Full text of DOI:10.1080/00036840050155940

This article was downloaded by: [Korea University]
On: 31 December 2014, At: 18:42
Publisher: Routledge
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Applied Economics
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/raec20
Willingness to pay for publicly-financed health
care: how should we use the numbers?
Phil Shackley & Cam Donaldson
Published online: 04 Oct 2010.
To cite this article: Phil Shackley & Cam Donaldson (2000) Willingness to pay for publicly-financed health care: how
should we use the numbers?, Applied Economics, 32:15, 2015-2021, DOI: 10.1080/00036840050155940
To link to this article: http://dx.doi.org/10.1080/00036840050155940
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any
purpose of the Content. Any opinions and views expressed in this publication are the opinions and views
of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources of information. Taylor
and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses,
damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions

±
Applied Economics, 2000, 32, 2015 2021
W illingness to pay for publicly-®nanced
health care: how should we use the
numbers?
PHIL SHA CKLEY * and CA M DONA LDSON
{
She eld Health Economics Group, School of Health and Related Research, University
of She eld, Regent Court, 30 Regent Street, She eld S1 4DA, UK, Health
{
Economics Research Unit, University of Aberdeen, UK and Departments of
Community Health Sciences and Economics, University of Calgary, Canada
This paper addresses the question of how willingness to pay (WTP) values in health
care evaluation can be used by policy makers. The way in which WTP values are
used depends on from whom values are elicited and whether the good concerned is
privately-®nanced or publicly-®nanced through taxation. Thus, four possible uses of
WTP values are identi®ed. The focus is on the two uses which arise in the publicly-
®nanced situation. `Conventional’ use of WTP values, where the decision as to
whether or not to provide a service depends upon whether or not WTP values are
greater than total cost, applies only in the privately-®nanced, and not publicly-
®nanced situations. The situations with publicly-®nanced goods are more complex.
The use of WTP values for publicly-®nanced goods is justi®ed and illustrated.
I. INTR ODUCTION
whether WTP values are greater or less than cost.
However, it is the authors’ view that the use of WTP values
in such a way can be questioned, especially when, as is
often the case in health care, values are often derived
from users of care (i.e. patients) (Johannesson and
Fagerberg, 1992) and the budget for health care is fre-
quently capped. In particular, it is argued that the way
WTP values are used will depend on from whom values
are elicited (patients or the public) and whether the good
concerned is privately-®nanced or publicly-®nanced
through taxation (hereafter referred to as `publicly-
®nanced’). This leads to four possible scenarios: (i) data
are collected from patients for a privately-®nanced good;
(ii) data are collected from the public about a privately-
®nanced good; (iii) data are collected from the public about
a publicly-®nanced good; and (iv) data are collected from
patients for a publicly-®nanced good.
The use of willingness to pay (WTP) in programme evalua-
tion has become commonplace in many areas of public
sector activity; most notably in valuing safety, environmen-
tal and health care policies (Jones-Lee, 1989; Diener et al.,
1998; Seip and Strand, 1992). The volume of empirical
research activity in this area is evidenced by recent attempts
to classify studies and even establish guidelines (Diener et
al., 1998; Arrow et al., 1993; O’Brien and Gafni, 1996).
Despite this growth in activity and the production of
guidelines, one important question appears to have been
overlooked; that is how the results of WTP studies can be
used by policy makers. The arguments presented are new in
terms of applying them to the speci®c context of WTP
valuations for health care. The mathematical theory under-
lying the arguments has, however, been presented in the
general economics literature (Baumol, 1972).
There is, of course, a ®fth possible scenario in health
care, and that is where the intervention to be evaluated is
a public good in the traditional economic sense. However,
as is well known, public good characteristics of non-rivalry
At ®rst sight, the answer to the question of how to use
WTP values may seem obvious since, conventionally, the
decision as to whether to provide a service depends upon
*Corresponding author. E-mail: p.shackley@she eld.ac.uk
2015
#
±
±
/
Applied Economics ISSN 0003 6846 print ISSN 1466 4283 online
2000 Taylor & Francis Ltd
//
/
http: www.tandf.co.uk journals

2016
P. Shackley and C. Donaldson
and non-excludability lead to problems in establishing
`true’ strength of preference for such goods through the
use of WTP. While there are examples of WTP being
applied to public goods in health care, e.g. water ¯ uorida-
tion (Shackley and Dixon, 1999), one would argue that this
®fth scenario is so rare in health care as to warrant no
further discussion in this paper.
In the next section, each of the terms `privately-®nanced
goods’, `publicly-®nanced goods’, `patients’ and `the pub-
lic’, is de®ned. A brief discussion of the conventional view
of how WTP values should be used is presented and, in
doing so, shows that the use of WTP in scenarios (i) and
(ii) is uncontroversial. It is then argued that the conven-
tional approach cannot be meaningfully applied in the sce-
narios that are of greatest interest for most developed
countries, i.e. (iii) and (iv). Although it may be thought
at ®rst sight that a notable exception to this is the USA,
this is not necessarily the case. With the growing import-
ance of managed care, for example, a main characteristic of
publicly-®nanced health care systems, i.e. a capped budget,
is assuming an increasingly important role in the USA. For
each of the scenarios (iii) and (iv) a modi®ed alternative of
the conventional approach is suggested and illustrated.
may be in the process of consuming health care, or are
waiting to consume health care (e.g. on a waiting list), or
who are immediately eligible to consume health care (e.g.
pregnant women o€ ered antenatal screening). When asked
to value the health care programme under consideration,
patients will be primarily concerned with the value to them
(and perhaps also their immediate family) from consuming
the health care.
O’Brien and Gafni de®ne an ex ante insurer as anyone
who is at risk of contracting a disease and therefore at risk
of consuming the treatment programme. For example, if
the treatment programme is cervical screening, then the
relevant population would be all women. The author’s de®-
nition of `public’ includes these individuals, but also
extends to include individuals who are not necessarily at
risk of consuming the programme. Thus, in the case of a
cervical screening programme, the de®nition of public
would include not only all women, but also men. When
asked to value the health care in question, the public may
be concerned with the value to them (and again perhaps
their immediate family) from consuming the care, but they
may also be more concerned than patients with wider ben-
e®ts such as option value and externalities (Weisbrod,
1964; Culyer, 1971).
II. DEFINITION OF TER MS A ND THE
CONVENTIONA L PER CEPTION
Breaking with the `conventional’
Having de®ned the terms, one now turns to how WTP
values can be used. One use of WTP values (the `conven-
tional’ view) would be to compare such values with the
resource costs of providing a service, the decision rule
being that if the former are greater than the latter, the
project should go ahead and vice versa. However, not all
WTP values can be used in such a simplistic way. It will be
seen below that applying WTP values in this way is not
appropriate in all situations. Speci®cally, the conventional
view covers only scenarios (i) and (ii) above, i.e. when deal-
ing with a privately-®nanced good.
`Patients’ versus `the public’ and `private’ versus `public’
®nance
Before proceeding any further, it is important to de®ne
what is meant by `privately-®nanced goods’ and
`publicly-®nanced goods’, and to clarify the distinction
made between `patients’ and `the public’.
Privately-®nanced goods are de®ned as units of health
care which are funded from out-of-pocket payments or
from private insurance payments (as in the USA).
Publicly-®nanced goods, on the other hand, are de®ned
as units of health care that are funded from tax revenues
(as in the UK and Scandinavian countries). It will be seen
below that it is not the means of ®nancing per se which is
important, but rather the implications the source of ®nan-
cing has for health care budgets.
The conventional use of WTP values is encapsulated in
the following quote from Pauly (1985):
The cost-bene®t approach selects projects . . . by initiat-
ing all projects for which bene®ts are greater than cost,
that is, all projects for which the ratio of bene®ts to costs
is greater than unity (p. 109).
To distinguish between `patients’ and `the public’, one
takes as a starting point the de®nitions o€ ered by O’Brien
and Gafni (1996). In the context of framing WTP ques-
tions, they distinguish between what they call an ex post
user based perspective and an ex ante insurance based per-
spective. Their ex post user is assumed to be `at the point of
consuming some unit of the program being evaluated’.
While one could de®ne patients as ex post users, it is not
clear what is meant by `at the point of consuming’. In order
to be more speci®c, patients are de®ned as individuals who
are `currently diseased’. By this it is meant individuals who
This criterion is acceptable provided that there is no expli-
cit health care budget constraint. When health care is pri-
vately ®nanced (scenarios (i) and (ii)), there is in e€ ect no
budget constraint speci®cally for the good in question,
since, if individuals are willing to pay for care (and if
total WTP exceeds total cost), then the service will be pro-
vided.
An example of scenario (i) is the study by Fisher et al.
(1985) who asked attendees at an adolescent health service
their WTP for con®dential health care; in e€ ect, a market-

W illingness to pay for publicly-®nanced health care
2017
ing exercise. The situation in scenario (ii) is almost the
same as that for scenario (i), except that this time the
good is one that is generally available for everyday con-
sumption rather than when one becomes ill, as in the case
of the general public being asked for their WTP to avoid
food-borne risk (Donaldson et al., 1997a), or for a new
health care programme to be added to the package already
paid for, say by a health maintenance organization. In the
case of the latter, enrollees could be asked what is the maxi-
mum they are willing to pay to upgrade their coverage to
include the new programme (O’Brien et al., 1998).
Circumstances change once there is an explicit budget
constraint for the good(s) under consideration, i.e. when
there is a ®xed budget to provide medical care services for a
given population. As Pauly (1985) notes, in this situation
(which applies to publicly-®nanced goods as de®ned
above) :
straint since the latter implicitly assumes no indivisibilities
and constant returns to scale (Baumol, 1972). While non-
linear programming does not require such assumptions,
there are likely to be di culties in applying it in practice
since it requires a knowledge of all the costs and bene®ts of
the programmes under consideration for a speci®c budget.
Alternative (and arguably more pragmatic) approaches
to dealing with ®xed budgets are suggested in the following
two sections. For ease of illustration, it is assumed that
there are two types of budget to consider. These are the
total health care budget (dealt with in scenario (iii) below)
and budgets allocated to the treatment of speci®c con-
ditions (scenario (iv)). It is with the ®rst of these that the
next section is concerned.
III. DA TA A R E COLLECTED FR OM THE
PUBLIC FOR A PUBLICLY -FINA NCED
GOOD
since costs are ®xed by the budget, the general cost-ben-
e®t decision rule is to choose the set of actions with
maximum monetary bene®ts (p. 106).
The relevant constraint here is the total health care budget
that is available to provide a whole spectrum of health care
services. If WTP values are elicited in this context, it is the
author’s view that it is the values of the public, rather than
patients, that are relevant since the public’s values re¯ ect
the true opportunity cost context of decisions at this level.
For example, given a ®xed budget, any decision to imple-
ment or expand a treatment programme A for patients with
disease Z is the bene®t forgone from not having those
resources available to implement or expand another pro-
gramme (e.g. treatment programme B for patients with
disease Y ). Priority setting in this context would require
two judgements: (i) what is the better or best alternative?;
and (ii) what is the bene®t forgone if this alternative is not
implemented? It is this paper’s view that patients of one or
other particular service are not well placed to form either
judgement. Although members of groups Y and Z, through
giving a WTP value for their care, will have revealed the
opportunity cost out of their incomes of providing care for
themselves as they perceive it, the opportunity cost of
investing in treatment A rather than B manifests itself
not at this `private’ level but at the level of the community.
The relevant question is whether one way of spending the
community’s resources should be expanded at the expense
of another. Thus, it is a community value-based decision as
to which treatment receives what level of funding.
This is di€ erent from the conventional criterion of select-
ing all projects for which total bene®t is greater than total
cost. Thus, it is argued that the conventional approach is
not appropriate when dealing with a publicly-®nanced
good, i.e. scenarios (iii) and (iv). Of course, as alluded to
above, it could be argued that decisions about publicly-
®nanced goods do not always take place under an explicit
budget constraint and those about privately-®nanced
goods sometimes do. For example, services like home
health care have been added to the US Medicare pro-
gramme without reference to an explicit budget constraint.
If this then adds to Medicare costs, an evaluation of its
bene®ts would have to take place under the `conventional’
approach as de®ned in this paper. Likewise, some private
managed care plans set prospective budgets, in which case
evaluations of new or expanding services would have to
take place under the publicly-®nanced goods scenario as
de®ned here. It is important to highlight, however, that
none of this changes the basic points made in this paper.
It has been suggested that one way in which the budget
constraint can be tackled is to rank programmes in terms of
their bene®t-cost ratios (assuming there is consistency with
respect to classifying costs and bene®ts) or net bene®ts,
thereby allowing net bene®ts to be maximized for the avail-
able budget. However, it has been shown that such an
approach is problematic and does not necessarily lead to
the maximization of net bene®ts (Birch and Donaldson,
1987).
At any one time, a public sector health care decision
maker will be faced with a number of proposals for expan-
±
sion or introduction of services (say, A D). Given limited
In principle, non-linear programming techniques, such
as integer programming, o€ er a potential means of max-
imizing bene®ts subject to a budget constraint (Misham,
1967; Birch and Donaldson, 1987). Non-linear program-
ming is superior to the more traditional linear programming
approach to maximizing output subject to a budget con-
funds, these proposed expansions would have to compete
±
with some proposed reductions in care (say, E H). What is
important, then, is to obtain, through WTP values, an
assessment from the community about their strength of
±
preference for each of the options A H. Although these
options will inevitably be of di€ erent scales, and hence

2018
P. Shackley and C. Donaldson
costs, the aim would be to choose that combination of
projects which maximizes welfare (as represented by
WTP) from the (marginal) funds at stake. Thus, it is not
adverse event occurring. Both future users and future
non-users derive utility from, and therefore are prepared
to contribute ®nancially to, health care. Taxation is di€ er-
ent, however, in that, being compulsory, some people may
be paying who would rather not. But, given that they are
paying, it is di cult to see why they should be excluded
from any valuation exercise. Furthermore, in such `com-
pulsory payment’ systems, some people will be exempted
from paying but will still be eligible for the use of the
services. They are also likely to value the service and should
be included in any valuation exercise. Taxation-based ques-
tions, in which values are elicited from members of the
community, may also activate elements in the utility func-
tion, such as `caring externalities’ (Culyer, 1971), thought
to be relevant to publicly-funded health care systems, but
which out-of-pocket framed WTP questions may not pick
up (Mitchell and Carson, 1989).
Another potential problem with the above approach
relates to the decision making context in publicly-®nanced
health care systems. As already outlined, a public sector
decision maker is faced with the task of deciding on di€ er-
ent ways of spending a ®xed budget. In view of the budget
constraint, the true opportunity costs of any service expan-
sions are re¯ ected in the bene®ts that would have been
obtained from corresponding service reductions. Yet
WTP values derived from individuals with respect to such
expansions and reductions re¯ ect opportunity costs to
those individuals from alternative uses of their private
incomes. This is not the same as asking such individuals
how a ®xed public sector budget should be allocated. Thus,
when involving individuals in giving values to alternatives,
the ideal would be for them to state which services should
be sacri®ced (at the margin), and to what extent, in order to
have more of other services. Two problems are associated
with this. First, individuals are unlikely to have knowledge
of the wide range of existing services to permit them to
make such a judgement. Second, even if they did, it
would be di cult to know how to use such information,
as di€ erent individuals would recommend giving up di€ er-
ent services. `Value’ would not be expressed in a commen-
surate unit such as money. Thus, the use of WTP values
from members of the community is a theoretical compro-
mise.
±
±
inevitable that A D will replace E H. Some of the latter
may be retained.
One way of approaching this problem is to present to
respondents a limited number of proposals for service
expansion and reduction. Respondents are then informed
that they will be asked valuation questions on each alter-
native, thus making them aware that the relative monetary
weight they attach to each, through WTP out of their own
pocket, is important. This is no di€ erent from methods of
valuing quality adjusted life years (QALYs) or healthy year
equivalents (HYEs) whereby respondents are asked how
much of some attributes of their own health they would
forgo in order to gain other health attributes, the results
being used in comparisons of di€ erent types of health care.
It could, of course, be argued that there is a potential
problem in that members of the public will often have little
or no experience of the options they are being asked to
value, and as a consequence the `validity’ of their values
is questionable. Indeed, such an argument may be cited to
support the case for asking patients (the alternative of
appointing (or electing) decision makers to allocate the
community’s health care resources does not solve the infor-
mation problem as it is unlikely that the decision makers
themselves will have experienced the types of care upon
which they are making judgements.) However, in insurance
markets (which would develop in the absence of a public
purchase of health care through taxation) the values that
count and determine WTP for various levels of service
provision are not just those of patients, but also those of
non-users in the broader community. This is because indi-
viduals who take out insurance presumably perceive them-
selves to have some positive probability of use of health
care. At some time in the future it will be revealed who the
patients and non-users of services are. But at the time of
purchase this will not necessarily be clear. Those who ulti-
mately will not use services may still derive utility (at the
time of purchasing insurance) from knowing that services
will be paid for should they need to use it (an example of
option value, Weisbrod, 1964). If non-users did not derive
bene®t from coverage of future possible expenses incurred,
and therefore did not pay insurance premiums which are
then used to buy services for users of care, many of these
services would not be provided or would be under-pro-
vided. Therefore, the values of both users and non-users
are relevant (Gafni, 1991; O’Brien and Gafni, 1996).
The arguments above concerning insurance and option
value do not apply just to private markets, however. They
can also be applied to public systems funded from taxation,
such as the UK NHS. Although having other attributes
(such as being compulsory), certain aspects of taxation
can be likened to an insurance premium which people
pay to cover the ®nancial consequences of an uncertain
Regarding empirical applications of the method, there is
currently one example of WTP values being elicited from
the public for rival publicly-®nanced goods (Olsen and
Donaldson, 1998). In the study, members of the public
were asked their WTP in extra taxation for each of the
following: the prevention of the closure of a helicopter
ambulance service to serve remote communities; an expan-
sion in the number of heart operations performed; and an
expansion in the number of hip operations performed. The
programmes were described in some detail and respondents
were told that each of the three programmes would cost the
same to implement and that only one could be implemen-

W illingness to pay for publicly-®nanced health care
2019
ted. Respondents were also asked the reasons for their
valuations. Mean WTP for the ambulance service was
316 Norwegian Kroner (NOK) per year. The correspond-
ing values for hearts and hips were 306NOK and 232NOK
respectively. The mean WTP for hips was signi®cantly dif-
ferent from that for the ambulance service and hearts,
although more detailed analyses did reveal di€ erent levels
of WTP according to factors such as age, sex and educa-
tion. This study demonstrates that eliciting preferences
from the general public about publicly-®nanced health
care programmes is possible, although further development
of the method is clearly warranted before ®rmer conclu-
sions can be drawn. By comparing the WTP values with
those obtained when valuing bene®ts through the use of
quality adjusted life years (QALYs), constructed largely
by assumptions based on evidence from other studies,
the study also demonstrated the potential for WTP results
to produce a di€ erent set of priorities from the more-
established QALY approach.
approach WTP values would be elicited for both the treat-
ment alternatives and total WTP compared with total cost.
In the not inconceivable event that the WTP values for
both alternatives are greater than their respective costs,
then applying the conventional view of WTP, both alter-
natives would be provided for those people whose WTP
/
values are greater than cost price. However, if the budget
is not su ciently large to provide both to such an extent,
then the decision maker would have to make a choice
regarding how much of each to provide. This decision
would depend on, inter alia, the relative costs of alterna-
tives and the relative strengths of patients’ preferences (this
issue is discussed in more detail below). The important
point to note is that a di€ erent outcome is obtained from
that which would have prevailed in the privately-®nanced
good scenario.
Since a decision to implement a particular treatment
programme in the face of a budget constraint can lead to
gainers and losers, there would seem to be a need to estab-
lish the extent of the gains and losses. In the event of there
being two treatment alternatives, A and B, one method
would be to ask patients to value each of A and B sepa-
rately and then to compare the total WTP for each pro-
gramme. This has been done with patients within
randomized trials (Johannesson and Fagerberg, 1992;
Donaldson et al., 1995; Ryan et al., 1997) and with patients
who are not actually participating in a trial but who have
the same condition as trial patients (Donaldson et al.,
1997c). Neither approach has proved to be satisfactory in
terms of establishing di€ erences in value between the treat-
ment conditions. The method proposed here is a three stage
process and focuses directly on the di€ erence in value
between the treatment alternatives. In the ®rst stage,
patients are asked to indicate which of the treatment alter-
natives they prefer. Stages 2 and 3 involve asking them to
place a value (in terms of WTP) on having their preferred
treatment rather than that which is less preferred, i.e.
patients are asked to state what could be termed a `mar-
ginal’ WTP value.
IV. DA TA A R E COLLECTED FR OM
PATIENTS FOR A PUBLICLY -
FINA NCED GOOD
The relevant constraint here is the budget that has been
allocated to the treatment of speci®c conditions. In this
context, it is the WTP values of patients which are relevant
since it is patients who bear the opportunity costs of any
decision. For example, suppose there are two alternative
treatment programmes A and B which could be used to
treat a particular condition. Suppose further that patients
have a preference for one or other of A and B (patient
group X prefers A and patient group Y prefers B), and
that these preferences are measured through WTP values.
The decision to implement one programme, say A, rather
than the other will mean that one group of patients will be
`gainers’ (group X if A is chosen) while the other group Y
will be `losers’ (the extent of the gains and losses being
determined by the WTP values). The opportunity cost of
implementing A is therefore the bene®t that would have
accrued had programme B been implemented instead.
Clearly this opportunity cost is borne by patient group Y
since they would have preferred programme B to have been
chosen. Because it is patients who bear the opportunity
cost of such treatment decisions, one would argue that it
is the values of patients that should be used to decide which
treatment is implemented.
The main theoretical reasoning behind the `marginal’
±
approach is the original Kaldor Hicks criterion that
implies that a proportion of people will have preferences
for one state of the world whilst a proportion will have
preferences for another state. The question then posed is
what is the intensity of preference of the gainers relative to
that of the losers. It is such changes in the state of the world
with which most economic evaluations are concerned, e.g.
through comparison of costs and bene®ts of current treat-
ment with a new alternative. Thus, in valuing the net ben-
e®ts of introducing a new alternative, the ®rst data required
are those on preferences, i.e. who prefers the alternative
(the gainers from any proposed change) and who prefers
the status quo (the losers from the change)? (stage 1). It is
then necessary to establish the WTP of the gainers to have
When choosing between treatment alternatives for a par-
ticular condition (for example, within a randomized trial),
one means of eliciting patient values is through WTP ques-
tions, where the values elicited provide an indication of the
patients’ strength of preference for the di€ erent types of
care. If the evaluation of two treatment alternatives are
considered for the same condition, under the `conventional’
±
the alternative rather than current treatment hence the

2020
P. Shackley and C. Donaldson
term `marginal’ WTP (stage 2). Finally, it is necessary to
establish the WTP of the losers to maintain the status quo
(stage 3). This estimate could, of course, be made by use of
a measure based on willingness to accept (WTA) compen-
sation (Shackley and Dixon, 2000). However, given prob-
borne in mind that this extra cost would be met by
resources being taken from (or not being available to)
another area of health care, for which (one can normally
assume) a WTP value has not been provided. (Ideally, of
course, if such a decision had to be made, it should be
preceded by a WTP analysis at the level of the community.)
The welfare maximizing position would be for both types
of care to be made available (although cost and practical
considerations may preclude this).
±
lems of WTP WTA disparities, WTP to maintain the
status quo can be seen as a possible way around this prob-
lem. It has to be remembered, however, that WTP to main-
tain the status quo is a measure of equivalent and not
compensating variation, whilst WTA not to have the status
quo and WTP for an alternative are measures of compen-
sating variation. It should be pointed out that the import-
The marginal approach has been investigated empirically
in a study of midwifery-led care versus consultant-led care
in the management of labour (Donaldson et al., 1998).
Women booking at a maternity hospital were sent a ques-
tionnaire in which intrapartum care in both a midwife-
managed delivery unit and a consultant-led labour ward
were described. Respondents were asked to state their pre-
ference for one or other type of care (they also had the
option of stating an equal preference) and then to state
their WTP for their preferred type of care. Total WTP of
those preferring the midwives’ unit was greater than that of
the group with a preference for the labour ward (not sta-
tistically signi®cant at the 5% level). Other things being
equal, if a decision had to be made to provide one type
of care, midwife-managed care would be the chosen option
since, in theory, those preferring this type of care could
compensate those preferring consultant-led care and still
remain better o€ . Had the same approach been taken as
in the previous applications of WTP in randomized trials
(Johannesson and Fagerberg, 1992; Donaldson et al., 1995;
Ryan et al., 1997), it would have appeared that women
were indi€ erent between midwife-managed and consul-
tant-led care. It should be noted that this study was a
®rst (somewhat crude) attempt to test the feasibility of
the marginal approach, and further testing of the method
is clearly required. The approach has also been applied in
an assessment of parents’ preferences for di€ erent con®g-
urations of health services for children (Donaldson et al.,
1997b) and in the assessment of public preferences for
water ¯ uoridation (Shackley and Dixon, 2000). As a result
of the former study, the customer for the research (i.e. the
main purchaser of health services in the north-east of
Scotland) concluded that:
±
ance of the Kaldor Hicks criterion here is not its strict
application, but rather that it highlights the need to
know who are the gainers and who are the losers from
any proposed change.
WTP studies in all ®elds to date have tended to focus
only on stage 2, without data being collected on preferences
(stage 1) or on WTP to maintain the status quo (stage 3).
Marginal WTP is more theoretically relevant (at least in
Paretian terms) and may result in a more practical format
of the WTP question from respondents’ points of view,
although it should be recognized that there are many
other sources of potential bias in WTP questions
(Mitchell and Carson, 1989). More generally, it seems
clear that what is required is for patients to be given a
choice between the treatments to be evaluated. This ®ts
better with the question which evaluations of close substi-
tutes seek to answer, i.e. what is the better (or best) way of
treating a given group of patients. The hypothesis here is
that making respondents aware of the fact that there is a
choice to be made will result in di€ erent valuations being
placed on care alternatives, or, at least, that a more dis-
criminating result will be obtained. By `discriminating’ it is
meant that if there is a di€ erence in patient utility between
each of the alternatives, it will more likely be picked up by
WTP.
Values elicited using the marginal approach can be used
in two ways. First, assuming that one of the options to be
valued is the maintenance of the status quo and that
options are mutually exclusive (i.e. only one can be imple-
mented), then, other things (including cost) being equal,
the WTP to keep the status quo of those preferring this
option should be compared with the WTP for change of
those preferring change. The option to be implemented
would be that preferred by whichever group could compen-
sate the other and still remain better o€ , i.e. the group with
the greater total WTP.
Where the WTP results demonstrate that a group of
people not only prefer one option over the other but
feel very strongly about it, then purchasers are faced
with either satisfying the expressed wish of the general
public or, if on clinical or ®nancial grounds, implemen-
tation is not feasible or desirable, consideration must be
given to undertaking a public information exercise to
explain in detail and to the public’s satisfaction why
their preferred option will not be selected. (Donaldson
et al., 1997b: 22)
Second, in situations where other things (in particular,
the cost of each option) are not equal, information on
preferences and strengths of preferences are relevant since
these indicate greater or less utility to individuals. If the
option with greater WTP (and hence utility) also has a
greater cost, the decision maker then has to decide whether
this extra cost is worth incurring. However, it should be
Of course one problem with such studies is in terms of
identifying and recruiting a representative and independent

W illingness to pay for publicly-®nanced health care
2021
Donaldson, C., Mapp, T., Ryan, M. and Curtin, K. (1997a)
Estimating the economic bene®ts of avoiding food-borne
risk: is `willingness to pay’ feasible?, Epidemiology and
group of patients. However, this is a problem for all studies
which attempt to elicit community values. These prelimin-
ary studies using the marginal approach have demon-
strated its potential to discriminate between options when
such di€ erences in preferences actually exist and to be use-
ful in health care decision making.
±
Infection, 116, 285 94.
Donaldson, C., Mapp, T., Farrar, S., Walker, A. and Macphee, S.
(1997b) Assessing community values in health care: is the
`willingness to pay’ method feasible?, Health Care Analysis,
±
5, 7 29.
Donaldson, C., Shackley, P. and Abdalla, M. (1997c) Willingness
to pay to value close substitutes: carrier screening for cystic
V. CONCLUSION
±
®brosis revisited, Health Economics, 6, 145 59.
Donaldson, C., Hundley, V. and Mapp, T. (1998) Willingness to
pay: a method for measuring preferences for maternity care,
It has been argued that the use of WTP has to take account
of the institutional arrangements of the market in which a
study is carried out. `Conventional’ use of WTP values is
only applicable to private goods. Once a budget constraint
is introduced for the health care evaluated or for total
health care expenditure, as is the case with publicly-
®nanced goods, the situation becomes more complicated.
The arguments justifying the use of WTP for publicly-
®nanced goods are not straightforward and neither are
the uses of these values. However, it has been argued that
the methods outlined for the elicitation and application of
these values re¯ ect more realistically the circumstances per-
taining to publicly-®nanced goods and that the values are
of use to decision makers.
±
Birth, 25, 32 9.
Fisher, M., Marks, A., Trieller, K. and Brody, R. (1985) Are
adolescents able and willing to pay the fee for con®dential
±
health care?, Journal of Pediatrics, 107, 408 83.
Gafni, A. (1991) Willingness-to-pay as a measure of bene®ts: rele-
vant questions in the context of public decision making about
±
health care programmes, Medical Care, 29, 1246 52.
Johannesson, M. and Fagerberg, B. (1992) A health economic
comparison of diet and drug treatment in obese men with
±
mild hypertension, Journal of Hypertension, 10, 1063 70.
Jones-Lee, M. W. (1989) The Economics of Safety and Physical
Risk, Basil Blackwell, Oxford.
Mishan, E. J. (1967) A proposed normalisation procedure for
±
public investment criteria, Economic Journal, 77, 777 96.
Mitchell, R. and Carson, R. (1989) Using Surveys to V alue Public
Goods, Resources for the Future, Washington DC.
O’Brien, B. and Gafni, A. (1996) When do the `dollars’ make
sense? Toward a conceptual framework for contingent valua-
tion studies in health care, Medical Decision Making, 16,
A CKNOWLEDGEMENT
The Health Economics Research Unit is funded by the
Chief Scientist O ce of the Scottish O ce Department
of Health. Cam Donaldson is Svare Professor of Health
Economics and Alberta Heritage Senior Scholar at the
University of Calgary. The opinions expressed in this
paper are those of the authors, not the funders.
±
288 99.
O’Brien, B., Goeree, R., Gafni, A., Torrance, G. W., Pauly, M.
V., Erder, H., Rusthoven, J., Weeks, J., Cahill, M. and
LaMont, B. (1998) Assessing the value of a new pharmaceu-
tical: a feasibility study of contingent valuation in managed
±
care, Medical Care, 36, 370 84.
Olsen, J. A. and Donaldson, C. (1998) Helicopters, hearts and
hips: using willingness to pay to set priorities for public sector
±
health care programmes, Social Science and Medicine, 46, 1
R EFER ENCES
Arrow, K., Solow, R., Portney, P., Leamer, E., Radner, R. and
Schuman, H. (1993) Report of the NOAA Panel on contin-
gent valuation, Federal Register, 58, 4601 14, Washington
DC.
Baumol, W. J. (1972) Economic Theory and Operations Analysis,
Prentice-Hall, New Jersey.
Birch, S. and Donaldson, C. (1987) Applications of cost-bene®t
analysis to health care: departures from welfare economic
theory. Journal of Health Economics, 6, 211 25.
Culyer, A. J. (1971) The nature of the commodity health care and
its e cient allocation, Oxford Economic Papers, 23, 189 211.
Diener, A., O’Brien, B. and Gafni, A. (1998) Health care contin-
12.
Pauly, M. V. (1985) Valuing health care bene®ts in money terms,
in V aluing Health Care: Costs, Bene®ts, and E€ ectiveness of
Pharmaceuticals and Other Medical Technologies (Ed.) F. A.
±
±
Sloan, Cambridge University Press, Cambridge, pp. 99 124.
Ryan, M., Ratcli€ e, J. and Tucker, J. (1997) Using willingness to
pay to value alternative models of antenatal care. Social
±
Science and Medicine, 44, 371 80
Seip, K. and Strand, J. (1992) Willingness to pay for environmen-
tal goods in Norway: a contingent valuation study with real
±
±
payments, Environmental and Resource Economics, 2, 91 106.
±
Shackley, P. and Dixon, S. (2000) Using contingent valuation to
elicit public preferences for water ¯ uoridation, Applied
gent valuation studies: a review and classi®cation of the lit-
±
±
Economics, 32, 777 87.
erature, Health Economics, 7, 313 26.
Weisbrod, B. A. (1964) Collective-consumption services of indi-
Donaldson, C., Shackley, P., Abdalla, M. and Miedzybrodzka, Z.
(1995) Willingness to pay for antenatal carrier screening for
vidual-consumption goods, Quarterly Journal of Economics,
±
78, 471 7.
±
cystic ®brosis, Health Economics, 4, 439 52.