Does it tell us whether Scotland can prosper as an independent nation? That is a no-brainer. Of course Scotland can prosper as an independent nation. That is not really an issue and GERS tells us nothing one way or another in that regard. Could an independent Scotland sustain the same level of spending and taxation as it presently enjoys? Probably not—not to start with, anyway. That is, essentially, what Dr Goudie said. However, the question is, would it want to? How the economy would develop is an issue that GERS does not help us to understand.

What is more fundamental is the fact that there are two important questions about Scottish public finances that the present treatment of GERS directs us away from and to which we should turn our attention. The first is whether the higher spending per capita in identifiable spending is appropriate. Professor Midwinter throws in a few interesting statistics to suggest that Scotland has greater needs, but I am sure that he would agree that that hardly represents a comprehensive analysis. Some of the difference is due to automatic mechanisms such as social security payments, and some of it seems to be the result of historical horse trading that has lived on into the present.

In 1997, I was involved in the production of the "Comparative Study of Local Authority Current Expenditure in Scotland, England and Wales" report—the LACE study—which was somewhat buried after the change of Government. It raised some interesting questions about the extent to which, across the board, Scottish authorities spent more than their English comparators. I suggest that questions about Scottish needs, in relation to public spending, are more important than the issues that are set out in GERS.

The other question is to do with revenue. If we are concerned about having a different fiscal base for the funding of devolved expenditure, there is a great deal to be said for considering the tax base of the Scottish economy and how tax could be raised from the Scottish economy under different systems of fiscal federalism. Again, a great deal more needs to be done in that area.

Possibly, we should ditch GERS as it stands or, at least, strip out the commentary and present a new document that examines various dimensions of public spending and the revenue base in the Scottish economy. Such a document could be used to inform important policy debates about the level of public spending and how we finance it in a devolved Administration.

Today's meeting has confirmed to me my view that I do not want to be a Government economist. I commend the Government economists on their attempts to deal with the issues and to correct and improve the series. I take the points that Dr Goudie made in response to questions.

As an industrial economist, I am interested not so much in the snapshot of the economy—which appears to be the emphasis of GERS—but in what might be described as the narrative that is described by GERS in terms of the long-term trends. Indeed, in that respect, I find myself in the peculiar position of arguing that GERS can say more than the Government economists say that it can. I noticed that there were reservations about stating that GERS can say much about the period outwith a four or five-year timescale, but I hope that we can stretch that timescale and examine trends over the longer period.

GERS throws up reasonable indicators of long-term trends in Scottish public finances, taking Scottish public finances as being indicative of deeper rooted aspects of the Scottish economy. I have produced a four-page submission to the committee that deals with that in more detail.

Table 7.3 in the GERS document outlines the relationship over time between Scottish Government spending and Scottish tax revenue, as percentages of the UK figures. The noteworthy aspect of that is that the gap between the two is widening. Earlier, during Dr Goudie's evidence, it was stated that, in particular, Scottish tax revenues tend to be declining quite substantially. As with any series, the interpretation depends on the timespan that one is looking at. However, whether one is looking at 25 years or 10 years, there has been quite a substantial decline.

If we accept, at least as a working hypothesis, that we have a broadly accurate picture of what is happening, the question is why tax revenues have been declining. In a sense, we have been asked to reverse the normal procedure for economics. We usually have a theory, consider the evidence and then use statistics to back it up. I suggest that we can try to validate the statistics using theory and evidence from what we know about the economy.

For example, in the past 10 years, Scottish tax revenues have declined substantially as a percentage of the UK figure. I wondered whether that is because of the decline in population but, in fact, as the budget adviser notes in table 6, paragraph 56 of his paper, Scotland's share of the UK population has not declined as fast as the Scottish tax share has. I note also the methodological points that the GERS economists mention, such as the one about corporation tax. From a quick check, the shares of all sources of Scottish taxes seem to be declining more rapidly than has the population share over the period. It seems as though something has been happening over the period.

Why should that be? To try to validate the statistics by citing theory and evidence, there is certainly evidence that the Scottish economy has been underperforming in several areas, such as business start-ups, productivity and, in particular, growth. Those seem to be a contributory factor in the declining tax revenue compared with the UK tax revenue. However, there may well be a deeper reason, which I mentioned in my paper as an element that should be introduced into the debate. It is a curious element that is well discussed outwith Scotland in the context of resource-abundant countries or regions. I am speaking about the phenomenon that is called the resource curse, which in essence is the apparent paradox that countries and regions that are blessed with natural resources tend to grow more slowly than countries or regions that are without the blessing of a windfall of natural resources.

Several reasons for that phenomenon have been considered. Jeffrey Sachs and a colleague at Harvard University have cited a number of such reasons, to which I refer in my paper. One aspect that may be relevant is the notion that growth of the natural resource sector might crowd out the development of growth in other generating sectors, such as manufacturing. That may well be an element in what has been happening in Scotland in the past 10 or 25 years. Whether or not the resource crowding-out effect is happening, the interesting point is that the arguments in the resource-curse literature are similar to arguments that Professor Ashcroft has cited for dynamic resource crowding out in the public sector. There seems to be a plausible connection between the statistics and what we might expect from theory and evidence. The matter obviously requires further research but, for now, I can say that the long-term trend of declining tax revenue in Scotland as a percentage of the UK tax revenue is expected, given what we know from the resource-curse literature.

We have been asked to say whether GERS gives a best estimate of the narrative of Scottish long-term trends in public finance. My answer, given the literature and the evidence and theory, is that it gives a plausible picture of the long-term deterioration in Scotland's tax revenue relative to that of the UK, which, in itself, has interesting policy implications.

As members know, I am from the Fraser of Allander institute at the University of Strathclyde, which has a reputation for and a continuing interest in considering the Scottish economy in general, particularly public finances. I checked the references in Arthur Midwinter's paper on GERS and found that, through our quarterly economic commentary, we have published six of the papers to which his assessment refers. From my position as editor of the commentary for many years—some would say for too many—I have formed a view from that work, and some of my own work, on how GERS stands as a comment on or, as the GERS economists point out, a dynamic statement of Scotland's public finances.

I agree broadly with the committee adviser's conclusion that GERS is

"valuable … for monitoring … flows between Scotland and the UK"

and that it is

"widely accepted by researchers as a reasonable estimate of the fiscal position, using realistic assumptions, and making defensible judgements."

That conclusion is supported by University College London's constitution unit, the UK's foremost independent research body on constitutional change, in a book published in 2002 by Murkens, Jones and Keating. They conclude that GERS is

"accepted by most independent commentators as the best available approximation to Scotland's public sector accounts".

We have heard similar points today.

GERS may not reflect Scotland's public finances if Scotland became a sovereign state or if it adopted another constitutional arrangement within the union, such as fiscal autonomy. However, as I think Andrew Goudie said, GERS shows us broadly what the position would be on day 1. Clearly, measures would have to follow from that and changes in the economy would have an impact. However, the position on day 1 is evident from GERS. In the constitutional unit's words, that position is that

"Scotland is in deficit. This excludes oil revenues which, when included, do not necessarily bring Scotland into balance. Relatively high oil prices and low US dollar values are required to bring present Scottish tax and spending patterns into balance".

That said, as we have heard today, GERS can be criticised and has not been without its critics, some of whom, particularly the Cuthberts, have been exceptionally constructive critics of the process, as Andrew Goudie acknowledged. However, although the Cuthberts' work on GERS is extremely valuable and should condition various changes in the future, the impact of that analysis on the overall balance of Scotland's public finances is comparatively minimal. As members may know, we published the Cuthberts' article on PESA in the quarterly economic commentary. I asked the Cuthberts, because this was not in the paper, to do a subsequent side calculation to show the implications of the £450 million-plus overcounting of identifiable expenditure in Scotland for new identifiable spending in Scotland relative to that in the UK. Their estimate was that aggregate expenditure would be 1 per cent lower and that identifiable expenditure would be 1.5 per cent lower, which would have taken relative spending for 2002-03—it has since changed—from 116.5 to 114.3. On the net borrowing requirement, which was more appropriately discussed earlier as the fiscal transfer, Scotland's deficit in GDP would change from 11.3 per cent to 10.6 per cent, which is important, but only marginal, given the overall figures.

I share the Cuthberts' concern with the PESA data. I am surprised that it has taken people from outside the system to identify the problems. I support their recommendations for change. My recent examination of the revisions in PESA of the Scottish identifiable expenditure for a given year shows considerable variations over five years. As we know, the identifiable expenditure relative is considered to be the most robust part of GERS but, if we look at the revisions, we see significant changes for a given year as we go through subsequent years. I can provide the committee with the data for that. For example, the 1998-99 identifiable relative was 3 percentage points higher in the GERS for that year than it became in the GERS for 2002-03. Other years show quite a bit of movement on the same basis. That is concerning, so we must ask what drives the revisions. We need more information on that. For example, we need to know whether the revisions are driven by the treatment of unallocated expenditure, which is an issue, given its allocation at the English level.

We need a basis for improving GERS and a way of checking how reliable it is compared with other data.

One piece of evidence that we can look at is GDP per head relative to that of the United Kingdom. My analysis of those data is that GERS is not very reliable on revenues in the period prior to 1996, but appears to be quite reliable after that. The revenue estimate does not pick up the fact that between 1988 and 1996 the Scottish GDP per head relative to that of the UK went from 93 to 101, which is a significant improvement in Scotland's relative performance. The revenue in GERS flatlines at around 8.6 to 8.9 per cent, which worries me. However, thereafter, the downward track broadly corresponds with the relative decline from 101 to about 96. Unlike Neil Kay, I do not take the statistics to indicate that there has been a long-term decline; I am sceptical about that.

Overall, the nature of the statistical deficit is a difficult issue. Economists often argue—some people will disagree with this—that, under certain assumptions, the trade deficit or current account deficit mirrors the fiscal deficit. We see from the latest input/output tables—this reflects what Hervey Gibson has done, although I am a bit uncertain about the analysis of the early years that he presents—that Scotland has a trade deficit for the fiscal year 2002 of £9.3 billion and a deficit of £8.7 billion in 2003. The annualised Scottish fiscal deficit is £9 billion in the first year and £10.7 billion in the second year. On that basis, there seems to be a fairly close correspondence between the trade deficit and the fiscal deficit, which offers qualified affirmation of the estimate of a deficit within GERS.

I do not want to linger on the technicalities of GERS, on which others have spoken with greater authority. I can clear up the business about Northern Ireland and GERS, if that would be helpful. Northern Ireland has replicated GERS, but has never published it. The most recent report of which I am aware was produced in 2002-03. It which showed a deficit, in our terms, of about £6 billion. The methodology used in GERS was followed carefully, because most people accept that it is pretty much at the cutting edge of this sort of work in unified fiscal systems. I understand that a revision is under way, which might well be published.

It is important to consider the sources of the data. I suggest that we take a closer look at the PESA database, not only at sub-programme level but even below that, at what we used to call unit-of-business level, especially where there are differences in institutional delivery of services—having different forms of institution can complicate the matter dramatically. In Northern Ireland, our local authorities have limited powers, so a lot goes through central Government departments, whereas in England it would go through local authorities. It is difficult to align those two things, particularly when considering the needs assessment, so it would be useful to revisit the matter.

We sometimes become obsessed with the detail. I find it much more interesting to consider how GERS can be helpful in considering the development of the economy. One of the reasons we were interested in it is that we did not have any estimate of the balance of payments. Brian Ashcroft has said that in certain circumstances there can be a degree of read-across from the fiscal deficit to the deficit on the current account for the balance of payments. That is fairly intuitive, given that a lot of things are imported here, not made here, so we have to consider how they are financed and how we are buying them.

One rapidly comes to the conclusion that, in a sense, the fiscal deficit is financing the balance of payments deficit. That takes us back into the real economy—if I can call it that—because it focuses our attention on why exports are not keeping up with imports and the fundamental issue of the rate of growth of the economy. On that basis, when we looked at the data we became concerned about the operation of regional policy generally in the UK and particularly in Northern Ireland. We are rapidly developing a new universal constant, which is the ratio of GVA per head in Northern Ireland, which is 80 per cent of that in the UK year after year. Scotland's GVA has been declining to 96 per cent of that of the UK over time. Solving one problem solves the other. If we can get export-led growth into the economy, the balance of payments comes down and, almost simultaneously, the fiscal deficit comes down with it.

GERS can be used to tell you what you should know in your heart you need to do, which is to boost the growth of the economy. Professor Kay mentioned the resource curse. We are concerned about the rentier economy—which is related to that—when the expenditure side of the economy gets divorced from the revenue side of the economy. It happens quite often in resource-rich countries, such as some of the gulf states that have large oil revenues and low taxation, where there is a weak link—if there is any link at all—between the taxes people pay and the public services they enjoy.

In a broad sense, we can view UK regions as partial rentier economies that are linked to the south east. In many areas, there is a dichotomy between expenditure and the revenue that is raised there, and what is funded by transfers within the system. I know that there is a sensitivity about language, but there is no doubt that there are substantial movements of money. The question is whether that is a good thing for any economy in the long term.

The other question on which to focus is what the public expenditure is being used for. I know that, in functional terms, it is being used to build houses, to pay nurses and doctors and so on, but public expenditure sits along a spectrum: at one end is pure investment and at the other end is pure income support. Most sits somewhere in the middle. Evidence from the north and south of Italy and east and west Germany after reunification suggests that the more public expenditure tends towards income support, the less dynamic the economy becomes.

In the early days of the transfers between the north and the south of Italy, most expenditure went into investment in roads and so on, and the economies converged quite rapidly, but from the mid-1970s, the trend started to move towards income support rather than pure investment and there has been no great convergence since. Something similar happened in eastern Germany. This is not an exact science, but it gives you some clues as to what might be going on.

By all means argue about the detail of GERS, but do not make the best the enemy of the good. GERS gives you some clues about the working of the economy, but it is more important not to lose sight of the underlying problems in the economy on which it sheds light and to address those problems than to argue continually about a few million pounds either way in the accounts.

I have been involved in this for rather a long time. I first published some figures of the GERS variety in the 1960s, in a book that I wrote, to which Professor Midwinter referred. I also did research on what was happening from 1888 onwards. The picture is fairly consistent. In the period to 1922, when Ireland left the United Kingdom, Scotland's public revenue and expenditure were broadly in balance, England was spending less than it was raising in tax and Ireland was spending substantially more than it was raising in tax. From the time Ireland cleared off out of the United Kingdom, the picture changed and Scotland began to get net transfers of public expenditure that were larger than the revenue that was raised. The figures are fairly sketchy for much of the period—funnily enough, they are very good in the 19th century, but rather unsatisfactory in the early part of the 20th century—but that is the broad picture.

What do the figures mean? One can chip away at things and say that this or that is not quite right, but it is difficult to argue about what the general picture is. There have been fiscal transfers to Scotland. Public expenditure here is higher than the revenue that is raised, but Scotland is not alone in that regard. There appear to be substantially larger transfers of money per head in Northern Ireland. Transfers in Wales are slightly less. People in the north of England periodically beef about the situation and say that that area does not get a fair deal, but it actually receives a higher level of public expenditure than the British average. That level is not as high as the level for Scotland, but it is around 10 per cent above the British average. Expenditure in the north-west of England is also higher than the British average.

Of the English regions, London has the highest level of public expenditure per head. Indeed, its figures are close to those for Scotland. It could therefore be argued that expenditure in Scotland is not too different from that in London. However, people in London would claim that they raise much more revenue. The figures are very unsatisfactory, but they suggest that London is in surplus.

What has happened is the natural consequence of revenue and expenditure decisions being taken independently and of no attempt being made to get the figures to balance in any part of the country, because there is no need to do so. Attempts are made to get the figures to balance in some way at the UK level so that there will not be too big a budget deficit, but, until now, nobody has cared about the position in individual regions. The position in individual regions has been an issue and the figures have been published only in Scotland.

I agree with what Peter Wood said about the GERS exercise—it is as good an exercise as can be done. Many figures on the revenue side in particular are estimates, which is inevitable; the figures on the expenditure side, which are derived from the Treasury figures, are a good deal firmer. The problem—if there is a problem—is that Scotland's public expenditure is pretty high; it represents around 50 per cent of gross domestic product. The figure for the UK is 40 per cent and the figure for Ireland is only around 30 per cent, I think.

In considering whether it matters that the figure is so high, we should ask whether the expenditure is justified. There is a lot of argument about the issue at the moment and the English are beginning to get much more excited about it because people in England have talked about Scotland receiving a subsidy. I reject the word "subsidy" because there has been no deliberate intention to subsidise—what has happened has simply been a natural consequence of the system. Fiscal transfers have taken place.

When I worked in the Scottish Office in the days before the Barnett formula, the Secretary of State for Scotland would go to the Treasury every year and argue with the Chief Secretary to the Treasury about individual items on the public expenditure list. Expenditure was supposed to be related to need, but secretaries of state tended to be persuasive and they usually did rather well in the exercise.

The Barnett exercise was intended to be used under devolution arrangements in the 1970s if the devolution legislation then proposed was passed, but the incoming Conservative Government thought that the exercise was a good idea because it meant that the secretary of state and the Chief Secretary to the Treasury did not have terribly painful arguments every year. The Barnett formula was therefore adopted.

At the time, Bruce Millan told me that the Barnett formula should result in some convergence, as it takes the level of public expenditure in the previous year and adds an increment, which simply represents the population share of the UK's increment. I will return to that issue in a moment. The Treasury did a needs-assessment study then, which, as far as I know, is the only such study that has been done. It showed that Scotland required more public expenditure per head than the British average because of its sparse population, the cost of providing services in remote areas, its deprivation and its bad health record, for example. Such factors justified considerably more public expenditure per head than the British average, but not as much as we were receiving. At the time, Bruce Millan told me that he was not too worried about the Barnett formula because it would take a long time to bring us down to the level of a needs-assessment study anyway. The process has therefore continued—many temporary things continue for a long time. The Barnett formula was never expected to last for more than a year or two, but it has been used since 1979.

Will the arrangements last? What can be done about them? I have written that I do not think that they can last because the arguments between England and Scotland will become increasingly raucous; indeed, we can see signs that they are doing so. Parts of England are exerting increasing pressure to reconsider the arrangements. Indeed, Lord Barnett has said that he no longer believes that his formula is appropriate. I am not sure whether he fully understands the formula, but that is another matter.

I think marked convergence is taking place, but it cannot really be seen in the figures for identifiable public expenditure that are published. Calculations of mine that are included in the paper to which Professor Midwinter referred show convergence in what is determined by the Barnett formula, which accounts for only 58 per cent of total aggregate expenditure, I think. The gap with respect to items in the Scottish block seems to have narrowed, particularly in health and education—in those areas, the difference in expenditure per head in Scotland and the British average has substantially narrowed over 20 years. That is not because expenditure in Scotland has decreased; rather, expenditure in England has increased as a result of big public expenditure increases in the UK.

For reasons that I do not understand, the difference between social security expenditure in Scotland—which is the largest component of the identifiable public expenditure budget, but not part of the block and therefore not determined by the Barnett formula—and that in the rest of the UK has increased, which has masked the decline in bits of expenditure within the block. Jim Gallagher, who worked for the Scottish Office and is now a professor at the University of Glasgow, has produced a paper that shows that considerable convergence has taken place. We should think about that convergence.

We should also consider what should replace the arrangements if something has to replace them. I do not see many alternatives to carrying out a new needs-assessment study and then having a prolonged period of adjustment to reach whatever level is required. As a member of the Arbuthnott committee on the distribution of resources in the health service, I was amazed by the enormously higher costs of providing similar standards of service in more rural areas. A needs-assessment study would show that Scotland required substantially more public expenditure per head than the British average, but goodness knows how much expenditure would be required.

The current system cannot last for ever. It constrains in one other respect. It is difficult for Executive ministers to do anything about tax rates, except income tax rates, not only because they are not permitted to do anything about them, but because if they cut tax rates, for example, there would be howls in England because it would be increasingly argued that we were receiving more money than we deserved.

I was struck by Gavin McCrone's comment that Lord Barnett may not understand how his formula operates. In that context, I make a plea to the experts. Your expertise is greater than ours—I speak on behalf of all members of the committee in saying that. We want to focus on the utility of the GERS document and how it might be developed.

I was also struck by what Victor Hewitt said about GERS being replicated in Northern Ireland, where a document has not been published. Perhaps having an unpublished GERS document is the only thing worse than having a published GERS document.

In the first part of our discussion, we hope to reflect on the GERS exercise. Eight years into devolution, is it the case that if GERS had not existed, we would have needed to invent it to give ourselves some sense of financial flows in the United Kingdom? To develop Peter Wood's point, are there new data on public spending the publication of which should be considered to augment what is already available? Does anyone think that if GERS had not existed, we would not have had to invent it? Perhaps Peter Wood could comment on what new data might be helpful and Victor Hewitt could say whether he thinks it is likely that Wales and/or Northern Ireland will have a published GERS or something comparable at some point. The floor is open.

My understanding is that the Department of Finance and Personnel is working on a 2004-05 or a 2003-04 version of the document, which it intends to put into the public domain, but I do not think that it will have quite the same authority that GERS commands in Scotland. It will be more of a technical document produced by economists in the department.

The same intensity of argument about the issue does not exist in Northern Ireland. When it comes to Northern Ireland's public finances, people simply accept that there is a huge deficit and get on with their lives because there are more interesting constitutional issues to fight about. A GERS document would be of interest because it would shed light on the underlying workings of the economy, but we are well aware of those, even without the detail of the public finances being published.

If a GERS assessment were published in Northern Ireland, what impact would that have on the debate about the differential corporation tax rate that your institute has recently instigated?

As I said, the proposal on a differential corporation tax rate relates to the fact that Northern Ireland's GDP has been stuck at 80 per cent of the UK average for a decade. We wanted to kick-start the process of achieving a quantum leap in performance. Essentially, the existing instruments are grant based. In the wider UK, a bargain is struck, whereby we have a unified fiscal system with the same tax and exchange rates. That may or may not be appropriate for every area, but everyone lives with it. The bargain is that the richer areas ensure that public services in the poorer areas are maintained at a sufficiently high level. The other side of the bargain is that particular areas might not be able to grow their economies as fast as they could if more flexible instruments were available to them.

In the future in the UK, given that the existing instruments do not seem to be working too well, we might move towards a more flexible taxation system. That idea is anathema to the Treasury, regardless of which tax one is talking about. However, it is not inconceivable that our proposal could be adopted. There are many technical difficulties with it, which we have discussed with the Treasury, but it is not impossible that we could have a differential corporation tax rate, if the will exists.

That is interesting.

As far as the technical improvement of GERS is concerned—in which context, I give due acknowledgement to the excellent work of Jim and Margaret Cuthbert and others—it must be nearly as good as it will get, especially as regards the facts of the case on the expenditure side. We have heard about the methodological improvements that have been made on the revenue and tax-raising side and I doubt whether significant further progress will be made on that.

My concern is more about what we do with the information and what it is for. It would be better if there were a move away from misleading spats about the extent to which the Scots are being subsidised to recognition of the fact that the analysis that is provided in GERS is a rich resource to help us understand how our economy and the public finances are working. The points that Neil Kay made about the time series are undermined by the lack of reliability in the time-series data, so perhaps some work could be done to help us understand how the finances have evolved over time.

In this discussion, a number of important points have been raised. Gavin McCrone made some interesting points about what we might call the Barnett and non-Barnett elements and why they seem to be moving in different directions. We need to understand that, because it will tell us something about how our economy is shaping up.

I agree with the requirement to consider needs again, because that covers a lot of issues that need to be addressed.

I do not want to repeat things that I said earlier but, on the revenue side, understanding the Scottish economy's tax base might help to inform a debate about how we could finance public spending in Scotland in future. The issue is not so much a need for methodological improvements but how to use all the work that has been done to understand better the way in which our economy is developing.

One of the purposes of this discussion is to record in the Official Report issues that we might want to pursue, which include the need for time-series data and how the Barnett and non-Barnett elements of spending are moving.

I detect that there is a desire to talk about what GERS tells us, but if anybody feels that it contains any assumptions that are unrealistic or findings that are seriously implausible, I give them the opportunity to put those points on the record before we move on. I have not heard much about that this morning, given the way in which the document has evolved.

Our reservations are that the report does not cover all the sectors of the Scottish economy and does not factor in the resource cuts. It tries to present a snapshot, but it fails to make the sensible adjustments that we would make or consider the upside of the growth of a more competitive Scotland. I am interested in what Victor Hewitt said about the sustainability of the rentier economy. I am also interested in examining countries such as Ireland, which has a much lower population than Scotland but where corporation tax and VAT, to take just two examples, materially outperform what we achieve according to GERS.

As far as I can see, nobody publishes the English comparator to which the Barnett formula is applied. That is a real lack because it means that it is very difficult to establish what has happened to the bit of the expenditure to which the Barnett formula applies and work out whether there has been convergence. The only way to do that is to take expenditure on social security and agriculture out, because the Barnett formula does not apply to either of them, and find out what happens to the remainder, but that is only an approximation. It would help greatly if we were able to see the English comparator and how the Barnett bit of public expenditure behaves compared with the rest. Once or twice, I have asked officials of the Scottish Executive whether the figures for the English comparator can be obtained, but they never have been.

You might have to try asking for them under freedom of information legislation.

I have already done my nitpicking for this morning. When we add the absence of oil-revenue calculations in the Scottish fiscal transfer but their inclusion in the UK budget deficit, the disparities on defence expenditure, issues on income tax, corporation tax and VAT and the errors that the Cuthberts identified—issues that I have placed on the record in the Parliament—to the cumulative position, they make a substantial impact on the veracity of the GERS figures. That must be reflected upon as part of a serious debate, particularly as Dr Goudie reinforced the caution with which we must treat some of the figures.

That is John Swinney's point of view, but the view that I am gathering from the evidence that we are taking in the discussion is not that GERS is so inaccurate as to make it useless and a source of major errors, as has been described, but that it is useful.

Perhaps the argument is more about how we politicians use GERS—which may not be surprising in an election year—than that it does not contain useful information, and that we could make better use of it. Perhaps that is something for our legacy paper.

Indeed.

I take the point that Peter Wood made about long-term trends and the points that the Government economists made about the unwillingness or reluctance to look at the issue in any great depth. I also accept the point that figures for a single year can be unreliable—that is the very nature of such a report. However, if we look at things over a longer period, and if we are consistent in our measurement, trends should become more visible and the figures more reliable than the estimates for any one year.

I am keen to move on to the next theme.

Yes. Let us do that.

I will kick off by returning to the interesting point that Professor McCrone made on what GERS tells us. I have a number of points to draw together. Subject to the same caveat that I made about my general attitude towards GERS, this genuinely intrigues me. Professor McCrone said—correctly, in my view—that there has been a clear convergence on elements of identifiable expenditure, such as education and health. In 1999-2000, the Scottish figure for identifiable expenditure per head on education was 126 against the UK position of 100, but it is now down to 106. On health, we were on 119, but we are now down to 110.

On page 30 of the GERS document, we find an acceptance that the convergence was driven by expenditure south of the border that

"grew faster than in Scotland."

Professor McCrone also helpfully said that convergence is not so acute overall as it is on health and education, which are, of course, the big spenders in Scottish public spending. If convergence on health and education is more dramatic than overall convergence—which is a fair reflection of the information that GERS shows us—that raises the question why. Again, Professor McCrone helpfully said that it was because of the increase in social security receipts.

Arthur Midwinter's paper shows that, according to GERS, in 1996-97, Scotland accounted for 8.6 per cent of all estimated revenues from UK taxes and 8.7 per cent of the UK population. The figures for 2005 show that our population has fallen only to 8.5 per cent of the UK population, but our share of all estimated revenues from UK taxes has gone down to 8.1 per cent. In paragraph 83, Professor Midwinter says:

"The Scottish fiscal deficit is … certainly not evidence of poor economic performance".

For me, all those pieces of information do not hang together other than by the explanation of poor economic performance. I think that that was Victor Hewitt's reflection on the debate from a Northern Ireland perspective.

In short, we have dramatic convergence in terms of identifiable expenditure on health and education, which is balanced out by social security spending that is not converging at as fast a rate—or may even be going in the other direction—and we have a tax take that is declining faster than our population is declining. Why? The only conclusion that I can arrive at is that Scotland's economic performance is poorer than that of the rest of the UK. I seek reflections from around the table on what all of that adds up to. I think that that is where Professor McCrone was going.

The issues have different roots. First, what is happening to the spend on health and education is an indication of what Neil Kay and others have called the Barnett squeeze. Clearly, Barnett is biting on the spend in those areas and we can make the comparisons. Secondly, between 2000-01 and 2004-05, the spend on the social security programme per head has gone up very slightly from 107 to 111. That is an interesting issue. My guess is that the spend has gone up for demographic reasons. We now have an aging population and inward migration that is positive for Scotland, but the inward migration tends to focus on the relatively young and the relatively old. Migrants are coming to Scotland for lifestyle reasons and that means that a pensions element is involved. Therefore, my guess is that the changes to spending on health and education and on social security have nothing to do with economic performance.

On the decline in Scotland's share of the tax take, my view is that the numbers are wrong. However, they are probably more right from 1995-96 and probably parallel a relative decline in the Scottish economy since 1996.

I think that the GDP per head figure has come down since that time.

The GDP per head figure has come down from 101 in 1995-96—when, in GDP per head terms, we were actually 1 per cent above the UK—so one would expect to see some impact from that on the revenue figures. What worried me was that the figures for the Scottish share of tax revenues flatline at around 8.9 per cent over the period before 1995-96 when there was a big hike—from 93 to 101—in the GDP per head figures. However, the revenue figures start broadly to track the decline in GDP per head figures after 1996. I think that that is because, as Andrew Goudie and his colleagues said, the numbers are better.

However, you think that the decline in the tax take reflects relatively low growth.

I think that the decline in the relative tax take is more likely to be related to economic performance. Not all tax receipts are responsive to the economy, but receipts for taxes such as income tax and, possibly, corporation tax are likely to be so.

I think that the relative changes to the items of expenditure that you mentioned are unrelated to that. Health and education expenditure is Barnett driven. Social security expenditure is driven more by demography. Social security spending can—as you rightly implied—be driven by the economy if there is a fall in employment. However, given our relatively good employment numbers, the increase in such expenditure is a bit of a puzzle. I think that it is down to demography.

I am trying to marshal all those indicators together to work out what they tell us. You mentioned that we have relatively good employment numbers, but our tax take is down. I cannot understand how the arguments that we are hearing about economic performance sit with the tax take numbers. Does Professor McCrone want to respond to that?

I think that the issue would need to be studied. Perhaps the Fraser of Allander institute would like to do that.

The decline in the tax take could be for a number of reasons. First, it could be due to changes in the population. Secondly, Scotland's GDP per head in comparison with the UK average has fallen, from what was more or less an all-time high, back down to about 96 per cent. Over the long term, the percentage figure has remained more or less in the mid-90s. One would expect that change to have some effect.

Another point, which occurred to me just as Brian Ashcroft was speaking, is that the decline may also be the counterpart to the increase in the social security ratio—if it is true that social security payments have increased because we have more old people and more immigrant young people. The natural population increase in England is higher than it is in Scotland. I think that that is partly because England has a huge immigrant population.

The change in the figures might mean that Scotland has a workforce that is not as wealthy as it was because we have more pensioners and more young people than previously. In England, there are more wage earners in the middle of the population who contribute more to taxation. However, the issue needs to be studied. At the moment, we are just speculating.

This may just be further speculation, but the decline in our share of the tax take might be a function of wage or salary levels. In the south of England in particular, income levels tend to be higher so people will pay more tax.

If that relationship has changed, which it may well have done, that would be true.

The reason might be that wages in Scotland have not kept pace with wages down south.

I note that the income tax position has improved over the past five years. Because of the recession in the electronics industry, 2000 was in some ways the low point for income tax receipts, but there has been an improvement in each of the past five years.

We have already put on record our desire for the Executive to think about how it might reconcile the figures on a year-by-year basis. Without that, we can only speculate whether the changes reflect improvements in the estimation techniques, changes to the population or a movement in the underlying economic performance. We will probably ask the Executive to consider how it could provide more direct comparability on a year-on-year basis and provide us with some guidance on these issues.

It is important that most of the shift, which was identified in Professor Midwinter's figures before it appeared in the table in GERS, occurred in the early and mid-1990s, when exchange rates were changing and the balance between profits and wage income was shifting significantly. Of course, that was also the period before devolution, which is an important issue for the committee.

We should remember that the figures for identifiable public expenditure are per capita in the total population. If the school population is declining, as is happening in Northern Ireland—I do not know whether that is happening in Scotland—the fact that per capita expenditure is going down does not necessarily mean that less is being spent on pupils. We are doing work on public expenditure on young people and have found it incredibly difficult to make proper comparisons of how much is being spent per pupil throughout the UK.

Employment has grown by 20 per cent or so in the past decade but, although we have done well in that regard, our GVA per capita relative to that of the rest of the UK has not moved one iota. The problem is that, although we have grown jobs, we have produced relatively low-value-added jobs. We are trying to break out of that cycle.

Your comments are helpful. It is perfectly possible to create employment, but if the wages are low and compound existing problems, GVA will not change.

In public debates, GERS is often regarded as a kind of proxy for and guide to the Scottish economy's performance—that view has been expressed during this meeting. If that is the view of the experts who are present, will you say whether a reduction in or elimination of fiscal transfers should be an objective of Scottish Executive policy or whether you think that such transfers will always be a feature of a unified state? There is a lot of public confusion about the interpretation of GERS in that regard, so it would be helpful to have some guidance on the record.

I agree with what Arthur Midwinter, Victor Hewitt and others have said. Given the current constitutional arrangements, the transfer is not of itself a particularly interesting matter that should be targeted. If the exercise were replicated for every region of Europe whose performance is below its country's average, very similar patterns would be identified.

The unified view of the experts round the table is that much of revenue and expenditure is simply an automatic feature of social security and other mechanisms that drive expenditure and income. There also seems to be agreement that there is a danger in presenting information in a way that does not focus on how expenditure is directed. Victor Hewitt made the good point that the composition of expenditure could be an important driver of economic development in regions. The danger is that we consider the bottom line but do not focus on how expenditure is spent. If the exercise is couched as a survey of Government expenditure and revenue for Scotland, it does not focus on the bigger picture about what is driving economic development. That is a difficulty.

Even more dangerous, GERS might be used as an instrument to change the way in which allocations are made and perhaps to rush us towards different solutions, as Gavin McCrone said. I have said on the record that I disagree with the Barnett formula. The approach should be focused on need—I agree with what Gavin McCrone said about that—but it would be wrong for it to be driven by the sorts of considerations that we have been discussing. An economic argument should be made and time should be taken to do a proper needs assessment before there is a move to a different system. We should not say, "There's a deficit; something has to be done quickly." The transfers are perfectly normal in a regional context in the European economy.

I reinforce that. As far as the performance of the economy is concerned, what matters is the rate of growth of GDP or GVA. Any Government ought to focus on how to enhance that, if possible. I do not think that this budget exercise helps in that respect, except in helping you to decide how best to spend public expenditure so as to enhance growth.

The important point in what we have been discussing this morning is that we must have a system that can be defended. There is a lot of criticism of the Barnett formula, which I think will increase. If there are Governments of different political complexions north and south of the border, that criticism will become much more intense. Therefore, I think that it is important to consider how we can move on from the present situation and what could replace the Barnett formula if we had to replace it. The danger is that, if we do not do that, we will be pushed into something all of a sudden and there will be a lot of political pressure to cut public expenditure per head in Scotland back to the level of public expenditure per head in the rest of the UK, or something like that. That would fly in the face of any proper assessment of need.

That is where some effort should be focused, and for that reason I would like to see a new needs assessment study conducted relatively soon. In saying that, I do not disregard the problems of conducting such a study. A needs assessment study contains quite a lot of subjective judgment, so it is not absolutely watertight. Nevertheless, we have done it in Scotland with the Arbuthnott committee for the health service, which is a large part of public expenditure. It could be done perfectly well on a UK basis, but I would like it to be done for all parts of the UK, not just for Scotland. If we are going to get into the business of deciding what fiscal transfers ought to be between different parts of the country, everybody should be in it together; it should not be just about Scotland.

On that last point, if there is to be a needs survey, I would prefer it to be carried out on a UK basis. However, if the rest of the UK did not agree to it, there would surely be some merit in proceeding alone in Scotland.

Well, yes—probably. However, the last needs assessment was done by the Treasury with Scottish Office help and if the same were done again—whether it was right or wrong—nobody would believe it. It is therefore essential that the assessment should be carried out by an entirely independent body of some sort. Also, in order to make the people in the north of England a bit happier, they should be told that they, too, have a net in-flow of funds. Many of them do not realise that at the moment. That is why I would like the study to be conducted more broadly than just in Scotland.

Although I agree completely with the principle that a comprehensive, UK-wide needs assessment is the right way in which to proceed, there is an anomaly in the fact that we do not have the same mechanisms in England to effect the fiscal transfers. We have the automatic mechanisms, but we do not have the Barnett formula equivalent. We have seen attempts to create regional government in England come to grief, so I am not sure how we would effect the fiscal transfers. It may be that, with enough ingenuity, bits of the budget could be carved up. Nevertheless, there seems to be a slight problem because we do not have a unified or comprehensive system of fiscal federalism in the UK.

The issue of needs assessment might, historically, have been reasonably straightforward to sort out if an assessment had been carried out within a small body of professionals. However, once the issue of needs assessment is raised with the public, all sorts of values and standards come into play. For example, it is probably reasonably straightforward to assess needs in the area of health, but if spending on school education in Scotland is compared with spending on school education in England—where there are very different systems, values and procedures—the matter becomes more complex.

I support that. I have some experience of needs assessment and know that you should be very careful what you wish for. It is an enormously complicated process. What was published in 1979 was only the tip of the iceberg of the work that was done at that time. The world has moved on and all sorts of new factors have come into play, which the Treasury will take into account. For example, the much greater need for payments for ethnic minority pupils in schools in the south-east becomes a factor. The whole thing starts to break down once we move away from the demographics. Needs assessment is really based on demographics—that is the only firm bit of it. When it comes to support for industry, for instance, there is no firm basis for comparison. Needs assessment can be done, but it is not an exact science by any means.

The experts will appreciate that it is not easy for us to hold a parliamentary committee hearing on a technical document. It might be easier to leave such matters to the headline writers, but we were anxious, as devolution matures, to start to examine documents such as GERS in depth using a similar approach to that of the UK Treasury Committee, which considers the pre-budget report and other UK budget documents.

This has been a learning exercise for the committee as we deepen our understanding of some of the issues. Either I or the clerks, or indeed Arthur Midwinter, the budget adviser, will be anxious to hear participants' observations on how scrutiny and exercises such as this are conducted through parliamentary hearings here in Scotland, rather than being left to the headline writers, as I said.

I thank all the independent experts. You will have a particular interest in our final item, on our draft report on the proposed legislation to establish a new statistics board, which represents the UK-wide effort to improve the statistical basis—although that item is to be taken in private.

All our experts should feel free to talk to us in private after the meeting about how this exercise could be conducted in the future.

Meeting continued in private until 12:41.