Benefits Transfer of Children’s Health Values

NCEE &

NATIONAL CENTER FOR

ENVIRONMENTAL ECONOMICS

BENEFITS TRANSFER OF CHILDREN'S HEALTH VALUES

Maria Markowski

Working Paper Series

Working Paper # 02-10
September, 2002

SS1

U.S. Environmental Protection Agency
National Center for Environmental Economics

# o %

° SwlZ. ^ 1200 Pennsylvania Avenue. NW(MC 1809)

^'I'^ > Washington DH 90460

^ Washington, DC 20460

pro^g
-------
BENEFITS TRANSFER OF CHILDREN'S HEALTH VALUES

Maria Markowski

Correspondence:

Maria Markowski
Industrial Economics, Incorporated
markowski@indecon.com

NCEE Working Paper Series

Working Paper# 02-10
September 2002

DISCLAIMER

The views expressed in this paper are those of the author(s) and do not necessarily represent
those of the U.S. Environmental Protection Agency. In addition, although the research described
in this paper may have been funded entirely or in part by the U.S. Environmental Protection
Agency, it has not been subjected to the Agency's required peer and policy review. No official
Agency endorsement should be inferred.

-------
BENEFITS TRANSFER OF CHILDREN'S HEALTH VALUES

Maria Markowski1

Abstract: One of the many difficult issues the Environmental Protection Agency's Office of
Children's Health Protection is addressing is the appropriate treatment of children's health effects
in the economic analyses performed by the Agency. Policy analysis efforts at the Agency often
rely on the benefits transfer technique, and very few of the Agency's benefit transfers have
explicitly addressed children's health issues. To assist the Agency in its efforts, this paper
discusses the benefits transfer technique as it applies to estimating values for children's health.

Subject Area Classification:

57 Benefit-Cost Analysis
63 Children's Health

Key Words:

Children, Benefit Transfer

This paper was funded by EPA's Office of Children's Health Protection and the National
Center for Environmental Economics.

-------
INTRODUCTION

EPA established the Office of Children's Health Protection (OCHP) in 1997 to support its
efforts to increase the protection of children's health throughout its programs. One of the many
difficult issues the Office is addressing is the appropriate treatment of children's health effects in the
economic analyses performed by the Agency. Policy analysis efforts at the Agency often rely on the
benefits transfer technique, and very few of the Agency's benefit transfers have explicitly addressed
children's health issues. In addition, no accepted systematic process for conducting benefits transfer
currently exists.

To assist the Agency in its efforts, this paper discusses the benefits transfer technique as it
applies to estimating values for children's health. The first section provides some general
background on the technique, and its application to estimate health-related values for children. The
second section raises important issues to consider when conducting a benefits transfer for children's
health values. The last section discusses the implications of using the benefits transfer method to
estimate values for children's health. The scarcity and state of existing child-oriented health
valuation literature suggests that it may be necessary to transfer adult-oriented values to estimate
child-related values (Neumann and Greenwood 1999). However, as noted in Agee and Crocker
(1999), several issues that play an important role in the economic valuation of health may differ
between adults and children. Analysts should acknowledge that the differences in these determinants
of value add imprecision to the transferred value estimates. The results of this paper suggest that
transfer of these value estimates to children at best provides estimates for a scoping analysis. In cases
where these scoping exercises indicate that children's health values may be a crucial component in
the policy analysis, primary research should be undertaken to explore how health value determinants
may differ between adults and children, and to estimate child-related values.

-------
BACKGROUND

Benefits transfer is an alternative to using primary research to value health effects. In
benefits transfer, valuation information from one or more existing studies is used to assess benefits
in a new policy setting.2 In the case of valuing children's health, much of the existing research does
not directly address child-related benefits. For this reason, it is important to explicitly consider the
issues involved in transferring existing adult-oriented health value estimates to estimate values for
child health effects.

Applications of benefits transfer to value children's health often have aimed to provide only
rough approximations of the monetary benefits of avoiding adverse health effects. For example,
EPA's retrospective cost-benefit analysis of the Clean Air Act (EPA 1997) estimates a range of
benefits associated with changes in children's IQ using a number of adult wage-rate studies. In
particular, the study estimates the value of neonatal mortality due to low birth weight caused by
maternal exposure to lead. The study applies a value of statistical life estimate of $4.8 million,
calculated from 26 contingent valuation and wage-risk studies of adults, to estimated mortality
changes.

In addition, EPA recently undertook an effort to present a set of "off-the-shelf' approaches
from the existing literature and a discussion of the use of these approaches for application in analyses
of environmental health risk reduction (EPA 1998). In this study, EPA assesses the practical issues
related to valuing non-cancer (i.e., morbidity) health effects by discussing the methods for applying
existing primary research through benefits transfer, including the key issues involved in applying this
technique. EPA provides case studies that identify important issues to consider when estimating
values, including case studies of childhood asthma and childhood lead poisoning. Several other
studies have discussed benefits transfer and the issues that may arise as a result of employing this

-------
technique for policy analysis; however, these studies do not specifically address the value of
children's health effects.3

BENEFITS TRANSFER CONSIDERATIONS

Because the number of existing child-specific value estimates is limited, transfers for
estimating the value of children's health effects will likely involve values developed for adults.
Because there are many aspects of children's health values that differ from those of adults (see Agee
and Crocker 1999), transferring these values is not a straightforward process. Children's health
values often reflect the values that adult caregivers have for children, and not values of the children
themselves. Many of the determinants of value for children's health are different from those
associated with adults, including: caregiver affection and empathy (i.e., altruism); parental
motivations to obtain future care from their children; desire to maintain potential for future
productivity of the child as an adult; public empathy; ability for caregivers to protect child health;
and availability of community-supplied child health protection. As a result, critical considerations
of children's health issues must enter into the following benefits transfer steps: describing the policy
scenario for transfer, evaluating the appropriateness of the existing studies, and transferring the
benefits to the policy case. Table 1 below summarizes the steps of the benefits transfer process and
the important characteristics and determinants of value to consider when using adult-oriented values
to estimate values for children's health. The remainder of this section discusses each of these
considerations.

Policy Scenario

The first step in conducting a benefits transfer is to carefully describe the policy case. The
ability to identify relevant existing studies, assess their suitability for transfer, and conduct the

-------
transfer first depends on an accurate and thorough description of how the policy will affect health
and economic well-being. Such a description involves: (1) consideration of how the health effects
of the policy are measured; (2) a thorough description of the characteristics of the health effect likely
to influence willingness to pay (WTP) values; (3) a complete accounting of how a change in the
health effect will affect well-being; and (4) a description of the population experiencing the change
in the health effect. Several considerations specifically related to children arise in each step of this
process of describing the policy case.

As a first step, it is important to consider the issues regarding the measurement of health
effects of the policy, and how they might be different for children than for adults:

• Would people perceive the effect as adverse? Health scientists may measure
effects that ordinary people might not notice or perceive as affecting their well-
being. It is difficult or impossible to place a meaningful economic value on
these effects. Health effects might be perceived as different for adults versus
children. Caregivers may perceive health effects to be more (or less) benign for
adults than for children.

• Does the measure reflect a health effect alone, such as an asthma attack, or a
behavioral response as well, such as a day of work loss? Depending on the
situation, this issue may not be different for children and adults. Caregivers still
may have to miss work to care for a sick child, but the measure may not reflect
effects that children experience (e.g., missed school).

What is the degree of uncertainty in the health effect measurement? Generally,
greater efforts at precision in benefit estimation are warranted when health
effects are measured with greater precision. In cases where less (or more) is
known about children's health effects than adults, this issue may differ from
adults to children.

-------
Table 1

BENEFITS TRANSFER STEPS
Using Adult-Oriented Health Studies to Estimate Child Health Values

BENEFITS TRANSFER STEP

SELECTED FACTORS TO CONSIDER

SteD 1: Describe the Policy Case

Perception of adversity; exclusion of child behavioral responses;
certainty of measurement different from adults

• Health Effect Measurement

• Health Effect Characteristics

• Weil-Being Impacts

• Population Characteristics

Certainty of occurrence; type of change; baseline health level;
frequency; duration; severity; ease of avoidance; nature of health
effect; latency

Exclusion of: lost school time; caregiver disutility due to child
pain and suffering; child's foregone future earnings; caregiver
foregone earnings

Age; health status; education level

SteD 2: Studv Suitability

Depends on study method's ability to account for determinants
of value, e.g., caregiver affection and empathy; parental future
care motivations; future child productivity; public empathy;
ability to protect child health; community-supplied child health
protection

• Study Quality

1. Cost of Illness

2. Contingent Valuation

3. Averted Behavior

• Study Applicability

Differences in: treatment; child-related costs; foregone earnings
length of illness

Issues are with respect to study applicability

Behavioral differences that depend on health effect, e.g., child
safety seat, nutrition needs

Differences in: policy and study cases; susceptibility to health
impacts; severity levels

SteD 3: Transferrins Estimates

Consider differences in: health effect measurement, health effect
characteristics, well-being impacts/determinants of value, and
population characteristics; need some understanding of child
health values relative to those of adults

• Point Estimate Approach

• Benefit Function Transfer

• Meta Analysis

• Bayesian Analysis

-------
As discussed in EPA's draft non-cancer valuation handbook (1998), a necessary precondition
for matching the policy case to existing studies is to account for the characteristics of the health effect
which influence WTP when describing the policy scenario. In any health-related benefits transfer,
key characteristics include:

1. The certainty with which the health effect is likely to occur, or is at risk of
occurring.

2. The baseline level and policy-induced change in the frequency, duration,
severity or probability of the health effects.

3. The ability to easily avoid or relieve the health effect.

4. The nature of the health effect as occurring in isolation or with other
symptoms.

5. The existence of a latency period associated with the health effect.

Although the importance ofthese characteristics will likely vary with the policy case, several
may have more relevance for children than adults, thereby requiring different considerations during
the benefits transfer. For example:

The duration of the health effect may have a serious effect on children's
health values (Moore and Viscusi 1988). Children have longer expected
lives than their adult caregivers, and therefore may experience chronic
health conditions that are longer in duration than what adults would
experience. Health effect duration may also be important if the long
duration of a health effect leads to significant and long-term indirect effects
(e.g., educational impacts).

Averting behavior may be more difficult to fully describe with children's
versus adult health values. The caregiver may try to control the child's
behavior in an effort to avoid or relieve a given health effect, and the child
may or may not respond to this intervention. In addition, caregiver
behavior or intervention may not accurately reflect child health values if the
caregiver is unaware of the risks to the child's health or does not perceive
the health risks accurately.

The severity of a given health effect could be more or less significant for
children, depending on the physical differences between adults and
children. Some health effects may have a relatively mild effect on adults
as compared with children, resulting in a more severe health condition for
children than adults (e.g., lead exposure). Alternatively, some health effects

-------
may have the opposite consequence. For example, liver toxicity from
exposure to acetaminophen is more severe in adults than children
(Kauffman, 1992).

• The delay from the time of exposure to the manifestation of the effect may
have complex effects on health values for children. Caregivers may
discount a child's future at a rate lower than that which they discount their
own future because they do not wish to preclude opportunities for their
children. For example, caregivers may de-lead their homes to prevent their
child from suffering from impaired cognitive development, leading to lower
IQ level and reduced educational attainment. In addition, caregivers
concerns for delayed effects may reflect selfish considerations if they expect
that their children would provide them with future care (i.e., the eventuality
of children caring for their caregivers).

The next element of describing the policy scenario is to categorize the health-related impacts
on children's well-being. For adults, several economic effects may arise from health decrements,
including increased medical expenses, foregone earnings, losses in nonmarket production, lost leisure
time, increased averting costs, and increased pain and suffering. These impacts may also occur
relative to children's health effects. For example, caregivers may incur lost earnings because they
are responsible for taking the child for medical care. Additional economic impacts may result from
poor children's health, including lost school time, disutility a caregiver bears from the pain and
suffering a child may endure, and children's foregone future earnings. Often it will not be possible
to value all of these effects without primary research.

Finally, a complete description of the policy includes a portrayal of the affected population.
This description typically focuses on personal characteristics likely to affect WTP, such as income,
age, education, and health status. In some cases, the characteristics used in the adult health valuation
context will be appropriate for transferring adult values to children. For example, in children's health
valuation, characteristics such as education and income are likely to representthe caregiver's income
and education level. These characteristics are appropriate for the child health valuation context
because these adult characteristics may affect the degree and type of mitigation, the amount of
information caregivers have concerning the health effect, and the accuracy of the caregivers'
perceptions of the risks to the child. However, missing from the adult-oriented study could be
characteristics such as education level or age of the child.

-------
Study Suitability

When applying the benefit transfer technique to value children's health effects, a necessary
first step is to identify existing children's health studies or adult-oriented studies which may be
relevant to the policy case. Using the description of the effects of the policy as a guide, the analyst
can search the literature for high-quality studies that value a similar health effect in a similar
population. In the case of children's health, however, analysts will be faced with the relative scarcity
of relevant primary research. Current efforts describe the existing studies that provide estimates of
children-oriented values related to mortality, childhood cancers, chronic effects, and acute effects,
prenatal exposure effects, and reproductive effects (Dickie and Nestor 1998; Neumann and
Greenwood 1999). These papers identify the relatively few (approximately 20) existing studies of
child-oriented values that value a broad range of effects. Because the studies cover relatively little
ground in comparison to the types of effects that may exist for children's health valuation, it may be
more appropriate to transfer adult-oriented value estimates to estimate child-related values. As a
result, the criteria for assessing study suitability must take this type of transfer into account.

Two major issues involved in reviewing the studies for suitability to the study case concern
(1) the quality of existing studies, and (2) their applicability to the new policy situation. Quality
refers to the defensibility of the research methodology employed, and the reliability and precision
of the estimates obtained. Applicability refers to the match between the study case (the situation
examined in the original study) and the policy case (the situation relevant to the new policy). A
summary of general criteria for evaluating the transferability of existing studies discussed in the
literature can be found in Desvousges, et. al (1998). Because child- and adult-oriented value
determinants differ, the evaluation of the guidelines for assessing the quality and applicability of
studies for children's health valuation may result in differing conclusions about the suitability of any
given study for transfer.

-------
Study Quality

In benefits transfer, the guidelines for assessing study quality should include some
assessment of whether the study employs "best research practices" for estimating health values (i.e.,
whether the study addresses the appropriate theoretical constructs necessary for valuing children's
health); has been peer-reviewed and is viewed favorably in the professional community; and provides
results that compare with results in other studies or conform with expectations from theory. EPA's
Guidelines for Economic Analysis (draft, 11/8/98) provides information on assessing the quality of
studies for benefits transfer. The Guidelines note that indicators of study quality will generally
depend on the method used.

In Agee and Crocker's discussion of the relative strengths and weaknesses of the methods
to value children's health benefits (1999), the authors state that a study is considered of high quality
if it addresses determinants of value for children's health. These determinants may include: the
motivations caregivers may have for the demand and supply of children's health; the context or
setting in which the adults make decisions regarding children's health; whether caregivers undertake
private protection or other mitigating factors to protect children's health; and if other important
parameters, such as the caregiver and sibling health states or the provision of local health and hospital
services, are included in the valuation. Because many ofthe value determinants differ between adults
and children, assessing the quality of an adult-oriented study for purposes of valuing child health may
differ from when valuing adult health.

Three nonmarket valuation methods typically have been used to estimate the value of health
risks (cost of illness, contingent valuation, and averting behavior/household production).4 Cost of
illness studies are used frequently for health valuation, however this method is problematic for
several reasons.5 These issues notwithstanding, using the adult-oriented cost of illness studies to

-------
estimate child-related values requires consideration of several child health-specific factors, for
example:

• The course of health treatments for a child may differ from those of adults,
resulting in differing costs (e.g., elderly individuals with kidney disease are
less likely to receive a kidney transplant as a treatment option than would
be young individuals with the disease). Similarly, the specialization in
treatment required for children may be different for adults (e.g., pediatric
oncology versus geriatric oncology).

• Adult-oriented cost of illness studies will not reflect the costs such as those
associated with school or future losses in labor market earnings.

• Foregone earnings from adult illnesses may be different for child illnesses.
For example, if the caregiver is a homemaker, a cost of illness study could
overestimate costs in this category. However, if the caregiver works outside
the home, the foregone earnings from the adult illness may be reflective of
the foregone earnings resulting from spending time to care for their child.

• The length of the illness and expenditures incurred as a result of the
duration of the illness may differ between children and adults.

When transferring adult-oriented values from studies using the contingent valuation
methodology to estimate child health values, the issues of study quality are the same regardless of
whether the value estimates are transferred for purposes of adult or child health valuation. In 1992,
the National Oceanic and Atmospheric Administration commissioned apanel of economists and other
experts to review the contingent valuation method and its application for measurement of values.
The panel provided guidelines for conducting contingent valuation studies (DOC 1993). If the
conditions for conducting these studies are met, the question of study suitability rests on study
applicability and whether the commodity is well characterized.

Transferring values from adult-oriented studies of averted behavior for children requires that
the analyst assess whether the steps taken to avoid a child-related health effect differ from those that
would be taken to avoid an adult-related health effect. For example, while adults might decide to
reduce their risk for heart disease and remain generally healthy by consuming less fat in their diet,
children may require certain quantities of fat to remain healthy. The preferred outcome is the same
in both cases, but the behavior differs between adults and children. Another example is one of safety

-------
in cars - young children require car seats, an expense not included as part of an adult-oriented averted
behavior study. In other cases, adults and children may undertake the same behavior to avoid the
same health effect. For example, both adults and children may purchase and wear bicycle helmets
to avoid injury.

Applicability

Applicability concerns whether available studies involve health effects and populations
similar to the policy case, and whether adjustments can be made for important differences. For any
type of health-related benefits transfer, similarity can be assessed by describing the health effects,
impacts on well-being, and affected population in a manner parallel to the description of the policy.
A careful comparison of the descriptions of the study and the policy case will reveal the
characteristics which are similar, and the nature and extent of differences. When transferring benefits
for children's health valuation, there will be some important differences in the health effects or
populations of the available adult-oriented studies versus those relevant to the policy case. Three
elements to consider when assessing the comparability of the study case to the policy case include:

1. Whether the basic commodities of the policy and study cases are equivalent.

2. Whether the baseline criteria (i.e., quantity and quality) and extent of
change are similar between the policy and study cases.

3. Similarity between the study and policy case populations.

While the last of the above three criteria raises the issue of similarity between adult and
children populations directly, the first two criteria must also take this consideration into account. The
basic commodities of the policy and study cases may appear to be equivalent for adult and child
health valuation, but the analyst must be sure to address fundamental health effect differences that
may exist between adults and children. Specifically, poor environmental health conditions may have
different health outcomes on children than adults, regardless of any mitigating behavior that may be
undertaken. Children may be more (or less) susceptible to health impacts from certain environmental

-------
conditions than adults (e.g., lead impacts are more serious for children than adults, air pollution
impacts may be more severe for the elderly than for children). Similarly, while the baseline criteria
and extent of change may represent one level of severity for adult populations, it may represent
another level of severity for child populations. For example, there remain significant uncertainties
that limit available information on dose-response relationships between environmental contaminants
and childhood asthma (EPA 1998); however, there is significant scientific evidence that indicates
children are more susceptible to adverse health effects caused by certain environmental
contamination (EPA 1996). As a result, the analyst must be sure to investigate the potential
differences in physical effects from contaminant exposure between children and adults for each
policy case.

In addition, when assessing study applicability, the analyst must consider how risk may be
incorporated into the existing studies and how that compares with the policy situation. For example,
analysts must consider whether caregivers are taking measures to reduce risk (i.e., an ex ante
behavior), or if they are attempting to lessen the effect on the child (i.e., an ex post behavior). Risk
may enter into ex ante valuation studies as uncertainty about whether the effect will occur. In
addition, in either ex ante or ex post studies, risk may involve the uncertainty about whether actions
taken to mitigate the effect will be realized. In child's health valuation, it is important to consider
how risk enters into the caregivers decision process, and how that compares with the risk described
in the existing studies.

-------
Transferring Estimates

The final step in the benefits transfer technique is to transfer the valuation information from
the studies to the policy case. In any benefits transfer this involves combining and applying estimates
to the policy case; aggregating benefits to the relevant population; and considering the uncertainties
and limitations of the procedure. At times, several relevant studies will be available, or a single study
will provide different estimates of the same value based on different subsamples, assumptions, or
estimation procedures. Although no benefits transfer protocol has been established when using the
results from many studies, several techniques exist to transfer estimates (see, for example,
Desvousges, et. al 1998). This section discusses the applicability of these different techniques to
transfer adult-oriented values to children.

Point Estimates

The point estimate approach involves taking the mean value (or range of values) from the
study case and applying it directly to the policy case. It is rare that a policy case and study case will
be identical when considering children's health values, thus this approach is most useful for
conducting preliminary benefits estimation. Rather than directly using existing values, analysts will
often adjust point estimates based on judged differences between the study and the policy cases.
Judgments of this type should be based on economic theory, empirical evidence and experience
(Brookshire and Neill 1992). When transferring adult health values for children, these judgments
should account for differences in the determinants of values that may exist for adults and children.
The point estimate approach is most appropriate for scoping and screening analyses (i.e., preliminary
analyses to characterize the magnitude of the benefit estimate).

Existing benefits transfer applications of point estimates that value health effects have aimed
to provide rough approximations of the monetary benefits of avoiding adverse health effects. For

-------
example, the retrospective analysis of the Clean Air Act (CAA) estimates the value associated with
reducing child lead exposure (EPA 1997). A major effect of certain elevated lead exposures is
permanently impaired cognitive development in children, however, at the time of the study, few
estimates of society' s WTP for improved cognitive ability were currently available. One recent effort
includes parental WTP for investment in child cognitive development (Agee and Crocker,
forthcoming). As a result, the CAA analysis estimates values for two effects of IQ decrements (i.e.,
reductions in expected lifetime income and increases in societal expenditures for compensatory
education) using existing studies. The effect of IQ on expected lifetime income comprises a direct
(the effect of IQ on income) and an indirect (the effect of educational attainment on income) effect.
In addition, the analysis assumes that part-time compensatory education is required for all children
with IQ less than 70. For these benefit categories, this analysis identifies and reviews several studies
for the transfer, and adjusts point estimates based on economic theory, empirical evidence, and
professional judgment. The result is that the analytic approach almost certainly understates the WTP
to avoid impaired cognitive development in children, and, as stated in the analysis, probably should
be considered a lower bound estimate.

Benefits Function Transfer

Another approach for transferring value estimates is benefit function transfer. If a study case
provides a willingness to pay function, valuation estimates can be updated by substituting applicable
values of key variables, such as baseline risk and population characteristics (e.g., mean or median
income, racial or age distribution) from the policy case into the benefit function. This approach has
received mixed reviews in the literature. Kirchhoff, et. al (1997) develop a methodology fortesting
the validity of benefits transfer for water-dependent recreation and reject using the (mean) point
estimate technique for transfer, preferring the benefits function transfer approach. By comparing the
transfer estimates to the site-specific estimates, the study rejects the validity of the point estimate
transfer, and cannot reject the validity of the benefits function transfer. However, Downing and

-------
Ozuna (1996) test the benefits function transfer approach for transferring fishing values. The results
indicate that the benefit function transfer approach overestimates benefits, and as a result the authors
find this approach "unreliable" in the case they considered. Without similar empirical research, it
is difficult to determine which would be the case for health valuation, either for adult-adult health
value transfers or adult-child health value transfers.

The relevant factor to consider in the case of children's health is whether the characteristics
of the policy case (i.e., characteristics specific to children) are significantly different from the
characteristics of the adult study sample. As Loomis (1992) notes, the validity of transferring a
benefit function relies, in part, on the equality of the coefficients of the study case with those of the
policy case (if such a function were to exist to describe the policy case). In this case, the existing
parameter estimates of the adult-oriented valuation model would only be peripherally useful for
valuing children's health effects. Crucial valuation elements, such as intertemporal effects, duration,
and altruism may play a significant role in children's health values that may not be represented in
existing models of adult valuation. As a result, the child-specific factors omitted from the adult-
oriented model have the effect of biasing the estimated coefficients for purposes of benefits transfer.

Meta Analysis

A more rigorous benefits transfer exercise uses meta analysis. Meta analysis is a statistical
method of combining a number of valuation estimates that allow the analyst to systematically explore
variation in existing value estimates across studies. As with the benefit function transfer approach,
key variables from the policy case are inserted into the resulting benefit function. As discussed in
detail in Desvousges, et. al (1998), there are several types of meta analysis models. This technique
requires analysts to systematically document the assumptions of the underlying studies, thus leading
to a greater understanding of the differences among value estimates.

-------
For example, Desvousges, et. al (1998) provide a benefits transfer case study utilizing the
technique of meta analysis. The authors construct a meta analysis model for short-term morbidity
effects associated with pollution resulting from electricity generation. Because much of the existing
literature on short-term morbidity effects measure values for a number of different, closely related
health effects, the authors use the meta analysis approach in combination with an index for "health-
state" to enable simultaneous consideration of these studies. The case study uses WTP for short-term
health effects from five contingent valuation studies that value 221 health conditions. The model
explains WTP values as a function of the health index, and variables describing the number of illness
days. This study is most useful for adult-oriented values and does not include demographic variables
that might be useful if the underlying studies included younger populations.

Meta analysis is possible when a number of health-related values exist. If analysts are
considering transferring a number of adult-oriented values to estimate the child-related benefits of
a policy case, they must recognize that the meta analysis will be limited to explaining the variation
among the adult-oriented values. To the extent that child-related values are of a fundamentally
different composition from adult-oriented values, meta analysis will fail to account for these
shortcomings. For example, in some cases child-oriented models may have different determinants
of value than adult-oriented values (e.g., caregiver empathy), and in other cases, the determinants of
value may be significantly different between the two models (e.g., the effect of latency or health
effect duration). In either case, a meta analysis model will not be able to account for these
differences.

-------
Bayesian Techniques

An alternative to the meta analytic approach for summarizing the results of several studies
is the Bayesian approach. Empirical Bayes techniques provide a systematic way to incorporate
existing study (i.e., prior) information with policy case (i.e., sample) information to estimate avalue
for transfer.6 The analyst assesses the extentto which transferability from existing studies to apolicy
case is appropriate by assessing the correspondence between the range of estimates provided by
existing studies (e.g., child health values, determinants of child health values) and a particular level
of uncertainty the analyst is willing to accept (i.e., a given confidence interval). Based on the results
of this comparison, the analyst can determine whether it is reasonable to incur costs to conduct a new
study. If in the use of this technique the analyst is more concerned about reducing the possibility of
understating value estimates (i.e., avoidance of a downside risk), the accuracy of the existing study
information plays a greater role in this comparison. In this case, analysts must consider obtaining
more accurate existing study information, or improved information on the policy case (i.e., better
sample information). As noted above, obtaining accurate existing study information for valuation
of children's health is difficult.

To the extent that child health values are generated similarly to adult health values, this
technique holds promise. However, given the few studies of child health values and complexity in
determining these values, it is not clear how to specify estimates of children's health values, or how
they may differ from those of adults. As stated above, this element of the analysis can play a major
role in the results of empirical Bayes techniques. Nevertheless, Kask and Shogren (1993) discuss
the potential importance of using the Bayesian framework to understand health values. In particular,
the authors note that this technique could be used to investigate how health risk reduction strategies
could influence health values. It is worth noting that this issue becomes even more relevant when
valuing children's health because the mitigating actions of the caregivers add a layer of complexity
to the link between risk reduction and value.

-------
IMPLICATIONS

Previous analyses of recreational behavior suggest that results from benefits transfers must
be interpreted with caution (e.g., Smith 1992, Loomis 1992). While these studies are not specific to
health valuation, they suggest that benefits transfers should be conducted and interpreted with careful
consideration of potential sources of inaccuracy or imprecision. At the same time, however,
inaccuracy and imprecision must also be considered in comparison to the uncertainty that might arise
in conducting incomplete primary research.7 In some cases, benefits transfer may be a more
comprehensive estimate than a new cost of illness study, while in others it may be no more
appropriate to conduct a benefits transfer exercise than it would be to conduct a new cost of illness
study. Depending on the policy case, analysts must consider how different adult-oriented values will
be from those of children, and which uncertainties are the greatest.8

This detailed assessment of using the benefits transfer technique to estimate children's health
values from adult-oriented studies indicates that there are currently significant uncertainties in
reliably transferring values to children.

The literature of child-oriented health values is not extensive, and the
majority of these studies were conducted using the less-preferred cost of
illness approach. The sparseness of the existing literature makes it difficult
to gauge to what degree adult-oriented health values reflect child-oriented
health values. As a result, there is no clear guidance for making factor
adjustments from adults to children in the transfer process.

Having to rely on adult-oriented studies to estimate child-oriented values
results in transfers that may not adequately account for important child-
related characteristics affecting values. For example, areas of inaccuracy
between the two contexts involve differences in the:

~ Health effect measurement (e .g., perception of adversity, additional
child behavioral responses, measurement certainty);

~ Health effect characteristics (e.g., risk perception, severity,
duration, latency, susceptibility);

-------
Impact on well-being and determinants of value (e.g., caregiver
altruism, parental motivation for future care, future productivity);
and

~ Population characteristics (e.g., age, health status, education level).

Because the structure of the value determinants appears to be quite different
between adults and children (e.g., public empathy), it is difficult to know
how values will differ between the two situations. Analysts must account
for policy-specific characteristics, characteristics of health outcomes, and
mitigating behavior that may differ between adults and children in order to
make appropriate transfers.

• The techniques available to transfer benefit estimates from adults to
children each hinge upon analysts having some understanding of how child-
related values compare to those of adults. Adjustments made using any of
the available techniques require prior information on economic theory,
empirical evidence, and experience. However, because children's health
values have been studied infrequently, it is currently difficult to make
adjustments with a given level of confidence.

The current practice of using benefits transfer to estimate rough approximations of the
monetary benefits of avoiding adverse health effects appears to be on target for children's health
values. However, because it may be important for some policy analyses to more accurately estimate
children's health values, analysts should weigh the potential benefits of conducting primary research
against the costs of uncertainty. In particular, in cases where the rough approximations of children's
health values indicate that this category of benefits may be a crucial component in the policy
analysis, it may be prudent to undertake additional primary research to estimate child-related values.

In addition, it may be useful to conduct more general research that systematically explains
linkages between adults and children. Such an effort could serve as a common foundation for
conducting primary research or identifying important considerations in benefits transfer. In
particular, studies of the health-related tradeoffs that adults make may be informative to this question.
There may be useful information on children's health values that could be obtained from a study of
the tradeoffs that caregivers make about their own and their child's health versus tradeoffs that adults
without children make about their own health. One possibility is to consider health studies that allow
an analysis of the marginal willingness to pay as a function of the presence of number of children in

-------
a household. Studies in other disciplines or areas of economics (e.g., labor economics) may provide
some insight. In addition, this research could include a more formal investigation of whether existing
adult valuations provide insight into the relative size and importance of children's health valuation
determinants for comparable effects.

-------
REFERENCES

Agee, Mark D. and Thomas D. Crocker, 1999. "On Techniques to Value the Impact of
Environmental Hazards on Children's Health." Issue Paper prepared for the U.S.
Environmental Protection Agency, Economy and Environment Division. February 1999.

Agee, Mark D. and Thomas D. Crocker, forthcoming. "Environmental Degradation and the
Intergenerational Transmission of Human Capital." Journal of Economic Psychology.

Atkinson, Scott E. and Thomas D. Crocker, 1992. "The Exchangeability of Hedonic property
Prices." Journal of Regional Science. Vol. 32, pp. 169-183.

Atkinson, Scott E., Thomas D. Crocker, and Jason F. Shogren, 1992. "Bayesian Exchangeability,
Benefit Transfer, and Research Efficiency." Water Resources Research. Vol. 28, No. 3, pp.
715-722. March 1992.

Brookshire, David S. and Helen R. Neill, 1992. "Benefit Transfers: Conceptual and Empirical
Issues" Water Resources Research. Vol. 28, No. 3, pp. 651-655. March 1992.

DOC, 1993. "Natural Resource Damage Assessment Under the Oil Pollution Act of 1990 -
Appendix I - Report of the NOAA Panel on Contingent Valuation," United States
Department of Commerce, National Oceanic and Atmospheric Administration, 58 Federal
Register 4601-4614, January 15, 1993.

Desvousges, William H., F. Reed Johnson, H. Spencer Banzhaf, 1998. Environmental Policy
Analysis with Limited Information: Principles and Applications of the Transfer Method.
Edward Elgar: UK.

Diamond, P.A. and Hausman, J.A., 1994. "Contingent Valuation: Is Some Number Better Than
No Number?" Journal of Economic Perspectives. Vol. 8, No. 4, pp. 45-64. Fall 1994.

Dickie, Mark and Deborah Vaughn Nestor, "Economic Valuation of Children's Health Effects: A
Survey of Empirical Evidence," draft, presented at the World Congress of Environmental
and Resource Economists, July 1998.

Downing, Mark and Teofilo Ozuna, Jr., 1996. "Testing the Reliability of the Benefit Function
Transfer Approach." Journal of Environmental Economics and Management. Vol. 30, pp.
316-322.

DuMouchel, W.H. and J.E. Harris, 1983. "Bayes methods for combining the results of cancer
studies in humans and other species." Journal of the American Statistical Association. Vol.
78, pp. 293-308.

Haveman, R.H. and B.L. Wolfe, 1984. "Schooling and Economic Weil-Being: The Role of Non-
Market Effects." Journal of Human Resources. Vol. 19, pp. 408-429.

Haveman, R.H. and B.L. Wolfe, 1995. "The Determinants of Children's Attainments: A Review
of Methods and Findings." Journal of Economic Literature. Vol. 33, pp. 1829-1878.

-------
Kask, Susan B. and Jason F. Shogren, 1993. "Benefit Transfer Protocol for Long Term Health Risk
Valuation: A Case of Surface Water Contamination." Journal paper #J-15670 of the Iowa
Agriculture Home Economics Experiment Station, Ames IA. Paper presented at the 1992
workshop for the Association of Environmental and Resource Economists. December 1993.

Kauffman, R.E., 1992. "Acute acetaminophen overdose: An example of reduced toxicity related
to developmental differences in drug metabolism." In International Life Sciences Institute:
Similarities and differences Between Children and Adults: Implications for Risk
Assessment. Edited by P.S. Guzelian, C.J. Henry, and S.S. Olin. Washington, DC: ILSI
Press, 1992, pp. 97-103.

Kirchhoff, Stefanie, Bonnie G. Colby, and Jeffrey T. LaFrance, 1997. "Evaluating the Performance
of Benefit Transfer: An Empirical Inquiry." Journal of Environmental Economics and
Management. Vol. 33, pp. 75-93.

Loomis, John B., 1992. "The Evolution of a More Rigorous Approach to Benefit Transfer: Benefit
Function Transfer." Water Resources Research. Vol. 28, No. 3, pp. 701-705. March 1992.

Moore, M.J. and W. K. Viscusi, 1988. "The Quantity Adjusted Value of Life," Economic Inquiry.
Vol. 26, pp. 369-388.

Neumann, James and Harriet Greenwood, 1999. "Existing Literature and Recommended Strategies
for Valuation of Children's Health Effects." Prepared forthe U.S. Environmental Protection
Agency, Economy and Environment Division. March 1999.

Smith, V. Kerry, 1992. "On Separating Defensible Benefit Transfers from 'Smoke and Mirrors.'"
Water Resources Research. Vol. 28, No. 3, p. 7-7. March 1992.

U.S. Environmental Protection Agency (EPA), 1997. The Benefits and Costs of the Clean Air Act,
1970 to 1990. Washington, D.C. October 1997.

, 1996. Environmental Health Threats to Children. Office of the Administrator. EOA 175-F-

96-001, September 1996.

, 1998. Handbookfor Non-cancer Health Effects Valuation. Washington, D.C., Social Science

Discussion Group. Internal Review Draft, September 30, 1998.

-------
ENDNOTES

The author would like to acknowledge significant contribution and review by Dr. Thomas Crocker
of the University of Wyoming.

The techniques of benefits transfer have been used for many years to evaluate public policy options
and to assess natural resource damages. Benefits transfer is often used when insufficient time or money is
available to gather the primary data required for a new valuation study. Typical steps involved in conducting
a benefits transfer have been discussed in the existing literature, including EPA's draft handbook for non-
cancer valuation (1998) and Desvousges, et. al (1998). Although no established protocol currently exists to
conduct a benefits transfer, steps to follow likely include:

1. Describing the policy case in detail.

2. Identifying existing, relevant studies.

3. Evaluating the suitability of the studies for benefits transfer. This step concerns
assessing the quality and applicability of identified studies.

4. Transferring the benefits to the policy case. This step includes determining the
extent of the market for the benefits transfer (i.e., the number of persons affected by
the policy), and describing the uncertainty associated with these estimates.

A variety of studies have discussed benefits transfer and the issues that may arise as a result of
employing this technique for policy analysis. For example:

• Desvousges, et. al (1998) discuss the transfer technique more broadly than just for
benefits transfer purposes. The authors define transfer as the "use of existing
information designed for one specific context to address policy questions in another
context" (p. 4), and assess the usefulness of this method for a number of linkages in
the overall process. For example, transfer methods may be used when describing
the physical effects of a policy (e.g., level of air emissions) as well as the behavioral
effects that have value implications. The authors illustrate the transfer method with
a health-related case study, considering the short-term morbidity effects, chronic
morbidity effects, and mortality risks associated with pollution resulting from
electricity generation, and employ simulation techniques to estimate "most likely"
health values associated with adult populations.

• Kask and Shogren (1993) discuss the benefits transfer protocol for long-term health
risk reduction, and provide a case study for surface water contamination. The
authors note significant challenges in overcoming the limitations of using secondary
research to value health effects for policy. The authors indicate that the most
significant limitations include matching the commodity to be valued associated with
the policy case to the set of existing valuation studies; and understanding the
relationship between how a health risk is reduced and what health risk actually is
reduced. The challenges in overcoming these limitations arise from the multiple
sources of risk, multi-symptomatic effects, the latency period between cause and
health effect, and the individual's ability (privately or collectively) to reduce risk.

• A special issue of Water Resources Research (1992) is dedicated to the topic of
benefits transfer. The papers in this issue investigate a variety of conceptual and
empirical issues associated with the benefits transfer technique, without specific

-------
application. The papers presented in this journal issue were prepared for the 1992
workshop of the Association of Environmental and Resource Economists on benefits
transfer issues.

Hedonic methods (wage orproperty studies) represent another valuation technique less preferred than
the three methods mentioned above. With this method, theoretically it is difficult to distinguish the
contribution that environmental quality makes to health (children or adults) for a given policy from the other
beneficial contributions that environmental quality may make to an individual. For example, the implicit
price of better air quality at a site may reflect the value of less soiling, less odor, or better visibility as well
as the value of better health. In practice, it is difficult to isolate the marginal WTP for health from the values
of other characteristics, including the contribution that environmental quality makes to the child's caregivers.

The theoretical basis of the cost of illness method is quite limited — the method does not measure
WTP for reduced health effects. In addition, cost of illness does not reflect the amount that will actually be
expended, rather this cost represents what could be spent. It is also important to recognize that because
treatments for illness change overtime, the cost of illness is a dynamic concept. Older cost of illness studies
may not account for changes in medical technology that might alter the cost or effectiveness of treatment.

Studies that have explored these empirical Bayes concepts include Du Mouchel and Harris (1983),
Atkinson, et al. (1992), and Atkinson and Crocker (1992).

The empirical Bayes technique may provide the analyst with a systematic way to consider this
tradeoff.

In view of the likely tradeoff between the convenience of benefits transfer and the reliability of the
resulting benefit estimates, the question of whether to apply benefits transfer may be best considered within
the context of the policy and the options available for assessing benefits. Factors worth considering include:
the accuracy required of the resulting estimate; the availability of relevant existing studies; the degree to
which additional primary research would improve the accuracy or reduce the uncertainty of the resulting
benefit estimate; and the time and financial resources available to conduct the analysis (Atkinson, Crocker
and Shogren 1992).

-------