Children's Health Protection
Background document supporting the workgroup's recommendations to CHPAC on children's health valuation issues
In April 1997 President Clinton signed Executive Order 13045 entitled "Protection of Children from Environmental Health Risks and Safety Risks," the ultimate objective of which is to ensure that federal health and safety regulations recognize and explicitly account for risks to children. The Children's Health Protection Advisory Committee (CHPAC) was subsequently appointed by EPA Administrator Browner to help the agency determine how best to ensure that scientific knowledge is used effectively to reduce environmental risks to children. Because some EPA standards and regulations require benefit-cost analysis, or related economic assessments, EPA also asked the CHPAC for advice regarding how to ensure that effects on children are adequately represented in economic analyses. To develop this advice, the CHPAC formed the Economics and Assessment Work Group, comprised largely of economists with expertise in benefit estimation, but also including experts in public health and ethics.
Why do Congress and the Executive Branch require EPA to consider the benefits and costs of some proposed standards and regulations? The essential objective of benefit-cost analysis is to avoid doing less than could otherwise be done, to ensure that society's resources are used to produce the greatest net benefit. If we cannot do all good things, we can at least try to do the most that is possible given limited resources.
We need to decide which environmental protections are the most worthwhile, relative to their costs. We want a systematic way to compare alternatives to decide which merit the highest priority. Benefit-cost analysis provides a systematic way to organize complex information and assess trade-offs, thereby improving the policy process.
The economic value to society of each kind of benefit (fewer asthma attacks, for example) is ultimately based on some expression of individual willingness to pay (WTP) for an environmental improvement. How are individual values representative of social value? Other than restrictions, such as speed limits or laws governing criminal activities and some environmentally harmful actions, we assume that what people choose to do accurately represents what is (individually) best for them, and by reference, for society. Logically, the sum total of value to society is the aggregate of value to the individuals who make up that society.
The question of how to value effects on children within the context of environmental risk does not have transparent answers, and the issues are not solely technical. The underlying theory and methods in the environmental health and resource damages literatures are applicable, but the question of how society values effects that impact children (contrasted with how adults value effects that may impact themselves, or with how parents value the same effects on their own children) proves more complex.
The focus of both the specific recommendations and this background document is to assist EPA in addressing these issues and improving the analytical tools used to conduct mandated economic analyses so that impacts on children can be better represented, and their health, therefore, better protected.
This background document is intended to provide information supporting the recommendations made by the Economics and Assessment Work Group to the plenary of the Children's Health Protection Advisory Committee. Since the Spring of 1998, this Work Group has considered how EPA can better reflect the economic value of protecting children's health in economic assessments, mandated or otherwise, of certain proposed regulations and policies. The associated recommendations letter sets for recommendations of the Work Group, as accepted by the Committee, in an effort to improve the ways in which EPA performs economic assessments and thus assist the Agency in protecting children's health. These comments include support for some current Agency practices as well as specific substantive and research recommendations for their improvement. The process used to develop these recommendations is attached to this letter.
Our society's ideal is to protect all children from the array of environmental health threats. We recognize, however, that society's resources are limited, that priorities for actions to protect children's health must be set, and that evaluation of benefits plays a key role in that process.
In preparing its recommendations, the Economics and Assessment Work Group evaluated: a) specific analytical and methodological issues in the economic evaluation of children's health effects, b) practical approaches to valuing the benefits and costs of children's health effects, and c) data gaps and research needs as they pertain to current EPA practice and the state of the available valuation literature. In particular, the Work Group considered the following:
- A series of commissioned issues papers (list attached) and presentations addressing topics germane to valuing children's health;
- Additional background papers, literature reviews, and presentations by EPA staff and others;
- Papers and discussion from the March 24-25, 1999 workshop, "Valuing Health for Environmental Policy With Special Emphasis on Children's Health Issues," which was jointly sponsored by the EPA Offices of Policy, Children's Health Protection, and Research and Development,
- Draft excepts from the Children's Health Valuation Handbook; and
- A draft chapter of the Guidelines for Preparing Economic
Recognizing that there is a very limited literature on the value of reductions in environmental risks to children, the Economics and Assessment Work Group focused its attention on consideration of seven substantive questions that encompass the primary concerns of the Work Group members:
- Are existing economic methods adequate for valuing health effects, including indirect economic effects, as applied to children?
- What value estimates are already available for specific children's health effects?
- Should socioeconomic characteristics of children be considered explicitly?
- Is there sufficient coordination between economists and risk assessors in developing appropriate data for valuation?
- Should EPA adjust values from adult-oriented studies when applied to children's health effects and, if so, how? Is it appropriate to use the adult value of statistical life for valuing reduced risk of children's mortality? If not, what alternative would be better?
- Is EPA's approach to discounting appropriate for children?
- How should the differences between individual and societal values of children's health effects be considered?
The Work Group compiled a series of substantive and research recommendations to address these questions.
Recommendations to the CHPAC for Transmittal to EPA
The Work Group has compiled a series of substantive and research recommendations to address the primary concerns. A more detailed discussion, including background information and the thinking of the Work Group that led to the recommendations, is presented in the next section.
- Agency economic analyses must clearly document all of the assumptions, implicit and explicit, that can affect the estimates of economic values.
- At present, valuation of children's health benefits by EPA typically reflects only avoided cost-of-illness, and these economic analyses often fail to recognize additional non-monetary benefits. The Work Group recommends that, where monetized values are developed, estimates of monetized value for health/wellbeing benefits should be expressed in terms of willingness-to-pay, rather than solely as avoided cost-of-illness.
- The existing literature used in monetized benefit estimation often pertains specifically to adults, few valuation studies on environmental risks to children are available. While research to estimate monetized benefits of children's health should be a priority, in the interim it will be necessary to consider transferring estimates of adult benefits to children. Benefit transfer is appropriate when three conditions are met: (1) generally accepted willingness-to-pay (WTP) estimates exist for adults; (2) the risk situation for which they would be transferred is substantially similar to the risk situation from which they were derived; and (3) the consequences of risks to health and well-being are similar in adults and children. Such estimates must be derived from risk situations closely related to the effect being valued. When any of these conditions is not met, cost-of-illness estimates may be used on an interim basis to represent a plausible lower bound, with a narrative description of the non-monetary benefits explicitly included. The process should also document how that measure of value is likely to vary as the size or character of the risk or benefit in question changes.
- In principle, any established method of non-market valuation could be used in children's applications. However, because every situation is different, and experience is limited, it would not be appropriate to specify a particular method or group of methods for use in valuing children's health. The key requirements are (a) that what is measured be appropriate for what is to be valued and (b) that the analysis be consistent with economic understanding of consumer behavior and market mechanisms.
- The lack of directly applicable values for many impacts on children means that omitted benefits are likely to have a greater influence on the results of economic analyses than is the case for impacts on adults, especially if adverse effects are severe, widespread, and unvalued. It is therefore especially important that non-monetized measures of changes in children's health and well-being, such as changes in risk, incidence or prevalence, and the estimated changes in numbers of children affected and the ways in which they are affected, be quantified and reported. After accounting for all monetized effects, the analyst should determine the minimum and maximum values of non-monetized effects necessary to affect conclusions about the efficiency of a proposed regulation or policy.
- We recommend that analysts recognize the wide potential range of socioeconomic characteristics when estimating both (1) relevant dose-response functions and (2) household willingness-to-pay functions, with effort devoted to including available socioeconomic or demographic variables that are indicators for income, minority, or immigrant status. We further recommend that the Agency report the available demographic characteristics (income, ethnicity, immigrant status, family structure, and so on) for any cohorts at risk.
- The usual challenges of coordination between risk assessors and economists are amplified in the case of children's health effects because of the paucity of information specific to children. To address this problem, we recommend that EPA structure the tasks associated with measuring the impacts of environmental policies and regulations using a "team" approach in which risk assessors and economists work collaboratively rather than sequentially in order to clarify the impact of possible assumptions and to ensure that both risk and economic analyses are as comprehensive as feasible. We further recommend seeking input from public health professionals, child health experts, clinicians, industrial engineers, etc., as appropriate.
- We recommend that estimates of economic benefits of improved children's health be based primarily on the value parents place on the health of their own children, rather than seeking to directly measure children's own willingness to pay to reduce environmental risks. Children usually lack the experience to make the types of informed choices economic analysis assumes will underlie a WTP, while parents often are uniquely positioned to know and to care about their children's health and well-being. We recognize, however, that parental values may be only one component of the overall social benefits of children's health. (Other possible components, for example, might include the general social benefit derived from children growing up healthy and employed, or symbolic or religious value that society might place on children's health.) Until we obtain better indicators of whether and by how much social benefits exceed parental values, the WTP of parents is probably the most defensible indicator of economic benefits of children's health, although we realize that there are a number of reasons to believe that it may be a lower bound.
- When EPA uses discounting in its economic analyses, the Agency should explicitly explain what rate or rates it uses, the basis for choosing the specific range if rates used, and the implications of those choices.
- Where an analysis places a monetary value on reducing risk to life (as contrasted with health/wellbeing), we consider the value of statistical life years (VSLY) to be a potentially superior valuation approach compared to the value of a statistical life (VSL), and that VSL values for adults are likely to be inappropriate for children. We recommend that the Agency develop appropriate VSLYs for children as soon as possible where monetized values are to be used. However, until appropriate estimates can be developed for children, the adult VSL should be reported as a plausible approximation, with the strong caveats and conditions noted above.
Referring to the recommendations above, the Work Group requests an opportunity to review the draft Children's Health Valuation Handbook as well as any economic analysis undertaken for the four regulations that EPA has decided to reevaluate on the basis of children's health effects: chloralkali National Emission Standard for Hazardous Air Pollutants; organophosphate pesticides tolerances for methyl parathion, chlorpyrifos, and dimethoate; atrazine tolerance and MCL; and Farm Worker Protection Standards.
- Research on the valuation (monetized or non-monetized) of children's health effects is underfunded. In FY'99, effectively $0 of the approximately $100,000,000 in research grants awarded by EPA's Office of Research and Development (ORD) were devoted directly to such research. Consequently, we recommend that ORD develop a detailed research agenda in this area and allocate significant resources to ensure that these research needs are met, to the extent feasible given competing priorities.
- In order to make better values for mortality and morbidity effect available to analysts, we recommend that research on parental willingness to pay to reduce risk to children receive priority attention. Since parental values may not completely reflect societal values, further research should also focus on societal willingness to pay and the magnitude and conditions under which divergence from parental values occurs.
- Because valuation methods based on wage-risk trade-off studies are unlikely to apply to children, we recommend that the EPA develop an accepted value for the loss of a statistical life year (VSLY) suitable for use in connection with children's health impacts, where monetized values are developed.
- We recommend that the Agency promote research to increase the body of knowledge about how willingness to pay for reducing specific environmental health risks in children varies with health status and with environmental and economic factors of the children and their families. In doing so, EPA should take note of the possible effect of income constraints, as note above, and the implications of reflecting such constraints in willingness-to-pay figures.
- We recommend that researchers in this area employ disaggregated household data, including detailed demographic variables, and attempt to control for the specific effects of the presence of children on household decision-making. It will be useful to other decision-makers involved in protecting children to understand how parents make economically relevant choices associated with their children, and what features of their children's well-being matter to parents and other adults in the household, even if these decisions are not directly associated with pollution or the environment.
- We recommend that the Agency promote research and public discussion on the appropriateness of discounting methods when undertaking economic valuations of children's environmental health benefits.
- To facilitate the implementation of these research recommendations, the Work Group recommends that they be integrated into the Economic Research Strategy. We also recommend that they be shared with other federal agencies, researchers, and institutions in the field.
Background Information - Questions Posed by the Work Group
Question 1. Are existing economic methods adequate for valuing health effects, including indirect economic effects, as applied to children?
Although it often may not seem so, economic valuation is something broader than looking up a price in the Wall Street Journal. It covers not only the valuation of market items, such as an automobile, a house, or a share of corporate stock, but also what is known as non-market valuation, such as placing a monetary value on a loss of beach recreation, an improvement in air quality, preserving a national monument, or on other items not directly traded in markets. Non-market valuation grew out of the conventional economic principles for valuing market items and extended the principles to other, non-market items. Many aspects of valuing children's health are examples of non-market valuation, and must draw on the techniques developed in that field over the last 40 years.
It is important to distinguish between the basic principles for non-market valuation, including the economic valuation of children's health, and the implementation of these principles to deal with specific undesirable health effects in children. The underlying principles are well defined and do not differ materially from those underlying any other type of non-market valuation. The implementation for specific health effects in children may sometimes be difficult or impossible due to the nature of the health effect, characteristics of children, or limitations of the available data.
Since the economic valuation of health, especially children's health, can sometimes be controversial, it may be useful to offer a brief description of the logic and purpose of this type of valuation. What is being measured here is the monetary equivalent of the loss of welfare associated with morbidity, risk of morbidity, or risk of mortality of children. Conceptually, there are two distinct components to the potential loss of welfare. One is various costs that illness imposes on the family and on society ? both monetary outlays for treatment-related expenses and, for parents if not for the children themselves, income or earnings foregone, at the time or subsequently, as a consequence of the illness. The second component of loss is the sheer disutility of the illness, due to factors such as pain, suffering, anxiety, or inconvenience. To the extent that they arise in any particular case, monetary valuation should cover both components of the loss of welfare. However, measuring the two components involves different considerations.
The market cost associated with an illness, whether income foregone or extra expenditures incurred, is usually fairly straightforward to measure ? it is simply the cost itself. This is known as the cost-of-illness (COI) measure. Since COI estimates medical and time costs of an illness and not the overall loss of utility, it is likely to underestimate the true welfare loss from illness.
The monetary measure of the disutility associated with illness is different because it involves an equivalence: one translates a loss of welfare (utility) into an equivalent loss of income. This relies on the economic concept of value in exchange. To say that, for some individual, X has a value of 50 in terms of Y means that the individual would be willing to exchange 50 units of Y for one unit of X. Here, X and Y are two different items, measured in two different units, with Y serving as the numeraire in terms of which X is valued. Y could in principle be anything, not just money. Monetary valuation, which arises when Y is money (income), is just one particular, but important, example of value in exchange. When one values the disutility of illness in monetary terms, therefore, this refers to the loss of money that an individual would consider equivalent to the illness, in the sense that this loss of money would reduce the individual's welfare by the same amount as the illness. Conceptually, there are two alternative ways to formalize this equivalence: (1) the maximum willingness-to-pay (WTP) measure of value ? 50 units of Y (e.g., $50) is the most that the individual would give up (pay) in order to avoid X (the illness); or (2) the minimum willingness-to-accept (WTA) measure of value ? 50 units of Y ($50) is the minimum compensation that the individual would be willing to accept in order to put up with X (the illness). All economic measures of value can, in fact, be reduced to an equivalence involving either a WTP or a WTA type of measure.
For the portion of welfare loss associated with an imposed cost or reduction in income, WTP and WTA are identical, both being measured by the cost or income reduction itself. Hence, COI measures both WTP and WTA for this component of the loss. But for the loss of welfare associated with the disutility of illness, as opposed to the imposed cost, WTP and WTA can differ. If they do differ, the typical pattern is for the WTA measure to be larger than the WTP measure. Which measure should be used depends on the initial conditions, whether the proposed change is a gain or a loss, and upon subjective judgments about the entitlement of the affected groups (historically, legally, or ethnically) to either the initial conditions or the conditions that would prevail after the change. For practical purposes, and to be conservative (in the sense of avoiding overstatement of benefits), most researchers tend to opt for the typically smaller WTP measure on the grounds that it is likely to be more defensible. However, the right policy decision requires that the appropriate valuation measure be used.
Economists employ several methods to estimate, either exactly or approximately, the WTP or WTA measures of disutility associated with illness. One approach is to elicit directly what people would be willing to pay (or to accept) through some form of survey or choice experiment; this is known as contingent valuation or, more generally, stated preference. Other approaches use more indirect evidence to estimate what people would be willing to pay (or accept) based on inferences from their observed behavior; this is known as revealed preference. An example of the latter is inferences based on defensive behaviors undertaken by people to avoid illness or mitigate its impacts, known as the averting behavior approach. Another example is an approach based on an analysis of differences in wages associated with riskier occupations, or differences in home prices associated with more polluted environments; this is known as the hedonic value approach.
These approaches to measurement share a basic similarity, in that they all involve identifying either explicit or implicit choices from which value in exchange can be inferred ? actual real-world choices with revealed preference, or choices made in an experimental or survey setting with stated preference methods. The challenge for a successful implementation, whether in a market or non-market context, is to find a choice that is both appropriate for what one wants to value and amenable to analysis, in that the requisite data are available or can be obtained. It is an empirical question whether this can be accomplished; there is no automatic guarantee of success for either type of valuation.
While the valuation of children's health is in principle no different than the valuation of adults' health, or any other non-market item, in practice there may be more empirical difficulties for at least two reasons. First, there are fewer existing applications of valuation specifically concerned with children's health, as opposed to the health of adults or the household as a whole. Because there is less experience, and perhaps less complete information, one might expect estimates of the value of children's health will be subject to more uncertainty than for adults or the household as a whole. Second, choices about children's health are almost never made solely by the children themselves; parents or other household members are almost always involved.
Resource scarcity is the fundamental justification for value in exchange: whenever a choice is being made to have more of something ? say, protecting children's health or improving their quality of life ? one must give up something else which could otherwise be chosen that also requires time, effort and resources. What is being given up is the cost of the item (whether direct cost or opportunity cost); this should be compared with the benefit from the item, measured in comparable units, in order to determine whether the choice is justified. There is a real trade-off, regardless of whether it is expressed in terms of people's time, money, or any other numeraire. This type of comparison of benefits and costs for policy alternatives related to children's health is inevitable, regardless of whether measures of the monetary value for health changes are used, simply because there will be a cost to any change. Making the change, by default, implies that it must be worth at least the cost to the decision-maker. Deciding not to make the change carries the implication that its value must be less than the cost.
Question 2. What value estimates are already available for specific children's health effects?
There is a very limited emerging valuation literature concerning environmental risks to children. Because adverse impacts on adults have been studied more extensively, and because adults can themselves express or reveal their WTP to reduce adverse effects, the existing literature that is generally accepted for use in benefit estimation reports values specific to adults. For mortality and for morbidity that affects children and adults similarly, careful transfer of adult benefit estimates to value impacts on children is preferable to omitting effects on children from economic analysis. In some cases, however, analysts may have to resort to using cost-of-illness estimates, or to presenting information on children's effects in non-monetized form.
EPA should give priority to research on parents' WTP to reduce risk to their own children and to contingent valuation studies that can help to determine whether parental values are good representations of social value (see also the discussion of Issue #7 below). In deciding how to structure this research, the health effects of greatest concern should first be identified. The attributes of children (as children and as future adults) that are altered by those effects should also be specified. Examples include intelligence, fertility, mobility, and life expectancy.
It may also prove fruitful to review existing and on-going research to determine where collection of additional data or further analysis of data already collected would provide information about WTP to reduce some effects.
Question 3. Should socioeconomic characteristics of children be considered explicitly?
There are at least two motives for considering socioeconomic characteristics including income, ethnicity and immigrant status in a benefit-cost analysis or other economic assessment of a proposed regulation. First, Executive Order 12898 directs the EPA to identify the costs and benefits to low-income and minority groups when regulations are changed. A recurrent observation in the context of environmental justice has been that the ability of a person to reduce exposure to environmental threats varies with socioeconomic status and political influence, and these factors in turn are related to income, ethnicity and immigrant status. Second, policy analysts within the EPA may find it helpful to know how income, ethnicity and immigrant status interact with environmental and other public policies (e.g., education, nutrition, income-support policies) to determine risk, exposure, health outcomes, and household WTP.
Socioeconomic heterogeneity should be recognized in the estimation of both dose-response functions and willingness-to-pay functions. Both dose-response relationships and willingness to pay may differ by age, gender, and ethnicity, as well as with the degree to which income, educational opportunities, and access to information constrain the choices that individual families are able to make. Keeping track of differences in dose-response functions and differences in WTP by socioeconomic groups also will allow policy makers to track the distributional consequences of policies or regulations.
For example, in estimating a dose-response function, a child's ethnicity may be correlated with baseline health status, nutritional status, housing type, or even variations in exposure levels. If ethnicity is included as an explanatory variable while housing type is not, the analyst risks attributing some of the effect of housing type to ethnicity. In the case of estimating household WTP for risk reduction, a household's ethnicity may also be correlated with parental comprehension of risks, age of parents, education, income, family structure, and other factors. Omitted variable bias in dose-response or WTP functions may be especially unfortunate if systematic differences are incorrectly attributed exclusively to ethnicity variables and therefore deemed to be beyond the reach of public policy influence. Ethnicity may capture differences due to other unmeasured influences, but it is obviously not a variable that can be manipulated by policy. In contrast, it may be possible for public policy to influence some of the other sociodemographic variables, such as nutritional status or parental access to information about and understanding of risks.
Researchers should also devote more attention to the differences between objectively measured risks and subjectively perceived risks. Parents may place unrealistically low values on measures to reduce risks to their children based upon the notion that they have control over these risks via averting behaviors. For example, everyone may feel that the risk facing their own child is lower than the "average" risk, because they will compensate by taking additional care.
More broadly, economists expect WTP to depend on many factors, including baseline health status and the initial level of risk, the availability, quality, and price of health care, household size and composition, and income. These and other determinants of WTP may vary markedly across population groups, potentially causing non-trivial differences in the economic benefits of environmental policies and regulations. Yet there currently is a limited body of quantitative evidence on how WTP for improved children's health varies with socioeconomic characteristics. Without detailed quantitative evidence on these issues, policy analysts can only guess about how benefits are distributed across population groups, or about whether benefits estimated for the general population of children accurately reflect benefits to specific groups.
Question 4. Is there sufficient coordination between economists and risk assessors in developing appropriate data for valuation?
Relatively few examples exist where risk assessors and economists have collaborated seamlessly in characterization and valuation of environmental risk. More typically, economic analysis follows risk assessment, but risk characterizations often do not provide information conducive to benefit estimation. As a result, benefit assessments cannot accurately or completely capture impacts on children, or the value of reducing risk. The limited collaboration between risk assessors and economists is not restricted to children's health, and often leads to three problems for economic assessments.
First, measures of risk conveyed by risk assessments often are not well suited for estimating individual benefits. Benefits arise from avoiding, or from reducing the probability or severity of, health effects that individuals recognize as adverse. Non-cancer risk assessments, however, generally characterize risks in terms of a Reference Dose (or the proposed Margin of Exposure). While threshold-type measures are important to establish health-based standards, they do not by themselves indicate how the probability or severity of illness varies with changes in exposure, and so provide a limited basis for benefit estimation. Risk assessments often measure subtle changes in physiological or neurological function, such as changes in lung function. But exposed individuals may not recognize a change in lung function as adverse without additional information about how lung function affects daily activities or long term health status. If people cannot understand how such an effect might impact them (or their children), they cannot make the informed choices that are associated with measures of their willingness to pay to reduce or avoid the effect. Economists are left to translate estimates provided by risk assessments into measures more suitable for estimating individual benefits, or else to omit known effects from the benefit analysis. Either way, there are increased prospects for significant errors.
A second difficulty arises when economists attempt to estimate aggregate benefits or to perform equity assessments. Aggregate benefits depend on the population distribution of risk. But risk assessments estimate individual risk, not a population distribution of risk accounting for differences in susceptibility and exposure (which may vary significantly among sub-populations of children). Similarly, equity assessments require identifying effects occurring in specific sub-populations, but this information often is not provided in a risk characterization.
Third, economic benefits and costs of regulations often depend on how people adjust their behavior in response to changes in their environment. For example, people may stay indoors more when ambient pollution is high, purchase organic produce to reduce pesticide exposure, or purchase water filters or bottled water to reduce risks from waterborne contaminants. These behavioral changes, which depend partly on economic factors like prices and incomes, influence exposure and risk. Risk assessments do not account for how behavioral adjustments to policy influence exposure and risk.
Greater communication, and especially collaborative team efforts, between risk assessors and economists would mitigate each of these three problems. Furthermore, input from public health professionals and clinicians, including experts on surveillance, would be helpful in alerting policy-makers to emerging threats to children's health, accounting for the full range of individual, family and community factors that affect child health, and identifying options for treatment. Industrial engineers, in turn, could provide useful information on options for prevention and mitigation of environmental threats to children's health. We recommend broadening the team of analysts charged with describing impacts of environmental policies and regulations to include these professionals as appropriate.
In view of effects that environmental policy may have on public health, coordination between EPA and the public health community would be useful in assessing, valuing and protecting children's health. Coordination of efforts across federal agencies should be promoted, perhaps through the Presidential Interagency Task Force on Environmental Health and Safety Risks to Children, as well as through other channels. Coordination with state and local public health systems should be undertaken to insure that the public health functions of epidemiology, community assurance and measures of population impacts are taken into account. The public health system is much broader than a community's public health department, and includes hospitals, clinics, laboratories, mental health facilities, epidemiology/toxicology departments, water and air regulatory agencies, and schools and social welfare agencies. Each of these agencies contributes to the health of a community and its children, and should be considered in assessing and valuing children's health.
Question 5. Should EPA adjust values from adult-oriented studies when applied to children's health effects and, if so, how? Is it appropriate to use the adult value of statistical life for valuing reduced risk of children's mortality? If not, what alternative would be better?
The existing literature that is generally accepted for use in benefits assessments often reports values specific to adults, while relatively few valuation studies on environmental risks to children are available. This situation creates a strong case for using benefits transfer ? that is, using information from existing studies on the value of adult health to estimate the value of children's health ? rather than omitting effects on children from the economic analysis. An obvious problem with benefits transfer is that values for children's health effects may differ from values for similar health effects in adults.
Children differ from adults in many ways, including, but not limited to, potential differences in the probability, severity or duration of a given adverse health effect, in life expectancy, and in economic status. This observation raises the question of whether, if values of adult health are to be transferred to children's health, some sort of adjustment should be made to account for relevant differences in exposure and consequences between adults and children.
Currently available evidence would not support the proposition that specific health effects would always have the same value for children as for adults, nor would it indicate how large any difference might be. The value of child health relative to adult health is likely to vary depending on the health effects considered, the populations affected, and other factors. Consequently, using benefits transfer to value children's health effects based on studies of adults' health may be quite unreliable. Furthermore, we cannot offer any general formula for adjusting values taken from adult-oriented studies before applying them to children's health.
While the current paucity of information on values associated with children's health effects makes benefits transfer more attractive, at the same time makes benefits transfer more speculative. The best way to derive a value for children's health is by measuring it directly using established techniques of non-market valuation. Funding for work in this area should be a top priority.
One area of benefits estimation where potential differences between adults and children are of particular concern is the valuation of reductions in premature mortality. Reduced mortality typically is monetized using the value of a statistical life (VSL). The VSL is based on estimates of the willingness to pay for small reductions in the risk of death (or on the willingness to accept compensation for small increases in the risk of death). For example, suppose the average person is willing to pay $58 per year to reduce his or her annual risk of death by 1/100,000. Imagine a group of 100,000 similar people each paying $58 to reduce risk by 1/100,000. As a group they would pay $5.8 million, and, statistically speaking, would experience one less fatality. The value of the statistical life saved in this example is $5.8 million. Estimates of the VSL are based primarily on studies that examine the higher compensation that workers accept for taking jobs with higher risks of death.
There are many potential problems with applying VSL estimates derived from the behavior of adult workers to value reduced mortality among children. There are no children in the samples of adults used to estimate the VSL, while available evidence indicates that the VSL varies among adults of different ages. The voluntary decision of an adult to accept a workplace risk in exchange for higher earnings seems far removed from the situation of a child who faces an increased risk of death from an environmental hazard. In addition, a child loses more years of life than does an adult when death comes prematurely.
One way to account for differences in life expectancy among persons affected by a risk-reducing regulation is to apply an estimate of the value of a statistical life year (VSLY). This measure makes distinctions between risk-reducing regulations based on their effects on longevity: if a regulation protected individuals whose average remaining life expectancy was 30 years, then a risk reduction of one fatality would be expressed as 30 life-years extended. If the average remaining life expectancy were 50 years, then one statistical fatality avoided would represent 50 statistical life-years extended. These life years extended would be valued using the estimated VSLY.
Typically the VSLY is estimated by annualizing an estimate of the VSL. In the undiscounted form, the $5.8 million VSL is divided by a life expectancy of 35 years. This is the average life expectancy of the subjects in the empirical studies used to establish the $5.8 million VSL estimate. This yields a VSLY of approximately $166,000. This VSLY is then multiplied by the average life expectancy in the subpopulation(s) affected by the proposed policy or regulation to produce an adjusted value of statistical life for that subpopulation.
Future life years can also be discounted before they are used with the $5.8 million associated with the value of a statistical life. Discounted remaining life years are given by (1/r)[1-exp(-rR)] where R is the study sample average life expectancy described above and r is the discount rate (Moore and Viscusi, 1988). For example, The Costs and Benefits of the Clean Air Act 1970 to 1990 (EPA, 1997) used a discount rate of 5% yielding a VSLY of approximately $354,000 (updated to 1997 dollars). Because of discussion surrounding the development of the Guidelines for Preparing Economic Analyses, the discount rate used in the future is likely to be 2% - 3%.
This strategy describes current practice, but the Work Group concedes that this approach is not yet entirely satisfactory. Further research to develop appropriate VSLYs for children in still necessary.
Question 6. Is EPA's approach to discounting appropriate for children?
One way to account for differences in life expectancy among persons affected by a risk-reducing regulation is to apply an estimate of the value of a statistical life year (VSLY). This measure makes distinctions between risk-reducing regulations based on their effects on longevity: if a regulation protected individuals whose average remaining life expectancy was 30 years, then a risk reduction of one fatality would be expressed as 30 life-years extended. If the average remaining life expectancy were 50 years, then one statistical fatality avoided would represent 50 statistical life-years extended. These life years extended would be valued using the estimated VSLY. Is EPA's approach to discounting appropriate for children?
The question of discounting arises whenever one faces outcomes ? whether costs or benefits ? that occur at different times. If a person is indifferent to the timing of outcomes and considers all time periods as equivalent, this corresponds to the use of a zero discount rate. By contrast, if a person weights current outcomes somewhat more heavily than future outcomes, this corresponds to the use of a positive discount rate. In the context of health valuation, the issue of discounting arises with greater force for children than for adults because the greater expected longevity can make effects with a longer interval between the exposure to an environmental hazard and any adverse health outcomes more important.
Discounting is sometimes used as a means of adjusting for other considerations unrelated to timing, some of which could also arise in connection with children's health. For example, there may be greater uncertainty about the health impacts on children than adults, and uncertainty is sometimes accounted for through the use of a higher discount rate. Alternatively, if there is a greater social concern for the vulnerability of children, this could be incorporated through the use of a lower discount rate for future adverse outcomes to children's health. We believe that it is preferable to deal with such considerations directly and explicitly in valuing the outcomes rather than through the surrogate of adjustments to the discount rate. We recommend that when discounting used, it be used purely and explicitly to account for differences in timing.
Discounting has been a source of controversy not only between economists and non-economists, but also among economists. This is because it involves reconciling two different aims. One is to strike a fair balance between the present and the future. The other is to use society's limited economic resources efficiently. Each has a different implication for whether one should discount, and how.
When focusing on the efficient use of limited resources, the consideration most relevant for discounting relates to the cost of capital. A positive discount rate is justified in order to account for the time value of money. The discount rate should reflect what capital costs, if one borrows, or what it earns, if one lends ? whether from an individual or societal perspective.
With regard to fairness in balancing between the present and the future, the relevant consideration is the rate of time preference of the decision maker ? whether an individual or society. The question of how one should weight the future in comparison to the present is a value judgment.
Which of the two aims is most relevant ? the efficient use of scarce economic resources versus striking a balance between the present and the future ? could vary with the circumstances. Geoffrey Heal, in his recent book, has suggested that considerations of the efficient use of scarce economic resources are more appropriate for conventional business and economic planning horizons, while considerations of inter-generational equity and the social rate of time preference are more appropriate for longer-run environmental problems with time scales of several generations or more. The valuation of children's health may involve intermediate time horizons between these two scales. Therefore, balancing the two alternative approaches to discount rate determination becomes more problematic.
The current EPA guidance on discounting recommends presenting both discounted and undiscounted benefits and costs. When discounting, EPA recommends the use of two alternative interest rates: an interest rate of 2 to 3 percent, to reflect the social rate of time preference, and an interest rate of 7 percent, which the Office of Management and Budget (OMB) recommends as a measure of the cost of capital. Presenting undiscounted benefits and costs means showing the benefits and costs in each year of the time horizon. Given the uncertainties about discounting outlined above, we believe that these three alternatives span the range of reasonable approaches, considering the various possible perspectives on discounting. In any case, the results will sometimes be insensitive to the choice of the discount rate.
Question 7. How should the differences between individual and societal values of children's health effects be considered?
Benefit-cost analysis of policies affecting children's health typically should not rely primarily on children's own willingness to pay as a measure of benefits. Children, particularly young children, often lack the experience and resources ? including information, judgment and income ? to indicate a meaningful willingness to pay. The resources and preferences of some third party must therefore provide the basis for estimating benefits.
In many cases, parents are uniquely positioned to know and to care about their children's health and well-being. Most parents invest substantial, though not unlimited, amounts of money, time and effort providing for the support, schooling and health of their children. Parents, therefore, are accustomed to making decisions about investing resources in children, and in facing the accompanying economic tradeoffs. For these reasons, and for consistency with social norms leaving parents considerable authority to act on behalf of their children, parental willingness to pay provides a natural starting point for benefit estimation.
Parental WTP, however, may not always accurately reflect social benefits, for at least three reasons. First, children may develop, depending in part on their health status, into adults who provide benefits to, or who impose costs on, society at large not reflected their parents' willingness to pay. Second, people may care about the children of others, and, under certain conditions, altruism would cause parental WTP to understate social benefits. Third, individual parents face constraints that society as a whole does not. Inadequate information about some health risks, low income, and limited access to insurance and to health care may prevent parents from making economically efficient investments in children's health. Society then has an economic incentive to raise the investment in children's health above the level chosen by more resource-constrained parents. The social income constraint being different from the sum of parents' income constraints, society at large may be willing to pay more for children's health than the sum of parents' private willingness to pay. Although these considerations may not be unique to children, the dependency, ongoing development and unknown future potential of children heightens concern.
Whether parental and social willingness to pay in fact diverge, and the size of any divergence, are empirical questions on which little evidence presently exists. Answers to these questions are likely to differ according to the health effects considered and the demographic characteristics of affected families. Quantitatively adjusting parental WTP, in an attempt to better reflect social benefits, is not in general advisable given the present state of knowledge. Even qualitative judgments must be made carefully to avoid implicit double-counting, for example, from consideration of changes in transfer payments as a source of social benefits. For example, a child whose adult earning capacity is impaired by environmental exposures may contribute fewer tax dollars to social insurance and welfare programs than an unimpaired person. These losses are not borne by the child individually, but by society as a whole. Whether this loss of transfer payments, which is borne by society as a whole, should be added to a measure of the child's own economic loss depends on how the child's loss was computed. If those losses consist of earnings losses gross of taxes, they already account for changes in taxes and transfer payments.
When parental and social willingness to pay diverge, however, we believe that parental willingness to pay is more likely to understate than to overstate social benefits. Therefore we suspect that social benefits are likely to be at least as great as the sum of parents' private willingness to pay.
EPA should promote research to articulate clearly how children and adults differ from the perspective of divergences between private and social willingness to pay. Little is known concerning how household income, family composition, parental education, access to insurance and health care, and other factors affect parents' willingness to pay for the health of their children. Likewise, there is little evidence concerning whether children's health values differ between individuals with and without children. When investigating this question, it may be important to recognize that fertility is not purely exogenous but rather partly a choice. Some insight on these issues might be obtained by considering the extent of community support for health programs for children.
The foregoing has provided, from the perspective of the Economics and Assessment Work Group, greater detail and explanation of recommendations communicated to the Children's Health Protection Advisory Committee.
Commissioned Issues Papers Addressing
Topics Germane To Valuing Children's Health
Benefits transfer of children's health values, prepared by Marla Markowski, Industrial Economics, Incorporated. March 1999.
Contingent valuation and valuing children's health, prepared by George Tolley, University of Chicago and Robert Fabian, University of Illinois at Chicago. March 1999
Data requirements for valuation of children's health effects and alternatives to valuation, prepared by Kimberly M. Thompson, Center for Risk Analysis, Harvard School of Public Health. March 1999.
Existing literature and recommended strategies for valuation of children's health effects, prepared by James Neumann and Harriet Greenwood, Industrial Economics, Incorporated. January 1999.
On techniques to value the impact of environmental hazards on children's health, prepared by Mark D. Agee, Pennsylvania State University and Thomas D. Crocker, University of Wyoming. December 1998.
Valuing children's health and life: what does economic theory say about including parental and societal willingness to pay? Prepared by William T. Harbaugh, University of Oregon. March 1999.
Valuing indirect effects from environmental hazards on a child's life chances, prepared by Jason F. Shogren, University of Wyoming. February 1999.