The Normative Economics of Health Care Finance and Provision: Oxford Review of Economic Policy January 1991

Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/242012274

The normative economics of health care finance and provision

Chapter in Oxford Review of Economic Policy · January 1991

CITATIONS READS

74 1,930

1 author:

Anthony John Culyer


The University of York
401 PUBLICATIONS 7,250 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Wefarism and welfare View project

Efficiency and Need View project

All content following this page was uploaded by Anthony John Culyer on 02 October 2014.

The user has requested enhancement of the downloaded file.


OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5, NO. 1

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


THE NORMATIVE ECONOMICS OF
HEALTH CARE FINANCE AND
PROVISION
A. J. CULYER
University of York}

I. INTRODUCTION insurance, private finance of other kinds (such as


out-of-pocket payments), and private provision of
Whereas in many countries the 'crisis' in medical health care itself.
care has been seen in terms of 'excess' spending on
health services, in the UK it has been seen (at least Few of these contributions have been informed by
by most of those who manage and work in the NHS) the work of health economists, particularly their
as a crisis of 'underfunding'. This has come about normative work. There have been several reviews
as the result of the government's successful at- of the empirical literature (e.g. Culyer, Donaldson
tempts to restrain the rate of growth of real spending and Gerard, 1988; Culyer, Brazier and O'Donnell,
on the NHS, which in turn reflects the govern- 1988). This essay is a review of the main contribu-
ment's belief that its principal effective weapon tions of a more conceptual kind.
against what it perceives to be inefficiency in the
NHS is to challenge management (itself reformed I have adopted the rather general term 'normative'
and to some extent liberated) by systematic finan- in my title, rather than 'welfare economics' for a
cial squeezes. The same concern has given rise to reason. This is that much of the modem systematic
a host of proposals for reform of the NHS most of approach to policy questions in the health territory
which involve a much greater role for private is based on an approach that transcends (and may

1
I am grateful for the helpful comments of Richard Amould, Gwyn Bevan, Paul Fenn, Alisudr McGuire, Gavin Mooney,
John Posnett, and Alan Williams. The usual disclaimer applies.

34 0266-903X/89$3.00 © OXFORD UNIVERSITY PRESS AND THE OXFORD REVIEW OF ECONOMIC POUCY UMITED
A. J. Culyer

come to supersede in part) the 'welfarist' tradition inputs than were technically necessary to produce
of normative economics. an output, that cannot be cost-effective) and the
cost-effectiveness version of efficiency is also
A key concept underlying the current discussion is embodied in the next concept of efficiency. Being
'efficiency*. This is, however, far from being an cost-effective necessarily eliminates as 'inefficient'
unambiguous notion and, because it is also so some technically efficient combinations of inputs,
central a concept (one indeed that pervades the but it leaves unsettled the question of the efficient
entire debate) it is the one that preoccupies me in rate of production. In economic jargon, cost-
this paper. The practical context in which the effectiveness means 'being where an isocost line is
concept can illuminate is, evidently, scarcely less tangential to an isoquant*.
important. For the purposes of this paper, the ideal output: obtaining when cost-effective out-
practical context is taken to relate to broad policy puts are produced at a rate that i s ' socially' optimal
questions about the 'efficiency' (somehow defined)

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


and allocated to individual members of society in a
of ways of financing health care (e.g. through 'socially' optimal fashion. This is 'top-level' effi-
private, competitive insurance or via compulsory ciency and arises from combining supply-side and
public insurance), and of the user-prices that it the demand-side considerations. In a society with
should carry (e.g. whether 'free', partially subsidized, limited resources it entails establishing rates and al-
not at all subsidized, differentially subsidized locations of outputs that are such that no alternative
according to particular categories of user). rates of reallocations can be perceived that are
'better'. In economic jargon it entails setting mar-
These are the principal 'demand side' contextual ginal rates of transformation on the production-side
questions whose discussion forms the bulk of the equal to marginal rates of substitution in consump-
paper. There is, of course, a host of related ques- tion.
tions that arises in specific policy formulation and
a whole raft of supply-side 'efficiency' questions. While this trio is an extremely useful sorting device
These may from time to time be alluded to here, or in discussions of efficiency, the main focus in this
cross-references made to other papers in this vol- paper is on the third, though (as will be seen)
ume. They are not, however, the main focus. ambiguities about the concept of' output' will force
us also to look to some extent at the other two
The conventional ways in which economists use notions of efficiency as well. Of the various tricky
the term 'efficiency' (which at this level of gener- things embodied in 'ideal output' as a concept of
ality are illuminating and ideologically innocuous) efficiency, one that has come to the fore in health
are the following: economics in a distinctive way concerns the mean-
technical efficiency: obtaining when for a given ing of 'socially' and 'better' (or 'optimal').
output the amount of inputs used is minimized or
(what is the same thing) when for a given combination (i) Welfarism and Extra Welfarism
of inputs, the output is maximized. This is a supply-
side concept that is a necessary condition embodied One approach in health economics, which has
in the subsequent concepts of efficiency. In economic become the traditional one in economics as a whole,
jargonitmeans 'being onanisoquant'. Since there is what Sen (1977) calls 'welfarist'. This is very
is usually more (many more) than one way of much in accord with liberal political opinion and
producing an output that meets this condition, the asserts that social welfare (any increase in which is
implied balance of resource use even for a given 'better' than none) is a function only of individual
target output is not unique. welfare (or utility) and judgements about the
cost-effectiveness: obtaining when for a given superiority of one state of the world (defined by
output the cost is minimized or (what is the same reference to these utilities) over another are made
thing) when for a given cost the output is maximized. irrespective of the non-utility aspects of each state.
This is also a supply-side concept It embodies Moreover, the individual welfares (or utilities) are
technical efficiency (clearly if one were using more a function only of goods and services consumed.

35
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5, NO. 1

The other approach, which might be termed 'extra- More recently an explicit departure from welfarism
welfarist', relaxes what its adherents see as an was advocated by Williams (1972) and discussed
undue information restriction in welfarism so that further by Sugden and Williams (1978) in the
other aspects of each social state are also embodied context of cost-benefit analysis. This is the 'deci-
in the judgement. Since, in the welfarist approach, sion-making' approach which is contrasted with
the basis of social welfare—or diswelfare—is only the 'Paretian' approach. The latter is one example
the utility got from goods and services (including of welfarist analysis in that social welfare is a
labour services), an important class of 'extra' welfare function only of the utilities to individuals of goods
sources is the non-goods characteristics of individuals and services. The former allows that governments
(like whether they are happy, out of pain, free to (and other 'decision makers') may have other ob-
choose, physically mobile, honest). Extra-welfarism jectives than the making of (actual or potential)
thus transcends traditional welfare: it does not Pareto improvements, which may involve not only

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


exclude individual welfares from the judgement imposing their own values on the consumption of
about the social state, but it does supplement them individuals (rather than those of the individuals in
with other aspects of individuals (including even question) as with merit goods, and their own view
the quality of the relationships between individuals, of the appropriate intertemporal discount rate, but
groups, and social classes). also takes into account some extra-welfarist ele-
ments of choices that pure welfarism excludes. In
This distinction is actually quite old in the history this context the answer to the question 'who decides
of (the so-called 'new') welfare economics. In what entities with what weights go into the social
Bergson's (1938) classic theoretical article, for welfare function?' is 'decision makers'.
example, his social welfare function included un-
specified terms that could be interpreted as extra- The bulk of the health economics literature is
welfarist elements of the sort just described. Buthe decidedly welfarist in orientation—and Paretian to
dropped them in favour of an explicit 'partial' boot. This is discussed in Section n. Section III
analysis after a page. addresses the more recent extra-welfarist tenden-
cies of health economists, and also (since the ap-
Another famous extra-welfarist strand in the wider proach lends itself so readily to this) looks at the
literature is the notion of merit goods (Musgrave, ways in which extra-welfarism has infiltrated the
1959)—goods whose consumption is considered supply-side notions of efficiency.
so meritorious (by government) that they are made
available on terms that are more generous than in Although the discussion is of a normative litera-
the market place. Despite (unsuccessful) attempts ture, it should be pointed out that in neoclassical
to bring merit goods into the welfarist scheme of economics it is easy to slip between positive and
things(e.g. Culyer, 1971), it seems altogether more normative (welfarist-style) according as to whether
preferable to adjust the scheme of things so as to one treats utilities as behaviour-generating indices
incorporate such considerations fully rather than or ascribes direct normative value to them. Thus,
leaving them as a kind of ad hoc 'escape clause' some of the literature on the NHS (e.g. Lindsay,
(Margolis, 1982) lying outside traditional theory; 1969; Culyer, 19716) was ostensibly explanatory:
not fitting into it but necessary in order to prevent the authors sought to provide empirically falsifi-
theoretical emasculation (inability to explain why able accounts of some of the institutional features
some common phenomena are observed, inability of the NHS by postulating the presence of particular
to discuss in a consistent normative framework arguments in individual utility functions. Under
some matters that are of evident normative impor- welfarism, it is quite easy to re-interpret these
tance). This plainly involves the possibility of theories as claims for the Pareto-efficiency of the
overruling individual judgements of value and raises NHS. In the positive interpretation, the institu-
the question not only of the weights to be attached tional features are phenomena shown (or so it is
to individual utilities in a social welfare function claimed) to be the predicted results of utility-
but of who should be assigning those weights. maximizing individuals' behaviour in particular
Should the values of some members of society environments; in the normative interpretation, the
count for more than those of others? institutional features are seen as desirable attributes

36
A. J. Culyer

rather than merely explained phenomena: the theory on its culture-contingency, I shall say no more on
justifies what is or implies what ought to be, pre- that here. The following sections embrace the usual
scribing rather than predicting. 'list' of factors thatmake health care'different' and
I shall indicate the kind of reasoning that relates to
As a final prefatory remark it is worth pointing out each.
that health economics (in common with many other
fields of application within economics) has by and (i) The Competence of the Consumer
large addressed only a part of the agenda of the
'political economy' of health care, to do with The welfarist approach requires that choices made
finance, public or private. There is very little on the by or for consumers be rational in the particular
normative implications of different forms of sense of that term used by economists. Should
ownership or organisation on the provider side. choices not be founded on axioms that include, for
Indeed, the aforementioned positive theories of the

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


example, an ability to compare alternatives and to
NHS (Lindsay, 1969; Culyer, 19716) are typical in rank them consistently (transitively, if weakly) in
that they are concerned entirely with the terms of order of preference, the entire edifice tumbles be-
access to health care. Although it may have frequently cause it no longer becomes possible to infer from
been thought in the past that arguments for' free' or actual behaviour that the choices made were (sub-
subsidized or publicly financed health care were ject to resource constraints) those most preferred.
ipso facto arguments for publicly provided care, Even if consumers were better informed than they
that is not so. Public finance can in principle be typically are about the pros and cons of alternative
combined with public or private provision, just as actions (for example, choices between alternative
privately financed consumption can be combined strategies of personal medical treatment) there are
with public or private provision. That is why it is occasions when even mentally healthy people prove
possible for people who support the general idea of incapable of choosing (especially between alterna-
free health care availabl e at the time it is needed also tives that involve horrid consequences). Mentally
to support private provision, or a public/private mix ill and mentally handicapped people may more fre-
in provision, or competitive 'internal markets' within quently be found to be 'irrational' in this sense.
the public system. However, the focus here is on There are also occasions, of course, in which they
the demand side, and so no further development of are incapable of making any choices at all (for
these possibilities is attempted. example, when they are the traumatic and uncon-
scious victims of accidents).

II. THE WELFARIST APPROACH TO It is common for the consumer, who is almost
HEALTH CARE FINANCE AND invariably incompetent in some degree, to defer
PROVISION some of the judgements involved in clinical choices
to professionals, especially medical doctors, who
It was (and is) common for these issues to be act as agents on his or her behalf: ideally choosing
addressed by considering a set of factors that make in the way the individual would, had he or she been
health care 'different' from other goods or services possessed of the same informational advantages as
and which therefore may constitute a reason for the professional, in a system characterized by what
wanting to allocate it differently from these 'other' Evans (1981) has termed 'incomplete vertical
goods or services (and to evaluate the efficiency of integration' between consumers and producers.
different allocations differently) (Culyer, 1971).
Implicit in this approach is the presumption that The particular technical skill possessed by the
goods and services are in principle best allocated by professional that is relevant here is a better know-
market mechanisms and that departures from that ledge of the effect that health care will have on
mechanism require special warrants. This seems to health. If one supposes that it is health which
be the product of a particular bit of cultural generates utility for individuals, both directly via a
conditioning to which most (Western) economists sense of well-being and indirectly in the sources of
are prone but, since it is my purpose more to outline welfare of which better health enables one to take
an approach than to make sociological comments advantage, then health care is itself only instrumental:

37
OXFORD REVIEW OF ECONOMIC POLICY, VOL 5, NO. 1

a means to an end. Even if the consumer is able and entry to the profession regulated, and for
rationally to value health, he or she will usually subsidized in-service training systems and other
have much less information about the process of professional activities. The existence of such market
medicine and the risks that different processes failures means that there are inevitably nice balances
imply. Contrary to the usual welfarist assumption, to be struck here between, for example, the self-
the buyer is not the best judge of his or her interests regulating monopoly awarded to doctors in medical
but must rely on the seller's advice. It is quite easy practice and its possible abuse by inhibiting
to imagine circumstances in which the selfish interest innovation (e.g. alleged but not actual quackery,
of the professional can conflict with the best interests inefficiency in doing whatever is done, and the
of the patient. Health itself is not a traded commodity, earning of monopoly rents). While economics is
yet that is what the rational consumer may be helpful in identifying the risks, and in predicting the
expected to seek. The traded commodities are consequences of policies designed to mitigate them,
information and health care, which the consumer its role is mainly qualitative: much judgement is

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


unaided cannot normally be expected to evaluate, required in making assessments of the pros and
and neither of which are valued for themselves. cons, and their respective strengths, of alternative
possible arrangements.
The technical expert is thus required to be more
than' merely' a technical expert. In acting as agent, (ii) Supplier-Induced Demand (SID)
the expert as it were 'enters the skin' of the patient
This is what seems to endow a professional Evans' idea that physicians have a target income
relationship with its most important—but elusive and adjust workload (under a fee-for-service system
and delicate—characteristic (Trebilcock et al., 1979). of paying doctors) in response to changes in the
Being delicate, it is vulnerable. It is vulnerable to environment, seems to have grown out of the
cultural assumptions not shared by agent and client empirical observation that regional utilization of
(the male-dominated specialty of obstetrics is often health care is positively associated with the regional
charged with sexist disregard of the interests of its stock of doctors, holding price and other variables
wholly female clientele). It is vulnerable to misplaced constant (Fuchs, 1978; Cromwell and Mitchell,
technological zeal, in which the expert subjects the 1986; Phelps, 1986). The thesis is that physicians
patient to painful diagnostic testing that yields little will induce patients to use more services in order to
usable additional information, or makes presumptions maintain income. A positive association has also
about a patient's trade-offs between short-term sometimes been found between physician stock
risks and long-run benefits. It is vulnerable to class and prices, though this result is even more disputed
bias, in which the professional, as a member of a than the fundamental utilization effect (Sloan and
higher class than at least some patients, is incapable Feldman, 1978; Auster and Oaxaca, 1981; Green,
of 'entering their skins'. It is also vulnerable to 1978). There are, of course, huge econometric and
financial distortions, whereby some systems of empirical problems in testing for SID: only one
medical remuneration encourage the supply of study for example (Pauly, 1980) controlled
services whose principal justification is the income (approximately) forpatienthealthstatus.butRice's
they bring the expert rather than the benefit they claim (Rice, 1983,1987) that experimental rather
bring the patient. This is sometimes referred to as than routine data strongly support an inverse
supplier-induced demand (Evans, 1974,1976). These relationship between reimbursement rates and use
'vulnerabilities' are forms of market failure and are of services seems persuasive.
widely regarded as justifications for regulatory
measures to protect both consumers and doctors; SID is an area in health economics where we suffer
the former from quacks and the exploitation of a an embarras de richesses. The target-income
professional monopoly, the latter also from quacks hypothesis is one possible explanation. But there
(unfair competition) and the unreasonable demands are others too that have never been rigorously
of dissatisfied (and possibly litigious) customers. compared and tested: increasing the numbers of
They also provide the basis for universally observed doctors increases their availability (less distance to
systems by which medical education is determined travel, lesstimeto wait) and hence reduces the 'time

38
A. J. Culyer

price' to patients; there may be 'excess' 2 demands patient (as is usually the case in general practitioner
at the existing stock and a non-market-clearing services in those medical care systems like the
price, so that increasing utilization following increases UK's that still have GPs), may also come to know
in the stock is simply the meeting of previously a good deal of relevant information about the
unmet 'excess' demand. Doctors may be less patient's values, financial circumstances, working
vulnerable to the effects of consumer detection of life, and family context In such circumstances the
SID in communities with a high doctor/population agency relationship is likely to be as perfect as it
ratio since, given extensive consumer ignorance, probably can be. But even in this situation, the
new patients can always be found to replace any agent too has his or her own values, financial
who leave because of the SID they detect—if so, interests, working relationships, family and social
cross-sectional analysis will produce the observed characteristics which need not be congruent to
correlation between physician stock and utilization. those of the patient (Evans, 1984). In both market
It is unfortunate that this unsatisfactory state of the and non-market systems of health care finance and

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


art persists—and is compounded by empirical provision, the medical monopoly is strong which
uncertainty (at least in some minds) about the very seems to produce higher monopoly rents in the
existence of SID at all—given the huge potential more market-orientated systems (but see Lindsay,
threat that SID constitutes to the welfarist approach 1973).
and especially to the pro-market camp within the
welfarist school. If demand curves really do reflect As regards the terms of access for consumers, much
the health-irrelevant preference of suppliers then, of this discussion casts doubt on the usefulness of
depending on the depth of this contamination, the marginal willingness to pay as an adequate repre-
use of willingness-to-pay as an indicator of consumer sentation of the marginal benefit to consumers, but
welfare is more or less illegitimate. Hence the it is not clear what the notion of welfarism requires
considerable amount of passion found in the literature to be put in its place, nor is it clear whether 'free*
on this subject. care under the NHS or under first-pound zero-
deductible health insurance encourages inefficient
rates of use relative to the 'true' optimum. On this
(iii) Where Does this Get Us?
issue, as will be seen, the extra-welfarists are able
to be more forthright.
The chief lesson for health care finance to be drawn
from this catalogue is not that the 'market' is inher-
ently more flawed than the 'state' in the way in (iv) Caring and Sharing
which medical care is financed (or vice versa) but
that there are no simple lessons to be drawn. Market In the early days of the welfarist literature, the idea
systems and their (usual) fee-for-service methods that health care was 'different' because the medical
of paying doctors are prone to violate the underpin- care consumption of others, or the health of others,
nings of welfarism in one way. State systems and may be a direct influence on the welfare of oneself,
their (usual) salary or capitation sytems are prone to was given short shrift. There was early recognition
violate the underpinnings in another. If fee-for- (e.g. Weisbrod, 1961) mat a direct physical externality
service may encourage an excess of interventionist might exist in cases of communicable disease (via
zeal then salaries and capitation fees may do the infection or contagion): an individual in choosing
opposite. Under either system it is common to see or rejecting vaccination may fail to take account of
a lot of interregional variation (McPherson et al., the benefits accruing externally (viz. to others) in
1981;Vayda«a/., 1982). the form of a reduced probability of the others
contracting a disease. This was conceded to provide
The agency relationship must inherently be incom- a case (abstracting from the problems arising from
plete. The professional may know best about the consumer incompetence) for subsidized prices for
instrumentality of health care and, if working in a such services and, in some cases, for making them
system that encourages a concern for the whole compulsory (e.g. for immigrants). But the scope

2
I put the term in quotation marks as I do not wish to imply that the demands are necessarily in excess of a Parcto optimal
rate of use, merely that they are excess in the (positive economics) sense of demand exceeding price at the going full user-price.

39
OXFORD REVIEW OF ECONOMIC POLICY, VOL 5, NO. 1

for interference with the market on this ground limited; the mention of things which 'may be' (but
alone was evidently highly limited and the marginal which are not) measured to find out what actually
externality probably falls quite fast as the proportion is.
of immunized individuals rises.
One of the ironies of the period was that these
The subtler kind of externality, direct interdepend- remarks and others in the same genre were being
ence of the form that makes one person's welfare made in societies that had substantially altered the
directly dependent on the consumption of others, terms of financial access to health care, possibly for
was not taken very seriously (e.g. Lees, 1960,1962; reasons (in part) of externality. At least, such a view
Jewkesera/., 1963; Klarman, 1963,1965; Bucha- would imply the possibility that the subsidy
nan, 1965). The view is well-encapsulated by Lees arrangements were a rational response to a real
(1967): situation rather than the irrational constructions of

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


ideologues unversed in the eternal truths of
It is argued that there is another relevant externality, neoclassical economics. It might be thought that
namely, the disutility felt by an individual at the thought the reasons for this intervention were like those for
that others are not getting adequate medical care. This similar interventions in education and housing.
is no doubt so and is the basis of philanthropic support Since even crude policies might have gone sufficiently
for health services. But the important point here is that far to eliminate most Pareto-relevant marginal
there is nothing special about health services in this externalities, casual observation could not have
regard. Similar disutility is felt at the thought of others revealed massive failures to internalize at efficient
not getting adequate food, clothing, housing, and other rates of utilization: one needed both observation of
goods commonly regarded as necessities. Apart from
what was and what otherwise would probably have
philanthropy, the community approach to this problem
has typically been public subsidies to those in need. Ex- been to make an assessment of the significance of
ternalities of this kind do not establish even a prime externalities; at the very least one needed the
facie [sic] case for the abolition of markets and the imagination to envisage what a counterfactual
substitution of collective arrangements. 'unintemalized' world might look like.

Even where externalities exist, recent studies have What sort of evidence might settle this issue of the
shown that the scope for government spending to cor- existence and size of what some have called 'car-
rect market imperfections and raise the general level of ing' externalities (e.g. Culyer, 1980)? One is intro-
economic efficiency is less than conventional analysis spective. Are sufficient numbers of introspectors
had led us to believe. The implications of these studies
may be summarized thus: prepared to sacrifice some of their own consump-
tion so that others may have more? If so, they care
(a) many externalities are irrelevant to human action in the externality sense. In particular, invite intro-
and the achievement of optimal solutions; spectors to try to imagine a society in which health
care was supplied entirely by commercial organiza-
(b) many relevant externalities can be, and are, dealt tions without public subsidy or by private charities.
with voluntarily; Ask each, given that they had imagined themselves
into such a society, whether they would be willing
(c) the costs of governmental intervention, even when to sacrifice some of what each has so that others
it is 'perfect', may outweigh the benefits; would receive more. If each would (and they may
want to add the side condition that each would
(d) the imperfections of government as a decision- sacrifice only if others similarly placed also made
making and choice-making process may make an sacrifices) then they would be providing evidence
'imperfect' market situation more imperfect. for the existence of caring externalities of this sort.

Such was the dominant idea at that time: the 'noth- One of the issues that arises in this context is the
ing special' about health argument; the distinction question of whether subsidies should be specific
between philanthropy approved at one level of (e.g. in-kind or directly tied to purchases of specific
collectivity (private charity) but not at a higher goods and services as with voucher schemes) or
(governmental) level; the reference to studies show- general (essentially income support). The basic
ing the scope for corrective government action to be line of argument has tended to be that it seems that

40
A. J. Culyer

generalized support may be appropriate for goods is not easily interpreted, partly because in those
and services that are highly income elastic at low systems which have been most studied—North
levels of household income, where there will be a American—fee-for-service and the possibility of
strong presumption that income supplements will supplier-induced demand can cause supply-side
be spent by most families on basic necessities like contamination of the 'pure' substitution effects of
food (so there is no need for a 'National Food changing user-prices. Beck (1974) estimated that
Service'), but that where this is not the case, or the reduction in use following the introduction of
where people are held to be poor judges of their own charges of $1.50 per surgery visit and $2.00 per
welfare, specific subsidies are a more cost-effective house call in Saskatchewan in 1968 was about 7 per
means of promoting consumption of particular cent for the whole population but 18 per cent for the
goods since they utilize substitution effects as well poor. Early studies of price-elasticity found values
as income effects. It should be clear that the in the range -0.4 to -1.0. Subsequent work by
potential superiority of specific subsidies arises Phelps and Newhouse (1974), Newhouse and Phelps

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


from the welfare gain to the subsidy-provider: in (1976), and, most recently and most authorita-
any comparison of alternatives, provided that the tively, the Rand study reported in Manning et al.
subsidy-receiver is not worse off under one than the (1987), found price-elasticities for all health care in
other, that yielding the largest benefit to the sub- the -0.2 to -0.1 range (i.e. a 10 per cent increase in
sidy-provider is Pareto-preferred. user price causes demand to fall by one to two per
cent). The effects in the Rand study were, however,
There has been some speculation about the way in much stronger for the poor, for children's demand,
which inequality in health care consumption enters and especially for the children of the poor (see Lohr
welfarist utility functions. This seems mostly to et al., 1986). These results have tremendous sig-
have arisen in the context of positive rather than nificance: the handicapping effects of, say, un-
treated otitis media in children far outweigh the
normative economics: to provide explanations of
short-term functional impairment, as further social-
the patterns of subsidy actually observed in various
izing and educational handicaps will almost inevi-
societies. From the late 1960s into the early 1970s,
tably become added for children already disadvan-
focus continued to be upon the consumption of
taged from birth.
health services as the principal source of external
'concern' (rather than, say, the effect of such con-
sumption on health), though for some the focus was The implications of the Pauly model are quite
on absolute rates of consumption while for others it explicit: given a similar cost of care per person in
was relative. a particular diagnostic group, efficient internaliz-
ing of the external effect requires a variable sub-
Pauly (1971) argued that the (negatively sloped) sidy: varying from 100 per cent in the case of the
external demand (marginal valuation) for a per- lowest income group to 0 per cent at a sufficiently
son's care was invariate with respect to the identity high level. This is archetypical 'selectivity' in
of the person of concern and that the demand for social policy.
health care was, in general, income elastic. De-
tailed micro empirical studies (all North American) The argument can be seen quite simply in terms of
on this latter question yield little unambiguous Figure 1. In this figure the three marginal willing-
evidence about direct income elasticities (though ness-to-pay curves (marginal valuation curves or
they seem to be positive) because of income-related real income constant demand curves) of individuals
upper limits on out-of-pocket expenses under health A, B, and C are shown. Each is supposed to have
care insurance (see, e.g., Manning et al., 1987), a common' taste' for health care and the difference
though aggregate studies show income elasticities between is postulated to arise from income differ-
of between 1.18 and 1.36, so that a 10 per cent ences (ability-to-pay). The curve EMV is the
increase in (aggregate) income can be expected to external marginal valuation curve reflecting the
lead on average to a roughly 12.5 per cent increase value that members of society other than these three
in health care spending (for a review see Culyer et place upon their consumption. It is assumed to be
al., 1988 and, for a critique, Parkin et al., 1987). the same for all (anonymous) individuals. At a
market price PA, A consumes xA, B consumes Xg
There seems little doubt about the overall negative and C consumes no X at all (when a particular event
slope of demand curves. The econometric evidence strikes, such as a specific illness). The socially

41
OXFORD REVIEW OF ECONOMIC POLICY, VOL 5, NO. 1

optimal amounts (assuming price represents mar- tive method of achieving a given level of equality
ginal social cost accurately) are, however, x^ x*, is by a combination of standard subsidy (but not in
and x*. determined by the condition MSC = MV, + general to reduce price to zero) and enforced denial
EMV for all i. This implies a variable subsidy such of access via nonprice rationing (' abstention' in his
that C faces a marginal price of P c , B faces PB, and terminology) to the better off.
A pays the full price: the subsidy varies inversely
with ability-to-pay. In practice, of course, fine One set of problems arising from this set of exter-
distinctions between individuals will not be pos- nality models is that, while each provides an ac-
sible and general approximations will have to be count of some of the features of health care subsidy
made—but on this rationale they should clearly (e.g. some selectivity, some 'free' care, some de-
have the selective character shown gree of reduced inequality in access), their practical
implementation (assuming that any one of them

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


Culyer (1971; for a simple version see Cullis and underlies the actual practice of democratic societies
West, 1979) attempted a model (again assuming seeking to implement institutional arrangements
conventional signs on the income and price elastici- that promote Pareto optimality more closely than
ties) that implied zero prices, as in the NHS, and would the market unaided) means that the appar-
showed, using a triangular Edgeworth Box, that in ently sharp distinctions between their implications
a two-person world free care will be preferred by at the theoretical level become blurred in practice.
both the subsidized and the subsidizing parties over It is consequently hard to discriminate between
cash transfers, provided that the preference func- them on empirical grounds.
tions of the two individuals produced an appropri-
ate configuration of offer curves. There are also more deep-seated difficulties. One
arises from the characteristic that the externality is
In contrast to these 'absolutist' externality models in each case public so that its internalization is of
of 'caring', Lindsay (1969; see also Cullis and benefit to all whether or not one contributes. It is in
West, 1979) offered a 'relative' model of 'sharing' everyone's interestto 'free ride'. While, on the one
in which equal treatment for equal (medical) need hand, the 'club' view of government as a mechanism
was implied. He showed that the most cost-effec- by which individuals voluntarily agree to be coerced

Figure 1
Consumption Externalities and Selective Subsidies

Price and
marginal
value MV A

_MSC

0=x

42
A. J. Culyer

to provide more optimal levels of public goods can public: if B gives to the charity rather than A, A
be seen as a collective solution to the free-rider derives no utility. However, this seems to lead
problem, it is quite clear that there are several types nowhere if one wishes to build a welfarist model of
of activity (most notably private charitable giving) collective health care subsidies, for it is the very
that are not collectivized and that are fully exposed publicness of the external effect that generates the
to the problem of free-riding. Strictly, the free-rider public subsidy argument. Perhaps a mixture of
problem implies that no-one (save one whose what Margolis (1982) calls 'goods altruism' and
marginal value of the external benefit most exceeds •participation altruism' might be developed that
its marginal cost) has an incentive to contribute. helped to account for the simultaneous existence of
Consequently, charities will have either no supporters collective and private acts of altruism.
or only one. Since this is plainly not so, there seems
to be something wrong with the theory. And if each Another problem has been pinpointed by Sugden

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


is therefore a wrong theory, none can be used to (1980,1982) again attacking the roots of the theory
justify the kinds of public subsidy often seen in of welfarist altruism. Again reverting to the private
health care. The free-rider theorem is an example charity case, suppose you have decided to donate
of'over-kill': its implications are too strong. The £100 to a charity. Under welfarism, you have
problem is further compounded by the well-known selected this as your 'best' choice given your in-
puzzle that a democratic collectivist solution requires come, what you know about the charity, etc. You
individuals to vote. Yet the same welfarist free- are on the point of writing out your cheque when
riding argument predicts that no one will ever vote! you discover that your neighbour has just posted a
cheque for £100 to the same charity. Since the
One solution to this sort of problem has been charity is now as well off as you thought it would
offered by Collard (1978) among others, which be with your own contribution, but you can now be
appeals to the rationality rather than the public- £100 better off, your welfare is undoubtedly higher
spiritedness of individuals. This is the Kantian if you no longer write the cheque. In fact, since you
principle: 'if the interest of the action can without are a bit richer than before, and charitable giving
self-contradiction be universalized it is morally has a positive income elasticity, you may want to
possible' (Kant, 1930). A non-Kantian altruist will revise your initial view about how much to give
consider only his own (usually) negligible contri- slightly upwards, but even if you gave £10 instead
bution to the financing of public good and free ride. of nothing, 10 per cent is a very large proportion of
If all are non-Kantian, all free-ride and the classic extra income (£100 in this case) for anyone to give
prisoners' dilemma result ensues. The Kantian away (see Collard, 1978 for a survey).
altruist, per contra, reckons with the behaviour of
others: if all do as he does and he free rides he Put it even more dramatically. Suppose the char-
knows (rationally) that the worst outcome will ity's annual income is £10,000 and you are contrib-
result, so he (and all other Kantians) behave mor- uting £100 each year out of personal income of
ally and pay up. While it is in many ways attractive £1,000 (you are unusually generous). Now sup-
to suppose that rules ofmorality actually may affect pose, entirely by coincidence, that your income
behaviour, it is plain that this approach is, in the falls by £1 at the same time as everyone else's
welfarist context, ad hoc. It is, in fact, extra- contributions rise from £9,900 to £9,901. With
welfarist. your preferences and the terms of trade constant,
your initially preferred combination (£10,000 for
Another, rather different, attempt to find a solution the charity and £900 for yourself) is still available
is to posit that individuals attach utility not only and will therefore still be preferred so you give £99:
(possibly not at all) to changing the consumption of a fall in an altruist's income that is exactly matched
goods and services by others, but to the act of by an increase in everyone else's contributions
contributing itself: utility is derived from what one means that an altruist, as welfarists see him or her,
gives away rather than that what it is used for. This will choose to reduce his or her contribution by the
would get one quite neatly out of the free-rider full amount by which personal income has fallen!
problem, for the source of utility is no longer

43
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5, NO. 1

Once again, a theory that has such counter-intuitive and s? Margolis suggests that the higher is g/s the
(and counter-factual) empirical implications has to greater the weight given to s and the higherthe ratio
be regarded as an unusually weak foundation upon G'/S 1 the greater the weight given to g. This he
which to build normative propositions. Yet this is terms the Fair Shares principle (FS). If the weight
the same theory that underpins all these externality given to g relative to s is W, then a stable equilibrium
arguments for public subsidy (of one form or an- occurs at E in Figure 2, with the consequent
other, depending on which of the rival externality equilibrium of g occurring at g*.
theories one is using).
Figure 2
Sugden suggests that it is preferable to invoke
another (extra-welfarist) theory (specifically, of
'duty'). Unfortunately, he does not develop this in
the context of individual charitable behaviour, let

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


alone use it to mount any arguments (welfarist or
other) for the welfare state or 'free' or subsidized
health care, etc. It is worth noting, however, that it
was precisely in the language of duty (as well as
altruism and reciprocity) that Titmuss couched his
defence of the NHS (Titmuss, 1970). This may not
be the welfarist way—but it may be (pace all the
economists who attacked Titmuss) the right way!
Thus we reach a rather destructive end of a chain of
welfarist attempts to build rationales based on
altruism or philanthropy.

One interesting possibility that may yet rescue


welfarism has been developed by Margolis (1982).
He postulates that individuals have a split prefer-
ence system in which one set of preferences relates
to group-interests and another relates to selfish As a model of altruistic behaviour, this has several
ones. This develops hints dropped by, for example, attractive welfarist features. Since it does not
Harsanyi (1955), Pattanaik (1968), Meade (1973), appeal to externality arguments, it is not flawed by
and Rawls (1972), in which the ancient distinction free-rider problems, nor does it have any of the
of Plato between man as citizen and man as individ- Sugden anomalies. It does not require that the
ual is developed in various ways, but each having individual be indifferent to the uses to which g is put
the characteristic that one (higher) set of 'prefer- (which are determined by the individual's prefer-
ences', or morals, constrains or interacts with an- ence over the set of possibilities that exist in the
other (lower) set. usual utility-maximiszing way once the size of g
has been decided). It does not elevate g-preferences
Let, then, a representative individual derive utility to a higher status than s-preferences (they are both
from s, expenditures on self, which are subject to preferences—one is not a set of 'moral rules') so it
the usual diminishing marginal valuation and from is amenable to the usual welfarist interpretations.
g, expenditure by the individual on the group The W function, while possible to derive from
(participation altruism). Since the individual's some form of 'super' welfare function, is not re-
contribution is small relative to the group's as a quired so to be derived (and Margolis advances
whole, the marginal utility of g can be taken as some attractive reasons for not deriving it in this
constant. Thus, as g increases (and s correspondingly way). It gives rise to the possibility of private
falls) the 'participation ratio' g/s rises, and the G7 charities (since gains from trade are implied by the
S\ the marginal rate of substitution (marginal utility g-preference function) and, via scale economies
of g-spending to that of s-spending) falls. What and the saving of transaction costs, to governmental
determines the individual's preferred balance of g social spending and 'coercive' taxation.

44
A. J. Culyer

The implications of the model have not been devel- literature affords of the concepts of 'altruism',
oped specifically for the health care sector (for a 'selfishness', and 'caring'. As an aid to clarity of
beginning, see Mooney, 1986) but, given the weak- thought, the welfarist literature is valuable; it cannot
ness of other welfarist models, it may be that it be claimed, however, that it has yielded settled
holds out the most promise for a satisfactory such conclusions on those issues I have identified as
model (Margolis himself developed it in a positive requiring judgement. The view that the externality
context, a positive-normative flip is thus required or group concern is derived not from health care
as was noted before in connection with Lindsay, consumption but from 'health' itself is not directly
1969 and Culyer, 19716). addressed in this literature but occurs in what I shall
discuss later in the extra-welfarist sectioa
( v ) . . . and Where Does this Get Us?
(vi) Uncertainty

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


Since the literature on externalities and 'caring' is
largely aprioriin nature, its value is primarily as an The stochastic nature of disease and ill-health has
aid to introspection and judgement. One needs first been held to be another respect in which health care
to form a view about the existence of the general is 'different' from at least some other goods and
events that may fall into this class by imagining a services. The standard welfarist economics of
state of the world devoid of attempts to optimize via health insurance (e.g. Arrow, 1963; Culyer, 1979;
subsidy and regulation, and to determine whether Evans, 1983,1984) is based on the expected utility
externalities are Pareto-relevant in total or at the maximizing model of risk-aversion (for a survey
margin (Buchanan and Stubblebine, 1962) or whether see Schoemaker, 1982). The aspect of health-
g-orientation is likely to be pervasive. Beyond this affecting events that is insured against is the finan-
there are questions as to whether group concerns are cial cost only of medical care (loss of employment
better conceived as relating to absolute or relative income is not considered here). In this analysis, the
levels of consumption and whether they are better welfare gain to the insured risk-averse individual
seen as questions of preference (arguments of utility arises from the implication that the welfare loss
functions) or of morals (constraints on selfish welfare from paying a certain premium is less than the
maximization). These rather useful sorting devices welfare loss of the expected financial loss (having
are further supplemented by the enrichment the the same actuarial value as the premium) of the

Figure 3

W(0

income

45
OXFORD REVIEW OF ECONOMIC POLICY, VOL 5, NO. 1

uncertain prospect if the individual 'self-insures' If an insurance firm offers insurance at an actuari-
by retaining net income in health and by paying ally fair premium pH (the cost of the event times its
medical expenses out-of-pocket if sickness strikes. probability), the welfare of the individual paying
This in turn arises because the m arginal welfare loss the premium is W(Io-pH). The choice confronting
from having to pay an uncertain but large sum is the individual is thus to self-insure or to purchase
greater than that from having to pay a certain but insurance, with payoffs as in Figure 4.
smaller sum, due to the assumption of a diminish-
ing marginal utility of income or wealth. It is It is clear that, with diminishing marginal welfare
usually assumed that the utility of income is inde- from income (risk-aversion), welfare is higher with
pendent of the stochastic states 'healthy' or 'sick'. insurance, at d, than without it, at c. The certain
If it is not, it is not clear what the optimal level of monetary loss pH is preferable to the uncertain loss
insurance is (Shavell, 1978). of H with the same probability. Fair insurance thus
increases welfare.

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


Thus Figure 3 shows an individual's welfare as a
function of income (diminishing marginal wel- The important question for health care financing is,
fare). Let the individual's income be Io, yielding a of course, whether a voluntary competitive health
welfare level Wfl^ on the vertical axis, and con- insurance market is better able to maximize these
sider a choice between insuring against an uncer- welfare gains than a compulsory system of social
tain event (illness), which willrequirean expendi- insurance or tax finance.
ture of H, or not insuring (self-insurance). If the
event occurs, expenditure will be H, income net of
(vii) Loading
expenditure Io-H and the associated level of wel-
fare is W(Io-H). Suppose the probability of the The analysis so far assumes that there is a competi-
event occurring is known to the individual and tive insurance market in which many insurers compete
insurance companies and is p. With self-insurance, for trade (or potential entrance deters oligopoly
income net of expected health expenditure is I0-pH. premium-setting) and premiums are set at actuari-
The expected welfare of this option is pW(Io-H) + ally fair rates: theproductoftheprobabilityandthe
( l - p ^ f l ^ : the sum of the expected welfare if estimated expense. In practice the genuine oppor-
sickness occurs plus the expected welfare if it does tunity costs of insurance provision, together with
not. The pdint c on the chord ab is located propor- any X-inefficiency and monopoly rent, will also be
tionately to p. If p is, say, 0.4, then bc/ac = 0.4. 'loaded' on to premiums which consequently rise
above their actuarially fair level. In large group
plans in the USA the loading is about 10 per cent
(Pauly, 1986). This implies that the welfare gains
Figure 4 for consumers of insurance will fall as premiums
Choice between Sure Thing and Uncertain rise and some risk-averse individuals will no longer
Prospect insure.
ewnl does not occur
Let loading charges, L, be proportional to the pre-
mium pH. The premium now becomes (1 + L)pH
and the insurance outcome (in this case) now has a
lower welfare for the individual, W[I o -(l + L)pH],
KU-insur« than self-insurance. Any loading large enough to
jwnloccura
bring the individual on to the section of the welfare
1=10 -H)
function below f on the horizontal through c will
reduce welfare relative to the self-insurance option.
Any loading less than cf will not

To the extent that the loading represents costs in


excess of opportunity costs, this will thus cause
Mfl-pH) deadweight welfare losses. To the extent that

46
A. J. Culyer

opportunity costs themselves can be lowered by United States, estimates vary of the numbers who
scale economies and other measures that might be are uninsured or inadequately insured, but seem to
taken to reduce costs, the competitive insurance indicate a total of around 50 million (Farley, 1985;
system itself may thus cause unnecessary welfare Mundinger, 1985). In any event, adverse selection,
losses. unless checked, is likely to cause massive externali-
ties of the sort described earlier and become a major
Loading also leads to a preference for deductibles affront to most principles of equity.
(Arrow, 1963). Suppose, as seems realistic, that the
loading is not simply proportional to the size of the Left to itself, adverse selection would destroy the
premium but is arelativelyhigh proportion of small market entirely. In practice this does not happen,
claims (say, checking for fraud, etc., is equally partly because insurance plans are often based on
costly for probable and less probable events). Then, employee groups in which a condition of employment
as the limit of acceptable (to the consumer) loading may be that each joins the company plan; so opting

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


is less than fc in Figure 3, as b is approached and as out is not possible unless the gains from doing so
the cost of the event falls (point a is closer to b on would outweigh the losses from changing one's
the utility function) the probability of the loading employment Partly it does not happen because the
exceeding the welfare gain rises: very probable presence of adverse selection itself affords gains
events (e.g. routine dental care) will be more self- from trade between self-insurers and profit-seeking
insured via deductibles and very expensive events insurance agencies (and non-profit seeking too).
(e.g. cancer care treatment) will be more likely to be
fully insured (no deductibles). It pays insurers to offer policies with premiums that
better approximate the true expected expenses of
(viii) Adverse Selection those driven out by adverse selection, even though
the acquisition of this information is not costless.This
Another problem that arises has its roots in another erodes the principle of community-rating which
kind of informational asymmetry—this time, how- becomes replaced by experience-rating and by pack-
ever, the informational advantage lies with the ages of cover that fall short of fully comprehensive
consumer rather than the supplier of the service (e.g. 'major medical' only). In this fashion 'good
(financial in this case). Premiums are set according risks' are creamed off. While there will remain
to the calculated risks for groups in the community. some inefficiency in that some very low risk groups
In the simplest case of 'pure' community rating, the may still fail to find a suitable package at an
premium is set according to the population-wide actuarially appropriate premium, experience-rating
probability of consuming health care. The typical is more efficient than community-rating. It differs
consumer will, however, usually know better his or also from community-rating in that, whereas under
her own probability. Those whose probability community-rating wealth is redistributed from low
times expected expenses is greater than the pre- to high risk individuals and families in advance of
mium will gain welfare in excess of that expected any health care consumption, under experience-
in the model (viz. > cd in Figure 3) and may gain rating, there is redistribution ex post from those
even if they are not risk averse! Those whose true (insured) who are well to those who are sick (as does
probability times expected expenses is less than the all insurance).
premium will tend (depending on their risk aver-
sion) to self-insure. This phenomenon is known as Though more efficient from a welfarist perspective
adverse selection. It leads to a progressive upward (Arrow, 1963), experience-rating is likely to violate
pressure on premiums (as low potential users are the usual distributional criteria of equity: those
driven out), with insurance cover increasingly re- with a history of sickness will face the highest
stricted to the very worst risks. The upshot is a pool premiums, have the less comprehensive cover, be
of uninsured people, many of whom are risk averse, most likely to pay deductibles (both of these latter
in addition of course to those who are uninsurable options have the effect of reducing premiums as
by virtue of chronic disease or other disqualifying they reduce probable pay-outs by the insurer)—and,
features, and a heavy financial burden on the sickest since ill-health and income are correlated, will on
members of society who remain insured. In the average be the poorer members of the community.

47
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5, NO. 1

Many, moreover, can be expected to find their risk- hazard is at best fuzzy. The other is that if hospital
aversion insufficient to warrant insurance of any and other services are priced at above marginal cost,
kind. second-best considerations would dictate that the
usual welfarist efficiency condition of P=MV may
(ix) Moral Hazard no longer apply. Consumers may be receiving false
signals about marginal cost from market prices and
Another difficulty with insurance systems is known should probably be encouraged to consume beyond
as moral hazard. One form of this has already been the rate at which P = MV, quite apart from any
met in the shape of supplier-induced demand, which externality considerations, even if this implies even
is a kind of producer's moral hazard, whereby bigger rents for monopoly suppliers of health care,
producers (on a fee for service system of reimburse- as income is redistributed from the insured to the
ment) have an incentive to adjust the client's de- suppliers via the insurers.
mand in pursuit of their personal income objec-

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


tives. This is easier done (or at least is less likely to In any event, the effect on the insurance market is
offend the doctor-patient relationship) if the cost of predictable: with moral hazard, consumption ex-
supplier-induced demand can be passed on to the ceeds the rate on which premiums have been (his-
insurer. torically) set, the (retrospective) reimbursement of
suppliers rises to levels higher than predicted, and
More commonly discussed in the literature, however, premiums rise. Higher premiums will drive out
is consumer's moral hazard: in which the fact of some risk-averse individuals from the insurance
being insured encourages the individual to take less market (dead-weight losses) and may (but only
care in ensuring that the undesired state (illness) ' m ay') cause an inefficiently large rate of consump-
does not occur (ex ante moral hazard) and, when tion.
sickness occurs, encourages the consumer to
maximize the consumption of services beyond the To retain their market, insurers offer packages that
point at which marginal cost (assuming that services do not include 'first pound coverage' and the con-
are priced at marginal cost) equals marginal value sumer has to pay deductibles (e.g. the first £x of any
(ex post moral hazard). In health insurance, ex post expense) or coinsurance (e.g. x per cent of the total
moral hazard seems to be the principal problem on bill). These out-of-pocket payments in the US
the consumer side. Two caveats are worth noting currently amount to about 30 per cent of non-
here. One is that consumer demand is, as we have hospital expenses and 10 per cent of hospital ex-
already seen, mostly interpreted by an agent, so the penses.
distinction between consumer's and producer's moral

Figure 5
The Welfare Loss of 'Excess' Health Insurance
price

demand or marginal value

health care (e.g. hospital days)

48
A. J. Culyer

While the likely offence to equity principles is reveal their preferences (viz. that the agency rela-
plain, the efficiency implications are less obvious tionship is perfect).
in a welf arist context. With full' first-pound' cover,
and zero marginal (money) user price, the sick Provided that it is not accompanied by compensating
individual may be expected to consume OQ units of X-inefficiencies, a system of compulsory universal
health care in Figure 5, given that he or she is sick, public insurance operated through the general tax
at which point MV = 0. This generates a welfare system may have substantial cost advantages over
loss of abQ: the amount by which the cost of competitive insurance: compulsion avoids adverse
providing Q*Q units exceeds their value to the selection and enables scale economies to be gained
individual (Q*abQ-Q*aQ). This analysis assumes (if they exist), reducing loading and increasing the
that consumer marginal willingness to pay is ade- welfare gains from comprehensive insurance cover.
quately reflected in the demand curve and that OP Universality and tax finance avoid the necessity of
is the marginal social opportunity cost risk assessment and premium setting, billing,

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


reimbursement, checking for fraud and so oa These
The excess 'burden' can be reduced by co-insur- costs appear to vary substantially across different
ance. Thus, ifthe consumer pays a proportion, p, of systems, being less than 3 per cent of total expenditure
the daily cost P, and that is 50 per cent, user price in countries with a tax-based social insurance system,
becomes pP, consumption falls from Q to Q' (closer like the UK and Canada, and in excess of 10 per cent
to the 'optimum' Q*), the excess burden falls by in private insurance systems, or in systems with
more than half to acd, and total expenditure falls public finance but with complex systems of billing
from OPbQ to OPcQ1 implying a lower (future) pre- and reimbursement (e.g. USA and France) (OECD,
mium. 1977). It may be technically possible for a large
country (like the USA) to operate a competitive
In one of the few thorough empirical attempts to system that would avoid monopoly exploitation
measure welfare effects in the literature, Feldstein and that would also realize scale economies (though
(1973) estimated that the maximum reduction in this does not seem to have happened) but may be
the excess burden of health insurance in the US, by implausible in smaller countries like the UK. Moral
raising co-insurance from an average of 0.33 to 0.5, hazard may be controlled in public or private insurance
would have been $10 billion in 1969 prices after sytems by the adoption of (private or public)
allowing for the loss of welfare to those who would regulatory schemes defining the appropriateness of
no longer choose insurance. However, the implica- packages of care, and by prospective reimbursement
tions of this analysis hinge crucially on the ade- acording to an agreed charge per case-type rather
quacy of prices as measures of marginal cost and on than retrospectively according to whatever the
the absence of externalities. They also depend on suppliers happen to have charged. Given a non-
the welfarist assumption that MV adequately repre- competitive health care industry (whether public or
sents the consumer's true estimate of the worth of private), such regulation is commonly observed,
medical care. (The actual calculation was also and it is a question of judgement whether the
based on elasticity estimates that were almost cer- regulation is better done by a publicly accountable
tainly too high by a large factor.) agency or by the (insurance) industry itself.

( x ) . . . and Where Does this Get Us? (xi) The Welfarist Approach: an Overview

Insurance is shown plainly to have welfare-increas- The welfarist techniques of analysis have been the
ing properties though it seems clear also that opti- traditional way in which issues of health service
mal insurance may be less than complete insurance: finance have been addressed in the economics
some events should be uninsured and for others less literature. Granted the acceptability of welfarism's
than 100 per cent of the risk will be optimal. This value assumptions, its implications for policy hinge
analysis does not, however, take account of any on judgements about the empirical significance of
contrary indications suggested by externality or consumer rationality, the 'purity' of the agency
equity factors and is entirely dependent on the relationship, the nature of any externalities (physi-
strong assumption that consumer choices adequately cal or utility interdependence or group-concern),

49
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5. NO. 1

the extent of adverse selection, moral hazard, sup- 'basic capabilities': a person being able to do
plier induced demand, unnecessary premium load- certain things. It is, he suggests, because a cripple
ing under insurance and the empirical validity of is unable to perform particular activities that he or
the neoclassical behavioural model that, in its nor- she is seen as having special 'needs' that are inde-
mative version, is welfarism's centrepiece. Each pendent of his or her total or marginal utility.
will draw his or her own conclusions based on the Culyer (1989) advocates the more general notion of
(patchy) evidence and more casual experience. 'characteristics of people'—for example, their ge-
netic endowment of health, their relative depriva-
tion independently of the absolute consumption of
III. THE DCTRA-WELFARIST APPROACH commodities or the characteristics of commodities,
theirmoral 'worth' and 'deservingness', whetheror
Whereas welfarism holds that standards of living, not they are in pain, or stigmatized by society.

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


the efficiency of social arrangements and the justice Characteristics may also relate to the character of
of distributions and redistributions are all to be relationships between people such as the quality of
evaluated in terms of individuals' utilities (or wel- friendships, community support for the individual
fares), an extra-welfarist approach (Sen, 1979) admits when in need, social isolation, or changes in them,
non-utility information about individuals into the such as becoming (as distinct from being) crippled.
process of comparing social states. Using an illus-
tration from Sen, consider three social states x, y, Only some of the characteristics of people (which
and z with the following (cardinal and even inter- will include some of their capabilities) will be
personally comparable) utility numbers for persons deemed relevant and the list of such relevant char-
1 and 2. acteristics is likely to vary between cultures, cli-
mates, historical periods, and so on. It is, in short,
x y z contingent It is related to a concept of need. If the
characteristics of people are a way of describing
person l's utility 4 7 7 deprivation, desired states, or significant changes
in people's characteristics, then commodities and
person 2's utility 10 8 8 characteristics of commodities are what is often
needed to remove the deprivation or to move to-
wards the desired state, or to help people cope with
In x, person 1 is hungry while 2 eats amply. In y, change. They are the necessary means to a desired
2 has been forced to surrender some food to 1 and end. To compare the ill-health of different indi-
2's utility loss is less than 1 's utility gain. Under viduals or groups is not the same as to compare the
welfarism, the factofcoercionisanirrelevance: the health care they have received (they could receive
sum total of utility is higher in y than x so y is the same amounts and still be unhealthy, or differ-
socially preferred. This is one example of the ent amounts and be equally healthy).
exclusion of non-utility information in the making
of social comparisons. In z, person 1 is as hungry (i) Need and Extra-Welfarism
as in x and 2 is as amply fed but 1, who happens to
be a sadist, is allowed to torture 2 (who is no Whereas the notion of 'need' has received a bad
masochist). It so happens that 1 's utility gain and press from many welfarist economists, extra-wel-
2's utility loss are the same as under the food farists have been able to use the term with some
transfer programme. Under welfarism y is socially precision and confidence. In its most rudimentary
preferred to x but z is the same, utility-wise, as y, form 'need' for health care would seem to imply
therefore z too is socially preferred to x. Again, the that someone is better off with the 'needed' treat-
relevance of non-utility information (in this case ment than without it (Williams, 1974) and that the
the fact of torture and my disapproval of it) is 'better offness' has to do with persons' health
denied. (Culyer, Lavers, Williams, 1971). Thus health
services are needed (viz. are a necessary condition
Sen (1980) argues that a particularly important for achieving a particular outcome) only if the
class of non-utility information about individuals is outcome is desired and there is no alternative (or

50
A. J. Culyer

more cost-effective) way of realizing it. Since age use of care that contributes to the objective of
inputs are nearly always substitutable, it will not maximizing health. There are also implications for
normally make sense to say that a specific resource rationing care (equalizing marginal products in
in a specific quantity is needed. Since there is no terms of health per unit of resource), selecting
effective treatment for some conditions, it is non- patients from waiting lists, conducting cost-benefit
sensical to say that persons suffering from such analyses in the health service.
conditions need health services (they may need the
fruits of research and they may need love and com- At the core of the extra-welfarist approach is, of
fort, but they cannot need ineffective care, even course, the issue of how the maximand is to be
though they may demand it, and such care may also measured. This has proved to be an issue involving
damage health). Since health services are needed much cross-disciplinary collaboration between (for
only for what they enable to be accomplished, in a example) economists, physicians, psychologists,
world of scarcity judgements must be made about and political scientists which has exposed the

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


the value of what might be accomplished. Some complexities involved in defining 'health' and
services are bound not to be supplied in the light of measuring it, and which has also developed a
such judgements and so it does not make much battery of experimental techniques designed to test
sense to require that all needs should be met. One theories and quantify the hitherto unqualified.
of the first lessons of 'needology' is the lesson of
the ethical acceptability of unmet need (Wiggins et (ii) Measuring Health
al., 1987).
Extra-welfarists identify 'health' as the principal
Since health services can be needed only if their output of health services and its efficient produc-
outcomes are desired, the important question arises tion (technical efficiency, cost-effectiveness) as an
as to whose judgements ought to be decisive in issue upon which its insights can be particularly
assessing desirability of outcome. While the judge- valuable, in contrast to welfarism, under which it is
ment of technicians (such as doctors) may be appro- natural to take goods and services as the natural
priate in evaluations of effectiveness, technical experts units of output- Much of the cost-effectiveness
have no particular authority for making value judge- literature in health economics is implicitly extra-
ments so it does not make much ethical sense to welfarist in seeking to identify the least-cost method
pretend that these essentially political decisions can of delivering a given health improvement (or pre-
be desanitized by being left to the experts. At the vention of deterioration) for a given patient group
centre of this problem lies the issue of how one (or across patient groups). In many cases, extra-
person's needs (viz. the potential improvement in welfarist economists have joined hands with non-
his or her health attributable to health care) are to be economists in the search for better measures of
weighed against another's, an issue taken up below. health and, of necessity, in identifying the produc-
tion functions that underlie cost functions and cost-
In health economics, the extra-welfarist approach benefit relationships.
has taken 'health' as the proximate maximand.
This does not imply the complete ousting of 'wel- One approach to health measurement has been to
fare' with its usual normative connotations, but the use ad hoc numerical scales to quantify the bundle
use of both sets of 'data' to evaluate alternatives. of characteristics (usually in the context of planning
'Health' is itself a descriptive characteristic of exercises or studies of the effectiveness of medical
people, but in practice has been interpreted in the procedures). This approach involves the individual
literature as a composite 'bundle' of other charac- being assessed in several dimensions, with num-
teristics such as pain and restriction of activity bers being associated with each assessment and the
(Culyer, Lavers and Williams, 1971). resultant scores (sometimes weighted) being added
up (e.g. Harris et al, 1971; Grogono and Wood-
The implications of the extra-welfarist approach for gate, 1971). The arbitrariness of such procedures
the finance of health care seem partly to take has been spelled out several times (e.g. by Culyer,
negative forms: insurance arrangements, user prices 1978a and b; also see Drummond, this issue):
(money, time, etc.) should not act so as to discour-

51
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5, NO. 1

(1) criteria for selecting characteristics were However, such measures present a potential prob-
usually unspecified lem. The literature frequently characterizes meas-
ures of health as utility measures (e.g. Torrance,
(2) the scaling systems often implied only 1986). Here it is important to distinguish between
order but were subsequently used to con- the welfarist notion of utility as welfare and the
struct a cardinal index extra-welfarist notion. Under Paretianism, for
example, the notion of welfare relates to goods and
(3) the possibility that combinations of char- services and is the utility of the individual affected
acteristics may have higher or lower num- by their consumption. Under extra-welfarism, while
bers than the sum of the separate scores was this notion of utility may still apply, there is the
often excluded further idea that uses utility theory in order to derive

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


measures of characteristics of individuals that are
(4) increasing marginal severity was rarely not goods, not services, nor necessarily having a
allowed value content that corresponds to the Paretian no-
tion that 'the individual is the best judge of his/her
(5) criteria for selecting those making these own welfare'. Confusingly, however, these too are
(value) judgements were usually unspecified. called 'utility' measures. The use of the standard
gamble based on Neumann-Morgenstem assump-
The approach that has most frequently been adopted tions described above is a good example. The
by economists has become known as the Quality- possibility that one can have utility measures that
Adjusted-Life-YearorQALY, and has been mainly are not welfarist is thus important to appreciate.
developed by Torrance in Canada (see Torrance,
1986 fora survey, and Drummond, this issue, for an (Hi) "Utility" is not 'Utility' is not Utility .„
extended discussion) and Williams in the UK (e.g.
Williams, 1985,1986). The QALY has two dimen- A related puzzle that pervades some welfarist theory
sions: life-expectancy (as a measure of the extra is the meaning to be attached to the word 'utility'.
life-years that may be procured) and a quality- While it is commonplace to distinguish between
adjustment (as weights indicating the 'healthiness' utility as 'welfare' and utility as 'an index of
of the expected life-years). In the context of the choice' in normative and positive analyses
measurement of output of health services, produc- respectively, some have difficulty in distinguishing
tivity is thus to be seen as the difference over a (in a welfarist context) between utility-maximizing
period of time between expected QALYs with a behaviour that is altruistic and that which is selfish:
particular procedure and without it (or with an is not one who maximizes utility by giving, in some
alternative). sense, selfish? In positive economics this poses
little difficulty: altruism means simply that a
The ways in which the quality weights have been person's allocation of wealth is influenced by the
developed and interpersonal comparisons made are effects that the allocation has on others and that the
of particular analytical interest. A historical review person in question has benign intentions. Since
of the health index literature (of which these quality both behaviour and motive are different from that of
weights form a part) is Rosser (1983) and a review one for whom neither of these things is true, this is
of the techniques for measuring health indices is a distinction that needs to be made. It fails, however,
Torrance (1982). to capture another distinction: that between fulfilling
one's selfish preferences and acting contrary (at
There are three commonly-used methods of meas- least in part) to those preferences by adopting a set
uring scales of health (on this and other important of moral rules that may constrain. Such altruism,
distinctions see Culyer, 1986): the rating scale, the that is contrary to preference, is of a different kind
standard gamble and the time trade-off (see and, perhaps, requires an appropriate term that
Drummond etal., 1987). These are defined in detail differentiates it from purely 'welfarist altruism': it
in Drummond (this issue). is, after all, extra-welfarisL

52
A. J. Culyer

As a result, and given the difficulties in defining the cycle relative to others. It is also planned to
relationship between health measures and utility unscramble the life-cycle phases into elements that
measures discussed above, one of the normative are age-, role-, and sex-related.
issues that the extra-welfarist approach identifies
(but does not resolve) concerns who shall decide the If ethical authority is to be accorded these (or
weights to be applied to different health states and similar) results, a departure is implied from the
to the components of health states. Who shall distributive value-judgement normally (if only
decide the categories of functioning, etc. to be provisionally) embodied in measures ofhealth such
considered? Who shall decide who shall decide? as QALYs: that a unit ofhealth' is of equal value
no matter who gets it. It thus seems possible that
The answers may well depend upon the nature of distributive judgements will be able to be built into
the problem under consideration: politicians, civil outcome measures and, via cost-effectiveness analy-
ses, into efficiency analysis (e.g. Culyer, 19886).

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


servants, managers, representatives of the public,
persons at risk of particular disease, patients, doc- Indeed, if all of the features of distributional equity
tors, nurses... all may have some claims by virtue that are of (legitimate) concern are built into out-
of identity, skill, or position of trust It does not, come data in this way, then a full integration of
however, follow that those judged best able to equity and efficiency will have been achieved in
exercise a judgement about, say, the effectiveness health policy: given routine information about popu-
of a medical procedure are necessarily those best lation characteristics (disease incidence, etc.), medical
qualified to exercise a judgement about how pain technology (the possibilities for changing health
and disability are to be traded off. states for the better), and cost information, it will be
possible to make informed routine judgements
Ov) Extra-Welfarism and Distributional Weights about resource allocations to providers of health
care.
The non-welfarist literature clearly sees informa-
tion of the cost-per-QALY sort as a means of (v) Objections to QALYs
guiding resource allocation decisions in the NHS,
whether on equity or efficiency grounds, and the There is, however, quite a long way before that
literature is quite explicit in its departure from dream becomes a reality—and some have seen it
consumer-based willingness-to-pay as the basis for more as a nightmare possibility. One type of
benefit valuation (Williams, 1988). objection isrepresentedby Smith (1987) who argues
that the use of a quantitative algorithm obscures a
The question that naturally arises next, of course, process by which essentially arbitrary assessments
relates to whose willingness-to-pay is to be substi- of the values of people's lives are being made. To
tuted for the consumer's. The most recent work has this Williams' (1987)retortseems compelling: that
addressed this issue in the following way: ' . . . is a far from obscuring the need for value judgement the
particular improvement in health to be regarded as procedure highlights the value-judgemental elements
of equal value no matter who gets it; and, if not, and offers techniques by which they could be made
what precisely is its relative value in accruing to one more explicitly; that far from imposing essentially
kind of person as opposed to another?' (Wil- arbitrary values, the process of quantification is, by
liams,1988). The current work has begun by seek- virtue of its explicitness, open to criticism and
ing to find out what views are actually held by change at every stage. Indeed, it is hard not to see
surveyed individuals on the matter. It would seem Smith's objections as being much more aptly directed
from the early results that there is a consensus that at present modes of making decisions (e.g. about
particular phases of the life-cycle are regarded as health service cuts across the board).
times when health is of greatest value. Two stood
out of the ten phases used in Williams' survey: 'as Rather deeperis an objectionofBroome (1985) that
infants' and 'when bringing up children'. The the measures ignore 'population': the outcome of
information available to date is only qualitative and health services may include not only the extra
a next phase of the work will address the issue of health for people who may also be having extra
how much good health is worth at points in the life time, but also more people if an additional outcome

53
OXFORD REVIEW OF ECONOMIC POLICY, VOL 5, NO. 1

includes children being bom who would not other- differences in measure that have been observed as
wise have existed In the welfarist tradition this between rating scales, standard gambles, and time
poses a problem because although it might be trade-offs (e.g. Torrance, 1976, Bombardier etal.,
possible to find out how extra QALYs are valued, 1982). His analysis hinges on a distinction between
it is not possible to find out how the unborn value the utility gained as the result of one's own choice
being born. But this also poses a problem for the and 'choiceless' utility: viz. the utility 'experi-
non-welfarists in so far as they too wish to ascribe enced* as the consequence of a happening gener-
a value to unborn lives—as well as the lives of the ated in any other way. The importance of the
children of the unborn, and so on ad infinitum. distinction lies in the fact that only in the former is
Unfortunately, we have no basis at present for there the possibility that, in an uncertain world, you
valuing population changes, therefore (says Broome) may come to regret or rejoice over a decision you
we have no basis for valuing life or QALYs. have made. The rating method does not involve

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


choice; it requires the subject only to locate a state
Earlier Broome (1978) argued that the only appro- on a utility scale. The standard gamble, by contrast,
priate value for a 'statistical' life was infinity on the does involve choice and so does the time trade-off
grounds that, eventually, statistical probabilities of method.
death (or of opportunities for life extensions not
taken advantage of) translate into deaths of actual 'Regrettheory' is recommended as thtprimafacie
individuals who might reasonably be expected to better foundation for future work in the QALY
exercise a veto. This clearly poses something of a territory. It seems clear that research should be
challenge to Paretian welfarists though less of one, expanded to incorporate regret theory into health
of course, to non-welfarists. A welfarist may find status and QALY measurement experiments in
something of a defence in the reflection that he order to compare results systematically with the
might himself agree to an option offering some other techniques. The potential from exploiting
benefit but with a very small prospect of its entail- other substitutes for expected utility theory (such as
ing his own death, so why should not a society of prospect theory) remains to be explored.
like-minded folk feel similarly? An extra-welfarist
might take the view that she would be guided by the (vi) Where does Extra-Welfarism Get Us?
majority view on the value of (or differential values
of) life. As far as the demand side is concerned, extra-
welfarism in health economics may be seen to take
It is worth, perhaps, reminding everyone (welfarist 'health' output as the maximand. The emphasis is
and extra-welfarist alike) that there is nothing to be not in principle exclusive, as extra-welfarism is not
gained in the context of resource allocation deci- exclusive, and it seems unlikely that any extra-
sion-making from taking an ontological view of welfarist would assign zero weights to such factors
QALYs, or life, or lives. One is not concerned with as consumer choice, privacy, speed of service,
the inherent cherishable worth of people but rather hospital hotel-services, and other factors that may
with the value of resources that we might spend in be only remotely causally linked to health.
order to gain better health or prevent (or postpone)
death or change the prospects of either for the Extra-welfarism thus immediately implies another
better. If we spend £2,000 per person to protect notion of efficiency, in which explicit (but not
them from the consequences of some risk that is welfarist) value judgements are incorporated into
fatal for say one in five hundred, it is merely the maximand which is a cardinal 'utility' index of
arithmetic that we shall spend, on average, £1 mil- health (the welfarist tradition, the Margolis model,
lion per life saved: only in this sense is a life'worth' and the insurance literature are also cardinal in the
£1 million. sense that 'utility' as used there is measurable up to
a linear transformation). This index is extremely
Loomes (1988) mounts a far more powerful assault useful in supply-side efficiency studies. It is worth
on both welfarist and non-welfarist traditions by noting that it is not uniquely applicable under
attacking the usual behavioural axioms that are socialized systems of care and may, forexample, be
shared in both. He focuses on the systematic used in market systems where insurance companies

54
A. J. Culyer

seek to control producer moral hazard by reimburs- justified by equity arguments but efficient policies
ing providers only for procedures that are demon- justified by heath maximization.
strably relatively cost-effective in restoring health
and in any clinical research with a similar objective. There is a danger in extra-welfarism of becoming
In this territory, extra-welfarists have been more too fixed on the' bottom line'. The great advantage
active empirically than welfarists typically have the approach can claim in issues like outcome
beea measurement is rather like the claim made earlier
on behalf of welfarism: it provides a conceptual
As a matter of necessity, the literature has focused framework for handling extremely complex issues
on the difficult issues of measuring health itself (or in a systematic fashion and that exposes each aspect
changes in it) in order to improve the ability of the of an argument clearly. It is less important what the
system to produce health cost-effectively. It is cost-per-QALY is, than that individuals with re-
quite possible for these efforts also to serve the sponsibility for resource allocation in health care

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


cause of welfarism: after all, better information on have a means of working through the issues so that
outcomes can enhance the physician's ability to act they can come to their own informed view about the
as a good agent for the patient. But it is quite clear pros and cons of different resource allocations. Itis
that other causes are served as well and that the in this sense that it is important to emphasize that
implications of extra-welfarist health maximization the method is intended (in an archetypical 'deci-
are not the same as welfarist analysis. sion-makers' approach) as an aid to, rather than a
substitute for, thought.
Extra-welfarism resembles the welfarist external-
ity arguments in implying free or subsidized terms
of access to healthcare. Under welfarism, however, IV. CONCLUSIONS
the reason for the subsidy lies in the optimal inter-
nalization of externalities. In extra-welfarism, the It will be clear from the foregoing that there is a
reason lies in more 'engineering' sorts of concern: paradigm clash in the normative economics of
optimal resource use is determined by equality of health, though it would be wrong to overdraw the
marginal health output per unit of resource in differences. The extra-welfarist approach is, after
various activities and across various client groups. all, inclusive of welfarism. There can be no ques-
Willingness-to-pay is an irrelevance and may be tion that the extra-welfarist approach is more toler-
directly counter to this objective if willingness-to- antof whatmay be seen as 'paternalism' and can be
pay is positively associated with ability to pay, but readily enlisted on behalf of the sorts of access
ability to pay is inversely associated with' potential terms and distributional issues that have lain at the
for health improvement'. The extra-welfarist ap- heart of the ideology of the British NHS—and in-
proach therefore attaches great importance to the creasingly of policy towards i t Extra-welfarism is
identification of potential for benefit. Indeed, it is also providing a theoretical basis upon which us-
possible to see some traditional policy arguments able output measures can be derived.
cast in equity terms as possibly extra-welfarist
health-maximization arguments. Assume, for Ultimately, however, neither approach can yield
example, that a (satisfactorily measured) QALY is final answers. They provide frameworks whose
judged to be equally valuable socially to whomso- usefulness remains a matter for judgement. An
ever it may accrue. Now allow that the sickest in earlier judgement by me upon the field, "The heady
society are by and large those for whom the mar- atmosphere of grand designs has to be replaced by
ginal product of health care in terms of QALYs is the mundane, but ultimately more fruitful, ground
highest, that these are also the poorest, and that of systematically applied economics . . . In this
when (ceteris paribus) health service per capita scheme of things the role of welfare economics is to
rises, the marginal product in terms of health falls. provide an appropriate theoretical base in which to
It evidently follows that efforts to equalise the geo- build empirical studies and not to prejudge the
graphical distribution of resources, to channel more facts' (Culyer, 1971) needs amendment in only one
of them to the sick and more of them to the poor, respect: add 'or extra-welfare' after 'welfare*.
might be seen not as distributional policies to be

55
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5, NO. 1

REFERENCES

Arrow, K. J. (1963), 'Uncertainty and the Welfare Economics of Medical Care', American Economic Review,
53,941-73.
Auster, R. D. and Oaxaca, R. L. (1981), 'Identification of Supplier Induced Demand in the Health Care Sector',
Journal of Human Resources, 16,327-42.
Beck, R. G. (1974), 'The Effects of Co-payments on the Poor1, Journal of Human Resources, 9,129-42.
Bergson, A. (1938), 'A Reformulation of Certain Aspects of Welfare Economies', Quarterly Journal of
Economics, 52,310-34.
Bombardier, C , Wolfson, A. D., Sinclair, A, J. and McGreer, A. (1982), 'Comparison of Three Preference
Measurement Methodologies in the Evaluation of a Functional Stratus Index', in Deber, R. and Thompson,
G. (eds.). Choices in HealthCare: DecisionMakingandEvaluationofEjfectiveness,ToTontQ,\JniveTsily

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


of Toronto Press.
Broome, J. (1978), 'Trying to Value a Life', Journal of Public Economics, 9,91-100.
— (1985), "The Economic Value of Life', Economica, 52,281-94.
Buchanan, J. M. (1965), The Inconsistencies of the National Health Service, London, Institute of Economic
Affairs.
Buchanan, J. and Stubblebine, W. G. (1962), 'Externality', Economica, 371-84.
Collard, D. (1978), Altruism and Economy. A Study in Non-selfish Economics, Oxford, Martin Robertson.
Cromwell, J. and Mitchell, J. B. (1986), 'Physician-Induced Demand for Surgery', Journal of Health Economics,
5,293-313.
Cullis, J. G. and West, P. A. (1979), The Economics of Health: An Introduction, Oxford, Martin Robertson.
Culyer, A. J. (1971), 'Merit Goods and the Welfare Economics of Coercion', Public Finance, 26,546-71.
— (1971a), "The Nature of the Commodity 'Health Care' and its Efficient Allocation', Oxford Economic
Papers,23, 189-211.
— (19716), 'Medical Care and the Economics of Giving', Economica, 151,295-303.
— (1978a), 'Needs, Values and Health Status Measurement', in Culyer and Wright (1978).
— (19786), Measuring Health: Lessons for Ontario, Toronto, University of Toronto Press.
— (1979), Expenditure on Real Services: Health, Milton Keynes, Open University Press.
— (1980), The Political Economy of Social Policy, London, Martin Robertson.
— (1986), "The Scope and Limits of Health Economies', Jahrestagung des Vereinsfuer Socialpolitik
Gesellschaft fuer Wirtschafts- und Sozialwissenschafien, Berlin, Duncker & HumbloL
— (1988a), 'Inequality of Health Services is, in General, Desirable', in Green (1988).
— (1988Z>), 'The Radical Reforms the NHS Needs—and Doesn't', in Minutes of Evidence Taken Before the
Social Services Committee, London, HMSO.
— (1989), 'Commodities, Characteristics of Commodities, Characteristics of People, Utilities and the Quality
of Life', in Baldwin, S., Godfrey, C. andPropper, C. (eds.), The Quality of Life: Perspectives and Policies,
London, Routledge.
— Brazier, J. E. and O'Donnell, O. (1988), Organising Health Service Provision: Drawing on Experience,
London, Institute of Health Services Management
— Donaldson, C. and Gerard, K. (1988), Financial Aspects of Health Services: Drawing on Experience,
London, Institute of Health Services Management
— Lavers, R. and Williams, A. (1971), 'Social Indicators: Health', Social Trends, 2, 31-42.
Drummond, M. F., S tod dart, G. L. and Torrance, G. W. (1987), Methods for the Economic Evaluation of Health
Care Programmes, Oxford, Oxford University Press.
Evans, R. G. (1974), 'Supplier-Induced Demand—Some Empirical Evidence and Implications', in Perlman, M.
(ed.), The Economics of Health and Medical Care, London, Macmillan.
— (1976), 'Modelling the Objectives of the Physician', in Fraser, R. D. (ed.), Health Economics Symposium,
Kingston, Queen's University Industrial Relations Centre.
— (1981), 'Incomplete Vertical Integration: the Distinctive Structure ofthe Health-Care Industry', in van der
Gaag, J. and Perlman, M. (eds.). Health, Economics, and Health Economics, Amsterdam, North-Holland.
Evans, R.G. (1983), 'The Welfare Economics of Public Health Insurance: Theory and Canadian Practice', in
Soederstroem, L. (ed.), Social Insurance, Amsterdam, North-Holland.
— (1984), Strained Mercy: the Economics of Canadian Medical Care, Toronto, Butterworths.

56
A. J. Culyer

Farley, P. J. (1985), 'Who are the Underinsured?', Milbank Memorial Fund Quarterly, 63,476-503.
Fuchs, V. (1978), 'The Supply of Surgeons and the Demand for Surgical Operations' Journal of Human Resources,
13, 35-56.
Green, J. R. (1978), 'Physician-Induced Demand for Medical Care', Journal of Human Resources, 13,21-34.
Grogono, A. W. and Woodgate, D. J. (1971), 'Index for Measuring Health', Lancet, 1024-6.
Harris, A. et al. (1971), Handicapped and Impaired in Great Britain, London, HMSO.
Harsanyi, J. (1955), 'Cardinal Welfare, Individualistic Ethics, and Interpersonal Comparsions of Utility', Journal
of Political Economy, 63,4, 309-21.
Jewkes, J.,Jewkes,S., Lees, D.S.andKemp, A. (1963), 'Ethics and the Economics of Medical Care—Discussion',
Medical Care, 1,234-44.
Kant, I. (1930), Lectures on Ethics, London, Macmurray.
Klarman, H. E. (1963), "The Distinctive Economic Characteristics of Health Services', Journal of Health and
Human Behaviour, 4,44—9.
— (1965), 'The Case for Public Intervention in Financing Health and Medical Services', Medical Care, 3,

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


59-62.
Lees, D. S. (1960), "The Economics of Health Services', Lloyds Bank Review, 56,26-40.
— (1962), "The Logic of the British National Health Service', Journal of Law and Economics, 5,111-18.
— (1967), 'Efficiency in Government Spending: Social Services: Health', Public Finance, 22,176-89.
Lindsay, C. M. (1969), 'Medical Care and the Economics of Sharing', Economica, 144,351-62.
— (1973), 'Real Returns to Medical Education', Journal of Human Resources, 8,331^*8.
Lohr, K. N., Brook, R. H., Kamberg, C. J., Goldberg, G. A., Leibowitz, A. et al. (1986), 'Use of Medical Care in
the Rand Health Insurance Experiment Diagnosis and Service-Specific Analyses in a Randomized Con
trolled Trial', Medical Care, 24 (supplement).
Loomes, G. (1988), 'Disparities Between Health State Measures: An Explanation and Some Implications', paper
given at the Summer 1988 meeting of the Health Economists' Study Group.
McPherson, K., Strong, P. M., Epstein, A. and Jones, L. (1981), 'Regional Variations in the Use of Common
Surgical Procedures: Within and Between England and Wales, Canada and the USA', Social Science and
Medicine, 15 A, 273-88.
Manning, W. G., Newhouse, J. P., Duan, N., Keeler, E. B., Leibowitz, A. and Marquis, M. S. (1987), 'Health
Insurance and the Demand for Medical Care: Evidence from a Randomized Experiment', American
Economic Review, 11,251-77.
Margolis, H. (1982), Selfishness, Altruism and Rationality, Cambridge, Cambridge University Press.
Meade, J. E. (1973), The Theory of Economic Externalities, Leiden, Sijthoff.
Mooney, G. (1986), Economics, Medicine and Health Care, Brighton, Wheatsheaf.
Mundinger, M. O. (1985), 'Health Service Funding Cuts and the Declining Health of the Poor', New England
Journal of Medicine, 313,44-7.
Musgrave, R. A. (1959), The Theory of Public Finance, New York, McGraw-Hill.
Neumann, J. von and Morgenstern, O. (1953), Theory of Games and Economic Behaviour, New York, Wiley.
OECD (1977), Public Expenditure on Health, Paris, OECD.
Newhouse, J. P. and Phelps, C. E. (1976), 'New Estimates of Price and Income Elasticities of Medical Services',
in Rosette, R. N. (ed), The Role of Health Insurance in the Health Service Sector, New York, National
Bureau of Economic Research.
Parkin, D., McGuire, A. and Yule, B. (1987), 'Aggregate Health Care Expenditures and National Income: Is
Health Care a Luxury Good?', Journal of Health Economics, 6, 109-28.
Pattanaik, P. K. (1968), 'Risk, Impersonality and the Social Welfare Function', Journal of Political Economy, 76,
6,1152-69.
Pauly, M. V. (1971), Medical Care at Public Expense, New York, Praeger.
— (1980), Doctors and Their Workshops, Chicago, University of Chicago Press.
— (1986), 'Taxation, Health Insurance, and Market Failure in the Medical Economy', Journal of Economic
Literature, 24,629-75.
Phelps, C. E. (1986), 'Induced Demand—Can we Ever Know its Extent?', Journal of Health Economics, 5,
355-456.
— and Newhouse, J. P. (1974), Co-insurance and the Demand for Medical Services, Santa Monica, Rand
Corporation.
Rawls, J. (1972), A Theory of Justice, Oxford, Clarendon Press.

57
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 5, NO. 1

Rice, T. H. (1983), 'The Impact of Changing Medicare Reimbursement Rates on Physician Induced Demand',
Medical Care, 21, 803-15.
— (1987), 'Comment on "Induced Demand—Can we Ever Know its Extent?" By Charles E. Phelps', Journal
of Health Economics, 6,375-6.
Rosser, R. (1983), 'Issues of Measurement in the Design of Health Indicators', in Culyer (1983).
Schoemaker, P. J. H. (1982), "The Expected Utility Model: Its Variants, Purposes, Evidence, and Limitations',
Journal of Economic Literature, 20,529-63.
Sen, A. K. (1977), 'Social Choice Theory: A Re-examination', Econometrica, 45,53-90.
— (1979), 'Personal Utilities and Public Judgements: or What's Wrong with Welfare Economics', Economic
Journal, 89, 537-58.
— (1980), 'Equality of What?', in The Tanner Lectures on Human Values, Cambridge, Cambridge University
Press.
Shavell, S. (1978), 'Theoretical Issues in Medical Malpractice', in Rottenberg, S. (ed.). The Economics of Medical

Downloaded from oxrep.oxfordjournals.org at University of Toronto Library on November 10, 2010


Malpractice, Washington DC, American Enterprise Institute.
Sloan, F. A. and Feldman, R. (1978), 'Competition Among Physicians', in Greenberg, W. (ed.), Competition in
the Health Care Sector: Past, Present, and Future, Germantown, Aspen Systems.
Smith, A. (1987), 'Qualms about QALYs', Lancet, (i), 1134-6.
Sugden, R. (1980), 'Altruism, Duty and the Welfare State', in Timms, N. (ed.), Social Welfare: Why and How?,
London, Routledge and Kegan PauL
— (1982), 'On the Economics of Philanthropy', Economic Journal, 92,341-50.
— and Williams, A. (1978), The Principles of Practical Cost-Benefit Analysis, Oxford, Oxford University
Press.
Titmuss, R. M. (1970), The Gift Relationship, London, Allen and Unwin.
Torrance, G. W. (1976), 'Social Preferences for Health States: An Empirical Evaluation of Three Measurement
Techniques', Socio-economic Planning Sciences, 10,128-36.
— (1982),'Preferences of Health States: a Review of Measurement Methods', in Sinclair (1982).
— (1986), 'Measurement of Health State Utilities for Economic Appraisal: a Review', Journal of Health
Economics, 5,1-30.
Trebilcock, M. J., Tuohy, C. J. and Wolfson, A. D. (1979), Professional Regulation, Toronto, Ministry of the
Attorney General.
Vayda, E., Mindell, W. R. and Rutkow, I. M. (1982), 'A Decade of Surgery in Canada, England and Wales, and
the US', Archives of Surgery, 117, 846-53.
Weisbrod, B. A. (1961), The Economics of Public Health, Philadelphia, University of Pennsylvania Press.
Wiggins, D. and Dirmen, S. (1987), 'Needs, Need, Needing', Journal of Medical Ethics, 13,63-8.
Williams, A. (1972), 'Cost-benefit Analysis: Bastard Science or Insidious Poison in the Body Politick?', Journal
of Public Economics, 1, 199-216.
— (1974), 'Need as a Demand Concept (with Special Reference to Health)', in Culyer (1974).
— (1985), 'Economics of Coronary Artery Bypass Grafting', British Medical Journal, 291,326-9.
— (1986), "The Cost-Benefit Approach to the Evaluation of Intensive Care Units', in Miranda and Langrehr
(1986).
— (1987), 'Quality-adjusted Life-years',Lancet, (i), 1327.
— (1988), 'Ethics and Efficiency in the Provision of Health Care', in Bell, M. andMendus, S. (eds.), Philoso-
phy and Medical Welfare, Cambridge, Cambridge University Press.

58

View publication stats

You might also like