Select Committee on Constitution Written Evidence

Memorandum by Professor Iain McLean, Professor of Politics, University of Oxford


  1.  The UK has developed two sets of formulae, one for the territories that now have devolved government, and one for the regions of England. Their conjunction leads to what Lord Barnett calls "terrible unfairness".

  2.  The Barnett Formula was designed as an anti-rounding-up mechanism and as a convergence formula. The first was needed because Scotland had done disproportionately well under the former "Goschen" arrangements. The second should have led to convergence in spending per head of population, but has only just started to do so. It has punished the territory which poses no credible threat to the Union, viz. Wales, while causing justifiable anxiety in Scotland that her favourable spending position is being eroded by a "Barnett squeeze".

  3.  The formula for transferring revenue to local authorities in England may have rewarded inefficiency and has been politically manipulated. Within England, territories with no credible threat have done badly. The Government has announced that it is scrapping the formula, but not what is to replace it.

  4.  In the light of the principles of devolution, it is inappropriate that finance disputes should ultimately be resolved in the United Kingdom Cabinet. I recommend a set of measures to tackle the defects of both formulae at once. Unless they are both tackled at once, Lord Barnett's "terrible unfairness" many remain. The measures include:

    —  a territorial grants board making allocations by unanimity rule;

    —  as a default should unanimity not be achieved, a rule that incremental grant in the next time period will be awarded by an "inverse GDP" formula.

1.  Introduction

  1.1  The following is evidence on topic 1(4) of your Call for Evidence, namely Finance, fiscal matters, and the Barnett formula. I have been researching these matters since the 1970s and have published extensively on them. A list of my relevant publications is available on request. I arrange my evidence according to your subheadings.

  1.2  This evidence is a summary of a much longer document, The Fiscal Crisis of the UK, co-authored with Alistair McMillan. The longer document contains full citations and methodology for the claims made here. It will shortly be freely available on the Web at and I will notify the Committee when it is available.

2.  How is funding allocated to the different parts of the United Kingdom; and what consequences arise?

  2.1  Formula funding is allocated to the different parts of the UK according to two quite different formulae. Allocation to the three non-English territories is by the Barnett formula. Allocation to the Government Office Regions of England follows the "York indices of health care needs" for NHS funding and the Standard Spending Assessments (SSAs) for local authority services. These funds are assigned to health authorities and local authorities, not to regions, but when aggregated to regional level, they comprise the formula funding to the nine standard English regions.

  2.2  The English formulae attempt to allocate funding on a needs basis. The health formula is generally regarded as a success, but SSA as a failure. Its sponsoring Secretary of State announced in December 2001 that it would be scrapped, but gave no clear signal as to what would replace it (Hansard, Commons, 11 December 2001, cols 713-28).

  2.3  The population-based Barnett formula has operated since 1978. It was intended to correct an anomaly in spending per head between Scotland and the rest of the UK, and then to be replaced by a needs-based formula. Because the Scotland and Wales Acts 1978 were never brought into operation, neither was the needs-based successor to Barnett. Therefore, for many years now, the Barnett formula has been inappropriate.

  2.4  In 1978, the idea of a mechanical formula for devolved spending was not new. Some Scottish spending (but never all) had been governed for decades by the "Goschen proportion", viz, that Scotland should be assigned 11/80 of the money assigned in England & Wales for the equivalent programme.

  2.5  Table 1, which gives the population ratios between Scotland and England & Wales for Census years 1891-71 inclusive, shows that the Goschen proportion was too mean to Scotland until 1901 and too generous thereafter.

  2.6  Scottish departments negotiating with the Treasury programme by programme could always claim at least the Goschen proportion, and more if there were grounds for making an additional claim for the programme in question. They could appeal to Ministers if the Treasury resisted. As there was always a Secretary (of State) for Scotland in the Cabinet, they had a reasonable chance of success. Therefore, by 1977, Scottish spending per head on the services which the Scotland and Wales Bill proposed to devolve was between 20 per cent and 30 per cent above English and Welsh spending per head.

  2.7  This caused resentment especially in the Northern Region of England, whose GDP per head was comparable to Scotland's but whose public spending per head was less. This resentment led to the defeat of the Scotland and Wales Bill in a guillotine vote on 22 February 1977.

  2.8  In response, the Treasury produced a Needs Assessment and the Barnett formula.

  2.9  The Needs Assessment was published in 1979 using data for 1976-77. It reported that actual spending compared to that needed to provide devolved services to the same standard throughout the UK was ahead of needs on Scotland and Northern Ireland, but behind needs in Wales (Table 2). A likely reason is that Scotland and Northern Ireland had had devolved administrations since 1885 and 1921 respectively; Wales only since 1964, and many programmes were still financed on a unified England and Wales basis.

  2.10  The Barnett formula was both an anti-rounding-up device and a convergence formula. By stipulating that the territories should negotiate for a block, not service by service, it succeeded in ending the rounding-up that had enabled Scotland to claim more than the Goschen proportion of spending on "devolved" services.

  2.11  The convergence component was designed to reduce this disparity gradually until spending was more closely aligned with "needs", when it would be replaced by a needs-based formula. But if the Needs Assessment was correct, the convergence element of Barnett should never have been applied to Wales, because Welsh spending per head was already below needs; to force it to converge on English spending per head would therefore worsen the situation there.

  2.12  Originally, for the services which the Scotland and Wales Acts 1978 proposed to devolve, each increment in funding in England was to lead to a proportionate increment in Scotland and Wales (Northern Ireland was added to the formula soon afterwards). For every £85 increment in spending on a "devolved" service in England, an additional £10 would be made available to the Scottish administration and an additional £5 to the Welsh administration.

  2.13  These population proportions were incorrect for 1978, and they worsened over time. Scotland's population in 1978 was less than 10/85 of England's; Wales' was more than 5/85 of England's. Therefore the formula tended to give too much of any increment to Scotland, and too little to Wales. The population proportions were corrected in 1992 and annually since 1997.

  2.14  For three administrative reasons, of which this was one, there was no convergence until the current Comprehensive Spending Review round. The other two are: (i) that until the mid-1980s, the formula operated on increments in volume (real increments) which were few or negative; only since then has it operated on cash (nominal) increments; (ii) that Governments could and sometimes did bypass the formula.

  2.15  The political reason lying behind all three administrative reasons is that Scotland and Northern Ireland (but not Wales) pose credible nationalist threats to the Union. Were Scotland to leave the Union, the interests of both the large UK parties would suffer.

  2.16  By 1999 therefore, the anomalies that had led to the 1977 Government defeat were as wide as ever. In a sense they were wider, as Scottish GDP per head is now almost at the UK level, but North of England and Wales GDP per head are still at 1978 proportions.

  2.17  Table 3 compares GDP per head with public spending per head on "devolved" services for the 12 standard regions of the UK. "Devolved" services means (as accurately as the source permits) those services which, outside England, are devolved to the three territorial administrations.

  2.18  The poorer a territory, the more a government should be spending there for both efficiency and equity reasons. Therefore there should be a strong inverse relationship between GDP/head and public spending/head. In fact, however, the correlation between GDP/head and spending/head is not even statistically significant.

3.  How satisfactory are these arrangements?

  3.1  Both the Barnett and the SSA arrangements are deeply unsatisfactory. Their conjunction is worse still.

  3.2  Their conjunction is most painful between the Northern Region of England and Scotland. According to the Journal, (Newcastle-upon-Tyne), health spending per head, in fiscal year 1998-99, was £692 in Northumberland and £945 in the Borders. The secondary school in Duns (Borders region) had a pupil: teacher ratio of 13:1 and one computer per five students. The secondary school in Alnwick (Northumberland) had a pupil: teacher ratio of 18:1 and one computer per 13 students. And yet GDP per head in Northern England is below that in Scotland.

  3.3  This conjunction has led Lord Barnett to say repeatedly, eg in a debate he initiated in the House of Lords on the inadequacies of his formula (Hansard, Lords, 7 November 2001, col. 228), that the operation of his formula has led to "terrible unfairness".

  3.4  However, this outcome is not due to Barnett alone, but to the operation of one formula in the Borders (Barnett) and another in Northumberland (SSA and York indices).

  3.5  As SSA is a needs-based formula, it ought to have led to a vertical distribution in favour of poorer regions within England.

  3.6  As with Barnett, there are both technical and political reasons for its failure to do this successfully. The technical reasons, now conceded by DTLR, relate to defects in its regression formulae. These are hard to explain in a non-mathematical way. But, in brief, they have (i) produced incorrect coefficients on the vector of services that go into SSA and (ii) conflated true costs of providing a service with inefficiency costs, thus tending to reward regions containing inefficient local authorities.

  3.7  Academic analysts have also shown that the weightings in SSA benefited two sets of clients: (i) local authorities that were "flagships" for the Government that set up SSA in 1990; (ii) regions containing a high proportion of marginal constituencies. As it happens, the Northern Region of England contains none of the first and few of the second.

  3.8  The needs per head of each non-English territory are above the UK average. Each of them has below average GDP per head, and each can plausibly plead special factors (sparsity in Scotland and Wales; the security situation in NI).

  3.9  But if Barnett operates mechanically, it will in due course lead to convergence on average public spending per head. For the mathematical reasons why this is so, see the evidence that I understand Professors David Bell and David Heald will submit.

  3.10  As explained at 2.11 above, this was never the correct policy for Wales. At some point it will no longer be the correct policy for the other territories either.

  3.11  Thus Barnett should be replaced by a needs formula, and SSA by a corrected needs formula. I offer some suggestions on how to do this in section 5 of this evidence.

4.  In particular, how, and in what circumstances, is funding allocated outside the Barnett formula?

  4.1  The official answer to this question is to be found in HM Treasury's operations manual Funding the Scottish Parliament, National Assembly for Wales and Northern Ireland Assembly Second ed July 2000, available on the Web at Briefly, Annually Managed Expenditure (AME) and a small amount of Non-Barnett determined spending within the Departmental Expenditure Limit (DEL) is allocated outside the Barnett formula. The significant items are all in AME. They include CAP; social security; and NHS and teachers' pensions. Any money to be raised from the Scottish Variable Rate of Income Tax would normally also be outside the Barnett arrangements, although the Treasury states in its manual (at 5.2) that it is "open to the [UK] Government to take into account" the resulting self-financed expenditure if it has grown so much faster than equivalent spending in England that it "threaten[s] targets set for the public finances as part of the management of the United Kingdom economy".

  4.2  However, the academic's answer is broader. In the past, the formula was periodically bypassed because, as stated above, both Scotland and Northern Ireland pose credible threats to the Union, and hence to the two parties that have formed the UK Government since Barnett began. Formula bypass in Scotland and Northern Ireland seems much diminished in the current spending round.

  4.3  However, in July 2000 the Welsh executive and the Secretary of State agreed to an allocation outside the Barnett formula, to guarantee the match funding for Objective One EU funding for West Wales and the Valleys. Wales had offered a credible threat to the Union for the first time since the creation of the Welsh Office. The previous failure to promise match funding had led to the resignation of the leader of the Welsh executive on a confidence vote. The new leader (Rhodri Morgan) had the backing of a National Assembly and popular majority in Wales for the claim.

  4.4  If the analysis above is correct, this extra-Barnett allocation was acceptable on equity grounds because the Barnett formula has always borne harshly on Wales.

5.  In the light of the principles of devolution, how appropriate is it that finance disputes would ultimately be resolved in the United Kingdom Cabinet?

  5.1  It is not appropriate. Finance disputes are disputes between governments. They should be resolved by intergovernmental bodies, not by a government that is one party to them. Although it is appropriate that most assignment should be by formula, the formula should not make the money available to the three non-English territories a mechanical function of decisions by the UK government on spending in England, as the Barnett Formula does. That is such a violation of the principles of devolution that I am surprised that any politicians in the devolved territories wish to retain Barnett.

  5.2  Below I outline a possible set of intergovernmental arrangements. They could operate either on a 4-territory (4T) or a 12-territory (12T) basis.

  5.3  In either case there is a non-partisan Territorial Grants Board. It would be an NDPB with an analogous constitution to the Electoral Commission, and for the same reason: that its operations must be totally immune from partisan politics. Its Director and staff would be public servants, who could be secondees from the Treasury, the Office for National Statistics, the DTLR, or the devolved executives. The Board would evaluate needs and resources in the territories, and annually propose a multiplier for each territory. The UK Government would continue to decide how much to make available each year from the UK Consolidated Fund for spending on devolved services. But the Board's multiplier would be applied to the average per-head value of this transfer to derive the per-head transfer to each territory.

  5.4  The basic model for this Board would be the Commonwealth Grants Board of Australia, which has operated such a mechanism successfully for many years. But it has to be adapted in two respects: (i) because Australia is a federal state and the UK is not; (ii) because in its early years it would have to correct the misallocations arising out of the defects of Barnett and SSA listed above.

  5.5  The Board must report to a joint ministerial council. In the 4T case, the council would comprise one representative of each of the devolved governments and one representative of the government of England (as distinct from the government of the UK). In the 12T case, each devolved government and each English Regional Assembly (or successor body) would send one representative. Each territory would have one vote, and all decisions of the council must be unanimous.

  5.6  The unanimity rule is needed to reverse the fundamental inequality that has led to Lord Barnett's "terrible unfairness" and much else, namely that some territories can exercise credible threats against the UK Government and some cannot.

  5.7  However, a unanimity rule on its own is insufficient. In any year, a territory that stood to lose from the proposed set of multipliers could threaten to veto it. Therefore, the unanimity rule needs to complement, and be complemented by, a publicly known default arrangement. If the joint ministerial council should fail to agree the allocation by a set deadline, then grant for the next period would be assigned according to the default arrangement.

  5.8  In the event of failure to agree, I propose that each territory should receive the same cash sum per head as in the previous time period plus an increment comprising the mean amount of new money available per head multiplied by the default multiplier. I propose that this default multiplier be 1/relGDPi, where relGDPi denotes (territory i GDP/head)/(UK GDP/head). If the ONS regional price level data are sufficiently robust, the multiplier should be converted to its value at (inter-territory) purchasing power parity. This formula is called "inverse GDP" below.

  5.9  I recommend inverse GDP over other possible defaults because:

      5.9.1  GDP per head is a reasonable surrogate for a territory's neediness. It is not the direct result of government policy, although it is highly correlated with things that governments must try to improve, such as human capital and health status.

      5.9.2  GDP per head is measured by an independent non-partisan agency (ONS), not by any party to the proposed negotiations.

      5.9.3  There would not, as now, be perverse incentives to become and remain "needy". In 2000, Wales gained from being "needy", and it will lose (at a marginal tax rate of at least 100 per cent) if and when West Wales and the Valleys cease to be "needy". By contrast, if a territory's government improves its GDP per head, then income per head must rise by more than grant/head would fall on an inverse GDP formula, as government spending is less than 100 per cent of GDP. Therefore the marginal tax rate on success would always be less than 100 per cent.

  5.10  If there is a Pareto-superior allocation to inverse GDP, the bargainers at the joint ministerial council will find it. A Pareto superior allocation is one that nobody likes less than the default and that at least one party prefers. If none exists, the inverse GDP default satisfies both efficiency and equity, and is cheap to calculate.

  5.11  If the inverse GDP formula were used over a number of years, it would bring about convergence. But, unlike the Barnett formula, it would converge on needs, not on population. Therefore, unlike Barnett, it would be just and efficient.

  5.12  In the 12T model, the formula operated by the Grants Board would supersede SSA and the York indices of health care needs for allocation of grant around England. Of course, the Grants Board would retain those parts of the present mechanism that work. Most commentators believe that the York indices work. Most, including the Secretary of State, believe that SSA does not work.

  5.13  In the 4T model, the government of England would have to decide on its own allocation formula. It would have an interest in aligning the formula to the UK-wide formula, so that Lord Barnett's "terrible unfairness" disappears.

  5.14  The Government is committed to replacing SSA, but has not yet said what will replace it. All commentators agree that the Barnett formula is under severe pressure. The proposals in this section are designed to tackle both of these problems at once. Unless they are both tackled at once, the scope for "terrible unfairness" will remain.

Table 1. Relative populations of England & Wales and Scotland, Censuses 1881-1971

Population, 000s
England and Wales
Scotland: England and Wales=80
Value of Goschen grants per head in Scotland (England=100)

  Source: British Historical Statistics, B R Mitchell (CUP 1988)

  1939: mid year estimate

Table 2. HM Treasury Needs Assessment, 1979 (data for 1976-77)

Northern Ireland
Relative needs assessment
Actual spending levels 1976-77

  Source, HM Treasury Needs Assessment—Report (1979) esp para 6.5.

Table 3. Public Spending and GDP per head, Regions of the UK, 1999-00

Public exp/head on `devolved' services, £
GDP per head, £
East Anglia
Greater London
South West
West Midlands
East Midlands
Yorkshire and Humberside
North West
Northern Ireland

  Source for column 1: HM Treasury, Public Spending Statistical Analysis 2001, derived from Tables 8.6b and 8.12. Column 1 reports (for the Barnett territories) "Identifiable total managed expenditure per head 1999-2000, and (for the non-Barnett territories) "Identifiable general government expenditure per head, by region and function, 1999-2000". In each case Social Security, which is a non-devolved function, is excluded.

  Source for column 2: Office for National Statistics, Regional GDP 1999, summary table.

previous page contents next page

House of Lords home page Parliament home page House of Commons home page search page enquiries index

© Parliamentary copyright 2002