Are government spending and taxes too high (or too low)?

Flemming, John ; Oppenheimer, Peter

The Review is pleased to give hospitality to CLARE group articles, but is not necessarily in agreement with the views expressed; responsibility for these rests with the authors. Members of the CLARE Group are M.J. Artis, A.J.C. Britton, W.A. Brown, W.J. Carlin, J.S. Flemming, C.A.E. Goodhart, J.A. Kay, R.C.O. Matthews, D.K. Miles, M.H. Miller, P.M. Oppenheimer, M.V. Posner, W.B. Reddaway, J.R. Sargent, M.FG. Scott, Z.A. Silberston, J.H.B. Tew, S. Wadhwani and M. Weale.

Targets and Accounting Concepts.

The UK Conservative government in the 1990s declared a central medium-term objective of its economic policy to be a reduction in the ratio of public (government sector) expenditure to GDP to below 40 per cent. Projections accompanying the 1996-97 Budget envisaged this target being achieved in 1997-98.

The target was expressed in terms of GGE(X), an aggregate which combines central and local government spending, but excludes items financed out of the proceeds of the National Lottery through the charity and heritage boards, and calculates government outlays for payment of national debt interest on a net basis, i.e. subtracting the government's receipts of interest and dividends. Whatever the justification for these adjustments in terms of accounting principles, they have the effect of making the proclaimed constraint on public expenditure (slightly) less severe than it otherwise would be.

On the other hand, GGE(X) also excludes receipts from the privatisation i.e. sales of (formerly) public enterprises. In other words, it does not treat them as a negative public expenditure item. Some presentations of the public-sector accounts still do show privatisation receipts in this way, by analogy with the netting out of sales of fixed assets (council homes, surplus land etc) in the accounting of public-sector capital spending. In the latter case, however, what is being calculated is the total of public-sector fixed-capital formation requiring to be financed. Enterprise privatisation sales are more usually and correctly classed as financial transactions which reduce public-sector assets, just as borrowing increases public-sector liabilities. See Tables 1a and b for the historical record of GGE(X) and other components of the Public Sector Borrowing Requirement (PSBR).

In terms of short-term policy decision making, the government focuses on a somewhat narrower expenditure aggregate, the so-called Control Total. This is arrived at by subtracting from GGE(X) those expenditure items which are most affected by cyclical factors and are therefore not subject to government expenditure policy in the short term (see Table 2). They comprise debt interest and cyclical social security (i.e. Unemployment Benefit and Jobseekers' Allowance plus income support paid to people of working age). There are also various accounting adjustments to adapt the accruals-based (national-income-accounting) concept of GGE(X) to the cash-based concept of the Control Total. All this means that the latter is some 15 per cent smaller than the former.(1)

Philosophy

The government's policy towards public expenditure reflects its wish to limit the economic role of the state and enhance that of markets and individuals. This broad objective impinges on public expenditure from different angles. First and foremost, public spending has to be financed by either taxation or borrowing. The government attaches high priority to limiting the weight of taxation (especially direct personal taxation), partly because of the distorting effects it is alleged to have on work incentives, and partly because it interferes with individual freedom of choice. The government attaches correspondingly less importance to the redistributive role of taxation and public expenditure in favour of the poor. [TABULAR DATA FOR TABLE 1A OMITTED] In addition the government aims to minimise the volume of public-sector borrowing in financial markets. The main reason is presumably that debt service will require taxes to be levied (sometime) in the future. At any rate there is no evidence to justify the other possible reason, namely that borrowing by the UK government in recent decades has in significant measure crowded out UK private borrowers from credit markets. Charts 1 and 2 reveal no positive correlation between the level of either government debt or borrowing and long-term real interest rates. The latter were particularly high between 1982 and 1989 when both the former were exceptionally low.

For whatever reason, the restriction and even elimination of government borrowing appears to be a subsidiary principle underlying the government's strategy. The cash measure of borrowing is the PSBR, which is projected to fall steadily from 7 per cent of GDP in 1993-4 to zero in 1999 and to become negative thereafter. The national-accounts measure is the Public Sector Financial Deficit (PSFD). This further differs from the PSBR by being struck prior to financial transactions such as privatisation receipts (which reduce government borrowing needs) and lending to other sectors (which increases them). In the past decade the scale of privatisation receipts has caused the PSBR to be considerably smaller than the PSFD, but this will cease to be true in the years ahead.

A second set of considerations underlying government [TABULAR DATA FOR TABLE 1B OMITTED] [TABULAR DATA FOR TABLE 1C OMITTED] [TABULAR DATA FOR TABLE 2 OMITTED] policy has been the belief that the profit motive is the only spontaneous and reliable spur to efficiency. Public production of goods and services which is not profit-driven has been viewed with suspicion as harbouring a chronic tendency to slackness and waste. This feeling seems to have become stronger over time, as a result of the performance of the privatised utilities and other enterprises such as British Airways. When the business privatisation programme first got under way in the early 1980s(2), the prominent objectives were to reduce the PSBR, to widen share ownership, and to shrink the public sector. Increased efficiency on the part of the industries concerned was well down the list, because the scope for it was not adequately visible. The scale of subsequent labour-force reductions and all-round productivity gains came as an eye-opener. The government was thereafter determined to reduce the number of its employees. One method of doing this has been to push the service activities concerned into the private sector - for which, on occasion, the government was prepared actually to increase its expenditure. Another method, in cases where it was unrealistic to privatise the service, is simply to enforce greater economy of staffing - with almost inevitable, though unstated, implications for reduced quality of service.

In general, what needs to be appreciated is the limited relevance of the privatisation experience to the control of public expenditure. This is essentially because ownership of enterprises is, in principle, quite separate from the channels through which money flows to purchase the output of those enterprises. Even when the utilities were publicly owned, the goods and services they produced were market products sold to individual customers at (in nearly all cases) ordinary market prices. By contrast, collective goods, such as defence and law-enforcement, or merit goods, such as health-care and education, are purchased for the most part not by individual customers but on their behalf by the government. The fact that the production activities involved are undertaken by government employees may be viewed largely as a coincidence. If these activities were privatised, in part or in whole, this in itself would have no first-order effect on the volume of public expenditure.

Indeed the government has privatised ordnance factories and naval dockyard facilities, and has 'contracted out', on a competitive basis, services such as cleaning and catering in schools, hospitals and military establishments - just as it has always purchased from private manufacturers hospital equipment, school desks, books and many other material inputs. Likewise assorted civil service functions have been put out to executive agencies with supposedly more cost-conscious modes of operation. In all these cases, however, the Government remains the final purchaser of the services being provided, which means that such privatisations or contracting out will, apart from timing, affect the public expenditure ratio only by altering production costs of the inputs involved. So far any cost reductions achieved by contracting out have been insignificant. Claims have been made from time to time that contracting out has achieved substantial reductions in the cost e.g., of hospital cleaning; but it has not been shown that such economies are big enough to have perceptible effects on aggregate expenditure, or that reductions in the cost of services have been achieved without equal loss of service quality.

It might be asked whether more radical or sweeping privatisations in these areas could produce better results. After all, in the inter-war period, when corporate control had only recently left the hands of owner-managers, new conglomerates or amalgamations such as the big railway companies, ICI and Unilever were being run by bureaucrats barely distinguishable from civil servants. The idea that public or private ownership would make little difference was natural. Since then, even large-scale private enterprise has learned to place a premium on entrepreneurship, while public-sector managers are seen to have become somewhat bogged down in the pursuit of producer interests (as predicted by Public Choice theories).

Alternatively, efficiency might be further enhanced by applying the new tougher private-sector managerial style to a public sector grown flabby. To do so may require replicating private-sector incentives and ownership. If a hybrid were to be sought in the interest of overall cost-minimisation, it should attempt to combine tough private-sector management with the cheap finance uniquely available to the Government as borrower. As it happens, however, the Government's Private Finance Initiative, whose projected scale is indicated in Table 2a, attempts to inject expensive private finance into a residual public sector! The merit, in the government's eyes, is that this reduces, in the short run, its need to raise finance in its own name either by taxation or through a higher PSBR. To be precise, the PFI, like the privatisation of dockyards, will affect the time pattern or composition of public spending by removing present capital investment outlays and subsequently adding the payments needed to furnish the privately owned capital with its normal remuneration.

Table 2a. Private finance initiative: estimated capital spending

[pounds] million

 1996-7 1997-8 1998-9 Total

Defence 30 80 210 320
FCO/ODA 10 10 10 30
Agriculture 10 20 10 40
Trade and industry(a) 10 10 10 30
Education and
employment(b) 20 40 50 110
Transport 1120 1320 1260 3700
Environment(c)(d) 30 30 30 100
Home Office 50 110 60 220
Legal departments 10 20 10 40
National Heritage 20 30 30 80
Health 170 200 300 670
Social Security(a) 130 70 100 300
Scotland(c) 140 360 420 920
Wales 60 150 150 360
Northern Ireland 50 80 80 210
Chancellor's Department 40 40 30 110

Total 1900 2570 2760 7240

Notes:

(a) Joint Bodies Agency/Post Office Counter project recorded against
Social Security.

(b) Excludes private finance activity in education institutions
classified to the private sector.

(c) Docklands Light Railway.

(d) In addition about [pounds]4-5 billion a year of private
investment is levered in through housing, urban regeneration and
other programmes.

(e) Includes Forestry Commission.

Source: Financial Statement and Budget Report, 1996-7, November
1995.

Choice May Prove Costly

Leaving aside, however, the particular shortcomings of the PFI, the case for state intervention in, for instance, education does not require that the state itself provide the education. Indeed the liberal case for intervention on the grounds that a private market would fail to take into account relevant externalities, limitations of the capital market and, in a few cases, parental neglect, is better met by vouchers than by public provision. Imagine, then, that all schools become private-sector institutions, and that all families with school-age children receive annual vouchers entitling them to some basic level of education for their offspring. Certain corollaries follow, especially if, as any politician would surely pledge, the standard of the education provided to those least likely to supplement their voucher entitlement was to be maintained.

First, standards would have to be independently monitored if the fulfillment of that pledge were to be open to assessment. More importantly, subject to this constraint, vouchers would probably result in a higher public-expenditure ratio for education than obtains today.

This is for three reasons:

1. those who presently opt out of state education would, in a fully liberalised scheme, be eligible to receive valuable vouchers - the taxpayer would, through vouchers, be paying for the (basic) education of more children than in present state schools;

2. many of those who do not at present opt out would choose to buy more when the quantity (or rather quality) was more subject to their choice. The supply of teachers is not perfectly elastic - more means a higher price per unit;

3. this variation in quality would be supplied by competing educational enterprises, which would break down the oligopsonistic and monopsonistic structures by which teachers' pay has been restrained in the past - though weakened trades unions might offset this.

Any private educational sector is likely to be dominated by charitable organisations not fully commercially profit orientated. This may moderate their response to the incentives represented by a liberal voucher system. It may also reduce the likelihood that efficiency would rise in such a system sufficiently to offset the factors listed above as making it more costly, in terms of public expenditure, for a given minimum quality of schooling.

The proposition that meeting assured minimum standards for all would cost the state more under a liberal regime of consumer choice applies to health no less than to education, and for largely the same reasons. The UK National Health Service incorporates a highly effective cost control mechanism - or at any rate did so until the recent restructuring of the system and the partial introduction of 'internal markets', which appears to have produced an enormous increase in administrative costs unmatched by savings elsewhere. NHS cost control has been achieved partly (as with teachers) by monopsonistic restraints on the pay of doctors, nurses and other staff, and partly by the rationing of access to hospital and nursing-home facilities. The suggestion that the present Government is going to introduce rationing to the NHS is absurd. Rationing has been endemic to the NHS from its inception, because it has not seemed socially desirable to meet the total demand arising at a zero price. The only question is how the rationing is to be done and on what criteria. The rationale of asking doctors to both establish and implement the criteria is that doctors are also responsible to a large extent for determining the demand that has to be rationed. Patients cannot know enough in most cases to decide what treatment they should be seeking - but even if doctors decide what demands should be met, it would be far too extravagant for them to be allowed to do so on medical grounds alone, without regard to cost.

Questionable Efficiency Gains

The foregoing argument concerning expenditure on education and health under a liberal regime of consumer choice is consistent with the view frequently reported in opinion polls that the electorate is prepared to pay higher taxes in order to finance improvements in services. Such a statement of preferences need not be wholly honest. But supposed lack of honesty may also reflect the fact that voters feel themselves powerless to ensure that higher taxes will in fact be put to their preferred uses. Moreover, the public would appear to be rather more honest than the politicians who, faced with such expressions of opinion, describe their policy in terms which both exaggerate the resources being made available and simultaneously imply that cuts are harmless and indeed helpful, because they bring about greater efficiency in service provision.

The verbal obfuscation in question involves two devices. The first is to talk about any item which retains its share of GDP, and also some which almost but not quite retain their share, as enjoying 'an increase in real terms'. This is at least defensible inasmuch as items of money expenditure do actually increase by more than the general rate of inflation. But then one needs to remember that, whenever average real incomes rise because of (genuine) national productivity improvements, any labour-intensive service with little or no scope for such productivity improvements has to have more spent on it 'in real terms' just to enable a constant number of its providers to obtain the same pay rise as everybody else. Hairdressing is a good - and uncontroversial - example.

The second device is to announce 'efficiency savings' or 'efficiency gains' when what is meant is reductions in funding. In cases where funding cuts are less than the chosen level of efficiency gains, the difference is actually claimed to have been 'ploughed back' into the service, as though there were an increase in funding. Whether one thinks this defensible or not depends on how far one considers politicians may legitimately go in putting a favourable gloss on their policies when presenting them to the public - a point exemplified further below.

If the feeling exists that we get insufficient value at the margin for our tax money, it may be reasonable for politicians to pledge to cut back (the growth of) public expenditure (relative to GDP), but such a pledge is essentially incomplete as a manifesto without spelling out how it is to be done and what is to be lost. To promise unspecified cuts in expenditure is as open to the challenge of 'where will the money come from (the cuts fall)?' as is a promise to raise expenditure without identified sources of finance. Reliance on 'efficiency gains' is merely an evasion of the issue, since the existence of hitherto unrealised opportunities to raise productivity in services without lowering their quality will be exceptional rather than the rule.

A similar evasion is implied by Mr Dorrell's instruction in June 1996 that the number of children's intensive care units is to be increased without provision of additional funds. Presumably he would say that local NHS managers know better than he does where economies can be made or lower value services displaced. But how does he know that these services are less valuable, as opposed to less publicisable, than those of infant intensive care?

Optimal Public Spending and Taxation

In a simple, probably oversimple, model of optimal public finance, the budget is balanced, with public expenditure and taxation being jointly determined as follows.

We start with a representative consumer and an arbitrary restriction to distorting taxes (such as a proportional income tax). In such a model there is no scope for redistribution, because all consumers are, by definition, identical; but there might be public goods. The effective price of such goods in terms of private consumption goods is the cost of the resources they incorporate plus the deadweight cost of the distorting taxation necessary to finance their public provision. The second component generally rises more than proportionately with the marginal tax rate. In the case of a tax on earnings, it depends crucially on the elasticity of supply of labour. The suggestion, or finding, that this elasticity had increased in the last few decades could therefore serve as an argument for reducing the level of taxation and of public spending. We know, however, of no study suggesting that such an increase has in fact occurred - although it may be higher for married women who now constitute a larger part of the labour force. The only other argument for lower taxes in this model would be that the preference for public goods had weakened. We do, of course, have no direct evidence of that having occurred either.

In any case there is a deep flaw in the representative agent model, or rather in associating that model with distorting taxation. In the absence of heterogeneity (e.g. skill and wealth differentials) there can be no argument against lump-sum taxes - such as the notorious poll-tax - which do not distort. In fact there may be scope for non-distorting lump-sum taxes even when people do differ. All poll taxes prior to Mrs Thatcher's differentiated by social status. Dukes paid more than knights who paid more than dustmen.

If differences in earning power rule out lump-sum taxes, we can save our model but we also have an opportunity to redistribute income by taxing the rich to give to the poor. The tax can no longer be proportional, but it can still be linear, giving a fixed amount to anyone with no other income. The extent to which such redistribution should occur will in turn depend on two parameters. One is the extent to which the taxpayers, probably the adult majority and thus the controlling power over decisions in a democracy, identify with the recipients. This is sometimes called 'social solidarity'. The other is the number of the deserving sick, old or unemployed eligible for support. The 'preference' for redistribution would typically take the form of saying that the least well off should have a particular proportion of what other people (on average) have. Such a preference implies that if there are more sick, more old, or more unemployed, taxes should be higher - even if, at the same time, the support given to each is indeed adjusted downwards.

It is sometimes said that taxes, such as those needed to finance redistribution or the supply of public goods, reduce the nation's competitiveness, and that in a world of increasing competition, particularly from the Far East, we can no longer afford to express our social solidarity to the same degree. This is not very convincing. Any fiscal handicap to competitiveness can be offset by an appropriate reduction in pre-tax wages. If external competition makes the nation less rich than it otherwise would be, that is no reason to impose disproportionate penalties on the retired, the unemployed or the disabled.

We have already seen that an increase in unemployment or inactivity would, by increasing the number of beneficiaries, increase the revenue necessary to maintain a given ratio of benefits to earnings. It would then be appropriate, other things being equal, to reduce the replacement ratio in order to limit the rise in the distorting tax. This could occur at a given level of social solidarity, regardless of the reasons for rising inactivity. If, however, rising inactivity is widely attributed to shirking on the part of adherents of an alternative or alien dependency culture, it is then also likely that social solidarity, the degree of identification of those in work with benefit recipients, will fall.

It follows from what has been said that in the model of redistribution we would expect the scale of transfers - and the level of distorting taxes - to be higher where there is more social solidarity, a higher incidence of incapacity or unemployment, and (again) a lower elasticity of supply of labour.

Where there are both disadvantaged people and public goods the two mechanisms interact in a fairly straightforward way. A greater demand for redistribution or for public goods leads, if met, to higher tax rates which raise the cost of meeting unchanged demands for the other uses to which public expenditure can be put. It follows that an increase in any one use should be met only partially by higher taxation and partly by a crowding out of other uses.

Thus, production of public goods should rise with:

the taste for public goods

and should fall with a rise in:

social solidarity the number of the disadvantaged the elasticity of aggregate labour supply

Transfers should rise with:

social solidarity the number of the disadvantaged

and should fall with a rise in:

the taste for public goods the elasticity of aggregate labour supply

while tax rates should rise with:

the taste for public goods (average tax rate - to increase revenue) social solidarity (marginal tax rate - to increase progressiveness) the number of the disadvantaged (both) the inelasticity of labour supply (both)

A persistent rise in the normal level of unemployment - say by 5 percentage points in a non-cyclical context, such as seems to have occurred since 1970 - would with an unchanged income replacement ratio, impose costs which could not be financed by borrowing. If the unemployed were representative workers and labour force participation rates were unchanged, output would fall by about 5 per cent. If the replacement rate was 50 per cent and labour's share in GDP 2/3, the job losers would receive 1.75 per cent of base GDP, 1.85 per cent of new GDP or 2.8 per cent of earnings.

The arguments given above imply that in these circumstances tax rates should indeed rise, and both generosity to the unemployed and the provision of public goods decline somewhat. Within this sort of comparative-static framework, to expect to be able to reduce tax rates while normal unemployment was rising presupposes changes in the taste variables - either social solidarity or the preference for public goods would have to be falling too.

These textbook exercises relate to the theoretically optimal levels of taxation and expenditure as determined by the levels of public preferences, solidarity and labour supply responsiveness. While one can infer a consequent relationship between changes in each set of variables, this may not do justice either to the incremental nature of policy decisions or to the perceived political cost of rapid adjustment.

Gradual adjustment has, for instance, been achieved in the ratio of pensions to earnings since the indexation of old-age pensions was switched from earnings to prices and as real earnings have risen. Changed perceptions of the relevant parameters may have led to the decision to change the policy, but they do not plausibly account for the path along which the ratio has, in the event, fallen.

The argument hitherto has assumed that we were comparing alternative equilibria of our political economy. If taxation and expenditure had each diverged from the trend that was optimal given evolving preferences etc., the attempt to adjust them downwards might merely be designed to correct previous failures to optimise. Evidence here too is inevitably hard to find. In its absence it is as likely that recent changes have represented a departure from an unchanged optimum as a return to it.

Trends in Total Government Spending and Taxation

The aim of keeping taxes, and hence government spending, as low as possible has acquired an air of a new political consensus. Despite doubts cast upon it by opinion polls, Tony Blair clearly believes that it is necessary to join the denunciation of earlier 'tax and spend' philosophies if New Labour is to have its turn in power. The reduced political acceptability of taxes seems to be just that, a diminished preference for public relative to private goods, or a decline in social solidarity, in which the role of politicians as leaders or followers is unclear. At any rate, there has been no demonstrated rise in the welfare costs of distorting taxes.

Politicians sometimes imply that there is a need to counter some inexorable tendency for the ratios to GDP of public expenditure, and hence of taxation, to rise, or to have reached, by inadvertence, insupportable levels. Such a tendency might be ascribed to the political consequences of universal suffrage. Taking a very long-term historical and wide-ranging geographical view, the twentieth century has seen a big rise in the public-expenditure and taxation ratios in all industrial countries. But if one adopts the more specific frame of the United Kingdom in the past 25-30 years, the picture is different (see Tables 1a, 1b and 1c). In particular:

* the trend of both expenditure and tax ratios since the end of the 1960s has been flat, with a range of fluctuation of about 6 percentage points of GDP.

* GGE(X) peaked in 1975-76 at over 47 per cent of GDP; the tax ratios peaked in the years 1981-83, total current receipts at 46 per cent and non-North sea taxes and national insurance contributions at 39 per cent.

* non-North Sea tax and NIC receipts at present amount to some 36 per cent of GDP, about 1 percentage point higher than when the Conservatives were elected to power in 1979. GGE(X) in 1995-6 was 42 per cent of GDP, the same as in 1978-79.

* compared with other OECD countries, the UK figures were about average in the mid-1970s but are now well below average. In the mid-1990s the only European industrial country with a lower public expenditure ratio than Britain was Switzerland at 38 per cent, about the same as Australia. Japan and New Zealand were fractionally lower at 37 per cent, and the USA lower still at 33 per cent.

The tax figures mean, of course, that income tax reductions in Britain during this period were one element of a restructuring from direct to indirect taxation, which involved large increases in VAT and excise duties. The old marginal income tax rates on high incomes (before 1980) of 85 per cent on earnings and 98 per cent on investment income were absurd, and probably did not contribute to revenue (cf the Laffer curve). Top marginal rates (with national insurance included) of 50-60 per cent are probably appropriate, but in fact that particular limit does not now bind on higher incomes. The highest net marginal rates - allowing for withdrawal of benefits as well as imposition of tax - unquestionably apply at the bottom end of the income spectrum. The Government has shown much less enthusiasm for remedying this than for reducing marginal rates at the top end, partly no doubt because it is much harder to achieve without either an unacceptable cut in all benefit levels or a serious sacrifice of revenue to the Exchequer. (See the article by Tony Atkinson on pp. 90 of this Review.)

The claim before the event that reduced marginal tax rates at the upper end would unleash entrepreneurial energies and reduce the highest pre-tax salaries has not proved to be justified. On the contrary, pre-tax hourly (or per-unit-of-effort) pay differentials have widened substantially. If highly skilled or energetic individuals supplied more input as the tax rate fell, the relative price of their time/effort should have declined. It has not done so. Demand must have shifted - or been manipulated - in their favour.

The widening of differentials benefited income tax revenues. The halving of top marginal rates has, in fact, been followed by a sharp rise in the share of income tax receipts from higher earnings. Some of this may be attributed to a lessening of income tax avoidance and evasion measures.

By far the largest single category of UK government spending is the social security budget. In the years 1992-95 it amounted to roughly one-third of the total or 13 1/2 per cent of GDP. This category consists of transfer payments, as opposed to purchases of goods and services. The major programmes in the latter category are healthcare, education and defence. Health and education together make up about 30 per cent of today's public spending or over 12 per cent of GDP (see Table 3).

More significantly, social security is responsible for much of the upward pressure upon the UK public expenditure ratio in the past two decades, its share of GDP having risen from 9 to 13 1/2 per cent. On the goods and services side, outlays on health and personal social services increased their share of GDP by 1 1/2 percentage points. Increases elsewhere, for example on law and order, were more than offset by cuts in the share of defence, housing and other areas. In what follows we comment on some aspects first of expenditure on goods and services and then on social security.

Education

In the case of UK public education, its statistical share of UK GDP has remained roughly constant in the past quarter century, at a little over 5 per cent, with some cyclical fluctuation (see Table 3). In the school sector, which accounts for the bulk of spending, this has been consistent - [TABULAR DATA FOR TABLE 3 OMITTED] thanks to a drop in pupil numbers following low birth rates after 1970 - with a modest decline in the pupil/teacher ratio, although that ceased in the early 1980s and at a level which compares very unfavourably with independent schools. Indeed, the pupil/teacher ratio in independent schools has continued to decline steadily, so that in the mid-1990s it was only half that in the state sector, a bigger gap both relatively and absolutely than that obtaining 20 years earlier (Table 4).

The pupil/teacher ratio, though acknowledged as an important index of quality, is not the only indicator of relevance. Direct measures of pupil attainment have shown that average English secondary school graduates are typically a year or more behind their continental counterparts in terms of mathematical and other competence(3). In addition, there are two areas where UK education provision has lagged well behind that of continental European countries, namely pre-school (nursery) facilities and part-time vocational or continuing education for early school leavers, particularly bearing in mind that a far higher proportion of pupils leave school at 16 in the UK than in the rest of Europe.

However, cross-country data on public education spending, whether as a share of GDP or per pupil in the respective institutions, do not suggest that the UK's shortcomings are a consequence of any overall financial [TABULAR DATA FOR TABLE 4 OMITTED] parsimony (Table 5). For example, UK expenditure per primary and secondary school pupil in 1992, though below those of Scandinavian countries, Austria and (surprisingly perhaps) Italy, looks not ungenerous by comparison with Germany or the Netherlands. Britain comes higher up the ladder with primary than with secondary education. All this suggests that a balance needs to be struck between the case for greater financial provision in [TABULAR DATA FOR TABLE 5 OMITTED] toto and other measures, such as redistribution of existing resources and stronger focus on academic attainment for pupils of average ability. The National Curriculum, and approximately quadrennial assessment of attainment levels, represent moves in this direction, but considerable further progress will be required to match continental European standards. (See the Commentary by Sig Prais on p.3 of this Review.)

When it comes to higher education, the 1990s have been a period of particularly rapid change in the UK, so comparisons need to be made with caution. Historically, the position has been that Britain sends a much lower proportion of its 18+ population than other industrial countries to higher education, and especially to university; and moreover that the length of their study courses is on average about one-third less than in other countries. However, the amount of public money spent per student has been at the top of the range, to a large extent because students have been more intensively and personally taught. One consequence has been a considerably lower drop-out and failure rate than in most other higher-education systems. Meanwhile, the fraction of GDP devoted to higher education has not been out of line with that in other countries (Table 5).

The changes that have come about chiefly since the mid-1980s have increased very substantially the proportion of school-leavers proceeding to higher education; the figure is still below that of most industrial countries, but by much less than formerly. In the mid-1990s about one in three 18-year-olds goes on to higher education, compared with one in six in the early 1980s and one in twelve in the early 1960s. There has, however, been no corresponding expansion of public spending on higher education since 1980. Consequently the student/teacher ratio, after rising very slowly for some fifteen years up to the mid-1980s has climbed steeply from about 8 1/4 (in universities about 10) to 13 1/2 in 1993-4 (Table 6). In addition, average pay in the sector has been kept significantly below the pay of other occupations formerly considered comparable.

The insistence on greater student throughput per unit of resource in higher education, together with a proliferation of bureaucratic control procedures, Teaching Quality Assessments, Research Assessment Exercises, and other investigations affecting the distribution of [TABULAR DATA FOR TABLE 6 OMITTED] funds, have intensified uncertainty and competitive pressures within the sector, and have weakened both professional solidarity and mutual commitment between employer and employee.

Health (NHS)

In the case of the National Health Service, a gently rising share of GDP has gone together with, and is to a large extent explained by a striking change in the aggregate production function of health-care services (presented in Tables 7 & 8). The number of hospital beds has been declining for several decades, and since 1971 alone has fallen by over 40 per cent. At the same time medical staff numbers are more than 20 per cent greater than they were in 1971. Staff numbers per hospital bed have thus more than doubled. In the case of doctors - both GPs and hospital doctors - the increase in numbers has been uninterrupted. In contrast, hospital technician numbers peaked in the early 1980s and numbers of nurses in 1989, not surprisingly given the continuing fall in the quantity of beds to be attended.

The underlying change may be summarised as labour-using technical progress in health-care - or to be more precise, in acute health-care. The phenomenon is an aggregative one, comprising many different strands. Some of these strands run counter to the overall trend, for example the development of expensive items of equipment such as scanners, or advances in the care of premature babies. Others are matters of fashion, for example the abandonment of routine tonsillectomy. The main strands, however, which have contributed to the aggregate trend include progress in anaesthetics, which has drastically shortened periods of post-operative recovery - more so than progress in surgery itself, though that too has been important in various fields (eg., ophthalmology) - wider use of chemo- and radio-therapy in cancer treatment and advances in pharmacology, such as the [TABULAR DATA FOR TABLE 7 OMITTED] development of psychotropic drugs for treatment of mental illness, which has been critically important in reducing the in-patient population of mental hospitals.

Separate from the technical development of acute health care is the growing need for chronic care of the elderly. In Britain as in other industrial countries, life expectancy has risen in the past few decades, while a generation of low birth rates is influencing the size of the younger adult population. The proportion of elderly people is rising, with particularly sharp increases in the oldest groups (over-80s - see [ILLUSTRATION FOR CHART 3 OMITTED]). To some extent the associated health-care needs are reflected in the total staff numbers already noted, since the elderly also make disproportionately heavy use of acute services. It is noteworthy, however, that staff numbers in personal social services delivered by the public sector have not increased in the past decade, although they rose quite steeply in the preceding fifteen years. A growing share of provision in this domain has come from private sources; and a growing [TABULAR DATA FOR TABLE 8 OMITTED] share of the costs has been carried within the private sector. This is in line with government policy objectives, besides being in many cases desirable. Care of the elderly is more suited to private-sector organisation and finance than some other aspects of social services, e.g. care of disadvantaged children.

It raises, however, the question of what proportion of total prospective demand in this area can realistically be financed from private sources, and of the proper contribution of public policy to the encouragement of such private finance. The 1996 Budget relaxed somewhat the asset rules governing eligibility for government assistance with the cost of residential care. The level of assets below which entitlement to assistance begins was doubled to [pounds]16,000, while the level below which assistance is comprehensive with no contribution expected from the recipient was more than trebled to [pounds]10,000. The government also announced its intention to encourage greater financial provision for long-term care, possibly via partnership schemes which would involve more generous asset eligibility rules for those with appropriate insurance arrangements in place. The difficulty will be to attract sufficient numbers into such insurance arrangements to get their cost down to a reasonable level. The optimum may be universal compulsory participation, accompanied by a measure of public regulation - as with the natural monopolies. If the government shrinks from this, it may be drawn into a measure of insurance subsidisation, at least in the earlier stages.

The obfuscation generated by references to 'efficiency savings' in public expenditure is particularly striking in relation to the NHS. As a specific illustration, consider the November 1995 post-Budget press release of the Department of Health. Referring to 'a substantial increase in resources for patient care', the Secretary of State, Stephen Dorrell, said:

'Current spending on health will grow by [pounds]1.3 billion in 1996-97, equivalent in real terms to 1.6 per cent or some [pounds]500 million. Hospital and community health services current spending will grow by 1.1 per cent in real terms.

Patients can look forward to further improvements in NHS services next year made possible through the combination of the new money announced today, through further improvements in efficiency - focusing resources from management onto patient care - and through ensuring that pay rises are earned through increased efficiency, in line with our policy on public sector pay.

For 1996-7, I am again setting health authorities a challenging target for efficiency savings of 3 per cent - equivalent to [pounds]650 million - which will be ploughed back into the service, in line with our past pledges. An important contribution to this will be the planned savings of some [pounds]130 million - 8 per cent in real terms - from a reduction in the management costs of NHS trusts and the costs of running health authorities'.

A rough explanation of this gobbledegook is as follows. Begin with the [pounds]130m planned cuts in administrative costs of NHS trusts and Health Authorities. This is a cash figure. The constant price ('real terms') equivalent is [pounds]50m. Deducting it for the moment from the total efficiency savings of [pounds]650m demanded of Health Authorities leaves [pounds]600m or about 2.75 per cent to find. Why choose this figure, given that efficiency savings are anyhow to be 'ploughed back' into health spending? The answer is that the government would like NHS pay rises to be limited to this figure, which is equal to the forecast rate of general price inflation. Assuming real wages in the NHS thus unchanged, the announced real-terms increase in current health outlays of 1.6 per cent for 1996-97 represents provision for a renewed rise in staff numbers or other current expenditures (including additional pay rises). Clinical staff numbers may further benefit from the aforementioned cut in administrative outlays, approximately equal to 0.16 per cent of total NHS spending.

In the long run pay rates must be expected to rise at the same average rate in the NHS as in other parts of the economy. It sounds impressive to say that from 1977 to 1994 spending on health 'increased by around two-thirds in real terms...., equivalent to an average growth of just under 3 per cent a year'.(4) About seventy per cent of this 'real growth', however, reflects the rise in real wages over the period as economy-wide productivity increased; that part only represents any increase in physical capacity in health care to the, probably very limited, extent that labour productivity has risen there too. The remainder is essentially accounted for by the modest net increase in NHS staff numbers and a shift in the composition of staff towards more highly qualified, and therefore expensive, individuals (i.e., more doctors relative to nurses, technicians etc).

The government's demand for visible 'efficiency savings' is met in the form of increased 'activity rates' which are supposed to be required by health-care purchasers (District Health Authorities and Fundholding GPs) as a condition of agreeing contracts with health-care providers at higher rates of pay for the forthcoming year.(5) The activities are measured in units such as finished consultant episodes (FCEs), community contacts, outpatient attendances and so forth. Contracts are settled in advance on the basis of forecasts and promises. This means that many increases in activity rates will be fictional, and that there are few if any genuine payments per item of service. Such payments arise systematically only in the case of extra-contractual referrals (ECRs) across Health Authority boundaries. Moreover, as with any physical planning system, activity goals create questionable incentives. The number of FCEs is increased by referring patients from one consultant to another. The average period on surgical waiting lists can be shortened by bringing in non-urgent cases after 6 months instead of 9 or 12, while urgent cases are made to wait for six weeks instead of two. The Patients' Charter (introduced in April 1992) guaranteed admission for treatment no later than two years after being put on the waiting list. It also stated that patients are entitled to information on maximum waiting times in their locality. Since then the proportion of patients waiting over twelve months has fallen (from 18.3 per cent in 1991 to 6.3 per cent in 1994 - see note 4), but - surprise, surprise - the total number of people on the waiting list has increased. Doubtless many of them are on the list for 11 1/2 months.

More important, the activities that are being measured are not final outputs but intermediate services, whose relation to the desired final output is far from being assured or constant. It is like rewarding financial portfolio managers by their volume of transactions or rate of turnover. In the case of portfolio managers a simple measure of final output is available, namely the return on the portfolio. But not, unfortunately, with health. Health outlays contribute to indicators such as infant survival, life expectancy, and incidence or prevalence of certain diseases. But the quantitative extent of their contribution is uncertain, and generally modest by comparison with other factors such as general prosperity, life styles and environmental conditions. Some of these factors are themselves influenced by public policy, but not from the side of health care: doctors may lobby against cigarette advertising or leaded petrol, and in favour of jogging, but they are not responsible for action in these areas. At the same time, health spending includes important activities which make no contribution to standard indicators of population health e.g. terminal care.

All this also means that the huge extra administrative expense involved in drawing up thousands of individual purchaser/provider contracts is bound to be largely wasted, because the objects of negotiation are the intermediate activities and not the desired final output, which is of course not readily defined in negotiable quantitative terms. The orders of magnitude involved may be gleaned from the twentyfold increase in the number of NHS managers in 1988-94, raising their salary bill in real terms from [pounds]26m to [pounds]500m (or about 1 1/2 per cent of NHS spending).(4)

Defence

Defence expenditure has actually fallen in recent years, as Table 3 shows, not only as a proportion of GDP but also in real terms. The ending of the cold war raised the prospect of a 'peace dividend' available either to reduce taxes or to permit higher public expenditure on other things.

An implication of the model presented above, where it was suggested that optimum expenditure on public goods and transfer payments would, at least in part, tend to crowd one another out, is a similar relationship between different public goods. Increased demand for defence means some extra taxation and some reduction in civil public expenditure. This linkage should obviously operate symmetrically when defence needs fall.

There are, however, two qualifications on the simple harvesting of the peace dividend. The first is that, although the ending of superstate confrontation probably reduces demand for nuclear weapons, they are actually cheaper to maintain in being than a comparably effective 'conventional' force. But the end of the cold war may increase the demand for conventional types of forces, by the unleashing of petty nationalism as we have seen not only in the former Yugoslavia but also in various parts of the Former Soviet Union. We may then need more of the forces deployed in the Gulf, or infantrymen to keep the peace in Bosnia or Northern Ireland. Thus the impact on the composition of our defence effort may be bigger than that on its cost.

The second point is that the resources committed to defence industries are very specialised. We know that defence conversion is a major task for the former USSR. It is only easier for us to the extent that we are cutting less and relied less on exclusively military and remotely located industrial complexes.

The lesson of US experience, and of our own earlier defence cuts, is that the labour freed from military production is not always rapidly reabsorbed into civil industrial employment. If there is a dividend, it is reaped not when a military factory is closed but when its former employees stop drawing the dole and start contributing to civil output and paying taxes.

The difficulties of this transition are likely to lead people to seek to sell abroad the military production no longer taken up by the British forces. Not only is this undesirable as it contributes to the probably destabilising proliferation of weapons, but it is also unlikely to be profitable given that Eastern Europeans too are slashing the prices of their arms exports. Depending on the scale of the arms concerned, such sales could at best buy a breathing space. The idea that the reduction in mainstream orders is likely eventually to be reversed is almost certainly an illusion.

Thus, all in all, we have reaped something of a peace dividend and may continue to do so for a little longer as those made redundant from defence industries find productive re-employment elsewhere.

Social Security

As mentioned above, a rising share of GGE(X) in recent years has been spent on Social Security which has risen to one-third of the whole. This in turn can be disaggregated into support for the unemployed, for pensioners and other items as in Table 3. It is likely that much of the extraordinary increase in sickness benefits since the mid-1980s represents a displaced form of unemployment assistance, consequent upon the tightening of criteria for payments of overt unemployment benefit. Certainly there has been no disproportionate increase in rates of sickness benefit, nor any reason to suspect a genuine deterioration in the health of the working population.

We have already noted the historical trend of apparently rising normal levels of unemployment which the UK has shared with much of Europe, and which we saw was likely to raise National Insurance Contributions by about 3 percentage points or taxation more generally by 2 per cent of GDP. The high cost of unemployment compensation intensifies debate about the form it takes. For how long should the able unemployed receive higher levels of benefit than other social security beneficiaries? How stringently should the tests of availability for work be applied? What training programmes should the unemployed be required to attend? We cannot pursue these questions here, but adjustment in these dimensions of our arrangements should probably precede those in the basic level of benefits themselves.

Lower levels of employment, lower birth rates and longer lives while retirement ages were fixed (at 60 for women and 65 for men) have inevitably increased the 'dependency ratio' of pensioners to the working population. Indeed during recessions, and as part of corporate 'downsizing', effective retirement ages have actually been falling while life expectancy and fitness among the elderly have been rising. While most of the financial burden of early retirement has hitherto been carried by private pension schemes, there is almost certainly some spillover onto state pension take-up as well.

In general, longer life and health (see [ILLUSTRATION FOR CHART 3 OMITTED] and Table 9) have pointed to a higher retirement age; recession, downsizing, and possibly more rapid obsolescence of skills, may have pointed to a lower age. The issue has also been complicated by the requirement of equality between the sexes in these respects, eliminating the earlier retirement age of women (who normally live longer than men) as well as by the argument that a lower retirement age would increase employment opportunities for the young. There may be some truth in this last argument, but it is rendered irrelevant by the fact that any reduction is likely to 'grandfather' most existing employees. This means that the number of retirements does not rise for decades to come, for which time the conjunctural appropriateness of the delayed impact is unknowable. In the short term there may be some reclassification of the older unemployed as retired; this, of course, is only a presentational effect, nonetheless attractive to politicians for that.

More generally, there is much to be said for greater neutrality in the effect of state pension schemes on retirement ages, by making contributions and benefits mutually more congruent at the margin. For example, persons who postpone drawing their state pensions for a number of years arguably do not receive adequate incremental benefit later as a result. Under present rules one needs, as a first approximation (or rule of thumb), to live to more than 88 years of age before a five-year deferment of state retirement pension becomes worthwhile. It will be interesting to see whether the Labour Party's promised anti-ageism will extend to banning compulsory retirement ages altogether, as has occurred in the United States.

On demographic grounds some rise in retirement age would seem desirable to reduce dependency ratios - which are also raised by the larger periods over which the young are now being educated. Recurrent education or retraining may also be the answer to the rapid obsolescence of skills acquired when young.

As shown in Chart 4, on present policies the prospects for public expenditure on pensions in the UK are very much more favourable to the low tax lobby than is the position of most other OECD countries. This is for three reasons. The first is that the demographic changes raising the dependency ratio are operating much more moderately in the UK than elsewhere. Among the G7 countries, the ratio is expected in the next few decades to be lower than in the UK only in the US.

The second reason is that in 1980 the basis for up-rating basic state pensions shifted from a link with earnings to a link with prices. In the recession of the early 1980s real earnings actually fell, so that the initial impact of the switch was the opposite of that intended i.e. it raised the real income of pensioners relative to the working population. Since then, however, earnings have outstripped prices by about 2 per cent per annum. Thus the basic state pension has fallen by 25 per cent relative to average earnings.

Table 9. Expectation of life(a) in the United Kingdom: by sex and
age

 years

 1901 1931 1961 1991 1993 1996 2001 2021

Males
At birth 45.5 57.7 67.8 73.2 73.6 74.4 75.4 77.6

At age:

1 year 54.6 62.4 69.5 73.8 74.1 74.8 75.7 77.9
10 years 60.4 65.2 69.9 73.9 74.2 75.0 75.9 78.0
20 years 61.7 66.3 70.3 74.2 74.5 75.3 76.1 78.2
40 years 66.1 69.3 71.4 75.1 75.4 76.3 77.2 79.3
60 years 73.3 74.3 74.9 77.7 77.8 78.6 79.5 81.4
80 years 84.9 84.7 85.2 86.4 86.4 86.8 87.2 88.2

Females
At birth 49.0 61.6 73.6 78.7 78.9 79.7 80.6 82.6

At age:

1 year 56.8 65.3 75.1 79.2 79.3 80.1 80.9 82.8
10 years 62.7 67.9 75.4 79.4 79.5 80.3 81.1 83.0
20 years 64.1 69.0 75.6 79.5 79.6 80.4 81.2 83.1
40 years 68.3 71.9 76.3 80.0 80.1 80.9 81.7 83.5
60 years 74.6 76.1 78.8 81.9 81.9 82.6 83.3 84.9
80 years 85.3 85.4 86.3 88.3 88.3 88.8 89.1 90.0

Notes:

(a) Total number of years which a person might expect to live.

Source: Government Actuary's Department; Social Trends, 1996.

We know from recent debates that the present Government adheres to an absolute concept of poverty. If the absolute standard of living obtained by pensioners was acceptable in 1980, the same absolute standard should be acceptable, they suggest, in 1996. This argument is highly controversial. For a number of reasons adumbrated above, social solidarity and tax tolerance might well point towards relative incomes as a basis for such transfers. Nevertheless, these same reasons would imply that adverse demographic developments (a fall in the share of the population at work) would lead not only to higher contributions/taxes, but also to lower pensions relative to earnings.

The other complication is that the government view relates not to what would be satisfactory standards of life but to acceptable guaranteed minima. This is the third reason for Britain's favourable-looking fiscal position on the pensions front. The tendency in Britain - with government encouragement - is for people to supplement the state minimum with their own private pension, either on their own account or, more usually, through their employers. Such supplements are relatively much bigger in Britain than in most other industrial countries and are not recorded on the graphs in Chart 4, which accordingly underweight the pensions sector of the British economy. An advantage claimed for the British system is that, unlike state pensions, private pensions are funded. As a result, they may raise total savings and thence national wealth. On the other hand, it would, in principle, be possible to fund state pensions and the administration of private schemes may be disproportionately costly. These costs may deter people from buying adequate pensions for themselves and/or represent social inefficiency.

The upshot of all this is that the freezing of the basic state pension at its real value of 1980 is likely to come under increasing pressure over the years, even if occupational and individual supplementary provision does increase. On the assumption, however, that the present policy is maintained, OECD projections suggest that the UK tax burden would fall further behind the European average over the coming decades.

Britain could, on its own, save more now, investing both at home and abroad to provide for better pensions in the future. If, however, all the industrialised countries together attempt to follow this path, they run two risks. The first is that pursuit of higher savings through a rising fiscal surplus may involve reducing aggregate demand and activity, thereby sacrificing output and, if anything, depressing savings. The other is that, notwithstanding recent theories of endogenous growth, a general rise in investment would encounter traditionally diminishing returns, unless matched by the extra labour input available from an extended working life and higher average ages of retirement.

Public Spending and Macroeconomic Prospects

From the wider macroeconomic viewpoint, the government is to be commended both for giving priority over tax cuts to the containment of the PSFD and for combining restrictive intentions in the fiscal sphere (even if these intentions have not yet delivered a decisively reduced deficit) with a relatively relaxed monetary policy. The background to this is that inflation has been substantially lowered and has reverted to the annual rates which characterised the 1950s and 1960s. At the same time, the economy retains a margin of spare capacity; the sterling exchange rate remains unpegged; and Britain's current balance of payments continues in deficit and looks vulnerable to further deterioration in the next consumer spending boom. While wage pressures remain moderate, it is appropriate to promote export-led expansion by means of monetary policy designed to keep the exchange rate at a suitably competitive level.

Success in permanently raising employment and lowering the average level of unemployment would both raise revenue and reduce social security spending. We have seen that the latter aspect alone might amount to 1 per cent of GDP if half the rise in unemployment between 1975 and 1995 were to be reversed.

We have referred to the likelihood that the government will come under pressure sooner or later to improve the real level of state pensions. Equally powerful cases can be made for allowing an extra 1/2 or 1 per cent of GDP to flow into public provision for health care and education over a number of years. This would not constitute a great bonanza but would permit very worthwhile, even urgent, improvements not presently attainable, while preserving appropriately rigorous control on total budgetary resources.

NOTES

(1) The government itself, or at least its printers, seem to have some difficulty remembering what is included where. In the 1996-7 Financial Statement and Budget Report, paragraph 6.73 on p.134 contradicts earlier paragraphs by stating (clearly incorrectly) that GGE(X) includes privatisation proceeds.

(2) Sales of local authority houses had begun under the preceding Labour government in the 1970s.

(3) On this and the following facts see the numerous articles on schooling, vocational training and economic performance in Britain and various continental European countries published in the Review over the past fifteen years by S.J. Prais and colleagues.

(4) Sally Prentice, 'Health Policy and Care' in D. Halpern et al eds. Options for Britain (Dartmouth 1996), p. 162.

(5) NHS Executive, Costing for Contracting Manual, Leeds, Department of Health 1993.