Introduction
Modern health care systems cannot be financed from the out-of-pocket expend-iture of patients. The mismatch between individual resources and health care needs dictates that the costs of individual care should largely be met from the pooled contributions of groups. In principle, these groups or third parties could take several forms, from extended families and voluntary associations, through commercial and social insurance programmes, up to the state at the national or regional level. In practice, however, the advantages of scale and the funda-mental limitations of private insurance markets have led to the predominance of public institutions. In almost all industrialized countries, most health care is paid for either by governments, with funds raised from various forms of taxation, or by social insurance institutions, largely or wholly outside the commercial market-place, which impose compulsory levies on all or most of the population.
All financing systems, whatever their structure, can be represented by a basic identity adapted from the fundamental income–expenditure identity of national income accounting. The total amount raised to pay for health care for a particular population, through whatever channels, must equal exactly the total amount spent on health care for that population, and that in turn must equal the total amount of income earned, in various forms, by those paid (directly or indirectly) for providing care. This identity of revenue, expenditure and income is not a theory but a logical necessity and is funda-mental to understanding both the effects of, and the controversies over, all financing and funding systems. This can be expressed as follows:
TF+ SI + UC + PI = P × Q = W × Z
where TF is the amount of revenue raised through tax financing, SI through social insurance, UC through private, out-of-pocket payments or user charges and PI through private insurance premiums. P and Q are vectors, listing the average prices (P) paid for and total quantities provided or used (Q ) of each of the various forms of health care. W and Z are also vectors, standing for the amounts of different types of resources (Z) used in providing care and the rates of payment (W) of those resources. One element of Z, for example, could be nursing hours worked, and the corresponding element of W would be the average rate of reimbursement per hour worked. P and Q correspond to units of output such as physician visits, surgical procedures or drugs.
Tax-financed systems are those in which most health expenditure is derived from tax payments (TF > SI + UC + PI), or at least the tax-financed component is substantially larger than any other component. The various questions that one might raise about the behaviour and relative performance of different financing systems can be posed, at a very general level, in terms of the compon-ents of this identity relation. Do predominantly tax-financed systems differ more or less consistently from those drawing more heavily on other sources of financing on some or all of the most significant dimensions of system per-formance as reflected in the internal structure of the identity?
The question is complicated, however, by the fact that tax financing is not a standard process. In principle, the government pays for health care from a general revenue fund into which all taxes flow. Most countries have several levels of government, and the powers and responsibilities of each level vary considerably by country. The characteristics of tax financing vary according to the amount of involvement of different levels of government.
Another question is how best to assess the performance of different financing systems. There are many ways to categorize the dimensions of health systems performance, including the distribution of burden and benefits across the population, the allocation of resources to the health sector and among its various sub-sectors, and technical efficiency and responsiveness. Assessing the strengths and weaknesses of different financing systems under each dimension would be a monumental exercise, especially because much of the necessary data currently exist in fragmentary form, if at all. Moreover, system perform-ance on several of these dimensions depends more on how providers are funded than on how payers raise revenue from the population. Providers are organized and paid in very different ways in Canada, Finland, Sweden and the United Kingdom, for example, although they all have tax-financed systems.
This chapter, therefore, focuses primarily on the first dimension of perform-ance, the distribution of burdens and benefits, for which considerable com-parative information exists.
The distribution of the burdens and benefits of a health care system can be represented along three axes:
•
Who pays – and what share?•
Who gets – what and when?•
Who gets paid – and how much?I explore these in the first three sections. This is followed by a section on the potential of tax-financed systems versus other financing systems to control
expenditure. Next, I explore the role of covert taxes. Finally, I discuss several policy implications.
Who pays – and what share?
The most clear-cut difference among alternative financing systems is how they apportion the total cost of health care among the national population. Tax financing includes this burden within the general tax system; in most high-income countries, tax liability is roughly proportional to high-income or mildly progressive (Wagstaff et al. 1999). People with higher income thus contribute, through tax financing, a share of their incomes that is the same as, or larger than, that contributed by lower-income people. Out-of-pocket payments, by contrast, whether co-payments or payments for private uninsured services (user charges, UC), are proportional to the use of care and not related to income. Accordingly, user charges for health care consist, on average, of a much larger share of the incomes of lower-income people.2
These generalizations are well illustrated in two North American studies, one in Manitoba, Canada (Mustard et al. 1998a,b) and one in the United States (Rasell et al. 1993, 1994). The Manitoba study is especially interesting, as it links individual-level administrative records from the universal public programmes covering hospital and physicians’ services with census records of family incomes and estimated tax liability for much of the provincial popula-tion. The distribution of expenditure and of corresponding tax liability by income decile (scaled up to the whole provincial population of about 1 million) is displayed in Figures 2.1 to 2.4, with the small but expensive institutionalized population as a separate category.
Figure 2.1 shows the amount (in Canadian dollars) spent by the public plans on the care of people in each income decile in 1994; Figure 2.2 shows the estimated amount of tax contributed according to income decile; Fig-ure 2.3 shows the difference, by income decile, between the total cost of care used and total taxes paid; and Figure 2.4 shows this gain or loss as a share of total family income. (Permanently institutionalized people have no significant income.)
The scale of the transfers is quite striking, especially from the top income bracket, making very clear that people in that group would benefit from lowering the tax-financed share and introducing some form of private pay-ment. Since people with very low incomes are unlikely to be able to bear a substantial portion of the costs of their own care, any shift in financing from tax financing to private payment would transfer funds primarily from the middle to the upper deciles of the income distribution.
The Manitoba study focused only on the public, tax-financed programmes.
In the United States, using survey data for non-institutionalized people only, Rasell et al. (1993, 1994) analysed the distribution by income decile of payments for a more comprehensive definition of health care through tax financing, user charges and private insurance. Their results are displayed in Figures 2.5 and 2.6; Figure 2.5 shows the pattern for the whole population and Figure 2.6 distinguishes households with heads 65 years or older and those
Figure 2.1 Expenditure (in millions of Canadian dollars) on publicly financed health care according to pretax income decile (10 is the highest income) in Manitoba, Canada, 1994
Expenditure in millions of Canadian dollars
400 350 300 250 200 150 100 50
0 INS. 1 2 3 4 5 6 7 8 9 10
Lowest Income decile Highest
INS.= institutionalized population
Sources: adapted from Mustard et al. (1998a,b)
Figure 2.2 Tax contribution (in millions of Canadian dollars) to publicly financed health care according to pretax income decile (10 is the highest income) in Manitoba, Canada, 1994
INS.= institutionalized population
Sources: adapted from Mustard et al. (1998a,b)
Tax contribution in millions of Canadian dollars
500 450 400 350 300 250 200 150 100 50
0 INS. 1 2 3 4 5 6 7 8 9 10
Lowest Income decile Highest
Figure 2.3 Net transfer (in millions of Canadian dollars) to (positive numbers) or from (negative numbers) each pretax income decile (10 is the highest income) from the public financing of health care in Manitoba, Canada, 1994
Net transfer in millions of Canadian dollars
400 300 200 100 0
−100
−200
−300
−400 INS. 1 2 3 4 5 6 7 8 9 10
Lowest Income decile Highest
INS.= institutionalized population
Sources: adapted from Mustard et al. (1998a,b)
Figure 2.4 Net transfer to (positive percentages) or from (negative percentages) each pretax income decile (10 is the highest income) as a share of pretax income from the public financing of health care in Manitoba, Canada, 1994
Sources: adapted from Mustard et al. (1998a,b)
Net transfer as a % of pretax income
20%
15%
10%
5%
0%
−5%
−10% 1 2 3 4 5 6 7 8 9 10
Lowest Income decile Highest
Figure 2.5 Expenditure on health care in the United States as a percentage of pretax family income according to family income decile (10 is the highest income) and type of expenditure, 1987
Source: adapted from Rasell et al. (1993)
Health expenditure as a % of family income
35%
30%
25%
20%
15%
10%
5%
0% 10a
(95–100%) 10b
(90–95%) 9 8 7 6 5 4 3 2 1
Highest Family income decile
Lowest
Taxes Premiums Out-of-pocket Figure 2.6 Expenditure on health care in the United States as a percentage of pretax family income according to family income decile (10 is the highest income), type of expenditure and the age of the head of household 65 years or older (left column in each decile) versus younger than 65 years (right column in each decile), 1987 Source: adapted from Rasell et al. (1993)
Health expenditure as a % of family income
30%
25%
20%
15%
10%
5%
0%
1 2 3 4 5 6 7 8 9 10b
(90–95%)
Lowest Family income decile Highest
10a (95–100%) Taxes
Premiums Out-of-pocket
under age 65 years. People 65 years or older are covered, for hospital and physicians’ services, by Medicare, which is national, universal and tax-financed.
These findings emphasize the contrast between the progressivity of tax financing in the United States and the regressivity of both user charges and private insurance. (The similar pattern for both modes of private financing is what one would predict a priori in an efficient competitive insurance market.) These highly regressive components of the financing mix, widespread in the United States, overwhelm the progressivity of the tax-financed component and make the whole distribution highly regressive. Perhaps surprisingly, how-ever, this pattern is found even among elderly people, who are covered by the universal tax-financed plan. The very substantial deductibles and co-payments built into the public programme, ostensibly to control overall costs, contrib-ute to making the overall mix markedly regressive. Individuals can and do buy Medigap private coverage for these charges, as do people in France to cover the co-payments in the statutory health insurance scheme. But Medigap cov-erage, being private, is also regressive in its distribution of financing burden (premiums are based on risk status, not income), as in France.
Tax financing and user charges thus provide the clearest contrast in who pays. Tax financing places a heavier financial burden on those with higher incomes, whereas user charges place more on those with lower incomes. This fairly obvious difference motivates much of the policy controversy over altern-ative forms of financing, generating a permanent tension in every national health care system. In addition, tax financing detaches payment liability from the experience of ill health, or at least the use of care, whereas user charges link the two directly. Regardless of income, sick people will contribute rela-tively less and healthy people more under tax financing than under user charges. Financially, extending the scope of user charges is a wise strategy for the healthy and wealthy, and extending tax financing reduces the share of the burden borne by the unhealthy and unwealthy.
Where private insurance is widespread, as in the United States and Switzer-land, it generates a highly regressive pattern of distribution similar to that of user-charge financing. Competition in private insurance markets forces insurers to adjust the premiums of enrollees according to their relative risk, which, in practice, means according to their past claims. Thus private insurance, like user charges, links individual contributions to illness experience rather than income; both are highly regressive compared with tax financing.
Social insurance, on the other hand, bases contributions on income, but the income base is not all-inclusive, and some systems place a ceiling on contribu-tions.A priori one might expect social insurance systems to be more progres-sive than private financing, but less so than tax financing – an early finding of the ECuity Project (Wagstaff et al. 1993).
Tax financing is predominant among the northern countries, including Canada, Denmark, Finland, Sweden and the United Kingdom, and in southern Europe – Italy,3 Spain and Portugal. Figure 2.7 presents summary estimates of the progressivity or regressivity of total health care funding in these and several other countries (excluding Canada), plotted against the percentage of health expenditure financed by taxes (Wagstaff et al. 1999). Unfortunately, the source data are now a decade or more old.
Figure 2.7 Estimates of the progressivity (Kakwani (1977) Progressivity Index) of total health care expenditure in 12 OECD countries in various years according to the percentage of health expenditure financed by taxation
Key: DK, Denmark (1981, 1987); FI, Finland (1990, 1996); F, France (1984, 1989);
D, Germany (1989); I, Italy (1991); NE, Netherlands (1987, 1992); P, Portugal (1990);
E, Spain (1980, 1990); SW, Sweden (1980, 1990); CH, Switzerland (1982, 1992);
UK, United Kingdom (1993); US, United States (1987).
Sources: Wagstaff et al. (1999) and calculations by U. Häkkinen for Finland
Kakwani Progressivity Index for total health expenditure
0.10
Percentage of total health expenditure financed by taxation
Although the financing mixes and structures may change, the implications of the ECuity Project about the general effects of different structures should hold. Increases in co-payments since the early 1990s (as in Sweden or Germany), the introduction of not-for-profit institutions with fiscal incentives to insure these and other private payments (as in France) and the expansion of oppor-tunities for private practice should increase the regressivity of the financing system, improve access for people with higher incomes and increase the ex-penditure on, and incomes earned in, the private sector and possibly overall.
The proportion of health spending financed by taxation is strongly and positively correlated with the progressivity of total health expenditure. Any statistically fitted relationship, however, would be dominated by the two outliers, the United States and Switzerland, in which private spending is both relatively high and very regressive. If these outliers are excluded, the correlation between tax financing and progressivity becomes less clear. Social insurance systems (low tax financing) may be either progressive or regressive, depending on their structure and policies. The key feature is comprehensiveness (Wagstaff
et al. 1999). Germany has a ceiling on social insurance contributions and, in both Germany and the Netherlands, more affluent people are permitted or required to opt out and buy private coverage. In France, however, social insur-ance covers the whole population, without premium ceilings or floors. The result is actually more progressive than tax-financed Denmark, Finland and Sweden, despite the user charges and the privately funded mutual benefit associations. On balance, the ECuity Project data show most tax-financed systems to be more progressive, or less regressive, than most social insurance systems, but predominantly tax-financed systems show no clear pattern of system progressivity rising with the proportion financed by taxes.4
The contrast between the United Kingdom and Denmark, Finland and Sweden is striking. All rely heavily on tax financing, but the United Kingdom has the most progressive funding system of all those reported, and Denmark and Sweden are mildly regressive overall. Finland is especially interesting, having had one of the most progressive funding systems in 1990. The eco-nomic and fiscal crisis of the early 1990s led to a sharp reduction in the proportion of tax financing, a rise in the private share and a corresponding abrupt move from progressive to regressive.
General taxation was progressive in all countries studied but much more so in some than in others. This variation results partly from the tax mix.
Tax revenue is raised through both direct and indirect taxation, with the former being consistently progressive and the latter regressive in all countries reported. The progressivity of direct taxation also differs across countries.
Direct taxes in Sweden and Denmark have a very low degree of progressivity (reflecting the importance of proportional income taxes at the local level), in marked contrast to the United Kingdom. Direct taxes in Finland were formerly more progressive but have become much less so during the 1990s. When weakly progressive direct taxes are pooled with regressive indirect taxes, there is little overall progressivity left.
In general, the extent of reliance on tax financing among the countries studied in the ECuity Project appears to be inversely related to the progressivity of the tax system. Denmark and Sweden, which rely heavily on direct taxes, have relatively little progressivity in their direct tax structure. The United Kingdom, with more progressive direct taxes, increased the share of regressive indirect taxation in its tax mix from 43.2 per cent in 1985 to 53.9 per cent in 1993. These three countries, along with Italy and Spain, had the least pro-gressive systems of general taxation of all countries studied.5 By contrast, the countries with the most progressive systems of general taxation – the United States, Switzerland, the Netherlands and Germany – make least use of tax financing to support health care.
This pattern strongly suggests a political compromise in the conflict of economic interest between the healthy and wealthy and the unhealthy and unwealthy, in which what is lost on the roundabouts is made up on the swings. Every financing system has this conflict, but the terms of the com-promise vary across countries not only in the extent of redistribution but also in the balance of financing sources through which it is achieved. A political coalition in support of tax financing can be assembled and maintained, so long as the redistribution is not too extreme.
The ECuity Project estimates suggest that the considerable egalitarian potential of tax financing is in practice much more limited.6 But limited is not negligible.
There is a sharp contrast between the tax-financed systems and the highly regressive distribution of the overall financing burden in the systems drawing heavily on private financing. Social insurance systems can go either way, depending on how they are structured.
Who gets – and when?
The experience of tax-financed systems since the early 1990s underlines the continuing tension between the economic interests of the healthy and wealthy and those of the unhealthy and unwealthy. Even proportional or mildly regres-sive tax-financed systems redistribute substantial sums, because the experience of illness and therefore the use of services (in the absence of financial barriers) is so much more regressive. Pressure for more private funding, for a shift of financing mix from tax financing to user charges, with or without private insurance, is therefore permanent in any tax-financed system. The arguments have changed little over the decades and the underlying economic interests have not changed at all.
Tax-financed systems seem to be especially vulnerable to this pressure, however, in times of general fiscal crisis. It is not difficult to understand why, when general incomes are falling, citizens should resist more taxation and govern-ments should therefore respond by controlling public spending more strictly, including health spending. Nevertheless, why these cuts should be associated with a shift in the redistributional compromise is not so obvious. The answer
Tax-financed systems seem to be especially vulnerable to this pressure, however, in times of general fiscal crisis. It is not difficult to understand why, when general incomes are falling, citizens should resist more taxation and govern-ments should therefore respond by controlling public spending more strictly, including health spending. Nevertheless, why these cuts should be associated with a shift in the redistributional compromise is not so obvious. The answer