Memorandum by Professor James Foreman-Peck
1. This note first shows the spending pattern
achieved under the current system of devolution financing. Consistent
with the untrammelled operation of the Barnett Formula, Welsh
public expenditure per head increased by a smaller proportion
than English spending between 2002 and 2007. But Scotland,
unlike Wales, obtained considerably more funding than predicted
by a simplified application of the Barnett Formula. In the second
section, an alternative rule is presented that maintains existing
proportionate differentials in public spending between UK countries.
This permits by comparison a calculation of the "Barnett
Squeeze" on different assumptions. For Wales, taking 2002-03 as
the baseline, the predicted "squeeze" on the 2007-08 budget
was about £630 million, and compared with actual (planned)
spending, it was around £860 million. The third section
suggests a simple needs-based rule to replace Barnett. Household
disposable income is a more appropriate needs indicator than GVA
per head (when both are adjusted for price differences between
regions) because GVA measures production which is less related
to need than is income. The rule takes as the needs standard variations
with household income of UK state spending on benefits in kind.
According to this approach a continuing "Barnett squeeze"
I. SPENDING PATTERNS
2. Under the Barnett Formula, once extra
spending has been agreed by central government, it is divided
between UK countries on the basis of population. Increases in
English public spending in relevant categories ("comparable
programmes") are matched by equal increases per head in devolved
administration budgets. But since the devolved administrations
spend more per head of their populations, their budgets are therefore
augmented by smaller proportions than England's. Hence their expenditures
per head tend to converge slowly to that of England (the "Barnett
3. Historically there was some "bypass"
of Barnett Formula (and use of incorrect population factors).
HM Treasury expected that the greater transparency introduced
with the Devolution Settlement would reduce such anomaliesand
certainly more information as to the operation of the Formula
is now in the public domain.
4. The present exercise attempts to assess
how much convergence of spending per head the operation of the
formula has produced during the period 2002-032006-07 (and
also to 2007-08, although only planned rather than outcome figures
are available for the last date at present). This allows a calculation
of (net) formula "bypass".
5. As a reasonable simplification, assume
of total identifiable public spending on services at the country
and region level, only "social protection" and "agriculture"
are not devolved functions. Then the growth of such spending per
head can be compared between England and the devolved administrations.
6. This calculation in Table 1 shows
that public spending in Wales increased about as much as predicted
by the Barnett Formula between 2002 and 2006. Despite Objective
One funding, there was minimal net "bypass". Welsh "devolved"
spending rose by a little more than 33 per cent, while comparable
expenditure in England increased by over 38 per cent. With
greater public spending per head, the Formula prescribes a smaller
percentage rise in expenditure for Scotland than for Wales (about
31 per cent compared with 33 per cent). Yet Scotland,
unlike Wales, appears to have "bypassed" the Formula,
almost matching England's percentage increase in spending per
7. Between 2006-07 and 2007-08, planned
spending was to rise by 8.6 per cent in England (Table 1).
Allocating the implied sum of money per head to the devolved administrations
according to the populations, should have given Wales a 7.5 per
cent increase and Scotland 6.8 per cent. However Welsh "devolved"
spending increased by 1.8 percentage points less than predicted
by the Barnett Formula, while Scottish spending increased by one
percentage point more.
8. These figures show that the "Barnett
Squeeze" took less from Welsh spending than other budgetary
influences between 2006-07 and 2007-08. For one way of calculating
the squeeze is to compare the actual budget increment with how
much money, say, Wales would receive if the percentage increase
in budget was the same as England's under the Formula. But this
approach would require a larger total UK budget. Calculating the
"squeeze" within the existing budget requires an alternative
budgetary rule against which the current Formula can be assessed.
IDENTIFIABLE PUBLIC SPENDING ON SERVICES
|"Devolved" per head (£)||3,662
|"Devolved" (£ million)||10,693
|"Devolved" per head (£)||3,922
|"Devolved" (£ million)||19,824
|N Ireland ||
|"Devolved" per head (£)||4,497
|"Devolved" (£ million)||7,751
|"Devolved" per head (£)||3,092
|"Devolved" (£ million)||153,522
Notes: Calculated from HM Treasury PESA. Tables 10.1-10.4.
"Devolved" = total identifiable spendingagriculturesocial
protection. "Barnett" predicted = increase over period*
average population share.
II. BARNETT WITHOUT
9. The objective of allocating a given sum of money without
the convergence property of the Barnett Formula is easily achieved.
Calculate the total devolved categories of spending for England,
as well as for the other administrations. Divide the total UK
figure into the sum to be distributed. Then allocate to each administration
a proportionate increase in their budgets equal to the resulting
fraction. This ensures the continuation of existing disparities
in proportionate terms. For any "comparable programme"
a similar exercise can be undertaken.
10. As an illustration, suppose the increased money available
is expected to be £15 billion and the current spend
is £300 billion on "comparable programmes".
Under Barnett the £15 billion is allocated according
to population. For ease of calculation we assume in the planned
expenditure period that England's population is 50 million,
Wales three million, Scotland five million and Northern
Ireland two million. Under Barnett England gets £(50/60)*15 billion
= £12.5 billion and Wales gets £(3/60)*15 billion
= £750 million. With "Barnett relaxed" each
country receives (15/300=) 5 per cent. The corresponding
number in sterling depends upon the country baseline. If England's
baseline was £235 billion (£4,700 per head
with no population growth according to these numbers) and Wales'
£20 billion (£6,667 per head), England would
receive (235*.05=)£11.75 billion, less than under Barnett,
and Wales gets (20*.05=) £1 billion, more than under
BARNETT RELAXED AND THE SQUEEZE 2006-07
Note: The "Squeeze" columns do not sum to zero
because of rounding errors.
11. Table 2 takes the figures for 2006-2007 and
2002-7 from Table 1 to calculate the difference the
"relaxed formula" would have made to the allocation
of 2007-08. Applying the "relaxed formula" for 2007-08 only,
England loses 0.3 percentage points of the increase, amounting
to almost £640 million. Wales gains £114 million
compared with the Barnett formula and £370 million compared
with the actual allocation. Scotland gains £406 million
from removal of the "squeeze", but only £135 million
more than planned.
12. The impact of the "squeeze" is cumulative.
Had the "relaxed formula" been adopted in 2002, the
effect in 2007-08 would have been much greater. A given percentage
increase exercises a larger absolute effect the higher the base
on which it is calculated. The earlier the base is raised (by
adopting the "relaxed formula") the higher will be the
base in any subsequent year. Table 2 shows that in 2007-08 England's
allocation would have been £2.6-2.9 billion less, if
the relaxed formula had been introduced in 2002, and Wales would
have been better off by £0.5-0.8 billion. As with the
one year exercise, the Barnett squeeze on Scotland is much more
severe in theory than in practice. The opposite is the case for
III. NEEDS AND
13. The "Relaxed Barnett Formula" shares with
Barnett the advantages of budgeting simplicity and ease of calculation.
It also removes the apparently arbitrary convergence property
of Barnett. But it embeds no less arbitrary differences in public
spending per head of population between the countries of the United
Kingdom. These differentials cannot readily be explained in terms
of variations in "needs" between the countries. One
reason is that there is no agreed measure of general public spending
per head has been suggested as an aggregate regional indicator
but GVA differences reflect divergent regional employment participation
rates. Wales's rate was 55.4 per cent at the end of 2006 compared
with the UK's 60 per cent. If Wales had achieved the UK rate
it is reasonable to expect Welsh GVA per head to be 81 per
cent of the UK average rather than 75 per cent.
14. Household disposable income (standardised for household
membership and including transfer payments) is not subject to
this difficulty and is more obviously related to need than GVA,
which concerns productivity and participation. Differences in
regional costs of living should be factored into the needs indicator
for an adequate comparison as well. Subject to these and other
adjustments, it is possible to obtain a relationship between household
income and utilization of public services for a cross-section
of the UK population and apply this to a modified income index
to obtain a public expenditure allocation model.
15. We use the approximation to "devolved"
expenditure under Barnett across UK countries, because England
is not subject to the Formula and the other countries do not have
identical devolved spending powers. For present purposes we will
assume that this measure of "devolved" spending is equivalent
to "state benefits in kind"of which health and
education are the largest components. We can compare the equivalent
benefit categories received by different income groups in the
UK with the comparable benefit possibilities allocated to devolved
administrations and English regions by average household income.
If the Barnett Formula is working well, we should find similar
relations between UK household receipt of state benefits in kind
by income as between devolved spending and household income across
UK HOUSEHOLD INCOME AND STATE BENEFITS IN KIND
disposable income £)
|State benefits in kind
Source: Economic and Labour Market Review, Vol
2 No 7 July 2008.
16. Populations of the devolved governments account for
only about one sixth of the UK total, so atypical spending allocations
to these countries will not significantly influence the total
distribution in the present exercise. Table 3 shows the average
state benefits in kind received by households by decile of disposable
incomes in the UK. As incomes rise, state benefits utilised fall.
17. In table 4 and Appendix figure 1, which show
the relation between household income by region and public spending,
no such negative relationship is observedonce the lowest
spending area (in figure 1 the bottom right), England, is
excluded. The two highest areas of "devolved" public
expenditure are Northern Ireland and Scotland. The lowest household
income area is the North East region of England, which received
higher public spending in 2007-08 than Wales. Consistent
with the negative relation of Table 1, Wales has a greater household
disposable income in 2006 than the North East. The anomalies
by UK spending standards then are Northern Ireland and Scotland.
These countries receive more in relation to household after-tax
income in public spending than the rest of the UK. But Wales is
slightly below a line connecting the North East and England as
a whole, at least hinting that Wales, unlike Northern Ireland
and Scotland, is receiving a slightly poorer deal than comparable
GROSS VALUE-ADDED, HOUSEHOLD INCOME AND PUBLIC SPENDING
|GVA per head 2007
Disposable Income 2006
spending per head,
|N Ireland ||16,170
Source: ONS 12 Dec 2008 Regional, sub-regional
and local gross value added; HM Treasury PESA Tables 9.10a and
Note: Definition of "devolved" as in Table 1.
18. Table 2 shows regional variations in the cost
of living. Whereas prices excluding housing costs in Wales were
about 4 per cent below the UK average, when housing costs
are taken into account, the Welsh cost of living was 7-8 per
cent lower than the UK average in 2004. How housing is taken into
account in the needs indicator is therefore critical.
PRICE INDICES IN 2004 RELATIVE TO THE UK (UK=100)
||Regional Weights||National Weights
Source:Wingfield, D. Fenwick, D. Smith, K. "Relative
regional consumer price levels in 2004" Economic Trends
615 February 2005.
19. The bulk of devolved spending is on state benefits
in kind. Figure 2 shows how these benefits vary with household
disposable income. One description of the relation between income
and benefits is that a one per cent rise in income is associated
with 0.39 per cent fall in state benefits consumption (Table
3) A fair allocation would then be obtained from the proportionate
deviation of devolved household disposable income from the UK
average, adjusted for differences in costs of living.
REGRESSION OF PROPORTIONAL RELATION BETWEEN HOUSEHOLD DISPOSABLE
INCOME AND BENEFITS IN KIND
|Adjusted R Square||0.98
20. In the case of Wales, household disposable income,
now 11 per cent lower than the UK average, must be divided
by 0.921 (Table 2) or 0.931 to correct for lower Welsh
prices. Real income per head is then approximately 4 per
cent lower than the UK average. Wales therefore requires 0.39*4 per
cent = 1.6 per cent more spending per head or per household
than the UK average, if the distribution of state benefits in
kind in relation to income across the UK is taken as the standard.
21. According to this needs measure Welsh devolved spending
per head should be reduced. Wales' identifiable state spending
net of social protection and agriculture 2007-08 was £5,051 per
head and the UK was £4,679. That is, Welsh state spending
per head was 7.9 per cent greater than the UK average, compared
with the 1.6 per cent warranted. For the other devolved administrations,
the excess of budgets over those warranted by UK spending are
22. This aggregate approach to needs assessment has the
advantage of relative simplicity, consistent with devolution.
Devolved administrations themselves employ needs-based rules to
allocate spending between health authorities and among local governments.
They make the decision about the resources to allocate between
spending departments. So for central government to employ a needs-based
formula that explicitly takes into account health, education,
transport and so on, would be second guessing the lower tier government.
The weights assigned to the various indicators would reflect the
central government's valuations not those of the devolved administrations.
23. A cost-based formula is unsatisfactory in removing
incentives to control costs and to adjust provision according
to circumstances. Scotland might have chosen a less expensive
housing policy before devolution had it faced a less accommodating
budget constraint. By contrast, a modified income approach provides
encouragement for cost-containment and flexible cost-minimisation,
at least in the short-run. If a devolved authority is less efficient
than the UK average at providing public services, it will be less
able to conceal the shortcoming by spending more on them.
24. While at first sight the Barnett squeeze appears
arbitrary, the application of an aggregate needs indicator suggests
it is a pragmatic solution to the excessive budgets granted to
the devolved administrations.
Particular needs indicators are more common but still controversial.
Free school meals are sometimes used as an indicator of educational
needs and the standardised mortality rate on occasion has been
adopted for health needs. Back
(GVA/Population.)= (GVA/worker).(workers/pop.). UK GVAP=100, Wales
GVAP=75. (100/75)= [UK/WalesGVAper worker]*[0.60/0.554.]. Implied
Welsh GVA per worker 81.2 per cent of UK. With UK employment rate
Wales would achieve the same GVA per head. Back